Marci n’ Fliptone Meet PhaseChange!

How-To mod a Prometeia – Marc Coyles

Fliptone and I have been working on the Prometeia Regassing stuff for what feels like a decade, but is only actually 5 months… 5 months of argument, debate, discussion, postulation, research and depression! Tomorrow (well, later today actually) we’ll be going for it!

We’ve been held up along the way by numerous points… the main one being how to extend a cap tube properly without it being detrimental to performance… and the second being how the hell do you pull a 29.X” Hg Vacuum with a vac pump that’s only rated for 26″ Hg…

Problem solved!

How? New Vac Pump!!! Well…. not “new” by any standard but…. in the pic below, on the right is the (tiny weedy and pathetic) old one, and on the left is the (huge and meaty) one we acquired today!


NB: Attaching to genitalia WILL kill you!

Pump 3

This puppy is an Edwards Speedivac; it’s reputed to be able to pull several Prommi’s simultaneously down to 29.x Hg Vacuum within an hour, as opposed to the regular vac pumps that take about 5 hrs to get near it on a single Prommi… this would be the case if the Prommi used ½” bore or larger tubing, but as such, pulling down thru ¼” tube means it’s gonna take some time….

Pump 2

So… that’s solved. Soon as you fire this puppy up attached to the gauges, it reads 29″ Hg instantaneously! So… what else do we have??

Brazing Gear is handy for a start…


Along with some nice tiny heads… so we can get a really fine flame goin’



Apparently, to do the job right you need a high percentage silver-based brazing rod. We acquired a load of 40% (about the best you can get and NOT cheap) silver rods…


And in case we need to braze anything other than copper to copper, we have some flux rods too…


How good are our brazing skills? Only one way to find out and that’s to leak test it all afterwards… and we now have a leak tester that emits a loud whiney noise when it cops a whiff of refrigerant…



We also have our trusty gauges that we picked up ages ago…


Problem being they’re for R12, R22 and R502 gas…. we’re using R404, R134 and R507


But never fear! We have a Comparator to convert R22 readings to R404 and 134 readings!!



To attach all this to the Prometeia’s low pressure line, we have line taps and control valve…. and also as an alternative to those, a static Schraeder valve (the brass one, bottom right):


Just in case… we have some copper pipe too…


And then there’s the bits left out! R404a gas for a start…


The Suppliers were shut today so we couldn’t get any 507, but we’ll be trying again first thing in the morning. If we can’t get any, then 404 it is for now, but we’ll get some 507 Tuesday anyways…

And finally, a baby tube cutter because you can’t fit a standard sized one in a prommi…. not enough room in the box…


All of which at the moment is sat in my basement! We’ll be attacking 2x Prommi MkII’s and a Prommi MkI tomorrow!


And why my basement? Mainly because it has a handy strip of concrete down one side of the room that we can safely braise on without torching the place!!! Secondly because it has a swing window as a convenient Fire Exit, and thirdly because I have several Fire Extinguishers and an Extractor fan down there… and the whole basement is already on one big RCD Trip… safety first!!!



We’ve changed the oil in the vacuum pump – once the pump has been used on a system, the oil captures the moisture dried out of that system. It then becomes creamy and milky and full of dirge, and eventually knackers your vacuum pump and makes it take longer to get down to -29.x” Hg. Not good. It should be completely clear and look a bit like baby oil…

So this morning, we emptied out the 5+ yr old oil that’s NEVER been changed and started flushing it through using Ester Oil.. Once we’ve got all the gunk out, it’ll get a blast through on the airlines and then filled with proper hivac oil… That should all be sorted by the morning so the vac pump will be ready to pull at full strength and speed… and off we go!!


Status update for this evening… Well, no pix I’m afraid! Unfortunately the camera software is on my PC, which is in the basement and I cba to bring it upstairs again to have to take it back down again tomorrow… and the only version of the software on the net available to download is a flaky revision that stops Windows booting on this rig! So, til my PC is done with in the basement… no piccies!

Anyhoo… today we successfully regassed my rig to 404a… WITHOUT a cap tube mod. Temp improvement of 13 degrees was achieved over the previous 134a charge. My XP-M at 2.8 Ghz with 2v was running at 100% load for over an hour at a level and stable evap head temp of -45 and core diode temp of -47. With the previous 134a charge, this was do-able but at -33 evap, -35 diode.

What have we learned? The charge is VERY specific to the load. If we drop the load, the suction line frosts up and onto the compressor… not ideal. To compensate, we have to reduce fan speed until the frost line is back at the original point…. 2″ from the compressor. SO… when the unit is optimally charged, the system must ALWAYS be run at the load used for the charge, or a higher load.

If charged properly, the unit should have headroom for a load increase without problems. If a lower load is used, you have to adjust to compensate to prevent floodback. Easiest solution to this? Always run the CPU at its stable oc’d limit at 100% load using F@H, or you will end up causing damage to the compressor in the long term.

Airflow over the condenser on this rig must also be increased so that the fans run at a minimum of 2000 rpm whilst clocked up to the limit at full load… the delta between air in the condenser and air out should be between 10 & 13 degrees. This confirms, as predicted, that this is NOT a mod for those wanting a silent system. Silence is simply not an option on a properly charged 404 system. If you want silence, stick with 134a.

Reason? The unit is designed to be silent at its factory specs. As soon as those specs change, the noise level can only increase, unless you change the specs to DECREASE performance. Only with a performance decrease from 134a will silence be achievable on anything other than the factory stock refrigerant.

Now… we have nailed the non-cap modded 404 charge and how to do it based on load, etc. What we plan to do tomorrow is find out exactly what benefits the cap-tube mod brings – if any. So we will be undoing everything we did today, starting from scratch by modding the cap tube and extending by 1.5 meters, recharging optimally to the new configuration, and comparing all recorded temps from today with the new temps from the capmodded system.

No 507 until after the Bank Holiday, sadly… once that arrives, we’ll be starting from scratch AGAIN on my unit, repeating all of the above with r507 so that we have a full set of comparisons. We also intend to investigate “CoolTop134” Substitute refrigerant. This is a DIRECT replacement for r134a but achieves approximately 10 to 15 degrees lower temps… thus, the same performance increase as non-capmodded 404a / 507 but adhering closer to factory specifications of the original unit, which should make the unit as a whole more efficient!

Pix tomorrow and results!!


Today we extended the cap-tube on the unit by 1.5 meters and gained even lower temps…. another 9 degrees lower to be precise…

We know that a regassed unit WITHOUT a captube mod is suited for a 200w heatload already. With the captube mod increasing performance for us, this shows that we aren’t applying a 200w heatload… we’re applying a much lower one – in the region at a guess of 170w or less… so the unit is now closer spec’d for day to day running conditions and as such is running much colder, and all the temps at each point across the whole unit are closer to their original tolerances. This is a good thing!!

At this point, the question was raised of using UV Dye mixed with the refrigerant to assist in leak detection. There are special UV Detector kits that you can get that allow you to inject the proper UV Dye into the oil of the compressor PRIOR to charging with refrigerant. Then any leaks can be traced with a UV lamp, but only whilst the system is running. Whilst running, the oil vaporises and mixes with the gas, so you detect the oil vapor leaking.

However, if the unit is off, the oil condenses again and settles, so no way to detect the leak. Ideally you don’t want to run a leaking unit as it’ll suck in moisture and mean more repair work: triple evac, both driers swapped out (there’s one under the insulation on the suction line), compressor oil-change, and then after the job, your vacpump oil will be contaminated so that’ll need changing too. To fill our vac pump up will take approx £20 of oil every change… Ester Oil for the compressor is more expensive… so ideally you want to be able to detect leaks with the unit off.

Auto & Domestic aircon rarely gets down to the vacuums and evaporating/condensing temps that we’re using (and therefore never get to the extremes in pressure that we’re using), so oil contamination by moisture isn’t AS harmful to the system but should still be avoided wherever possible. So yes, in those fields regular UV Dye can probably be used (check with a qualified HVAC Engineer first tho), but within these systems, it’s the proper made-for-the-job stuff or don’t bother…

Anyways, we haven’t given many temps etc yet, so here they are. All the work so far has been done using an XP-M 2500+ @ 2.8 Ghz with 2v CPU Core @ 100% load & DFI LanParty nForceII UltraB Mainboard. Load inflicted by F@H (Team 3365 of course)

Start of the project… first charge of r404a WITHOUT cap tube mod:

Diode: -47
Socket: +13
Case: +34
Room: +19
Evap Head: -45
Condenser Air In: +16
Condenser Air Out: +36
Condenser Coils: +34
Liquid Line: +32
Hi Pressure Line: +53
Compressor Top: +48
Compressor Bottom: +41
Suction Line: Can’t measure due to naff thermometer…

We then improved airflow over the condenser by cleaning it and adding another 120mm fan stacked behind the original (testing by Bill Adams on ThermoChill rads showed 2 fans stacked on one side tends to be better than push/pull… so we figured we’d stick with that theory even though the condenser is most definitely NOT made by ThermoChill)

Diode: -48
Socket: +16
Case: +33
Room: +18
Evap Head: -46
Condenser Air In: +17
Condenser Air Out: +33
Condenser Coils: +27
Liquid Line: +29
Hi Pressure Line: +54
Compressor Top: +52
Compressor Bottom: +43
Suction Line: What? You think thermometers magically “un-naff” themselves???

Then we decided to perform the cap tube mod. We’ve tweaked the charge a few times, so there’s a batch of temps not shown, but here is the best we’ve managed so far… this is exactly the same as above in terms of load applied, 1.5 meter extra cap tube, and a lower refrigerant charge….

Diode: -55
Socket: +12
Case: +35
Room: +20
Evap Head: -57
Condenser Air In: +19
Condenser Air Out: +29.8
Condenser Coils: +29
Liquid Line: +30
Hi Pressure Line: +50
Compressor Top: +49
Compressor Bottom: +41
Suction Line: Sorry, Thermy the thermometer isn’t home right now. Please leave a message after the beep.

So, as you can see the cap tube mod offers a significant performance benefit… we’re still unsure whether this is the lowest we can achieve at this load. We need to continue reducing the charge (which means remove it all, vac down and put it back in again, but using less 2nd time round and-so-on) to the point where the temp reaches its lowest, then starts coming back up again. So far, when we reduce the charge, it’s just getting lower, not coming back up over the performance peak, so we’re assuming it will go colder still… until it peaks we can’t tell.

Previously, at 2.7 Ghz the CPU Diode temp was at -39 deg C. At 2.8 Ghz full load on 134a, it was normally at about -36 deg C full load. Compare these to the temps above… we have achieved a 19 deg C performance increase (temp decrease) over 134a.

We’ve taken the rig off the lines now… brought it upstairs, and I’m about to hook it all back up again and give it a damn good thrashing with Battlefield: Vietnam… after reducing the charge a teensy bit more we ended up with:

Diode: -57
Evap Head: -60
Ambient: 19
Condenser Delta-T: +8 (air in – 19, air out – 27, coils – 28)

All whilst at 100% load at 2.8 Ghz with 2v.

We’re gonna’ run it at that til Saturday and see how she does! Now to get it hooked back up, get the pix dumped n’ resized and hopefully post pix filling in the gaps from first post to present tomorrow, then spend the rest of the day with the family.

Next plan – we have the OTHER MkII still to do… that one’s holding out for r507 when the supplier reopens after Easter Holidays on Tuesday morning. We’re gonna’ do that n’ compare the two as on stock 134a both rigs performed identically with same CPU at same OC etc etc… Then we’ve got 2x MkI’s to do… one with 404 and cap, t’other with 507 and no cap mod.

We’re hoping by the end of it all that we’ll have 4 more or less identically performing units if we’ve done it all right… then we’ll start on the Vapo SEs and PEs that we have access to… LOTS more work to do though on this rig before any of that begins… so far we’ve only used 1.5 meter cap extension. We’re not sure what effect lengthening it or shortening it will have yet…

So I now theoretically own a unit that can just beat a MkII GT! The MkII GT theoretically maintains -49 deg C evaporating temp (-28 deg C CPU Temp) at 150w. Well, we’re producing slightly more than that from what we can work out, at -60 deg C evaporating temperature and -57 deg C diode temperature… HOWEVER, we all know how inaccurate both the prommi temp sensors and mobo diode sensors can be… I can hand on heart say I think we’ve at least matched a GT tho…

Righty then… tis piccie time!!

First up, here’s the rest of the bits… cap tube, valves, and replacement suction line (if necessary)…



Adhesive armaflex insulation tape, again if necessary… shouldn’t be required tho…


Flaring and swaging tools for joining piping…



So, on we go then! First things first: Disconnect all power to the prommi. This unit has been idle for 24 hours to pressurise and cool down and has been sat in the same roo for 12 hours to get to ambient temp of the room. The room is well ventilated. All safety precautions are adhered to.

Next thing we do is fit a linetap to the low pressure side of the compressor… and onto this linetap we fit the control valve, which is removed once the job is finished.


Now, we attach the control valve to the fill line (blue) which is controlled by the left hand knob on the manifolds, and is read by the left hand gauge. With all valves shut, we open the control tap, then SLOWLY open the low pressure valve on the left. Once fully open, we can get an idle pressure reading (irrelevant) of the stock 134a charge… in this unit it’s 32psi, 2.1 bar:



We then slowly open the right hand valve (labelled Ref in the photo above… we had them hooked up wrong to start with) on the front of the manifolds… yellow… which runs to the vacpump/recovery unit… start the vac pump and remove the old refrigerant.


Once the unit is empty of refrigerent and repressurised, we begin attaching our static valves. We cut the charging line before the linetap…


We then flare the end


and push in the static valve!


For safety’s sake, we have a fire extinguisher nearby at the moment. Fliptone is an experienced brazer. I’ve done some in my time, but it’s been a while – I’ll get in on the action later on. We reach for the brazing torch, open a 6″ flame on the butane, slowly add the oxygen and bring the blue flame down to its smallest, sharpest finest tip possible…



Once the joint has cooled, we attach the blue line to the valve… and start up the vac pump to pull the unit down into a deep vacuum…


Vac 1

This took a matter of seconds. So we left it half an hour and returned… a few minutes later, things looked like this:


Notice the return of the linetap?? Well guess what… the static valves we’ve got won’t work with our fill lines. The fill line doesn’t have the correct center pin to open the valve.


So… back to the control valve method… we hook things up and start it back off pulling a vacuum!


At this point, we disappear upstairs for a good few hours… it takes a while to pull down to such a deep vacuum with 1/4″ lines… and we also want the moisture that has collected in the unit to burn off. As you lower the vacuum, the temperature at which water boils also drops…


To ensure as much moisture is boiled off as possible, we MUST get water’s boiling point to below ambient temperature – my basement is 19 deg C. The unit is a tad cooler at 17 deg C, so we need to pull at least a -29.3″ Hg Vacuum. Now, our gauges unfortunately aren’t that accurate. We couldn’t afford to buy or find to borrow a torr gauge to allow us to measure the degrees of the range between 29″ and 30″, so we just had to hope and pray that what we had was good enough and leave it for plenty of time to ensure as much moisture as possible was removed.


That’s it for part 1! I’m still sorting through pix and typing up the rest!! Stay t00ned!

For regular status updates on this project, head for:

(All pix by Marci / Fliptone… reproduction permission granted)

Marc Coyles

Be the first to comment

Leave a Reply