Tri Voltage Monitor

[MrAl]

Hi,

I was looking for a hardware based 3 voltage monitor, to monitor three voltages inside the PC computer. Didnt find one.
This would monitor the +12v, +5v, and +3.3v power lines coming out of the power supply. Even the first two, +12 and +5v might be good enough.
This would not be software monitor like CPUID, but a true hardware monitor, a hardware monitor that monitors the hardware, not a software monitor that monitors the hardware. Maybe with lighted LCD display.

Anyone see one of these anywhere?

If i dont find one i'll have to fire up an A. Nano and get busy with the code. Was hoping to find one though.

Alternately, anyone here ever use the IT87xx chip on most motherboards these days? (that provides voltages as long as the code can read them). This would take some code writing for the computer maybe at least in part in assembler.

 

[KeepItSimpleStupid]

https://www.linear.com/product/LTC2910 Also look at the recommended suggestions/

https://openhardwaremonitor.org/

Then this https://www.matrixorbital.com/index.php?cPath=45&osCsid=lk4pdg2l2vk4hh8gut5ln48gp4 if you want to roll your own.

 

[MikeMl]

How about three of these?

 

[tcmtech]

Yep I second the three mini LED volt meter method.

 

[KeepItSimpleStupid]

AeroCool X-Vision 5.25" Bay control 5 sets of Fans and Temp and voltage display. Color LCD Monitor

www.mewegg.com Item#: 9SIA3913V16645

Another drive bay naked kit: http://www.crystalfontz.com/search.php?q=CFA631&submit=Search

 

[dr pepper]

Google for pc power supplies, some of the older ones have a tl494 and a seperate voltage monitor chip in them.

 

[MrAl]

Hello again,

Some interesting suggestions, thanks a bunch.

I'll have to think this over now, but the three mini displays look good because it would come out being pretty cheap (around $10 USD) and easy to implement.
I looked at OpenHardwareMonitor and the code looks fairly complicated, not sure if i want to take the time to delve into that or not yet, but it sure is interesting.
The Aero thing only monitors the voltage of the FAN itself, as you adjust the setting.
The CrystalFontz site looks interesting too, although the prices are a little steep around 25 dollars USD and up, and it is unclear if they monitor anything or you have to supply your own voltage measuring hardware (like an uC chip) and send the readings to the display. A Nano with standard LCD would be much cheaper.

 

[JimW]

I am curious about what you are trying to accomplish. I used to design fault tolerant computer systems that had circuits for voltage monitoring. But there were completely stand alone (battery backed up) single board computers with remote communication and the ability to shut down and switch between N+1 power supplies.

In general, the 5, 12, and 3.3v supplies generally have a fairly large tolerance in most computers without failure. So are you looking for the something drifting over time/temperature towards the tolerance limit? Are you thinking about having a under/over voltage glitch detect circuit (because that would be the cause of most hardware system failures)?

And a final point, the most sensitive voltage to be monitoring is the CPU Core voltage. Depending on what your goals are, this is not something you want to ignore.

-Jim

 

[MrAl]

I am curious about what you are trying to accomplish. I used to design fault tolerant computer systems that had circuits for voltage monitoring. But there were completely stand alone (battery backed up) single board computers with remote communication and the ability to shut down and switch between N+1 power supplies.

In general, the 5, 12, and 3.3v supplies generally have a fairly large tolerance in most computers without failure. So are you looking for the something drifting over time/temperature towards the tolerance limit? Are you thinking about having a under/over voltage glitch detect circuit (because that would be the cause of most hardware system failures)?

And a final point, the most sensitive voltage to be monitoring is the CPU Core voltage. Depending on what your goals are, this is not something you want to ignore.

-Jim

Hello there Jim,


Some good points.

I was just looking to monitor the lines so i know if something is going wrong before it goes wrong completely.
I have the VCore voltage covered, so i dont need to measure that, but the +12 and +5 lines would be best for me to measure.

I just got another idea from Nigel however, which changes my total outlook on this matter now. He mentioned that measuring the ripple would tell us if the caps were going bad (something that i definitely want to know because i've had failures in the past only because of the large filter caps).

So now i think i might just build one, probably with a Nano because that would go quickly. What i plan to measure is at least the +12v and +5v lines, and also provide AC measurement for the ripple on the two lines. If the AC measurement changes too much i know the caps are going. What happened last time the caps went bad was the power supply would shut off as soon as it tried to start the computer, so the computer would not boot anymore. On the TV, the TV would not boot anymore due to the automatic shut down due to the bad caps. On the TV however i probably wont install a measuring unit, just on the PC.

 

[JimW]

I understand better now. The problem is that it is just not that simple. The outputs are from high frequency switched mode power supplies. So the the AC ripple, is really more of a high frequency noise. And the bigger problem is that the different components drawing high instantaneous currents are what cause the voltage drop outs that result in failures. So a failing capacitor on the 3.3V rail might not have any indication of failure at the 500KHz switch frequency, but could not prevent a 10 nsec. drop out caused by a data bus driver switching.

I am not trying to be overly negative, and I think your idea has merit. Getting a picture of the DC voltage and the measurable AC noise level would be a good data point to have. And then deviations from that over time might be a good indicator of an looming failure.

Historically, most of the failures we noted were complete power supply failures of one or more of the voltages. Most other hardware failures that were power supply related were random hangs or reboots that went away when the power supplies were replaced, but it took high end digital storage scopes to detect any problems with the voltages.

-JimW

 

[KeepItSimpleStupid]

When I did in-house repair of micro-computers that were ties to our laboratory measurements, the failures fell roughly into the following categories:

1. Power conditioning - I used OneAC power conditioners with an Isobar. Got like 18 years out of a Mac computer with just a floppy and fan failure.
2. Mechanical - Dust. Mainly caused fan failure. Fixed with PM's
3. Mechanical Wear - 8" floppies, regular floppies, fans.
4. Heat - The corollary of dust.

Once I did #1 which was about $1000 down time really dropped. The conditioning was implemented as the company (Digital Equipment Corp) went under.

One vacuum gage we had many of always had capacitor failures about every 4 or 5 years. It was really environmental. Another Sorenson power supply which we had many of basically needed a re-cap at the 10 or 15 year mark.

So, temperature, power supply voltage and ripple are really good things to monitor.

Crystalfontz and one of the other ones was just the display and/or buttons and everything else to be able to drop it into a drive bay, You get to roll your own,

While your at it connect it to ethernet with SMNT traps and or email/txt messages.

 

[MrAl]

I understand better now. The problem is that it is just not that simple. The outputs are from high frequency switched mode power supplies. So the the AC ripple, is really more of a high frequency noise. And the bigger problem is that the different components drawing high instantaneous currents are what cause the voltage drop outs that result in failures. So a failing capacitor on the 3.3V rail might not have any indication of failure at the 500KHz switch frequency, but could not prevent a 10 nsec. drop out caused by a data bus driver switching.

I am not trying to be overly negative, and I think your idea has merit. Getting a picture of the DC voltage and the measurable AC noise level would be a good data point to have. And then deviations from that over time might be a good indicator of an looming failure.

Historically, most of the failures we noted were complete power supply failures of one or more of the voltages. Most other hardware failures that were power supply related were random hangs or reboots that went away when the power supplies were replaced, but it took high end digital storage scopes to detect any problems with the voltages.

-JimW

Hello again,

Well most of the power supplies i have seen so far still use the 20kHz or even 10kHz switching frequency, but this could be changing so i'll be on the lookout for a higher frequency like 150k or 500k. The frequency is a good point, and so the measuring equipment will have to be able to deal with it.

Yes the way it changes over long periods of time is what i am looking for.

That's an interesting history lesson. I am not sure why my stuff has failed differently where the caps went bad and when replaced the device works like new again.

 

[MrAl]

When I did in-house repair of micro-computers that were ties to our laboratory measurements, the failures fell roughly into the following categories:

1. Power conditioning - I used OneAC power conditioners with an Isobar. Got like 18 years out of a Mac computer with just a floppy and fan failure.
2. Mechanical - Dust. Mainly caused fan failure. Fixed with PM's
3. Mechanical Wear - 8" floppies, regular floppies, fans.
4. Heat - The corollary of dust.

Once I did #1 which was about $1000 down time really dropped. The conditioning was implemented as the company (Digital Equipment Corp) went under.

One vacuum gage we had many of always had capacitor failures about every 4 or 5 years. It was really environmental. Another Sorenson power supply which we had many of basically needed a re-cap at the 10 or 15 year mark.

So, temperature, power supply voltage and ripple are really good things to monitor.

Crystalfontz and one of the other ones was just the display and/or buttons and everything else to be able to drop it into a drive bay, You get to roll your own,

While your at it connect it to ethernet with SMNT traps and or email/txt messages.

Hi,

Yeah i know what you mean about the dust and i found one other problem possibly with all AMD processors and heat dissipation.

I just did a major case cleanout with vacuum and all, and took out a lot of dust.
I have had so many fans fail i wonder it they are just made cheap sometimes.

The problem with the AMD processors looks like the type and/or amount of thermal paste they ship with. It looks like a very gray color not white. The color probably means it is a better type, but the main problem looks like it might be too thick. When i took my CPU out the other day, i found the layer was thick and dried out. Not super thick, but what i call too thick for comfort. A very thin layer works better. I was seeing constant 45 degree C temps in the CPU and that was even with no load. Now it is down to less than 20 deg C no load and that's much better. This might be something to check now and then if the CPU seems to be heating up too much.
Also, and maybe or not the fault of AMD, is there was thick dust buried in the plates of the heat sink. This happened to be worse than usual because the way the heat sink is designed it has very thin fins and they are very close together. Yes, it works better that way, but collects the dust unlike any heat sink i've ever seen before. I had to pop off the top fan to get to the heat sink fins and clean them out good. What a pain.

So the consensus seems to be that the power supply failures are not usually due to the caps going bad. Funny how my devices failed for that very reason though. Maybe it is related to the hours run time vs the amount of times the device is turned off and then back on later, so maybe my usage is very different than the consensus group's usage. Just a guess

 

[KeepItSimpleStupid]

So the consensus seems to be that the power supply failures are not usually due to the caps going bad. Funny how my devices failed for that very reason though. Maybe it is related to the hours run time vs the amount of times the device is turned off and then back on later, so maybe my usage is very different than the consensus group's usage. Just a guess

I said the caps go bad, but the caps going bad was purely environmental, The first line of defense is adequate cooling. That instrument had high humidity, high temperature, motors nearby and lots of other nasty things.

I had laptop power bricks go bad too. There was no power protection.

We had a Multi-channel analyzer that was hooked to a Scanning Electron Microscope. It kept blowing one of the boards. We did have schematics and i noticed zero power protection and I mean zero. I think we just used an ISOBAR and the issues went away. In any event, i called up the manufacturer and may have said it died during a storm and that they should know that power lines have problems and their wasn't a ZNR to be seen anywhere.

They promptly told me that the instrument requires 120 VAC at 60 Hz to function. Essentially, No spikes. I was definitely annoyed. Another $1500 USD for a repair.

There's a LOT of stuff here: **broken link removed**

Look at the Demo Room. OneAC is now PowerVar.

Somewhere there is a video using PowerVAr with an ISOBAR. The same system I used. It actually failed their tests.