PA120.3 vs HE120.3 vs PA160

[Marci]

Go on then guyz... tear it up...!
Posted in here as the end result is the resulting temps for the rads... and yes, I'm aware my testing methodology sucks :P This is just for my own amusement whilst we wait for rads to land with Cathar...
================================================== ===

A combination of unfortunate events recently resulted in an old familiar friend making an appearance on my doorstep... my old TEC rig had returned home...!



The guy I sold this to (R.I.P. d00d) sadly couldn't get his head round the TEC side of things so ran it as a standard watercooling setup. Once it got back to me however, twas time to get the TECs back on the job.



226w TEC on CPU (A64 Clawhammer 3200+), 172w TEC on GPU (X800ProVIVO >> X800XT-PE). Figured whilst I was in there I'd strip out the previous 240v Eheim1250 pump... hell... all the watercooling got ripped out basically... and refit with more up-to-date equivalents....



The radiator in this rig is the first ever ThermoChill HE120.3 prototype from way-back-when. Still going strong...



So... what was I gonna do with the rig now? Well, with the new PA Series rads finally getting sorted, figured may as well do a comparison of HE120.3 vs PA120.3 with the same fans... to do so required some monitoring tools in the form of the mCubed T-Balancer and some handiwork with silicon... thus giving me liquid in and out temps...



And air in and out temps





Add in the handy feature of the tbalancer software being able to talk to LCDC and we get...



MBM5 will collect CPU temps, and ATITool will handle GPU temps. So, that was the rig all ready... time to case it all up but leave rad external (just easier for swapping it over later)



Started off with unit under full load with as much heat as possible, whilst keeping speeds at stock.

CPU was set to 1.75v. AGP VDDQ to 1.6v. 100% load inflicted.

Temps were meaured every 20mins. Surveying them, they levelled out rapidly, within an hour, and hovered around the same points for the rest of the day (controlled ambient as best possible).

Max variance in ambient temp was 1.5 deg C. This was reflected across the board.

Note, temps shown are just a quick cross-section of readings recorded, not complete end-to-end results...

HE120.3 with 3x Panaflo M1A@7v

Air<: 23.0 | 24.0 | 23.5 | 23.0 | 23.5 | 24.5 | 24.0
Air>: 26.0 | 27.0 | 26.5 | 26.0 | 26.5 | 27.5 | 27.0
Liq<: 33.5 | 34.0 | 34.0 | 34.0 | 34.0 | 34.5 | 34.5
Liq>: 33.0 | 33.5 | 33.5 | 33.5 | 33.5 | 34.0 | 33.5
CPU: 17.0 | 17.0 | 18.0 | 18.0 | 18.0 | 19.0 | 18.0
GPU: 19.0 | 19.0 | 20.0 | 19.0 | 20.0 | 20.0 | 20.0

< = In
> = Out

So, averages are being used.

I'll summise testing of the HE120.3 with Panaflo M1A (fans at 7.04v by multimeter) as follows:

Avg Air <: 23.6
Avg Air >: 26.6
Avg Air Dt: 03.0
Avg Liq <: 33.9
Avg Liq >: 33.5
Avg Liq Dt: 0.40

Next step - remove HE120.3, install PA120.3 with same fans set to same voltage by Rheostat (measured with multimeter). This gave me the following recordset...

PA120.3 with 3x Panaflo M1A@7v

Air<: 22.5 | 24.0 | 22.5 | 23.0 | 23.0 | 23.5 | 22.5 | 23.0 | 22.5 | 22.5
Air>: 27.5 | 28.0 | 27.5 | 28.0 | 28.0 | 28.5 | 28.0 | 28.5 | 28.0 | 28.0
Liq<: 30.5 | 30.5 | 31.0 | 31.0 | 31.0 | 31.0 | 30.5 | 31.0 | 30.5 | 31.0
Liq>: 30.0 | 30.0 | 30.0 | 30.0 | 30.5 | 30.5 | 30.0 | 30.0 | 30.0 | 30.0
CPU: 15.0 | 15.0 | 15.0 | 14.0 | 15.0 | 15.0 | 15.0 | 15.0 | 15.0 | 15.0
GPU: 15.0 | 15.0 | 15.0 | 16.0 | 16.0 | 16.0 | 15.0 | 15.0 | 15.0 | 15.0

Avg Air <: 22.9
Avg Air >: 28.0
Avg Air Dt: 05.1
Avg Liq <: 30.8
Avg Liq >: 30.1
Avg Liq Dt: 00.7

Already, we see a substantial drop in CPU and GPU temps between the two rads, with the PA120.3 shaving 4 or 5 degrees off each over the old HE series.

Figured I'd go for some self-amusement next...

PA160 with 1x Panaflo M1A@7v
Well... I'm impressed...

Air<: 24.0 | 23.5 | 23.0 | 23.5 | 23.5 | 23.0 | 24.0 | 23.5
Air>: 31.5 | 31.5 | 31.5 | 31.5 | 31.5 | 31.5 | 32.0 | 31.5
Liq<: 37.0 | 37.5 | 37.5 | 37.0 | 37.0 | 37.5 | 38.0 | 38.0
Liq>: 35.5 | 36.5 | 36.0 | 36.0 | 35.5 | 36.0 | 36.5 | 36.5
CPU: 21.0 | 22.0 | 21.0 | 21.0 | 21.0 | 21.0 | 21.0 | 21.0
GPU: 21.0 | 22.0 | 22.0 | 22.0 | 22.0 | 22.0 | 22.0 | 22.0

Avg Air <: 23.5
Avg Air >: 31.5
Avg Air Dt: 08.0
Avg Liq <: 37.4
Avg Liq >: 36.0
Avg Liq Dt: 01.4

Only 1 or 2 degrees hotter than the HE120.3 manages on CPU & GPU temps, with 2 less fans = less noise. Always a bonus!


Break it down...

PA160 with 1x Panaflo M1A @ 7v: CPU - 21 GPU - 22
HE120.3 with 3x ^^ ^^ ^^ ^^ : CPU - 18 GPU - 19
PA120.3 with 3x ^^ ^^ ^^ ^^ : CPU - 15 GPU - 15

[j813]

Just gotta ask, what Motherboard are you using?

[Cathar]

Results (looking at air/water temps only) seem to be indicating a ~37% performance improvement with the PA over the HE at the fan power provided. That's even better than the results I was seeing with the old coarse prototype I had here. Should be even bigger gains on tap if you ran the fans at 5v and did a comparison there. Looks like the final production version has come on song very nicely indeed. Results are rather wild though, and swinging by quite large amounts, so hard to be really confident about what's being reported, but there can be no doubt that the PA is kicking the HE's butt by significantly more than any apparant margin of error.

[Marci]

Abit KV8 Max 3 (aka a horrible example of VIA Chipset misery)

[freecableguy]

Your CPU and GPU temps are less than the water temps. This is not right.

[Marci]

READ the thread Dingbat.

[j813]

What temps per Peltier? Sorry 0 knowledge & experience. :shrug:

[ricecrispi]

Very impressive work guys. Someone has been working hard as the results show.

Marci, what are you powering the pelts with and at what voltages? Could you do a quick run with two fans on the pa160. I always wondered what a pa160 could do with fans in push and pull config.

We need a rad round up with thermochills, coolingworks, BI, and some bonne and chevette heatercores. Maybe some other rads from Europe as well.

[nikhsub1]

Originally Posted by Marci

READ the thread Dingbat.

LMAO! Yeah, guess he missed the whole part about the TEC's

[Marci]

We need a rad round up with thermochills, coolingworks, BI, and some bonne and chevette heatercores. Maybe some other rads from Europe as well.

What we need is Swiftech to publish the report Bill wrote for them during his time there... */me prods at gabe with a pointy stick... c'mon d00d.... u know u wanna...

what are you powering the pelts with and at what voltages?

226w TEC - Meanwell s320-12 - 11.98v according to MultiMeter.
172w TEC - Tagan 480w (ie: The PC PSU) - 11.98v according to MultiMeter. (How convenient!!)

Could you do a quick run with two fans on the pa160

In progress, but it's a bit cooler here today - not sure if I can get ambient upto and keep it at same as previous tests were done at.

[Marci]

First observation - 2x M1A's @ 7v makes much more row than one of em on it's own...

[Cathar]

Originally Posted by Marci

First observation - 2x M1A's @ 7v makes much more row than one of em on it's own...

Need to get you a pair of Yate-Loon/Tricod's. At 12v, they'll perform much like the M1A's at 7v, and your ears will thank you.

[Les]

Originally Posted by Marci

...
226w TEC - Meanwell s320-12 - 11.98v according to MultiMeter.
172w TEC - Tagan 480w (ie: The PC PSU) - 11.98v according to MultiMeter. (How convenient!!)....

Observation - produce a very inconstant "Heat Dump"; HE120.3: PA120.3: PA160 appears to be 1:1.75:3.5

Edit. Inserted correct quote

[Marci]

S'looking like push-pull basically scrapes another degree off CPU & GPU in the current scenario...

Found the most annoying source of racket in this setup - MicroCool NorthPole... even 5v'd it gives off a distinct "I'm a small Delta" style drone... unavoidable - no alternative to hand.

Les - elaborate... you're saying my heatload is too variable?

[Les]

Originally Posted by Marci

....Les - elaborate... you're saying my heatload is too variable?

Only emphasizing Cathar's point(" Results are rather wild though, and swinging by quite large amounts, so hard to be really confident about what's being reported,..")
) and indicating that, to me, "Heat Dump" stands out
Perhaps, also,suggesting using TECs(for heat load) maybe part of the problem

[Marci]

Yep - heatload reproduction is an issue... hence not using this as any serious testing... just informing reported temps for folks to draw what they will. Anyways.... results for PA160 in push-pull... 2x Panaflo M1A's at 7v.

Air<: 23.0 | 23.0 | 24.0 | 23.0 | 23.0 | 23.5 | 24.0 | 23.5 | 24.0 | 23.5
Air>: 27.0 | 28.0 | 27.5 | 27.5 | 27.5 | 27.5 | 27.5 | 28.5 | 28.0 | 29.5
Liq<: 34.5 | 35.5 | 36.5 | 36.5 | 36.5 | 36.5 | 37.0 | 37.5 | 37.5 | 38.0
Liq>: 34.0 | 35.0 | 35.0 | 35.0 | 35.0 | 35.0 | 35.5 | 36.0 | 36.5 | 36.5
CPU: 19.0 | 20.0 | 21.0 | 20.0 | 21.0 | 21.0 | 21.0 | 21.5 | 21.5 | 21.5
GPU: 19.0 | 21.0 | 21.0 | 21.0 | 21.0 | 21.0 | 21.0 | 22.0 | 22.0 | 22.0

Push-Pull appears to be no significant benefit over push. (Well, nothing I can measure anyways)

[DaveMac]

Photo above has the fans pushing through the rad.
Was this on the PA, or both PA and HE?

Would pulling through give better results?

Were shrouds used on either set-up?

[Marci]

All rads were tested with fans pushing. Shrouds were not used other than on PA160 as requirement for fan mounting (as shrouds don't exist for PA120 series)

Doubt pulling over pushing would make any significant difference to the avg user.

Originally Posted by Cathar

The differences between pull and push have always been very minor, ~2% or so by those who have actually measured in the instances where pull was found to be better.

For the PA rads I noticed that push was better, and again by a small margin, ~2%. We're not talking BIG differences here.

I suspect that the reasons for why are two-fold (or perhaps more-fold).

1) The air coming off the fan blades is fairly turbulent. This is perhaps helping out the more open-finned radiators slightly. Perhaps those who measured better performance in pull-mode were noticing that against more densely finned radiators where laminar flow is fairly quickly established once the air enters the fins (ignoring the effects of fin louvring here).

2) The Thermochill radiators have a better default fan stand-off distance from the core, and so the differences between shrouded and unshrouded are fairly small as noted, as well as perhaps evening the score between pull and push modes somewhat?

I'm just reporting what I saw. Again, the differences were not big, and it may be measurement error, but I saw the same phenomena repeated across the various PA prototype cores I have here. Go figure.

[WarlordBB]

Originally Posted by Marci

All rads were tested with fans pushing. Shrouds were not used other than on PA160 as requirement for fan mounting (as shrouds don't exist for PA120 series)

Doubt pulling over pushing would make any significant difference to the avg user.

Just purchased my PA120.2 from the Thermochill webstore. I'm not gonna wait till some US store decides to stock them .

Thanks for doing this review Marci, I had already decided to look into a PA120.2 but your review made me want it now.

My intention is to replace the standard radiator on the Swiftech Extreme Duty Storm kit with this radiator. I'd love to hear some opinions on how much of a difference this will make.

Themochill's website had a question on checkout that said, in essence, "What made you decide to buy this product" to which I answered, "Cause Cathar said it's great".

I was also wondering if this radiator would be any benefit to my secondary loop which will all be Aqua-Computer and 6mm ID based. I know you'd probably need to know the components to judge this but I haven't decided yet. It will be a GPU block, Ramplex, Northbridge and possibly one or two HD coolers. I will either use a MCP350 or an Aquastream (Eheim 1046).

[shadowing]

When exactly are US stores going to stock them?

Btw, do the M1A's make a 'whiny' sound at 7V?

[ricecrispi]

They do make a whiny motor noise but get more air flow than nexus fans. I don't think the extra cfm will make much of a difference with the pa120.x because the fin pattern. If you seen the new pa120.x, the fin pattern is less dense. Very simliar to pa160 fin pattern. Marci has pictures available I believe.

Nice to know you guys are focusing on more towards silent WCing. If i have to buy a new rad for my pelt setup looks like I have an easy decision.

[ProHandyman]

Not to go off topic, but is Thermochill thinking/designing shrouds for the PA120.x series? Or is it that the overall design of the PA120.x series doesn't require a 30-40mm lift of fans off radiator core?

[Cathar]

Up close shot of the PA120.2 here:

http://canislupy.com/p/albums/userpi...01/h2o_009.jpg

Beware: BIG file

[j813]

Oh my that is a BIG picture file, 5.29mb

[Marci]

When exactly are US stores going to stock them?

When they actually get round to ordering some... out of our hands... hence why we've allowed ordering direct from ThermoChill...

They do make a whiny motor noise

USED to get the whiny noise, but with Panaflos received in the past 6 months, whine appears to have vanished. Dunno if Panasonic have changed anything in the motor windings etc but they're certainly quiet enough with no audible whine for me...

is Thermochill thinking/designing shrouds for the PA120.x series?

Shrouds aren't a requirement or a necessity, but some folks will still want them either way so yes, we will have shrouds available for them soon.