===== Using Medical / Personal Oxygen Concentrators and Compressors for SCUBA Applications:  Practical Examples =====

In this document I will outline how I am using medical/personal oxygen concentrators and compressors for use with my personal technical diving needs.  I use them to supply high-oxygen gas directly into a SCUBA cylinder.  I use this process for a couple of different purposes:  as a means of directly filling cylinders with oxygen for oxygen-accelerated decompression dives ('deco bottles') or as a source of oxygen for a Nitrox/Trimix stick supplying gas to my high-pressure breathing gas compressor.

To generate oxygen at home, you will need an oxygen concentrator to generate a stream of oxygen-enriched gas, and an oxygen compressor to take that gas and push it into a cylinder at high pressure.  It turns out that there are medical devices that can do each of these things.  They are designed for patients that need additional oxygen to breathe properly, both in and out of the home.  We can re-purpose these devices for our own needs.

This is not my original idea.  A number of other people have done this, and a few of them have written up their experience on ScubaBoard.  I first heard of this idea from @tbone1004, and since then also from @Tracy, from whom I purchased one of my oxygen compressors.  But there aren't a lot of specific details on the setup or process of using these devices.  So I thought I would take some time and write up what I have learned.

Unfortunately, it's taken me a great deal of frustrating experimentation to pull all of this together.  I swear I should have started a Patreon for this:  I've spent way **way** more time, effort and money figuring all of this out than I thought I would -- and probably more than I should.  By the end, I just wanted to make it into something that mostly works for my needs, sunk cost fallacy be damned.  :)  Hopefully, this can provide a good idea of what such a system can and cannot do, and maybe even save others some time and effort in the future.  And if it does, please let me know:  I would appreciate knowing that my efforts were not in vain.

Like everything I write, this will be long.  Really, really, **really** long.  Be prepared.

===== Warnings and Disclaimers =====
I am not an expert in **anything** I am going to discuss here.  I am simply a crazy technical diver who likes to experiment.  I have zero training or instruction in any of the concepts or devices described here.  This is merely a documentation of what I have done for myself.  This is not an instruction manual.  It is a collection of ideas that you can use to develop your own strategy, using your own knowledge and expertise.

This entire process involves a number of dangerous and potentially deadly items.  First of all, we are dealing gas enriched to nearly 100% pure oxygen.  This is **NOT** at **ALL** the same as "Enriched Air Nitrox".  Recreational Nitrox has a limit of 40% oxygen.  This number was not chosen by accident.  Research has found that the risk of fire increases significantly above 40% oxygen concentration.  Even a 50% concentration has almost the same risk as 100% oxygen.  (Reference PDF's:  [[https://ntrs.nasa.gov/api/citations/20090001157/downloads/20090001157.pdf | Page 6]] or [[https://www.sciencedirect.com/science/article/abs/pii/S0165587610005252 | Results section]])  And this equipment produces just-about-100% oxygen.  This creates **significant** risk.

Second, we are dealing with high-pressure gasses.  These oxygen compressors can produce 2000 or even 3000 PSI of pressure.  These pressures are significant -- 20 or 30 times the pressure of the average shop compressor.  Of course, SCUBA divers deal with these pressures in our cylinders all the time.  But this is a lot different than just putting a commercially-produced regulator on a commercially-produced cylinder, both of which were specifically designed for this purpose.  There is a lot of DIY in re-purposing medical equipment for SCUBA applications, and none of this equipment is designed or tested for these applications.

And of course, this process has the special risk of **combining** both of these risks.  This process takes effectively-pure oxygen and compresses it to 2000 psi or beyond.  The US military has found that the risk of ignition was "essentially the same" between oxygen at 2000 psi and 7500 psi.  (Reference PDF:  [[https://apps.dtic.mil/sti/pdfs/AD0608260.pdf | page iii ]] )  This is wading into the deepest end of the pool.

For these reasons, I will not be including specific details of the parts I have used.  As I've described, this series of posts will outline what I have been able to accomplish and the obstacles I have had to overcome.  I will give enough information that an experienced technician will be able to determine for **themselves** what parts will best meet **their** requirements.  But you will not find a list of items to buy and instructions on how to put them together.

Because of the risks involved, it is vital that you are fully aware of the dangers of high-pressure gasses and high-percentage oxygen -- and both combined.  You also need to be fully aware of the requirements to address these risks.  This includes making sure to select materials that will handle the pressures involved as well as being compatible with pure oxygen.  These are most certainly **not** parts you will get at the local hardware store.  In addition, you must use the proper techniques and procedures to make sure that all items are oxygen-clean and stay that way.

This document will not cover any of this information here.  To cover this information takes multiple manuals and literally hundreds of pages.  To get you started, here are a few references:
  * [[https://ntrs.nasa.gov/api/citations/19690017609/downloads/19690017609.pdf | NASA Compressed Gas Handbook]]
  * [[https://ntrs.nasa.gov/api/citations/19960021046/downloads/19960021046.pdf | NASA Safety Standard for Oxygen and Oxygen Systems]]
  * [[https://files.caireinc.com/Docs/Cleaning_of_Equipment_for_Oxygen_Service.pdf | EIGA Cleaning of Equipment for Oxygen Service]]
  * [[https://apps.dtic.mil/sti/pdfs/AD0608260.pdf | Compatibility of Materials with 7500 PSI Oxygen (Referenced above)]]

Please make sure you understand this information thoroughly before you attempt to use any of this information for any purpose.

===== An Introduction to the Equipment =====

To generate and store oxygen for diving purposes, you will need a couple of devices:  an oxygen concentrator to generate the oxygen, and an oxygen compressor to store it.  Let's take a brief look at how they work and how they were designed to be used.

==== Oxygen Concentrator ====

=== Design ===

An oxygen concentrator is a device that takes atmospheric air (roughly 78% nitrogen, 21% oxygen, 1% argon, water vapor and other trace gasses), and compresses it slightly (to between 15 and 25 psi or so) in cylinders of zeolite, a molecular sieve designed to adsorb nitrogen.  This  leaves behind the remaining gasses.  With the almost 80% nitrogen removed, that leaves a mix of about 93% oxygen, 5% argon, and a balance of residual nitrogen and water vapor and trace gasses.  Much of this oxygen-enriched gas is then able to be released to a downstream product tank, while some of the gas is used to flush out the adsorbed nitrogen back to the atmosphere.  There are two such cylinders, so that one cylinder can be capturing nitrogen while the other cylinder is releasing its nitrogen.  There is a regulator installed on the product tank that drops the pressure to a low pressure (5 psi usually) and releases the gas on to the patient.

This is a very simple process with very few moving parts.  It's little more than a glorified aquarium air pump, a pair of zeolite-filled metal cylinders, a few valves (usually combined into a single 4-way package), some tubing and a timer.  The machines also often have certain sensors to make sure that they are working correctly, producing gas with a sufficiently high level of oxygen (usually above 80%), and ensuring a sufficient volume of gas is flowing out of the machine -- to alert you to a pinched hose for example.

A note about the gas you can expect to get from these devices.  As described above, they function not by selecting oxygen but by rejecting nitrogen.  Removing the nitrogen **does** increase the fraction of oxygen in the output -- but also increases the fraction of everything **else** in the output.  That creates two concerns.  The first is that the amount of argon will increase as well:  up to as high as 5% or so.  This resulting gas is affectionately nicknamed "Argox" because of this extra argon.  Please note that argon is a highly-narcotic gas, much higher than nitrogen.  For open-circuit, this is not an issue.  Either the argon is re-diluted to a much lower level (in the case of using this oxygen to supply a Nitrox/Trimix stick), or the gas is used at a much lower maximum depth (in the case of an oxygen deco cylinder).  However, this is **not** true for rebreather divers.  If the output of a concentrator is used to fill oxygen cylinders for rebreathers, the argon will build up in the loop quickly over time.  This makes Argox unappealing for this application.  If you are a rebreather diver, make sure you understand this issue and the potential for risk.

The other concern from using gas from an oxygen concentrator is that it will concentrate everything in the air and put it into your cylinder.  There is effectively zero filtration in this process.  Everything your concentrator is exposed to is, well, concentrated and pushed into your cylinder.  Carbon monoxide, exhaust gasses, off-gassing materials, even things like radon will be passed right along and amplified by a factor of 5.  Then, when you breathe this gas at depth, the exposure will be amplified by another factor of 5 or so.  That means you'll be breathing 25 times as much of these chemicals on a dive as you do at home.  A little bit of CO that might just give you a headache at home could kill you at the bottom of the lake.  Make sure the concentrator is getting clean air.

=== Variations ===
For this application, we are focusing on home-based oxygen concentrators.  When they were first developed, they were called "portable oxygen generators", because at the time a 35-or-so pound machine was a **lot** more portable than what came before it.  While these machines are able to be moved (and therefore called "portable"), they are not able to be operated while mobile.  More recently, truly portable battery-operated oxygen generators that are small enough to be carried in a pack have been developed.  Those devices, though, are too small for our purposes, and are not discussed here.

Home-based oxygen concentrators come in two different capacities.  The majority of devices are 5 liters-per-minute (5 lpm) concentrators.  There are also 10 lpm devices, though these are much less common (and much more expensive).

Another feature that separates concentrators is the output pressure.  The majority of devices have only a low-pressure (usually 5 psi) output.  Many 10 lpm devices have a slightly higher output pressure (8-10 psi), while a few devices have a specific high-pressure (20-25 psi) output.

One final feature that many concentrators have is an output designed to supply gas to an oxygen compressor.  This is almost always a quick-release connector with a check valve that is plumbed internally ahead of the metered flow valve to give the compressor priority over the patient output.  This is usually supplied at the same pressure as the patient output, though sometimes can be higher.

There are certainly other features as well:  as mentioned, some units have oxygen sensors or flow sensors; some have mechanisms to add humidity to the output gas;  and there are other features as well.  For our purposes, though, the above are the important features to pay attention to.

In my case, I only own a single model of oxygen concentrator:  the Invacare PerfectO2 V, a 5 lpm concentrator that outputs low pressure O2 through a metered flow valve, as well as a dedicated compressor output at 5 psi.  I have also carefully examined the service manual of the Respironics EverFlo concentrator as well, to compare and contrast its capabilities compared to the PerfectO2.  Its capabilities are very similar, though almost always without the dedicated compressor port.

=== Service Manuals ===
Speaking of service manuals, they are a great way to get comfortable with these machines.  They show you a great deal of detail, including extensive disassembly and assembly instructions as well as diagnosis and troubleshooting processes.  Even if they aren't the same as your machine, it is quite likely that many of the details will carry over so they are worth your time.  

  * [[https://www.invacare.no/sites/no/files/csv_migration/product_docs/technical_docs/DTEC009879_Perfect_O2_V_Servicemanual.pdf | Invacare PerfectO2 Service Manual]]
  * [[http://www.frankshospitalworkshop.com/equipment/documents/oxygen_concentrators/service_manuals/Invacare%20Platinum%205%20+%2010%20Concentrator%20-%20Service%20manual.pdf | Invacare Platinum Service Manual]]
  * [[http://www.frankshospitalworkshop.com/equipment/documents/oxygen_concentrators/service_manuals/Philips%20EverFlo%20Oxygen%20Concentrator%20-%20Service%20manual.pdf | Respironics EverFlo Service Manual]]
  * [[http://www.frankshospitalworkshop.com/equipment/documents/oxygen_concentrators/service_manuals/Respironics%20Millennium%20Oxygen%20Concentrator%20-%20Service%20manual.pdf | Respironics Millennium Service Manual]]

=== Maintenance Needs ===
There is very little to go wrong with these machines.  They are designed to run continuously for many thousands of hours.  About the only requirements for the machine is that they be kept clean, dry and cool.

  * **Clean:**  They have external air filters that should be washed regularly (every few weeks) and internal HEPA-grade filters on the air intake that needs to be replaced periodically (every 6-12 months).  Environments with lots of dust or pet hair will need more frequent attention.

  * **Dry:**  Moisture can damage the zeolite, so these should not be operated in high-humidity environments.  Practically, this means that they should be used in an air-conditioned indoor environment.

  * **Cool:**  An air-conditioned space is plenty cool enough -- as long as the machine has enough space around it.  Don't do what I did and place the machine right up against a wall.  It sure seems logical, but the machine does not like it.  After a while (which in my case was years), your machine will no longer produce oxygen properly.  You'll then get in touch with someone who knows more about these machines than you do who will ask you about white powder around the machine.  You will then notice that the entire wall behind the machine is covered in white powder.  You will then be told that heat destroyed your zeolite packs (they chemically break down and no longer do their job) and $200 later, you will have newly re-packed zeolite cylinders...  Or at least that's what I'm told. :)

=== Sources ===
Speaking of that expert:  Bradshaw Oxygen Supply (https://bradshawoxygen.com) is the company that helped me to repair my concentrator when it stopped producing oxygen.  He was easy to get in touch with, very knowledgeable, and got me my parts quickly and easily.  He specializes in supporting people using oxygen equipment in creative applications.  For example, glass artists will use these devices to produce oxygen for "lampwork" (low-temperature glass melting and shaping), and Bradshaw Oxygen supports them.  So you're not going to be asked for your insurance information or doctor's prescription. :)  And his prices are quite reasonable:  if you want a concentrator quickly, just check out Bradshaw Oxygen and buy one.  You can get one cheaper, but the only way you're going to do that is to search high and low on Craigslist or Facebook Marketplace (or, sadly, estate sales).

=== Designed Usage ===
This machine is about as simple as it gets.  There's basically a switch to turn it on, and a knob to twist to determine how much gas you get out.  Turn it on, and set the machine to put out how much gas you need.  That's it.

You might wonder why it has a knob to adjust the gas:  why not just crank it all the way open and let it go?  Mainly because these machines cannot actually put out maximum oxygen at the machine's rated flow.  For a 5 lpm machine, you can really only get maximum oxygen concentration (93% or maybe a bit higher) up to about 3 lpm.  Above 3 lpm, the oxygen fraction begins to drop.  At 4 lpm you will be at approximately 90%, and at 5 lpm you could be as low as 85% or even less.  For a 10 lpm machine, the point of diminishing output is approximately 6-7 lpm.  If you want maximum oxygen content, you will need to keep the output at or below that cutoff.

Also, if your machine has a flow sensor, you may need to make sure that your flow is high enough not to set off the alarm.  For such machines, this usually means a minimum flow of about 1 lpm.  If you're using the metered output only then it's very likely you will set the flow higher.  This is more an issue for machines that have two outputs, one metered output for the patient and an unmetered output for an oxygen concentrator.  In some cases, even if you want to use only the compressor output, you might need to make sure that you have at least 1 lpm or so coming out the metered output -- or listen to the machine beep the whole time.  Some machines have the flow sensor before the compressor output so they won't beep if you're using the compressor output even when the metered flow is zero -- until the cylinder is full and the compressor shuts off.  Hopefully, that won't wake you up in the middle of the night...

In any case, using the concentrator is about as easy as it gets:  turn it on and set the flow to the correct level.  It should simply sit there and do its job.

==== Oxygen Compressor ====

=== Design ===
An oxygen compressor is a small and simple reciprocating compressor.  An electric motor turns the compressor head, which might have three to five stages.  Each stage gets progressively smaller, which allows it to apply more and more force to the gas, progressively increasing the gas pressure.  They also have a number of safety features:  they monitor the oxygen content to make sure it's high enough (usually over 80%) and they monitor the output pressure to automatically shut off when the cylinder reaches the target pressure (usually around 2000 or 3000 PSI).  They may also have other safety and convenience features that may make things easier in its normal application, or might prevent the thing from working the way you want it to in its new application.

=== Variations ===
The biggest and most noticeable variation between compressors is the maximum output pressure.  Some have a maximum of 2000 PSI, and others have a maximum of 3000 PSI.

The other variation is the expected input pressure.  Some work with or even expect low pressure (5 PSI) and others require high pressure (20-25 PSI).  They all claim to require a concentrator with a dedicated compressor output.  You will need to pair them with a concentrator that supplies the correct pressure, or you will need to figure out a way to deliver that correct pressure and volume between the concentrator and the compressor.

I am aware of only two home medical oxygen compressors:  the Invacare HomeFill II, and the Phillips Respironics UltraFill.  I own one of each.  The HomeFill II seems to work fine with a supply pressure of 5 psi and can compress up to a maximum of 2000 PSI or so (though the manual states it expects a pressure of 14-21 psi).  The UltraFill requires a pressure supply noticeably higher than 5 psi (and will not operate at all with only 5 psi), and can compress up to 3000 PSI or so.

=== Service Manuals ===
Again, the service manuals are a great resource.  Just like with the concentrators, they show extensive disassembly and assembly instructions as well as diagnosis and troubleshooting processes.

  * [[https://invacaredocs.com/assets/documents/PIM/Invacare_HomeFill_2_AW_Compressor_Service_Manual_.pdf | Invacare HomeFill II Service Manual]]
  * [[https://www.cryois.com/wp-content/uploads/2021/04/Respironics-UltraFill-1.pdf | Respironics UltraFill Service Manual]]

=== Maintenance Needs ===
These machines aren't as durable as the oxygen concentrators.  They are not designed to run hour after hour like the concentrators are.  Most compressors are designed to run for no more than a handful of hours at a time.  Despite this, there are people who do run these devices practically continuously:  for example, lampwork users.  Unfortunately, they have found that these compressors will only last a few years of nearly-continuous operation.

Also unfortunately, there's not much you can do to extend this life.  Like the concentrator, you will want to make sure that the machine stays clean and cool.

  * **Clean:**  They too have external air filters that should be washed regularly (every few weeks or months).  They do not have user-serviceable filters for the incoming gas.  They may have an inline micro filter inside the unit, but it's not really designed to be serviced.  The idea is that the gas should be filtered by the concentrator before it moves on to the compressor.

  * **Cool:**  The compressor will generate heat as it compresses the gas, so it will need to be kept cool as well.  It should be designed to have sufficient cooling, but once again you will want to run this unit in an air-conditioned space, and make sure the machine has enough space around it.  Pay attention to where the intake and exhaust vents for the cooling is:  they may not be in the back.  No matter where they are, you will want to make sure to leave plenty of space around them.

=== Sources ===
The compressors are harder to come by:  there just aren't as many of them in the used market.  Bradshaw Oxygen Supply does have some HomeFill II units from time to time (though you may have to search their website or call/email to see if they are available).  The UltraFill units are not as common.  I purchased mine from @Tracy.  He purchased them as Covid-surplus, new in box.  I don't know exactly how many he has, but as of 2024, it's more than one, and they're available for purchase.  The price was very reasonable as well.  As long as he has them I would start by getting in touch with him -- though read this entire document before you do, so you know what you're getting into.  You're unlikely to find it for a lower cost elsewhere, and his are brand new.

Just like the concentrators, the other way to go is to search high and low on Craigslist or Facebook Marketplace.  The compressors don't come up often, but they do appear.  Unfortunately, they can also have a very high price:  people see what the cost of these are new and think that they can get a sizeable fraction of that price.  That's what happened when I managed to get my HomeFill II and PerfectO2:  I saw a Craigslist listing from a local pawn shop.  Their price was high, but I offered a little more than half of their listed price and they took it.  Sometimes it works, sometimes it doesn't.  Hopefully you can find one where it does!

=== Designed Usage ===
Once, again, using these machines is designed to be as easy as possible.  Connect the plastic quick-disconnect supply tube between the concentrator and the compressor using the dedicated compressor output on the concentrator.  Connect a medical bottle with the proprietary valve fitting to the output.  Turn the compressor on and maybe hit the start button if it has one.  That's it.  Let the machine run until the bottle is full, and it will stop pumping automatically -- though you will actually have to hit the switch to completely turn the compressor off.  Of course, the concentrator won't shut off at all -- someone might still be breathing from it!

===== Using the Equipment for SCUBA Without Modification =====
These machines are **very** simple to use -- when you use them for the task for which they were designed.  But what if you want to use them for SCUBA?  Then things work a little differently.  Let's take a look at how things change.

Note that in this section, my goal is to connect the devices without modification.  Additional parts and equipment may be needed, but no real permanent changes or modifications will be made to the concentrator or compressor.  I've been able to make this happen mostly, though not completely, successfully.  In some cases, this process might be easier if you were to permanently modify the equipment.  At some point I will probably move into modifications, but I have not yet done so.  I will speculate about possible modifications below, but know that none of them have been tested by me at this time.

==== Concentrator ===
For the most part, the concentrator works the same when used for medical applications as it does for SCUBA applications.  The two biggest issues to address is what options you have to get the gas out of the concentrator and into the compressor, and at what pressure the concentrator will deliver it.

=== Dedicated Compressor Output ===
If you have a concentrator designed for the compressor you will be using, that's easy:  do it the way the manufacturer intended.  For example:  using a HomeFill-compatible concentrator with a HomeFill II, or an EverFlo concentrator model with compressor output (it seems most EverFlo units do **not** have a compressor output!) with an UltraFill.  Both of these concentrators have a plastic quick-disconnect socket, as do both of these compressors.  You simply use a tube with a quick-disconnect plug on each end, snap them into place and turn the machines on:  you're done.

=== Metered Output Only ===
What if you your concentrator doesn't have a compressor output?  Well, you'll have to use the metered flow valve output.  If you were to take the cover off of a HomeFill-compatible concentrator, you will see that the HomeFill socket is simply teed off right before the metered flow valve, downstream of the 5 psi regulator.  Basically, this allows the compressor to take as much gas as it wants **before** it gets to the flow meter.  That way, you can't starve the compressor, no matter how much you open the metered flow valve.

However, in this case, we're not worried about sharing the gas with a person, so we can use the metered output without worry of being starved.  You will want to open the flow valve up to minimize the pressure drop.  Both the HomeFill II and the UltraFill will compress about 2 lpm, so make sure that the flow meter shows that it is supplying more than this.  Of course, we don't want the oxygen percentage to drop, so make sure it is not flowing more than 3 lpm or so.  (If you have a 10 lpm concentrator, the flow can be much higher without reducing the oxygen concentration, enough that you can likely run two compressors from a single 10 lpm concentrator.)

=== Low Pressure Target:  HomeFill II ===
If you are connecting to a HomeFill II, this will likely work with no other equipment.  The service manual states that the compressor wants 14-21 psi input pressure, but as far as I can tell it works acceptably with 5 psi -- at least it has for me for years now.  If you are using a compressor output, the compressor will be getting the maximum pressure the concentrator will supply and it should work correctly, if a bit more slowly.  And if you're fortunate to have a concentrator that supplies higher pressure (like a 10 lpm unit), the  standard output pressure might already be near-ideal for the compressor.

If your concentrator doesn't have a compressor output and you have to use the metered output, you will want to adjust the flow meter to be as open as you can without overdrawing from the concentrator.  Once the compressor is hooked up, if you don't have gas leaks, you should be able to open the flow meter wide open:  the compressor will only accept so much gas at a time anyway, and there's nowhere else for the gas to go.  Because of the way reciprocating compressors work, there is a rapid intake portion and a portion where the gas is being compressed and no longer accepting more gas.  This will make the flow meter bounce between 1 lpm and 3 lpm or so.  That's exactly what you would see if you were to put a flow meter between a concentrator with a compressor output and the HomeFill II, so it's what we should see if we are properly using the metered output as well.

However, if the metered output is spending too much time above 3 lpm flow, you may need to use the flow meter to bring that maximum down to 3 lpm or so to keep the oxygen concentration up (on a 5 lpm concentrator, anyway).  You might want to find out why the flow is higher than it's supposed to be, and fix the leaks that are allowing that to happen.

For the HomeFill II, while there doesn't seem to be an actual requirement to increase the pressure -- the compressor is seemingly happy with even 5 psi -- there may be a benefit to doing to.  The compressor might fill the destination cylinder more quickly, and it might save some wear and tear on the compressor.  If you want to pursue that, consider the process used to increase the supply pressure to the UltraFill below.

=== High Pressure Target:  UltraFill ===
Concentrators usually supplies a constant and often low pressure.  Unfortunately, the UltraFill requires a higher pressure.  This is due to a couple of reasons.  First, the unit will alarm if the input pressure is below 7 psi or so, which is higher than nearly all 5 lpm compressors can supply.  But even if you supply that minimum 7 psi pressure, the compressor clanks and rattles, which probably means that it is very unhappy operating at that pressure.  To get the machine to quiet down, it seems to want at least 10 or so psi.  It might be possible to use certain 10 lpm concentrators that can supply a little bit higher pressure, but even that might be too low for quiet and reliable operation.

There is a second reason why the UltraFill requires a higher input pressure.  There is a very annoying feature within the UltraFill's control software:  it tracks the output psi increase as the compressor is running.  The purpose is to detect if the fill bottle's valve is open (and leaking) and alert the user.  Unfortunately, if the output psi does not increase fast enough for the software, it will alarm **and** stop pumping.  And the lower the supply pressure is, the less gas it is pushing into the fill cylinder per cycle and the slower the fill rate.  Depending on the size of the cylinder you are filling, the fill rate may be too slow and the compressor will stop.  Increasing the input pressure can increase the output fill rate and can allow the machine to fill cylinders that it otherwise wouldn't at a lower input pressure.

<alert type="warning">Please pay attention to this limitation!  This pressure-tracking feature is by far the most annoying feature of the UltraFill and is the most difficult to overcome.</alert>

To avoid these two issues, we need to increase the pressure being supplied to the compressor.  Ideally, we want to supply to the compressor the highest pressure that the compressor will reasonably accept.  The compressor will displace the same volume per stroke, so the higher the input pressure the more molecules that are being moved in each cycle of the compressor, and the faster the destination cylinder will fill -- hopefully fast enough to avoid the too-slow-fill shutdown.

== Additional External Equipment Needed ==
To increase the pressure, we will need to add an external pump between the output of the concentrator and the input of the compressor.  This will take our low-pressure concentrator supply and boost the pressure for our compressor.  In addition, because of the highly-variable rate at which the compressor accepts gas, we will also want to add an accumulator as a buffer between the pump and the compressor.

A note about measuring supply pressure:  the only number that matters is the lowest pressure measured during a rotation of the compressor.  What happens is this:  while the compressor's first stage input valve is closed, the external pump will increase the supply pressure, because no gas is being taken by the compressor.  When the compressor's first stage input valve opens, the supply gas rushes in and the supply pressure drops until the first stage valve closes again.  It's that lowest pressure that shows what the pressure in the first stage starts at, and it's the only pressure that matters for how much gas the compressor will deliver to the cylinder for that stroke.  

That's part of why we need an accumulator to act as a buffer.  If there is a very small volume of gas between the pump and the compressor (like just a hose), the compressor will draw a relatively large volume and significantly drop the supply pressure.  If there is a very large volume of gas between the pump and the compressor (by adding an accumulator), the drop will be relatively small -- as small as a bit less than 1 psi for the larger accumulator I eventually built.

== Pump: Rated Pressure / Volume ==
What size pump should you select?  It depends on which concentrator and compressor you are using and the cylinders you want to fill.  As was mentioned, if you're going to bother to increase the pressure, you might as well increase it as much as you can without exceeding the capabilities of the compressor **and** without exceeding the maximum flow rate recommended for the concentrator to provide maximally-enriched gas.

Let's start with the maximum target pressure determined by the compressor.  For the HomeFill II, the specified acceptable range is 14-21 PSI.  For the UltraFill, there isn't a specified input pressure; however, the machine will alarm with an input pressure below 7 or above 34 PSI.  The middle of that range is 20 PSI, and something around 25 PSI seems like a reasonably conservative target.

In the beginning, I had no idea what PSI or flow I needed to be able to fill different sized SCUBA cylinders.  So, I had to purchase several pumps to try to figure this out and measure each combination.  My first attempt was with an 80 kpa pump rated for 3 lpm, which seemed like it should be sufficient.  Testing it, it sure seemed like this was going to work:  it quickly built up to 25+ psi into a closed accumulator!  Unfortunately, as soon as I turned the compressor on, the pressure dropped to less than 10 psi.  This isn't enough to fill even an AL40.

That pump was clearly inadequate, so I purchased a 120 kpa pump rated for 5 lpm.  I thought for sure this would do the trick.  Sadly, while it was better, it just didn't deliver enough:  at about 3 lpm flow from the concentrator it was delivering only 13 psi.  This was not enough to fill even a medium-sized (AL80, 11 l water volume) cylinder.

Next I purchased a 200 kpa pump rated for 10 lpm.  It did quite a bit better, if you can give it enough flow.  At 3.0 lpm, the maximum concentrator flow guaranteed to provide maximum oxygen levels, my initial attempts reached a pressure of 14 psi or so.  That was enough to fill an AL80.  If I bumped up the flow to 3.5 lpm (still pretty close to maximum oxygen concentration on my concentrator), I was up to 16 PSI or so, which is enough to fill an HP 100 (15 l water volume).  If I supplied 4 lpm flow from the concentrator (the maximum I could deliver from a wide-open metered flow valve from my 5 lpm concentrator) I could generate up to approximately 18 psi.  Unfortunately, that wasn't always enough to allow filling of the large cylinders I previously used with my HomeFill:  LP104/HP130 (16 l water volume) and LP121/HP149 (19 l water volume).  Unfortunately, that's the maximum output I could get!

I went ahead and tried one more pump:  400 kpa / 40 lpm.  This actually fared **much** worse:  it tended to want to overheat badly, draw way more power than its rating, stall at pressures that the smaller pumps handled successfully, etc.  It performed so badly I actually returned it and got a different one -- mainly because the first pump I received was **obviously** previously used:  silicone adhesive residue on the barbs and soldered wire bits on the electrical tabs...  (Thanks, Amazon!)  I thought maybe it was worn out by the first user, but the second pump operated just about as poorly.  Maybe it was a poorly-designed pump?  Or maybe it's just too big.

Finally, for testing, instead of using a concentrator with a pump, I used cylinders I had previously filled with the HomeFill with an old SCUBA first stage hooked up to a shop air regulator.  I found that if I could supply 26-28 psi to the UltraFill I **was** successful in filling cylinders all the way up to the LP121!  Unfortunately, my pump would not allow me to do that right from a 5 lpm concentrator.  So more work was going to be needed.

In the end, I focused on the 200 kpa pump:  it was the biggest pump I could find that could reliably supply elevated pressure to the compressor, even if it wasn't as high as I really wanted.

<alert type="info">I still think that the right pump might be able to deliver a higher pressure for the same flow volume!  If you find a better pump, please let me know so I can let others know as well.</alert>

So, at the end of a several weeks of frustrating testing, I had a pump that was able to help, but not quite deliver what I need.  I could fill medium-sized SCUBA cylinders, but not the larger cylinders that would make this setup much more useful.  But as it was, I was at the limit of what both the pump and the concentrator could deliver.  What else could I do?

It turns out there were still a couple of things to play with.

== Accumulator:  Reduce Pressure Drop ==
One thing that I added fairly early in the process was an accumulator.  This is simply a buffer of gas between the pump and the compressor.  As previously mentioned, reciprocating pumps tend to demand input gas in bursts.  When it would demand gas, the pressure from the pump would crash to a low level, then build back up between strokes.  But it doesn't matter what the pressure is at the start of the demand burst, but at the **end** right before compression starts, and that end was **much** lower than the beginning -- easily 5 PSI or more.

To help the pump smooth out these demands, I built a simple accumulator.  My first design was both simple and somewhat poorly suited to the task.  It was a 20" piece of 2" PVC with a cap on each end.  In one end I drilled and screwed in a 1/4" NPT to barb fitting (for gas input) and in the other I drilled and screwed in two similar fittings (one for gas output and one to a 50 PSI pressure gauge).

(Why 2" PVC?  Because that was the largest size at Home Depot that is actual PVC, **not** DWV-only 'PVC'.  DWV-only is not rated for pressure, while PVC is.  At 25 to 30 PSI, it's not likely that even DWV will explode; however, just remember that PVC is invisible to x-ray, so when you end up with shrapnel in your body, the only way to find it is exploratory surgery...  I decided to stick with pressure-rated PVC, even if it is overkill.)

This design had advantages and disadvantages.  The big advantage was that it worked:  the compressor draw would now only drop the pressure by about 2.5 PSI or so.  This certainly helped boost the pressure into the compressor -- and gave me a nicely-installed and easier-to-read pressure gauge.  The disadvantage was that it was awkward to work with:  I now had this 2-foot-long stick of PVC that had to be dealt with between my concentrator and my compressor, and because I had put fittings in both ends I couldn't even stand it up between them.  Dumb.

In the end, I had to build a better one.  In order to get even that last bit of pressure from my setup, I about quadrupled the size of the accumulator:  3 x 24" with elbows in an S.  Plus, instead of drilling and screwing NPT fittings into the cap, I used actual PVC parts to adapt down to 1/4" NPT, which allowed me to use gas-tight fittings as much as possible for the pressure gauge as well as a 35 PSI pressure relief valve I also installed.  Also, I installed only a single barb connection.  You don't need an 'in' and an 'out':  one connection to the rest of the tubing is just fine.  (Think home hot water expansion tank...)  This got the pressure variations down to about half a PSI, which is about as good as that's going to get, I think.

== Watch Those Leaks! ==
The second improvement was also the hardest.  To improve the performance of the entire system, I needed to improve the efficiency of the pump.  And that meant making sure that all of the gas that left the concentrator had to find its way to the compressor.  Any gas that leaks will result in a lower pressure at the compressor input.  And this is especially important because we are limited on how much gas we can get out of the concentrator.  For a typical 5 lpm concentrator, the maximum flow guaranteed to achieve maximum oxygen levels is usually 3 lpm.  Often the drop between 3 and 4 lpm is relatively low, such as 94% to 90%.  Ideally, we would keep the flow below 3 lpm, though if necessary we might have to go as high as 4 lpm.  Beyond 4 lpm, you will absolutely have to compromise on the oxygen concentration -- or upgrade to a 10 lpm concentrator.

It seems like that should be plenty of gas:  after all, the manual says that compressor is only supposed to draw 2 lpm, and we should have 3-4 lpm out from the concentrator.  But that ignores a couple of issues.  First, increasing the pressure increases the number of molecules we're pushing into the cylinder per stroke (which is why we're doing this!) so higher pressure will demand more volume from the concentrator.  And the second issue is maybe the most annoying part of this process:  leaks.  We do not have a lot of extra gas volume from the concentrator to play with in the first place -- there just isn't a lot of margin for leaks.  Even a small leak can easily waste 1 lpm or more.  And that problem is only compounded by now working at 20+ psi instead of 5 psi.

If we waste even 1 lpm of gas, what happens?  One of two things.  Either we have to demand more gas from the concentrator to make up for the wasted gas, or if we don't (or can't) increase the concentrator flow the leaking gas will lower the pressure that the pump can supply.  If we increase the gas flow, we will likely lower our oxygen capacity.  If we lower the pressure into the compressor, we limit how large of a cylinder the compressor will allow us to fill.  Neither is a good situation.  It is very important that you address these leaks!

The problem was not immediately obvious.  My pump would build pressure just fine to some maximum, which seemed OK -- in fact, without the compressor running it would have no problem hitting 25 or even 30 PSI.  Of course, once the compressor started demanding gas the pressure dropped noticeably -- too much.  It seemed that I had enough pumping power, just not delivered fast enough to beat the compressor.

There weren't any obvious leaks:  you couldn't easily hear or feel anything.  But as I started to really dig into this, the leaks became more obvious.  If you bent a tube near a connector you might feel something.  If you put your ear up to a fitting you might hear something.  And with no volume to spare, it doesn't take many of those even tiny leaks to add up.

The desire to reduce leaks caused me to re-work just about everything between the concentrator and the compressor.  I was using a collection of 1/4" OD (1/8" ID) and 3/8" OD (1/4" ID) tubing, with a variety of barb and push-to-connect fittings.  This wasn't by choice:  some devices had larger barb fittings and some had smaller.  I wanted to be able to put flow meters in various places, and that mean that there were multiple separate pieces of tubes between different points, even when the flow meters weren't in place.  I couldn't find push-to-connect fittings for 3/8" OD tubing, only 1/4", and there weren't too many places where I needed to connect 1/4" tubes together in the first place.  I had ordered barb adapters to go from 1/4" OD to 3/8" OD, but they weren't great:  the tubes were somewhat loose on them.  Plus my original accumulator had three barb fittings on its own, one of which was going to an NPT pressure gauge, held to the tube with just a hose clamp.  Like I said, lots of tiny leaks.

I ended up standardizing as much as possible on 1/4" OD tubing and push-to-connect connectors wherever possible.  The push-to-connect fittings use an o-ring and do a very good job of sealing.  The concentrator and compressor use either a quick-disconnect fitting (which also use an o-ring) or a specific silicone cone fitting that fits very well (and is on the very low pressure side).  Unfortunately, the pump and accumulator are stuck using barb fittings, but I made sure to use quality fittings and properly-sized and -installed hose clamps.  I'm using standard American perforated hose clamps because I already have bunches of them; but if you need to buy some, consider the non-perforated European hose clamps, which do a better job of holding evenly without deforming the tube.  And I've already described the improvements I made to the accumulator above, making it not only larger and more convenient to use, but also much more gas tight.

There are probably more extremes that I could go through, but I'm not exactly sure what they would be.  The improvements I did make provided a noticeable boost.  Before, the best results I could deliver were about 18 PSI to the compressor at about 4 lpm from the concentrator, which couldn't fill an HP120;  now, with the larger accumulator and careful leak management, I was getting about 22 PSI to the compressor for the same concentrator flow.  That's enough to fill an HP120.  Still not quite enough for everything, but much better than before.

Fix those leaks!

== Push the Pump Pressure ==
The final area where I spent even more time was in trying to get every last PSI out of the pump and into the accumulator.  If you can't get a bigger pump, you need to make sure that you're not wasting any bit of capability.  Of course, I was already focusing on that:  that's why I added the accumulator and jumped on the leaks as best I could.  But I needed more.

The obvious way to do that was to increase the volume going into the pump.  Supplying more flow to the pump allowed it to increase the resulting pressure -- for the same reason why we needed the higher pressure for the oxygen compressor in the first place!  Unfortunately, I was already well above the limit of my concentrator.  I was already pulling 4 lpm from the concentrator, which is going to produce a flow that is less than 90% oxygen.  Ideally, I would like that to be closer to the 95% maximum concentration if I can, but without the extra flow it won't fill the larger tanks at **all**, so I'm going to have to live with that now.

Could I at least temporarily go higher than 4 lpm?  I thought maybe if I had another concentrator to run in parallel I might be able to.  So after scouring Craigslist and Facebook Marketplace for a couple of months I was able to buy a second concentrator.  Imagine my surprise when I hooked it up and it made almost zero difference in both flow and pressure!  It probably shouldn't have been a surprise:  even a single concentrator will generate more than 5 lpm if unrestricted.  The fact that it wasn't with the pump in place is probably because the pump can't handle additional flow.  This might be able to be improved if I use shorter or larger-diameter tubing or other such improvements, but I find that unlikely, given the relatively small amount of flow we are talking about.  But this is an area for future investigation.

The final area where I was able to push the pump's capability was to increase the voltage to the pump.  The pump is designed for 12 V, and that was exactly what I was supplying it, confirmed with a voltmeter.  However, the pump is rated to handle up to 14 V.  So, what would happen if I increased the voltage to 14 V?  The pump runs faster, and therefore pumps a bit more gas.  This did not create a huge improvement, but it was measurable:  between 10% and 15% or so.  

== Final Fill Capabilities ==
Here are the final fill results I experienced.  These results were found using a Invacare PerfectO2 V 5 lpm concentrator via the metered flow valve into a 200 kpa / 10 lpm pump running on 12.0 V (actually measured), drawing 1.0A, through final large accumulator and into a Phillips Respironics UltraFill.

^  Flow (lpm)  ^  Pressure (PSI) ^  Largest Successful Fill  ^  Could Not Fill  ^
|         2.00 |            <7.0 |           -- --           |  Any cylinder <sup>1</sup>  |
|         2.25 |             8.0 |        AL40 (5.7 l)       |      ?LP50       |
|         2.50 |            12.5 |       ?LP50 (8.0 l)       |       AL80       |
|         2.75 |            15.0 |       AL80 (11.1 l)       |      HP100       |
|         3.00 |            16.5 |       HP100 (12.7 l)      |      HP120       |
|         3.25 |            18.0 |       HP100 (12.7 l)      |      HP120       |
|         3.50 |            20.0 |       HP120 (15.3 l)      |      HP130       |
|         3.75 |            22.0 |       HP120 (15.3 l)      |      HP130       |
|  4.00 <sup>2</sup> |     23.50 |       HP120 (15.3 l)      |      HP130       |
| <sup>1</sup> 2.00 lpm failed with low pressure error on compressor.  ||||
| <sup>2</sup> 4.00 lpm is the maximum flow I could get from a single Invacare PerfectO2 V 5 lpm concentrator with this pump.  ||||

The following results were found using a Invacare PerfectO2 V 5 lpm concentrator via the metered flow valve into a 200 kpa / 10 lpm pump running on 14.0 V (actually measured), drawing 1.2A, through final large accumulator and into a Phillips Respironics UltraFill.

^  Flow (lpm)  ^  Pressure (PSI) ^  Largest Successful Fill  ^  Could Not Fill  ^
|         3.50 |            23.0 |       HP120 (15.3 l)      |      HP130       |
|         3.75 |            24.0 |       HP120 (15.3 l)      |      HP130       |
|         4.00 |            25.0 |       HP130 (16.6 l)      |      LP121       |
|  4.25 <sup>1</sup> |      26.0 |       LP121 (19.0 l)      |      -- --       |
| <sup>1</sup> 4.25 lpm is the maximum flow I could get from a single Invacare PerfectO2 V 5 lpm concentrator.  ||||

A few words about the results.  First, don't take the level of precision very seriously.  The measurements will vary based on a number of factors even throughout a single fill.  The purpose was not to be highly precise, but even a half a PSI will make a noticeable difference, and rounding to a single PSI was just too coarse.

I would suggest that you run your own testing and document your own specific results.  My suggestion is to start with your smallest cylinder at the lowest flow you wish to work with and start the fill.  Be aware it can take a bit for the system to reach a steady point:  usually 10 minutes or so.  As the pressure in the accumulator increases the flow from the concentrator falls, and you will need to adjust the concentrator flow to return to your target flow rate.  You will keep doing that until it finally steadies.  Once it does, note the lowest pressure for a cycle.  Also note that the accumulator pressure will increase as the cylinder fills, so pick a consistent fill range to record your pressures.  For the table above, that is in the range of 2000 to 2500 psi or so.

Let the fill continue and see if the fill completes.  If not, note the failure, increase the flow, note the new accumulator pressure and continue the fill until it completes.  Then, note that successful fill.  Repeat with your next largest cylinder, starting with the pressure that successfully filled the previous cylinder.  You'll end up with your own version of the chart I have above.

Also, just because a cylinder is listed as being filled or succeeds in filling once, it is not a 100% guarantee that the cylinder will never error out for a too-slow fill.  It has happened that a setting that has previously filled a cylinder 100% successfully might rarely error out, usually during the last few hundred PSI or so.  If you simply restart the fill it will then complete successfully, especially if you let the pressure build in the accumulator before you restart the fill.  Most likely it's because those settings are right on the edge of a successful fill and a small variation in concentrator or pump performance might be just enough to time it out.

What if you want to fill larger tanks or sets of doubles?  Most likely, any set of doubles will have a water volume too large for the UltraFill to fill without erroring out.  The easiest answer is to simply close the isolation valve and fill them as separate singles.  As for people who want to consistently and reliably fill HP130s or LP121s without stopping:  you are probably going to have to investigate a larger pump, and if you want greater than 90% O2 concentration, either a 10 lpm concentrator or a pair of 5 lpm ones.  The larger or additional concentrator won't help directly, but it will allow you to supply the larger flow rates (4 lpm and above) and yet still deliver the highest concentration of O2.  And if you pursue this, let me know what configurations you try:  I would love to see a better solution for this.

==== Compressor ====
At this point, we now should have a setup that will supply the proper oxygen-rich gas at the proper pressure to our oxygen compressor.  But now what is the compressor going to do with it?

Normally, the compressor would put the gas into a medical cylinder using a proprietary connector.  But most likely we don't want it in a medical cylinder:  we want it in a SCUBA cylinder.  How are we going to accomplish that?  The answer depends on which compressor we have.

=== HomeFill II:  Use the QR ===
In the case of the HomeFill II we don't have a lot of choices on how to get the gas out.  On the HomeFill, there aren't really any fittings on the block that provides the cylinder Quick Release fitting.  In the service manual, there are flared tube fittings shown; on my actual compressor, however, those tubes are actually brazed in place.  There is a burst disk, but that isn't something that's easily used by another fitting and you probably don't want to remove the burst disk in the first place.  There is also a pressure switch that might provide a fitting we can use, but there isn't a lot of space in there to add a tee and high-pressure hose in that area.  In short, it's not an appealing prospect to tap into the high-pressure gas at that point.

For me, I was fortunate to purchase my unit with a medical cylinder on it as well.  That cylinder came with the quick-release plug on it.  That is threaded into the valve as a simple SAE ORB -6 (9/16-18 UNF).  You unscrew that from the cylinder valve and then purchase an adapter to go from ORB to whatever you want (1/4" NPT in my case).  You can then add a tee, pressure gauge and a 4' HP hose to a valve and DIN filler -- which is exactly what I did.  Basically, the QR plug will just stay in there permanently:  there's really no reason to take it out at that point.

What if you don't have a cylinder for the HomeFill?  Fortunately, the HomeFill is a relatively common setup in the world of home O2 cylinders.  Buying one is not that hard, especially on eBay.  And you don't even need the whole cylinder:  all you need is the valve, which can save you on shipping, too.  And that will be a lot easier than trying to muck around inside the HomeFill trying to find a suitable HP port.

Now, I admit that the fact I had the QR plug means that I did not put a lot of effort into trying to find a way to avoid using it.  I did look, though, and did not find it appealing.  However, if you do not have a QR plug and you are motivated to avoid buying one, I would start with exploring the pressure switch.  It is definitely a threaded connection designed to be replaced, and the low torque requirements called out in the service manual means that it is probably an ORB-style connection.  In fact, it's most likely an SAE ORB connection.  However, exactly what type of fitting and how to get a tee and hose of some sort into that space is left as an exercise for the motivated reader.  See how this same type of solution was implemented on the UltraFill below for ideas.

=== UltraFill:  Crack the Case ===
In the case of the UltraFill, we have a bit of a different situation.  Unfortunately, this compressor is not nearly as popular as the HomeFill.  That means that it can be **very** difficult to find a valve with the QR plug.  In fact, I could not find one for less than $200 or so, which seemed way too much to pay for a small piece of brass.  However, there are a couple of advantages that the UltraFill has here over the HomeFill.  For one thing, there is a great deal more space inside the case.  And for another, there is a perfectly usable fitting just sitting there for us to use.

On the block that provides the cylinder Quick Release, there is another component installed there:  the pressure transducer.  That part uses a simple SAE ORB -4 (7/16-20 UNF).  In fact, that is the same size as something us SCUBA divers use:  first stage HP ports.  I tried to find a standard (industrial) fitting that I could use to tee off and adapt to something else.  In the end, I found that a SCUBA part actually made the most sense:  a HP port 1-to-3 adapter.  These are usually used to allow someone to add a transmitter to a first stage that might only have a single HP port.  I chose the 1-to-3 because it is compact and had a fitting in-line with the original fitting as well as at a 90-degree angle to it.  This allowed me to screw the pressure transducer right back where it came from, just about an inch farther away.  There was enough space for it to fit without interfering with anything else.

To one of the perpendicular ports I used a SAE ORB -4 to JIC -4 fitting, and then used a 16" JIC-JIC high-pressure hose.  From there, I used a JIC -4 to 1/4" NPT **bulkhead** fitting.  I then drilled a hole in the case a bit above the low-pressure oxygen in port for the bulkhead fitting.  I recommend the JIC hose here:  there is not that much space, and the JIC fitting allows you to install the bulkhead fitting without worrying about how the hose will lay or how exactly it will end up when you tighten it like happens with NPT fittings.  The ability to swivel while tightening that JIC provides is very useful in this case.

Once you have this NPT fitting on the outside, you can do whatever you need to from there, such as a tee, pressure gauge and a 4' HP hose to a valve and DIN filler, just like the HomeFill above.

One note:  if you purchased a new machine like I did, you may want to make sure that your machine is working correctly before you begin this process.  You have the advantage of this write up when you are working on yours;  I did not.  I ran into all kinds of problems, mainly related to my cylinders not filling either at all or only for a short while.  I now know that this is likely because of me not supplying enough pressure to the input.  But in the beginning I thought it might be because my machine was broken.  I ended up tearing out all of the parts, re-assembling everything as it was and making sure that the compressor would actually run properly.  Then when I started to experience the error in filling large cylinders, I thought it might be because of leaks on the HP side.  So, I went back and removed everything again and re-tightened everything.  When you install everything, make sure everything is tight.  There's no need to overtighten:  ORB connections do **not** need much torque at all, and JIC doesn't need that much more;  there's no NPT inside to worry about, either.  But just like on the supply side, it doesn't take much of a leak to steal enough gas to matter.

I know I said that my goal was to do this without modifying the equipment, but I drilled a hole in the side.  I'm OK with that modification:  it's not like I was going to be able to return the thing anyway.  I'm also OK with adding these extra fittings and hoses:  they are easily removable to go back to how the machine came from the factory.  I just didn't want to alter the machine in an irreversible way or change the way it works from its actual design.  But if you want to avoid drilling a hole in the machine, you can:  one of the screws holding the case together uses a strange sleeve that creates a larger hole in the case.  That hole would be large enough to use to get your HP hose out of the machine, if you use a longer hose.  And then zero irreversible modifications!  However, hose routing and other issues for this solution are an exercise left for the reader.

==== Ready to Operate ====
At this point, you should have everything you need to start filling SCUBA cylinders with near-pure oxygen.  You have a concentrator generating low-pressure oxygen.  If you have a compatible compressor, you just plug it into the concentrator.  If you need to give your oxygen a pressure boost, you should now have a pump and accumulator plumbed in between the concentrator and the compressor.  Your compressor is now connected to a SCUBA cylinder by plumbing it to either the specific quick-release fitting or to an additional HP hose connected inside the unit.

=== HomeFill II:  Plug and Play ===
To operate the HomeFill, it's easy:  open the SCUBA valve, turn on the concentrator, turn on the compressor.  Wait about 5 minutes for the concentrator to be happy with the gas it's getting and it will automatically start operating.  Wait a bunch of hours (something like an hour per 100 PSI depending on cylinder size) and the compressor will stop operating when it hits roughly 2000 PSI.  Turn the units off and close the cylinder valve.  You're done.

About the only error you might encounter is O2 too low.  The proper light will light up and the compressor will stop.  Make sure the gas you are supplying is correct, both content and pressure.  Most of the time, that error happens because something is not connected properly -- or your zeolite packs have disintegrated...  :-(  Other than that, it should simply run until it's done.

=== UltraFill:  Lots of Moving Parts ===
The UltraFill is a little tricker.  First of all, because we're not using the HP output quick-release plug, the QR fitting needs to be manually engaged for the compressor to operate.  Use a small screwdriver or other tool to push down the flat ring in the bottom of the QR socket.  It will latch in the down position, which makes the machine think the cylinder is installed.  Don't push the square blue button below the QR socket **or** the socket itself, or the QR socket will release and your fill will stop.

Second, the compressor will fill both 2000 PSI and 3000 PSI cylinders.  It knows which is which because there is a small magnetic (hall-effect) switch installed in the QR fitting for 3000 PSI cylinders.  Of course, we don't have the QR plug or the QR valve...  No problem:  just place a magnet right over top of the QR socket.  It doesn't seem it has to be a particularly strong magnet:  refrigerator magnets work just fine.

It's best to do both of those things as well as open the SCUBA cylinder valve before you power up the compressor.  It checks for the magnet and cylinder pressure fairly early in the process and can get confused -- or at least react unintuitively -- if those things change later, so it's easiest to just have everything set up and then turn on the unit.

Once the unit is on and gone through its brief self-test, the blue pressure meter LED's will light up and show the general range of destination pressure.  The lights will be solid.  If you don't have the external pump running, a yellow light will blink.  That is telling you the input pressure is too low.  Build up the pressure and it will stop blinking.  When you're ready to start, press the large square black button.  The compressor will start and the last lit pressure LED will blink, showing that it's filling.

There are a few error conditions I have run into during a fill.  If the yellow light starts blinking again during fill, that means that your supply pressure is too low -- or too high (above 37 psi or so).  If the yellow light comes on solid with no beeps, that means that the incoming O2 content is too low.  The fill will continue for a short period of time (a few minutes), and if you don't correct the condition, the red error light will light up and the fill will stop:  it thinks the O2 content of the final cylinder is now too low.  You will have to stop and restart the fill -- and you're supposed to dump the contents of the destination cylinder to ensure the proper O2 percentage.  If the yellow light comes on solid and beeps, that is the dreaded 'the cylinder is not filling fast enough' error.  You're going to need to supply more pressure to the compressor or limit yourself to smaller cylinders.  Of course, see the manual for more details and for other lights and conditions.  But hopefully you just get a quietly-running compressor (and a likely not-so-quietly-running external pump) slowly filling your cylinder.

===== Equipment Modifications and Additional Uses =====
My goal when I started this was to be able to make these devices work without actual modifications to the equipment itself -- except for adding an adapter to a threaded  parts and maybe drilling a single hole in the case.  For the most part, I succeeded:  I can fill even reasonably large SCUBA cylinders with nearly-pure oxygen.  That allows me to create and compress sufficient gas for me to add oxygen to a variety of SCUBA cylinders, either to use them directly as a deco cylinder with high-oxygen gasses or supplying my Nitrox stick with O2 so my breathing gas compressor can create EAN32 to fill other cylinders.

However, leaving the equipment in its original condition brings along with it some limitations.  Others (like @Tracy and @Tbone1004) have suggested other uses that can benefit from or even require making permanent modification.  I had been reluctant to make changes to my equipment, but this is slowly changing.  I am slowly moving into making some of these modifications, but I'm fairly early into that process..  This section is a combination of documenting changes I have made, as well as untested possibilities that I wish to explore further in the future.

==== Compressor:  Bypass O2 Sensor to Pump Trimix ====
Not only are these compressors designed to pump oxygen, they are specifically designed to shut down if they are pumping anything other than near-oxygen.  One of the most useful modifications would be to adapt the devices to compress non-oxygen gasses such as Trimix.  Let's imagine you have two sets of doubles that you used with Trimix.  Now that the dives are done, you now have two sets that are each 1/3 full.  Most of the time, that means that you either let the doubles sit there waiting for your next Trimix dive, or you fill them with Nitrox and use the helium on a shallower dive.  But what if we could use the contents of one of the sets to supply the compressor and pump that gas into the other set of doubles?  That would free up one set of doubles with little waste, and we'd have a nearly-full set of Trimix doubles that will be a lot cheaper to fill up all the way.  Sounds like a win-win!

It doesn't seem like it would take much to make this work:  connect an inexpensive shop regulator to a BCD hose of a spare first stage, put that on one of the sets, set the pressure to the proper input pressure for your particular O2 compressor, and feed that into our compressor instead of O2.  Easy!  And it would be -- except for one thing:

The compressors include oxygen sensors to monitor the input.  If the oxygen concentration drops below a certain point (80% or so usually), the compressor will stop running.  And our Trimix will have a **lot** less than 80% oxygen.  So what to do?

The easiest answer is to bypass the oxygen sensor.  Unfortunately, it's not usually as easy as just removing the sensor:  that will cause the test to always **fail**.  We want it to always pass.  That usually means removing the sensor and replacing it by supplying the specific voltage signal that the computer is expecting.  In fact, that is exactly what @Tracy did on his UltraFill:  he used a USB charger to start with a 5V signal and used a potentiometer (variable resistor) to dial the voltage down to supply what the compressor wanted.  That was in his own words a pretty ugly brute-force solution, but it was straightforward and it worked just fine.  Users looking for a more contained and sophisticated solution might consider tapping off of the controller low-voltage power rail and using a voltage divider (maybe using that exact same potentiometer!).  If anyone does this, please let us know what voltage you need to supply:  it would make it a lot easier on the rest of us who want to go down that path.  Once you know what the supply voltage and signal voltage is, you can implement this with just a pair of properly-chosen fixed resistors.

==== UltraFill:  Replace the Electronics ====
The UltraFill is a complex device.  It monitors everything.  Unfortunately, this can get in our way -- especially the oxygen-sensing and 'speed-of-fill' function.  What about just bypassing everything and running the compressor in a much more manual way?  That would allow us to eliminate the features that get in our way, but will also eliminate some features that we really want, like pausing if the input pressure is too low or stopping when we reach the destination pressure.

There are a few ways that this can be done.  One is to simply apply AC power directly to the compressor motor.  This guarantees that the motor will run no matter what the input pressure is, what the output pressure is, or how slow a fill might take -- for good or for bad.  This is very simple:  splice into the wires that go from the controller to the motor, and wire them into a power cord, ideally with a power switch inline.  Supply gas to the compressor, hook up the destination cylinder and plug in the pump.  Of course, this will be a very manual setup:  you're likely going to need to sit there and watch the machine the whole time.  In addition, there will be **no** air cooling for the pump:  you will either need to use an external fan and leave the case open while it's running, or also wire into the cooling fan -- it too runs directly on AC power.  Either way you will lose the thermal monitoring and shutdown that the integrated controller provides.  You will also need to monitor the output pressure:  the machine will **not** shut down when it reaches the target pressure and will **very likely** damage itself if you exceed the proper fill pressure, especially if by more than a little bit.  @Tracy reports that even 200 psi is too much -- don't mess around with this.

Some of these features can possibly be restored.  For example, it may be possible to find a pressure cut-off switch rated for 3000 PSI, though is not a common off-the-shelf part.  A thermal cut-off switch could be added as well, if desired.  People who are familiar with embedded electronics may be able to envision an Arduino or such solution that might simplify and consolidate some of these features and might allow you to re-use some of the existing sensors (such as the Wika pressure transducer).  But these are not straightforward additions or changes:  it will require the proper electrical and gas-management configuration in order to do this, and it may make the compressor much more difficult to manage unless you also do additional work to add switches, indicators and other such features.  However, this will also **greatly** simplify the ongoing running of the compressor:  make sure you supply a sufficient pressure to keep the compressor happy (ideally 20 PSI or more), and it it will compress gas for you.  If all you want to pump is oxygen and you can provide enough pressure and volume to keep the compressor happy while successfully filling your desired cylinders, this is almost certainly not worth the hassle.  But if you can't keep the compressor happy -- likely because the target cylinders are too large -- this may be the only way to perform your fills.                                                                                                                                                                                                                                                                                                                                                    

==== Concentrator:  Bypass Regulator to Increase Output Pressure ====
There is a concentrator modification that can greatly simplify using an UltraFill compressor directly without using an external pump.  Internally, concentrators work with slightly elevated pressure:  in the neighborhood of 20 psi.  In fact, the gas from the zeolite canisters is pushed into the 'product tank' at that pressure.  A regulator is then used to drop the output to 5 psi for the user.  The product tank is there for largely the same reason as why we use the accumulator:  to give an even, consistent pressure to the end user, while still having the pressure needed to flush the zeolite free of nitrogen.  And we might be able to take advantage of that pressure, too.

There are a couple of ways that we could get access to this pressure.  One way **might** be to adjust the regulator to supply us with more pressure.  Most of these regulators are externally adjustable.  This adjustability is more to allow service techs to adjust the regulator to be at the proper 5 psi pressure, but they can allow a larger range of output pressure if you want.  Unfortunately, it's unlikely that they will allow you to adjust the pressure to a significantly higher pressure, but it might allow you to increase the pressure to a range that makes a compressor like the HomeFill happier without using an external pump.  10 or so psi might be a reasonable expectation for that modification.  (The identification sticker on the regulator on my Invacare PerfectO2 V seems to imply it has a range of 0-15 psi but I have not tested this.)

That might be enough to help an already working setup to work a little better, but it's not likely enough to help us to push the full 20+ PSI the UltraFill needs to fill larger cylinders.  So the other modification we could attempt would be to take gas directly from the product tank.  We might accomplish this in a few different ways.  Your product tank might have an unused fitting to use:  sometimes these are provided for service techs to check product tank pressure.  Another way might be to tap into the supply into the product tank from one of the zeolite cylinders after the check valves.  If there's nothing between the tap and the product tank, it is just like tapping into the product tank.  In fact, this is sometimes the procedure outlined in the service manual to check the product tank pressure, so examine the service manual and see if it might give you a clue.  In addition, there might be a tube that goes from the product tank to the pressure sensor on the controller board:  you might be able to tee into that hose.  One of these options **should** be available to you; but if it's not, you can always drill and tap into the product tank itself.  Obviously, this is the most risky -- it's hard to un-drill a hole -- but sealing even a badly drilled hole against 25 or so PSI is not the most difficult task, and there is much less danger from a catastrophic failure as there might be at 3000 PSI.

On paper, this seems to be a very useful modification:  let the big pump inside the concentrator deliver our 20-or-so PSI, doing the job it already needs to do.  Cleaner, simpler, quieter, even cheaper.  In fact, I believe this is exactly how the EverFlo concentrators designed to supply gas to the UltraFill work.  If you dig into the service manual, you will find that EverFlo units designed or upgraded to work with a compressor have a different regulator, one with an extra port.  I doubt they are adding an extra actual pressure regulator in there -- there's no procedure for adjusting the compressor output pressure, for example.  Most likely, that regulator just has an extra port into the product tank upstream of the regulator itself.  It might also include a check valve -- and if you decide to pursue a DIY version of this solution I would recommend putting a check valve in yourself as well.

I was able to pick up a second PerfectO2 V concentrator, which made me a lot more comfortable with making some modifications.  This is the first modification I made.  I added a tee into the hose from one of the zeolite packs to the product tank, downstream of the check valve.  I then plumbed this into the accumulator (with pressure gauge) and the UltraFill.  Note that in the future, I would suggest tapping into the hose that goes from the product tank to the controller board pressure sensor (the small square sensor, not the larger round O2 sensor).  This is the same hose that the service manual recommends for measuring the pressure of the product tank.

Tapping into the product tank worked reasonably well.  I was able to get just shy of 20 psi.  I was really hoping for a little more pressure -- I really need every psi I can get to fill those larger tanks!  I then hooked up an LP121 to the UltraFill and turned it on.  Unsurprisingly, it did not complete the fill:  20 PSI won't get that done.  I also tested with an HP133 with the same result.  But it **was** enough to complete a fill into an HP120!  For me, HP120's are the tanks I use for my Nitrox stick, so that is a huge help.  No external pump:  just the concentrator feeding into the accumulator and into the UltraFill.  Note that I did try it without the accumulator and it was unsuccessful at filling the HP120.  So it seems that the accumulator still gives us some benefit.  It's possible that a smaller accumulator might be sufficient, but I already have the larger one and I'll stick with it.

I have found one noticeable issue with this:  tapping into the higher-pressure gas seems to throw off the control electronics.  On the PerfectO2 V, the biggest issue is with the minimum-flow monitoring.  The concentrator will beep repeatedly unless it detects at least 1 lpm of flow.  I'm not exactly certain how it measures that flow:  there's no actual flow sensor.  Rather, I think it is using the pressure sensor to heuristically determine the flow rate.  And our tapping into the product tank throws that off.  Oddly, that effect does not seem to be consistent.  Sometimes, adjusting the flow meter output (connected to nothing) to even 0.25 lpm will stop the beeping.  Other times, even adjusting the flow to 2 or more lpm won't stop the beeping.  And that's with the same configuration in place.  I have not been able to determine a pattern to this:  sometimes it's easily happy, and sometimes it's impossible to make it happy.  Fortunately, it doesn't seem to affect actual operation, so you might have to ignore the beeping (or disconnect the speaker, I guess...).

Unfortunately, this clearly isn't a complete solution for all tanks.  At 20 PSI, my concentrator just won't supply enough pressure to fill an LP121, or even an HP130.  In addition, my examination of the service manual doesn't seem to offer a way to change the pressure of the pump.  It's possible that a different machine might supply a little more pressure:  this machine is pretty worn, and the service manual states that the target pressure should be 21 psi +/- 0.3 psi.  But I'm fairly certain that even a new machine supplying 21 psi will not supply enough pressure to fill that big of a tank.  In addition, you're not going to get much higher pressure without modifying the electronics:  the service manual states that there is a high-pressure limit set to 23 psi.  And to add insult to injury, the external pump I'm using simply stalls when supplied with nearly 20 psi, so I can't even use my external pump to increase the pressure.

Other concentrators may have different capabilities.  For example, the Respironics EverFlo service manual suggests a **maximum** pressure of 25 psi, but the expected nominal pressure might be noticeably lower.  In fact, the manual states a rather large range of pressure for the product tank:  14-25 psi.  Maybe the device will have a high enough pressure.  But maybe not:  testing and research will need to be done.

===== How I Use the Equipment =====
For me, I have been able to create a configuration that successfully fills sufficiently large tanks with near-pure oxygen without extensive modifications.  Originally, I used a pair of LP121 cylinders with my HomeFill, but these proved too large to be filled reliably by the UltraFill without constant intervention.  Fortunately, I was able to use a set of three unloved narrow-neck HP120 cylinders instead.  I use the HomeFill to fill them to 2000 PSI, then use the UltraFill to fill to 3000 PSI.  When the first cylinder is filled by the HomeFill and moved to the UltraFill a new cylinder is attached to the HomeFill.  That way, filling three smaller cylinders to 3000 PSI doesn't take that much longer to complete than filling two larger cylinders to only 2000 PSI.  In addition, I end up with more oxygen this way than before with just a pair of LP121.  The LP121 are now used to store helium.

Operating the HomeFill is straightforward:  the concentrator is connected to the compressor using a hose with matching male plastic quick release connectors at each end.  Connect the destination cylinder, turn on both devices and away you go.

Operating the UltraFill is not as straightforward, but not awful:  with an unmodified concentrator, take the metered output of the concentrator and feed it into the 200 kpa / 10 lpm pump.  Feed the output into the large accumulator.  Feed that into the compressor.  Turn everything on, and adjust the metered output to supply 3.5 lpm.  This supplies sufficient gas to successfully fill the HP120 to 3000 PSI, without over-driving the concentrator and dramatically lowering the oxygen percentage.

With a concentrator modified to supply higher-pressure output gas, operating the UltraFill has now become simpler.  I have tapped into the product tank upstream of the regulator, which allows the concentrator to supply 19-20 psi directly.  Feeding that into the accumulator and then into the compressor will also supply sufficient gas to successfully fill the same HP120 tanks.

Unfortunately, cylinders larger than HP120s are difficult to fill reliably from an oxygen concentrator.  Fortunately, there isn't that much of a requirement to fill those larger cylinders with nearly-pure oxygen;  you are much more likely to want to transfill Trimix into such cylinders.  In that case, you will likely be supplying the compressor from a regulator capable of supplying sufficient pressure and volume to keep the compressor happy, though to make it happy with the low oxygen content you will either need to bypass the oxygen sensor or drive the motor directly with your own controls and safety features.  In my case so far, for the very occasional times I want to do that, I have simply supplied power to the motor directly and then watched it like a hawk.

===== Conclusion =====
I hope that this document will be useful to you.  Hopefully this will help you to see how a medical O2 concentrator and compressor might be useful to you as a technical SCUBA diver.  Or, maybe this will help you to see that it might just be a lot cheaper and easier to have your local dive shop or compressed gas vendor take care of filling your bottles... :)

I have literally hundreds of hours and probably thousands of dollars into the research, testing, writing and editing of this material.  I hope that it will help you to avoid the wrong turns and rough edges that I have encountered.  But it's entirely likely that despite all this effort there are a number of refinements and improvements just waiting to be tried.  If you find one, please share them!  I would be grateful for your thoughts and experiences, especially as you go through this process for yourself.  You can only be a new user once, and your feedback is invaluable.  Please let me know what you liked about this process, and what could be improved in the future.