Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
2-3 week inspection & PSU temp comparison
#11
(16-Oct-2015, 03:38)Manoet Wrote: First temps after 3.5 hours even tho they've remained unchanged for the last 2.5 hours. As I figured, temps run right at 10% hotter with a full compliment of thermal transfer pads in place. Pads that are 17X more efficient than OEM pads! Proof the PSU enclosure dissipates heat faster thru the cover via convection cooling than it can possibly achieve via conduction thru its top only. I doubt many are surprised by that. What I am surprised about and still do not understand is how/why A/D board temps remain a degree or two and inexorably tied to the power supply temps!?! They did with OEM 200 before any modifications (43/44/45) and they've done it since in E-V-E-R-Y subsequent temp test, ie; vertical, horizontal, no convection, passive convection, forced air convection pushing or pulling. I'm completely stumped! I fully anticipated this test turning out the way it did except for the A/D boards. I seriously thought they'd stay pretty close to their original/normal 29/29C I've been running for two weeks or maybe climb a single degree, two tops. But they continually mimic the PSU temps in every test percentage-wise and I haven't the foggiest notion why that is or how it can be. I'm open to well thought out opinions as this has dogged me since reassembly.

Testing is such an obsessive pursuit! After another day of testing this configuration I've got to see if just a sliver of heat transfer pad I used originally on the rebuild can achieve cooler temps than 29/29/31C if I just use a single 2 x 25mm strip running centered vertically (in wall-mount position) on the ceramic pad which will increase flow thru the PSU housing significantly due to reduced obstruction, ie; cleaner airflow. It may not manifest itself as even a single degree reduction on the GUI but it could! Of course then the A/D temps will follow-suit and I'll be more flummoxed than I am now. See!?! It never ends!!

Have you found the actual temperature sensors on the board?
If you have how close are they physically to the item they purport to measure the temperature of?
The hottest bit of each of the three sections will be inside one of its components. There is a temperature gradient between this internal spot and the external part of its casing which the thermal sensor may be measuring, if it is directly mounted on the actual surface of hottest component. If it is mounted on the PCB near the hottest component, rather than on it, it will definitely measure a lower temperature when conduction de-coupled from the hot spot even if the actual hot component temperature does not change, particularly if there is circulating air too.
Devialet Original d'Atelier 44 Core, Job Pre/225, Goldmund PH2, Goldmund Reference/T3f /Ortofon A90, Goldmund Mimesis 36+ & Chord Blu, iMac/Air, Lynx Theta, Tune Audio Anima, Goldmund Epilog 1&2, REL Studio. Dialog, Silver Phantoms, Branch stands, copper cables (mainly).
Oxfordshire

Reply
#12
Pim-
What I sense I'm hearing is akin to edgy/raspy but I can't attribute it to a certain temp degree. It sounds best for about the first 5 minutes where everything is far smoother, clearer and more detailed. My mind wants to associate it with temp change but too many times having sat and watched the temp changes on the GUI I find its gone after a period of time but temps haven't changed. Weird I know but palpable. And I don't know how to deal with an SQ loss in a purely time domain scenario without temperature changes creeping into the equation. I work on it daily. Its constant and there-at-initial-power-up-then-gone. Its the best I can do today.


thumb 5-

I apologize for the percentage I originally used. Now I can see the error of my ways but I had to convert those number to Fahrenheit to do that. I'll try to not let that happen again. Bear in mind there was no effort on my part to enhance or exaggerate any findings... simply not fluent with Celsius as a 2nd language.


F1eng-

No, I haven't found the actual sensors but would dearly love to! I'm using sensors post modification as you or everyone else might sans modifications. Its the only game in town! Tho I do have the infra red and surface probe sensors for surface to balance and in some ways validate internal findings.

Now really, doesn't ANYONE have a clue why A/D temps parallel S temps so closely in all manner of different scenarios/situations?
Statements in my posts are opinion only, not to be construed as fact. Any projects I engage in are at my own risk! Their outcome cannot be assured and may result in success, small/no change or catastrophic failure. I encourage no one rely on anything I say or do as gospel and to realize your mileage may vary!
Reply
#13
(16-Oct-2015, 13:00)Manoet Wrote: Now really, doesn't ANYONE have a clue why A/D temps parallel S temps so closely in all manner of different scenarios/situations?

Manoel, this is pure reasoning and hypothesis from my part, but having A/D and PSU temp closely correlated make sense to me: the A/D part of the Devialet is by far the most power hungry part of the system, therefore it make sense that the more the A/D needs power which translate by a temp increase of A/D the more the PSU heats up delivering that power.

Basically, and again as an approximation, you can consider that heat is the result of inefficiency of the functions and therefore proportional to amount of power drawn from the A/D.

 my two cents,

Jean-Marie
MacBook Air M2 -> RAAT/Air -> WiFi -> PLC -> Ethernet -> Devialet 220pro with Core Infinity (upgraded from 120) -> AperturA Armonia
France
Reply
#14
Thanks Jean-Marie! Admittedly I have to squint less to see your explanation than any I've come up with. Any thoughts on why some others seem to have significantly higher spreads between board and supply numbers? I've seen many that approach 10C board-to-board and/or board-to-supply temp differentials and a few that exceed it. I'm not at all unhappy with the consistency/closeness of my 'temp trio' but wonder about the significantly higher spreads I'm seeing in previous posts. Realizing of course certain unmentioned environmental conditions, equipment settings etc in some of those posts when left unanswered don't help in trying to find or identify common cause/effect.
Statements in my posts are opinion only, not to be construed as fact. Any projects I engage in are at my own risk! Their outcome cannot be assured and may result in success, small/no change or catastrophic failure. I encourage no one rely on anything I say or do as gospel and to realize your mileage may vary!
Reply
#15
The last test I'll conduct in this series is one to try to confirm real world benefit, if any, of the uber expensive thermal transfer pads. Plan is to seal the 200 back up to OEM spec by taping over all convection intakes/outlets that were added during the chrome-to-anodizing conversion. That when coupled with the current PSU convection provision blocked off and running full OEM thermal transfer pads for yesterday's testing "should" provide a reasonably direct comparison between an OEM D200 with factory thermal transfer pads and a D200 with the high-end German thermal transfer pads without benefit of convection cooling. As close to apples-to-apples as is possible today. Only difference worth mentioning is the added benefit of radiation cooling provided by the black anodized Devialet enclosure vs the multi-plated chrome finish on the OEM.

This ones for you kenreau!
Statements in my posts are opinion only, not to be construed as fact. Any projects I engage in are at my own risk! Their outcome cannot be assured and may result in success, small/no change or catastrophic failure. I encourage no one rely on anything I say or do as gospel and to realize your mileage may vary!
Reply
#16
Manoet,

On my system too the temperatures are within 1C (33, 32, 34) at the moment.
I would tend to think that the mix between A and D depends on the speakers themselves since the lower the impedance the more work the class D amplifiers will have to do for the same energy from the class A amplifier. Also I suspect that the type of musical program can influence that ratio too: the class A amplifier works mostly to compensate the inaccuracy of the class D amplifiers which are higher for high frequencies than for low frequencies. So I would tend to think that a solo of crash cymbal would be more demanding to the class A amplifier than playing a pipe organ tube of 16 feet which would be more demanding on the class D part.

My point being that given a pair of speakers and a particular type of music, I would expect all temperatures to be linearly related, which mean that if they are within 1C around 30C it would require them to be in the 60s range to see a delta of 2C.

Jean-Marie
MacBook Air M2 -> RAAT/Air -> WiFi -> PLC -> Ethernet -> Devialet 220pro with Core Infinity (upgraded from 120) -> AperturA Armonia
France
Reply
#17
Thanks again Jean-Marie... that helps my understanding tremendously! Now some previous quandaries/paradoxes are beginning to come together, starting to make sense. Making order from chaos finally.
Statements in my posts are opinion only, not to be construed as fact. Any projects I engage in are at my own risk! Their outcome cannot be assured and may result in success, small/no change or catastrophic failure. I encourage no one rely on anything I say or do as gospel and to realize your mileage may vary!
Reply
#18
    All planned heat testing is concluded. I'll post that info when time permits. I immediately jumped into the next round of modifications this morning which includes making my Devialet wall-mount less restrictive to convection inlet airflow as well as the massive thermal transfer pad between the digital board and belly pan more efficient with regards to radiant cooling. The Devialet wall-mount besides being over-priced even after the price reduction of a few months back is still a POS! Sorry but no other way to say it. It showed up here with 2 mounting screws stripped which had to be repaired. No way I was sending it back as it took 2 months to get originally! And on my latest dis-mount for more testing I found another mounting screw had fallen off sometime. Sure I R&R this thing from the wall often but no way it should be stripping & shedding mounting screws at this rate. I know its frailties and am always careful. My guess is this one was stripped at the same time as the originals but since it wasn't lying loose in the original plastic bag like the others I presumed it was good. Bad assumption! These things should be able to stand up to thousands of R&R's rather than a dozen or so. While I was repairing the latest one I also prophylactically went ahead and did the 4th one.

The OEM wall-mount was never intended for convection cooling so I'm pointing no fingers of accusation at Devialet for any responsibility on this one. But if you look at the pic you can see the acrylic mount has three vertical rows of 52mm hexagon cut-outs. And as luck would have it (Grrr) the observation inlet ports in the Devialet belly pan fall squarely between the hex dividers on both hex's outside edges in the center row. Couldn't be more restrictive to those convection inlets in my case if they'd planned it!! I'm sure they also affect the radiant area of the large internal thermal pad between those two ports as well. I mean there's a tad over 5mm there for air to squeeze thru but why limit myself cooling-wise when you consider I can make that gap over 3 times that to an area that's 15.875mm X 160mm X full width of the Devialet? And here's the kicker; I can do it without raising the Devialet off the wall even a hair's breadth... none, nada, zero, zilch... bupkus! To me and my pursuit where everything's a compromise that's a MASSIVE freebie. And I'll greedily snarf it up!

No idea if it'll net me even a single degree of cooler temps. Don't much care. It'll help convection and radiant cooling and that's good enough for today. And it opens the door to perhaps CNCing an additional convection cooling inlet or two in the belly pan down the road. Those that made their own DIY wall-mounts using single pads at each corner netted even more cooling air if pads are 10mm or thicker. A project I may well find myself taking on down the road.
Statements in my posts are opinion only, not to be construed as fact. Any projects I engage in are at my own risk! Their outcome cannot be assured and may result in success, small/no change or catastrophic failure. I encourage no one rely on anything I say or do as gospel and to realize your mileage may vary!
Reply
#19
Good luck and I agree with your statement not really caring if it will give you an other degree cooler. When you look at the figures you have achieved in a room that is probably between 20 to 23°C, I think you are already quite close to the theoretical limits: less than 10°C above ambiant temp is really impressive!

Jean-Marie
MacBook Air M2 -> RAAT/Air -> WiFi -> PLC -> Ethernet -> Devialet 220pro with Core Infinity (upgraded from 120) -> AperturA Armonia
France
Reply
#20
I wish our onboard temps sensors had better resolution. I never know when a number is showing, say 30C if its 30, 30.1, 30.01 30.5 etc or if its 30.9 or 30.99. No way of knowing if when I get a degree drop if it really dropped .1, .01 degree or .9 or .99 degree. I'm sure its something that could be done with a FW upgrade but I won't hold my breath. The primary reason I'd like that is I've done things that I thought should easily net me a degree or two but doesn't. Conversely I've done things I didn't think would do anything and netted two degrees! When I removed the OEM thermal pad layout I used for testing I also removed the single largish pad I've been using since reassembly between the power board and cover. I replaced it with a 2mm wide sliver of thermal transfer pad thinking this would allow a lot more air 'thru' the PSU cover. Originally I wanted to eliminate that pad altogether but its required due to the ceramic heat generating pad being free-floating. Without a small piece of heat transfer between it and the heat-generating pad on front there would be nothing to prevent it from moving away from the identical large thermal pad in the rear which is the ceiling of the Devialet where heat is conducted into the Devialet aluminum case in quadrant 3 on the D's top surface. What I achieved by doing that was a whopping two degree temp reduction!! Down from 29/29/31 to 27/28/29. What I was hoping was maybe a single degree reduction in the GUI or even a half degree that I'd never know I'd gotten due to lack of sensor resolution. Now did I really get a full 2 degree decrease, maybe more or just over a single degree decrease? Frustrating not knowing as it makes trending much more difficult.

Last night before going to bed I set up the following test. I lowered my room thermostat from 78*F/25.55*C to 76*F/24.44*C and reversed the summer "pull" direction of my 4-speed ceiling fan to "push" and raised its speed from #1 to #3.

This pic shows the only sliver of thermal pad remaining in my PSU on the top side:

   
Statements in my posts are opinion only, not to be construed as fact. Any projects I engage in are at my own risk! Their outcome cannot be assured and may result in success, small/no change or catastrophic failure. I encourage no one rely on anything I say or do as gospel and to realize your mileage may vary!
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)