[time2throttle] Optimize XU3/4 cooling solution

Post Reply
User avatar
fse
Posts: 42
Joined: Tue Feb 03, 2015 8:40 pm
languages_spoken: english
ODROIDs: XU4, C1
Contact:

[time2throttle] Optimize XU3/4 cooling solution

Unread post by fse » Fri Aug 14, 2015 1:57 am

Hi,

the stock cooler on the XU3/4 gives me very little time (a few seconds) of raw, full power, something I am not used to coming from x86 hardware.

I wrote/attach a simple shell script which measures the time in seconds before the CPU is throttled: time2throttle.sh

I get the following consistent result:

@cold start 20C air temp

Code: Select all

root@odroid:~# ./time2throttle.sh 4
RUN: instances=4 startTemp=42C time=3.91380643845s
How it works:
  • It uses burnCortexA8 and kernel output to detect the throttle event, it starts 2 parallel instances.
  • It PreCools the CPUs with 100% FAN-setting before each RUN to ensure consistent results.
  • A command line option (number) allows you to add more instances.
I tested a slightly larger cooler, with no pad, but thermal paste and smaller fan: Here I can get ~26s with instances=4 consistently.
DSC_0455.jpg
(51.9 KiB) Downloaded 4404 times
I wonder what other ppl get with their custom cooling solutions. Your are welcome to improve the script and try it out.

My cooler design goal is to have it run at 4 instances without throttling ever. :D

time2throttle.sh: https://raw.githubusercontent.com/f-s/t ... hrottle.sh
GIT repository: https://github.com/f-s/time2throttle

Thanks
Last edited by fse on Wed Aug 19, 2015 9:03 pm, edited 6 times in total.

mikronauts
Posts: 225
Joined: Wed Aug 27, 2014 4:28 am
languages_spoken: english
ODROIDs: odroid-w, odroid c1
Location: Langley, BC, Canada
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by mikronauts » Fri Aug 14, 2015 3:07 am

Interesting result!

I agree that the XU4 cooling needs improvement, and I also find the fan a bit too loud for my comfort.

Having said that, I REALLY like the performance :)
http://Mikronauts.com ... Home of RoboPi, Pi Rtc Dio, Pi Jumper, EZasPi

User avatar
mrneilypops
Posts: 95
Joined: Thu Mar 12, 2015 6:29 am
languages_spoken: english
ODROIDs: XU3-lite,XU4,C2(x2)
Location: luxembourg
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by mrneilypops » Sat Aug 15, 2015 6:24 am

Interesting idea...I will give it a try when I have some spare time...

User avatar
fse
Posts: 42
Joined: Tue Feb 03, 2015 8:40 pm
languages_spoken: english
ODROIDs: XU4, C1
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by fse » Mon Aug 17, 2015 5:34 pm

Here my results for the stock cooling (@21degree C, room temperature)
root@odroid:~# ./time2throttle.sh 4
RUN: instances=4 starttemp=57C time=1.47591018677s
root@odroid:~# ./time2throttle.sh 3
RUN: instances=3 starttemp=56C time=4.37799525261s
root@odroid:~# ./time2throttle.sh 2
RUN: instances=2 starttemp=55C time=23.3243393898s
A drop by 10degree in env/room temp gives me an additional 10sec for 2 cores. But the 1.4s b4throttle for 4 cores is pretty bad.

TbirdMan
Posts: 39
Joined: Thu Jun 18, 2015 2:38 am
languages_spoken: english
ODROIDs: C1, C2, XU4
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by TbirdMan » Tue Aug 18, 2015 1:42 am

something to account for that i found on my XU4. due to the offset position of the SoC and lack of another component of the same height under the heatsink area, the stock heatsink tends to not sit flat on it. Mine was making maybe 25% contact. I put a small piece of aluminum from the top of a drink can on between the USB3.0 hub controller and the heatsink and that made it much better but this only works if you have something like the adhesive on the stock heatsink or maybe sticky thermal compound to hold the shim in place. If you're mounting the bigger sink the same way it will have the same issue. I made odroid aware of this and they are looking into it.

User avatar
fse
Posts: 42
Joined: Tue Feb 03, 2015 8:40 pm
languages_spoken: english
ODROIDs: XU4, C1
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by fse » Tue Aug 18, 2015 4:23 pm

TbirdMan wrote:I put a small piece of aluminum from the top of a drink can on between the USB3.0 hub controller and the heatsink and that made it much better but this only works if you have something like the adhesive on the stock heatsink or maybe sticky thermal compound to hold the shim in place.
Hi did you get anything better with 4 instances and the correct placement of the stock heatsink? So far there is no confirmation of my results ...

tebruno99
Posts: 16
Joined: Fri Jul 24, 2015 3:13 pm
languages_spoken: english
ODROIDs: ODROID-XU4
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by tebruno99 » Thu Aug 20, 2015 12:55 am

TbirdMan wrote:something to account for that i found on my XU4. due to the offset position of the SoC and lack of another component of the same height under the heatsink area, the stock heatsink tends to not sit flat on it.
I've been thinking about this as well. I already replaced my heatsink and fan with great results and I've been checking under mine and it seems to be making good contact. I think I will take it apart though and put 2 layers of thermal tape under the heatsink where the offset is and press pretty hard to set it so I'm always sure to be making full contact.

TbirdMan
Posts: 39
Joined: Thu Jun 18, 2015 2:38 am
languages_spoken: english
ODROIDs: C1, C2, XU4
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by TbirdMan » Sat Aug 22, 2015 12:02 am

fse wrote:
TbirdMan wrote:I put a small piece of aluminum from the top of a drink can on between the USB3.0 hub controller and the heatsink and that made it much better but this only works if you have something like the adhesive on the stock heatsink or maybe sticky thermal compound to hold the shim in place.
Hi did you get anything better with 4 instances and the correct placement of the stock heatsink? So far there is no confirmation of my results ...
I have not tested it, I will see about it this weekend.

TbirdMan
Posts: 39
Joined: Thu Jun 18, 2015 2:38 am
languages_spoken: english
ODROIDs: C1, C2, XU4
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by TbirdMan » Sat Aug 22, 2015 10:16 am

Ok, I ran the test a 4, 3, and 2 instances, 23C ambient in stock plastic case.
RUN: instances=4 startTemp=51C time=2.10887074471s
root@odroid:/home/odroid# ./time2throttle.sh 3
RUN: instances=3 startTemp=51C time=7.60197615623s
root@odroid:/home/odroid# ./time2throttle.sh 2
RUN: instances=2 startTemp=51C time=36.9071581364s
An improvement, but the stock HSF is just too small. with the embedded fan there is very little fin area.

JagDoc
Posts: 61
Joined: Sun Feb 24, 2013 2:06 am
languages_spoken: english, deutsch
ODROIDs: ODROID-X2, ODROID-U2, ODROID-XU, ODROID-U3, ODROID-XU3, ODROID-XU3lite, ODROID-XU4, CloudShell, ODROID-C2, ODROID-HC1, ODROID-MC1
Location: Germany
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by JagDoc » Sat Aug 22, 2015 7:05 pm

With orginal cooler.
Cut off the thermalpad from the cooler where it sit on the SoC. Then put some thermalpaste on the SoC and mount the cooler.
Now it have perfect contact.

root@odroidxu41:/home/odroid# ./time2throttle.sh 4
RUN: instances=4 startTemp=37C time=24.9458475113s
root@odroidxu41:/home/odroid# ./time2throttle.sh 3
RUN: instances=3 startTemp=37C ^C

With 3 instances the max temp was 85 C and i killed the task after 10min.

Edit:
With scaling_max_freq 1900000 the max temp was 91 C with 4 instances and i killed the task after 15 min.

dertester
Posts: 120
Joined: Fri Feb 20, 2015 8:51 pm
languages_spoken: english
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by dertester » Sat Aug 22, 2015 7:32 pm

RUN: instances=4 startTemp=43C time=133.966139078s

With northbridge heatsink and stock fan.
The heatsink is thermalglued on the chip.

It helps to cool the back of the board.

If i didnt screw up the glue, i think the chip itself cant transfer the heat fast enough. One of the cores is 95°C while the others are at 83°C. Also the ram-on-package and hot backside bring me to this conclusion,

User avatar
fse
Posts: 42
Joined: Tue Feb 03, 2015 8:40 pm
languages_spoken: english
ODROIDs: XU4, C1
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by fse » Tue Aug 25, 2015 8:54 pm

dertester wrote:RUN: instances=4 startTemp=43C time=133.966139078s

With northbridge heatsink and stock fan.
That sounds like a giant jump up to me, is that @2Ghz :?: Do you have a photo of the setup? Do you cool the backside?

With thermal paste and startTemp <<33C and low fan I can improve things to 75sec @4instances at best.

Applying extra pressure/downforce to the cooler (for better contact to die) and polishing the base, I cannot improve the heat transfer much at all.

The deciding factor for me is the startTemp: e.g. 43C will already give me only 26sec @4instances :roll:

dertester
Posts: 120
Joined: Fri Feb 20, 2015 8:51 pm
languages_spoken: english
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by dertester » Wed Aug 26, 2015 8:00 pm

I used arctic alumina as themal glue and the stock thermal adhesive on the stock cooler for the pmic next to the cpu.

The tape helps to direct the air through the heatsink untill i get one of these noctua 40mm ones.
For normal work i just keep the fan at a fixed 70-80 pwm value, to keep it cool and quiet. 60-70 gives me some additional rattle, above 80 its some high pitch noise..

For the test post above a had a very slow 140mm Fan @ 5V blowing from the side to cool the underside . Thats also why i use these standoffs.
Without its 100-120s.

I guess with very strong fans on both sides it might be possible to keep it throttle-free. But on the oder hand i can live with short throttles just to keep it quiet.

My config:
Image

TbirdMan
Posts: 39
Joined: Thu Jun 18, 2015 2:38 am
languages_spoken: english
ODROIDs: C1, C2, XU4
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by TbirdMan » Sat Aug 29, 2015 11:09 pm

My next attempt will be to use the C1 heatsink (it is the same heatsink without the material removed to embed the fan) with a 40x40x6.8mm Sunon fan to see if more heatsink will improve the time. My goal is to find a solution that will fit in the stock case that will cool adequately.

rowan194
Posts: 146
Joined: Tue Jun 02, 2015 1:43 am
languages_spoken: english
ODROIDs: C1, XU4, C2
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by rowan194 » Sun Aug 30, 2015 3:38 pm

Hmmm, this is not good news. I guess it doesn't matter too much when someone is using an XU4 as a "desktop", where high loads are short and transient, and throttling may not even be noticed, but my app is CPU bound and ideally should be working with all (or most) cores at full load 24/7/365. Seems that this may not be possible on an XU4, even with a reasonably decent heatsink upgrade. :(

Ambient temp 20C, XU4 with stock fan in stock case:

Code: Select all

RUN: instances=2 startTemp=47C time=113.465072393s
RUN: instances=3 startTemp=48C time=13.6849861145s
RUN: instances=4 startTemp=48C time=3.17732214928s
I already have a 4 x C1 cluster, was considering building a 4 x XU4 cluster, but if there's no practical heatsink solution then I will have to limit the application load, or let the CPU throttle itself at alarmingly high consistent temperatures. Will continue watching this thread.

dertester
Posts: 120
Joined: Fri Feb 20, 2015 8:51 pm
languages_spoken: english
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by dertester » Sun Aug 30, 2015 7:56 pm

The throttle starts 95°C and temp limit is 115°C. Its quite a difference. I wouldnt be worried to much.

Im really interested how good the internal heat transfer is (maybe due to the ram on top). Even with my big heatsink i see temerature rising way to fast. At least faster than i would expect.
With a better fan and maybe a better (copper) heatsink it probably wont throttle.

I see that the fins of my heatsink warm up quite a bit before it throttles. With a little more air moving i could probably prevent throttle.

@rowan
If you add 4 good heatsinks and 4 good fans and can live with the high temperature, i guess it will be fine.
Also it should be possible to downclock just a little but still blow the C1 away.

TbirdMan
Posts: 39
Joined: Thu Jun 18, 2015 2:38 am
languages_spoken: english
ODROIDs: C1, C2, XU4
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by TbirdMan » Sun Aug 30, 2015 10:32 pm

I suspect the PoP RAM is the culprit here. It is likely the heat cannot transfer through the RAM fast enough to be dissipated by the heatsink. These SoC's were not designed for 24/7 number crunching, they were designed for mobile computing which is a low duty cycle situation. Unless you can do something like active cooling via a Peltier or maybe water cooling you are gonna face this throttling eventually.

User avatar
fse
Posts: 42
Joined: Tue Feb 03, 2015 8:40 pm
languages_spoken: english
ODROIDs: XU4, C1
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by fse » Mon Aug 31, 2015 3:15 pm

dertester wrote:I see that the fins of my heatsink warm up quite a bit before it throttles. With a little more air moving i could probably prevent throttle
This is my finding too. Using the sink heatsink pictured in the first post and the https://www.pollin.de/shop/downloads/D320493D.PDF fan (>6CFM) I get no throttle with thermal paste. (no glue required, but extra 5V powersupply ~7000rpm)

Important: Running this fan from the on board 100% PWM supply the speed (voltage) drops on high load and slows the fan ever so slightly (using the recommended power supply): the throttle kicks in at ~240sec.

rowan194
Posts: 146
Joined: Tue Jun 02, 2015 1:43 am
languages_spoken: english
ODROIDs: C1, XU4, C2
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by rowan194 » Sat Sep 05, 2015 2:59 pm

Slightly crazy idea - I wonder if undervolting would be possible? In terms of CPU, undervolting means to reduce the voltage, which will reduce consumption (and heat). I've had a lot of success with this on X86 motherboards - for example, I reduced the power on a Q6600 system at 100% load from 139W to 63W, simply by dropping the CPU from 1.2875V to 0.975V - but it's a lot easier to do there, because the BIOS lets you directly set the CPU voltage. On a tightly integrated system like the XU3/XU4 you would be undervolting everything, which would include the CPU, RAM, integrated SoC devices, even external USB devices. You would have a lot less room to reduce voltage before the first subsystem started failing, but it may still be possible to shave off a watt or so.

I am using a DC-DC converter so I can set the input voltage precisely, but even a crude inline diode to drop 0.2 - 0.3V may work. Will try some experimenting later.

BTW the beauty of undervolting is that you are not affecting performance in any way - the CPU still runs at the same speed. All you are doing is reducing the safety margin between absolute stability, and failure.

User avatar
rooted
Posts: 6440
Joined: Fri Dec 19, 2014 9:12 am
languages_spoken: english
Location: Gulf of Mexico, US
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by rooted » Sat Sep 05, 2015 4:09 pm

Actually you can undervolt each aspect of a SoC, it's not linked. I haven't checked the source closely but I'm fairly certain that you can change the voltage in code, you could even expose the DVFS to userspace and control it in real time....just like most Android phones.

See this to see where the DVFS values are stored: https://groups.google.com/a/chromium.or ... LS-XrRrSBg

None of this is advisable but hey, it's a developer device...go for it if you want to.

rowan194
Posts: 146
Joined: Tue Jun 02, 2015 1:43 am
languages_spoken: english
ODROIDs: C1, XU4, C2
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by rowan194 » Sat Sep 05, 2015 4:17 pm

Quick undervolt experiment:

I have reduced the input voltage to 4.5V at some load (2 cores at 100%) and it's still stable.

Catch is that the fan speed also reduces slightly with voltage, which may negate the advantage of less power = less heat.

I had a look at the schematics, couldn't find any dropout circuit (resets the CPU when voltage is too low) but it is possible it's integrated into the SoC. So I guess at 4.5V I'm not far off failure.

User avatar
odroid
Site Admin
Posts: 30028
Joined: Fri Feb 22, 2013 11:14 pm
languages_spoken: English
ODROIDs: ODROID
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by odroid » Sat Sep 05, 2015 4:47 pm

The Brown-out detecting level is around 3.5Volt in the PMIC as far as I remember.
If your source of 4.5Volt is stable, there will be no negative side effect except for a few voltage sensitive USB devices.

User avatar
rooted
Posts: 6440
Joined: Fri Dec 19, 2014 9:12 am
languages_spoken: english
Location: Gulf of Mexico, US
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by rooted » Sat Sep 05, 2015 5:44 pm

Interesting, my Pi 2 freaks out at anything less than 5.13v according to the SmartPower. Perhaps I should try running the XU4 @ 3.6v and see what happens to the watt/hr usage.

rowan194
Posts: 146
Joined: Tue Jun 02, 2015 1:43 am
languages_spoken: english
ODROIDs: C1, XU4, C2
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by rowan194 » Sat Sep 05, 2015 6:41 pm

rooted wrote:Actually you can undervolt each aspect of a SoC, it's not linked. I haven't checked the source closely but I'm fairly certain that you can change the voltage in code, you could even expose the DVFS to userspace and control it in real time....just like most Android phones.

See this to see where the DVFS values are stored: https://groups.google.com/a/chromium.or ... LS-XrRrSBg

None of this is advisable but hey, it's a developer device...go for it if you want to.
This is intriguing. Not sure I'd be comfortable with hacking at the kernel, in fact I'm not even sure how to compile a Linux kernel. ;)

I see from page 26 of the XU4 schematics that it has multiple buck/boost converters, so maybe undervolting the input won't make that much difference, since the converters will maintain the same output power ratio, drawing more amps when the input voltage droops. The buck/boost converters may have to be adjusted in software as per your suggestion.

Some quick testing confirms that reducing the input voltage makes very little difference to overall consumption, in fact as you go low enough it actually increases it...

Code: Select all

Load: 2.00

5.0V x 1.42A   = 7.100 VA
4.5V x 1.57A   = 7.065 VA
4.0V x 1.79A   = 7.160 VA
3.5V x 2.17A   = 7.595 VA    <-- yes, it works at 3.5V! The heartbeat LED is a bit dimmer than usual. :D

User avatar
rooted
Posts: 6440
Joined: Fri Dec 19, 2014 9:12 am
languages_spoken: english
Location: Gulf of Mexico, US
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by rooted » Sun Sep 06, 2015 6:33 am

I figured it must be boosting to be stable that low.

TbirdMan
Posts: 39
Joined: Thu Jun 18, 2015 2:38 am
languages_spoken: english
ODROIDs: C1, C2, XU4
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by TbirdMan » Tue Oct 20, 2015 11:31 am

as promised, i installed a C1 HS and a sunon fan. I had to cut a hole in the top as the heatsink and fan was about .75mm to tall. reran the test, don't have the exat figures but with a start temp of 45c it made it to about 2.9-3sec on 4 instances. not sure what more can be done short of a heat pipe.

hbieganowski
Posts: 21
Joined: Thu May 28, 2015 3:06 am
languages_spoken: english
ODROIDs: XU3-lite
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by hbieganowski » Sun Jan 24, 2016 7:20 am

TbirdMan wrote:I suspect the PoP RAM is the culprit here. It is likely the heat cannot transfer through the RAM fast enough to be dissipated by the heatsink. These SoC's were not designed for 24/7 number crunching, they were designed for mobile computing which is a low duty cycle situation. Unless you can do something like active cooling via a Peltier or maybe water cooling you are gonna face this throttling eventually.
I made some tests and I'm not sure if even with Peltier I could get good enough rresults... :(

I put the cloudshell outside ( it's minus 7.0 deg. C right now here!) for 15 minutes and then I run tests:
4 instances / governor: PERFORMANCE / start temp.: 20 deg.C - only 27.79 sec.
3 instances / governor: PERFORMANCE / start temp.: 20 deg.C - no throttle / running stable (even 30% more load because of running MATE Monitor tool) - test killed after 10 mins.

In a room temperature (about 21 deg.C):
4 instances / governor: PERFORMANCE / start temp.: 42 deg.C (with precooling) - 4.70 sec. !!!
3 instances / governor: PERFORMANCE / start temp.: 39 deg.C (with precooling) - 15.40 sec.
2 instances / governor: PERFORMANCE / start temp.: 41 deg.C (with precooling) - 250.79 sec.

Hardware modification: northbridge heatsink + Noctua NF-A4x10 at CPU + AAB Cooling thermopad glue 6W/mK 1.0mm, Cu heatsink at eMMC
Images:
https://drive.google.com/file/d/0B-HnD_ ... sp=sharing
https://drive.google.com/file/d/0B-HnD_ ... sp=sharing
---
XU4/cloudshell/32GB eMMC/1TB HDD
C2/VU7+/16GB eMMC/Wifi 4/BT
XU3-lite - with broken PMIC - anybody could help? :D

dertester
Posts: 120
Joined: Fri Feb 20, 2015 8:51 pm
languages_spoken: english
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by dertester » Sun Jan 24, 2016 10:19 am

I think the chip Material itself has a Bad thermal resistance.

What definitely gets you a few seconds is a strong fan on the back side.

rowan194
Posts: 146
Joined: Tue Jun 02, 2015 1:43 am
languages_spoken: english
ODROIDs: C1, XU4, C2
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by rowan194 » Thu Jan 28, 2016 7:04 pm

TbirdMan wrote:as promised, i installed a C1 HS and a sunon fan. I had to cut a hole in the top as the heatsink and fan was about .75mm to tall. reran the test, don't have the exat figures but with a start temp of 45c it made it to about 2.9-3sec on 4 instances. not sure what more can be done short of a heat pipe.
Just out of interest, I tried a C1 heatsink with a 60mm 7000RPM fan blowing downwards. time2throttle.sh lasts about 4.5 seconds with 4 instances @45C start temp, and 7+ seconds if the CPU is allowed to cool down a bit between runs (start temp mid to high 30s)

3 instances lasts around 45-50 seconds before throttling.

2 instances go better, with the temp staying fairly constant at around 80C, so it seems it may be able to run indefinitely. With the stock fan this load will cause throttling after about 15 to 20 seconds.

So this is a definite improvement over the stock heatsink + fan, but at a reasonably high load (between 2.0 to 3.0) it is still not sufficient to prevent throttling. The fan is also 25mm thick, so there is no way everything would fit in the case.

dronus
Posts: 87
Joined: Fri Jul 25, 2014 7:24 pm
languages_spoken: english
ODROIDs: U3, XU4, C1+
Contact:

Re: [time2throttle] Optimize XU3/4 cooling solution

Unread post by dronus » Fri Apr 29, 2016 8:14 am

I think undervolting need to be done by software, if possible.

Undervolting the whole board below the usual 5V will have almost no effect at all: Despite the USB, almost anything on the board uses lower voltages, there is a complex regulator for the CPU core voltages too, and those regulators would keep the internal voltages up as long as the input voltage is high enough.

So to lower the CPU core voltages, DC input voltage would have to go to 2V or even lower, which most likely will not work because some peripherial complex would stop working before.

Post Reply

Return to “Projects”

Who is online

Users browsing this forum: No registered users and 1 guest