Tomahawk X870 + RM850x 3.3V voltage

[Forceflow]

Hey,

My 3.3V according to the bios and HWinfo is fluctuating between 3.240 and 3.260.

Is this normal? Had another RM850x lying around, swapped it to see if that made a difference, and the interval was now a bit tighter (3.248 to 3.260), but basicly the same.

Rest of system:

CPU: 9800X3D
CPU cooler: NH-D15 G2
Motherboard: Tomahawk X870 (latest BIOS)
Ram: 32 GB Kingston CL36
SSD/HDD: Lexar 4 Tb
GPU: Geforce RTX 5080 Asus Tuf
PSU: RM850x (both under a year old)
Chassis: Fractal Torrent
OS: Win11
Monitor: LG850

 

[Bill_Bright]

2/100ths of 1 volt is nothing, and frankly, quite stable.

The ATX Form Factor standard allows for a deviance of ±5%. So the +3.3VDC can be between 3.14V and 3.47VDC.

You are good to go.

 

[Forceflow]

It's not an indication that there's something wrong with my motherboard or PSU(s) that out of the box it reads that value?

 

[Bill_Bright]

Once again, 2/100ths of 1 volt is nothing.

 

[Forceflow]

Once again, 2/100ths of 1 volt is nothing.

Thank you - I mean: the fact that out of the box, it STARTS at 3.25-ish, is not a sign of degradation / a bad unit?

Switched the ATX cable as well: same result.

It's not an indication that something's off or degrading with the PSU's? The fact that I get the same readout from two identical units points at a slightly inaccurate motherboard sensor, right?

 

[Bill_Bright]

3.25-ish, is not a sign of degradation / a bad unit?

You are still only talking about 5/100ths of 1 volt.

3.25/3.30 = .985

That's less than 1% deviation from the standard 3.30V.

Is a deviation of less than 1% within the allowed tolerance range of +5% or -5%?

I bet if you look at your +12V and your +5V values, they are not exactly +12.00V or +5.00V either. And that's okay. The new ATX Form Factor Standard for the +5V is also ±5% tolerances. In fact, the new standard allows the 12V output to vary even more with a range of +5%/-7% (+11.16V to +12.6V).

Note as long as your PSUs output voltages stay within those allowed tolerances, the regulator circuits for your motherboard (and graphics card, drives, an other connected devices) are designed to easily compensate, if necessary. They are all specifically designed to work with those slight variances with no problems or degradations in performance.

Note the standard is based on an input voltage of 115.00VAC (230.00VAC). Odds are your input voltages are not exact either. And once again, that okay. Electronics are designed to, and can easily tolerate and adjust for slight variances such as this. And let me emphasis one last time, yours are slight, nothing to worry about, variances.

So seriously, Forceflow, while it is good you check your voltages and I recommend you keep doing so every 5 - 6 months. But note yours are perfectly fine and you need not obsess over them.

 

[Shrek]

I concur, your supply is just fine.

 

[Bill_Bright]

Thanks for stepping in with your second opinion, Shrek! I was beginning to think the OP didn't believe me.

 

[_roman_]

HWinfo is fluctuating between 3.240 and 3.260.

That is not a measurement. hoax value of some hoax software with hoax mainboard. You could borrow proper a, b, c and get a proper measurement setup to figure things out. Do not forget calibration first.

 

[Bill_Bright]

hoax value of some hoax software with hoax mainboard.

Nonsense.

It is not hoax software. And to suggest the Tomahawk X870 motherboard is a hoax is just being trollish! Come on, _roman_! Even your "proper a, b, c" does not make sense. What is that anyway?

Try to be helpful, okay?

The only thing you said that was true is the reading is not a true "measurement". But that does not suggest the value is not accurate enough for these purposes. It is.

What these HW monitors do is read a hexadecimal value/number from a voltage sensor, then convert that hexadecimal number into a decimal number. Simple, low-tech (a good thing in this case) and more than adequate for these purposes.

For a true "measurement", one must insert voltmeter probes into power supply cables or onto motherboard test or lead points, then read the reading on the meter. Yes, that is, technically more accurate but not necessary here.

And FTR, the meter does not need to be calibrated either.

 

[_roman_]

@Bill_Bright Did you understand what i wrote with measurement?

The only thing you said that was true is the reading is not a true "measurement". But that does not suggest the value is not accurate enough for these purposes. It is.

I will not here explain on how to measure properly. Or the basics. I did that too many times already. Even igor does it wrong sometimes in my point of view.

I do not get paid to educate other people on basics in electronics. I do not get paid to educate others on how to quantify values. Igor tries to explain things but sometimes fail miserably. He is a starting point for maybe 35% accuracy in my point of view on the topic. You may look up what i wrote even here about similar topics months ago.

And FTR, the meter does not need to be calibrated either.

Well - low quality post. Have fun on my ignore list. FTR Fuck the rest?

 

[mclaren85]

Congrats, for contributing my shitty obsession about my hardware. Now, I have one more parameter to check and monitor:/

 

[Marcus L]

Software values can be wildly inaccurate so yes for a proper reading you would need a multimetre to get a true value, seen some wild 3.3v rails reading 2.5** in HWMon/BIOS, 12v reading 9.***v etc, that said, if the values are true as mentioned the variance is very minute and WELL within ATX specs as Bill pointed out and could even be margin of error the % is that small, carry on

 

[Bill_Bright]

Have fun on my ignore list.

That would be wonderful. Thank you!

Software values can be wildly inaccurate so yes for a proper reading you would need a multimetre to get a true value

Very true. However, I have learned over the years through countless times double-checking with my meter, when reputable HW monitoring programs like HWiNFO64 or HWMonitor indicate the +12V, +5V and +3.3V voltages are correct (within allowed tolerances), they really are correct - and identical or nearly identical as measured by the meter.

But, when those software monitors indicate the voltages (or temperatures) are "wildly" incorrect (or missing), it is typically due to there being no sensor where the program is looking (there is no industry standard for sensor placement), or the sensor is faulty. After all, if the 12V voltage really was 9V, (I've seen anywhere from -5V to over +20V), odds are the computer would not be working at all. Motherboard regulator circuits are designed to compensate and adjust for minor deviances, not big ones. And if the temperature said the CPU was 190°C, it would have melted (or burned up) long ago.

One more thing to consider with software monitors (for both voltages and temps too). It is possible - even likely - if you run two monitoring programs at the same time, HWiNfo64 and HWMonitor, as examples, the voltages and temps will be slightly different. This is because (1) they may be reporting the values as seen by different sensors (again, there is no standard for this). But even if monitoring the same sensor, the values may be different because (2), due to constantly changing loads and other potential factors, the values may vary slightly, (3) it is unlikely each program started monitoring at the exact same instant in time, and (4) it is possible they have different sample rates (one may sample every 5 seconds, the other every 10, for example). So as long as they are close, and still within allowed tolerances, then all is good.

Unless qualified and understand the risks and precautions, I do not recommend users stick sharp, hardened metal, highly conductive meter probes into PSU cable connectors, onto motherboard circuit points, or anywhere into the heart of a computer case.

 

[Forceflow]

I thank all of you for your insightful comments. I've plugged the PSU into a different motherboard, and that indicates a 3.23V value.

So yes, this boils down to the quality and positioning of voltage sensors on the motherboard, and probably the attached components (the original system has 3 M.2 drives, which I think use the 3.3V rail).

I have a tendency to obsess over these things (and I sympathize with people who do too ), but this is really a case of "let it be" - it's a non-issue.

 

[Bill_Bright]

I have a tendency to obsess over these things

Perhaps but admitting that to the world is admirable.

but this is really a case of "let it be" - it's a non-issue.

Agreed.

and probably the attached components

To a degree, maybe. But as long as the load they present is within the capability of the supply, the regulator circuits are there to compensate for any differences different devices (or the number of devices) put on the supply.

So more likely you were more correct when you said "quality". And to be sure, this does not suggest these sensors are of poor quality. They are "cheap" (to make) but not cheap in quality. The use of these is, of course, about money but really simply indicates higher quality (read: more expensive) just isn't needed. This isn't critical, life supporting medical monitors, or "smart" missile guidance systems where true precision monitoring is needed to save lives.