1 (edited by electdaw 2023-10-04 22:21:45)

Topic: PCIe,PSU, Earth, common ground issue = noise!

hello

I have been using RME products for maybe 15 years, from 9636/52 , 9632, Raydat and now an AIO and I love them allot.

However, I have been having trouble with system noise, such as  mouse movement windows open, rendering files) mainly from the headphone out but also the unbalanced RCA (not so much with RCA as we can set hot levels) but with in-ear,  low impedance headphones it is very bad, in fact!!

Test room is very quite, no fan noise, all fans were reduced to almost inaudible levels for the test , so we can hear well.

2 x AIO's 
3 x PC's
3 x sets of various impedance headphones,

we were able to test various configurations and have found some interesting things.

we have gone as far as using 'insulation tape' wrapped around the AIO mounting bracket, so to isolate
the AIO from the chassis (As mains Earth is bolted to the Chassis of the internal PSU)

In the UK we have Earth in all our mains sockets/outlets, this is then connected to the chassis of the ATX  PSU. (standard) and that PSU is then connected with 4 conductive bolts/screws to the chassis of the PC and the AIO is also bolted directly to the chassis of the PC. So, the AIO has two grounds connected, the Earth from mains and the ground from the output of the PSU.

All my ATX PSU's have these 4 screws.

The AIO is designed to be used inside an ATX chassis and use ATX PSU's yet, there seems to be an issue with (common) ground.

1. when I power up the PC with the AIO bolted to the PCIe slot ( as it should be) and with all 4 screws of the PSU tightly connected to the PC chassis( as it should be) the interference noise is quite loud with various windows open and rendering.

2.When I power up the PC with the AIO 'isolated' with 'insulation tape' all over ('not' as is should be), so it does not touch the PC chassis/earth at all, but still with all 4 bolts of the PSU connected to the PC chassis(as it should be), There is no noise.  (if i wrap the AIO in tape = no noise)

3.When i power up the PC with the AIO correctly bolted to the chassis with no tape(as it should be) but with the PSU 'not' bolted, instead with the PSU loose inside the PC chassis and so with its own chassis isolated from the PC chassis in the same orientation, in the same place, just minus the bolts( 'not' as it should be)  the noise is no longer there. it is quiet. ( so no bolts = no noise)

So there is a definite problem with these two grounding points and the AIO

1.  The ground from the output of the PSU  ( black wires)
2,. The ground/Earth that is bolted to the PSU chassis > AIO mounting bracket.

The AIO seems to only like the one ground from the output of the PSU ( the black wires) if one allows the PSU's chassis/mains Earth to connect to the AIO through the PC chassis the interference noise is again present.

I have tried 3 x PSU's and my friend also has an AIO and PC! So that is 4 x PSU's and 3 x AIO's all tested in the same way, they all fail whilst bolted up correctly inside ATX systems.

Has the AIO got an Earth lift switch? Maybe it has, i can't see it? Would it be good if it was able to electrically isolate itself from the ATX chasis? future revision, maybe?

Yet, the newer AIO PRO version has the same issue. we tested that too.

Even when we disconnect the mouse,keyboard, Monitor, every add on peripheral = disconnected, with only the headphone connected whilst standing on rubber, in rubber shoes, in a quiet room, this issue is still present.

The tests were done with low impedance headphones approx 32ohm (in-ear), but also 64ohm , with 64ohm the noise was, of course, not as loud, but still present in a quite room.

This is maybe one reason why some people hear noise and some do not. also most people output to a mixer or headphone amp or receiver and use that headphone port instead , also many people use spdif or balanced out which is not effected!

There are many reasons why many people do not hear this noise, especially if sitting near to more than one pc fan, whilst overclocking.

I notice some people say " turn off power saving c-states" that reduces the noise, But one should not have to do that. Because With the AIO disconnected from the mains Earth, I can have c-states on, power saving on, and there is next to 'no noise' !  Whisper quiet!   Even as the CPU  clocks up and down ( we see in CPU-z.)

As soon as we reconnect the Earth back up to the AIO , by bolting the PSU back on/ pulling off all the insulation tape from the AIO = All the noise comes back!!

This did not take an afternoon to figure out. it was very stressful.

Any info would be gratefully received.

Thank you

Re: PCIe,PSU, Earth, common ground issue = noise!

Wow, great work! But I do not have an opinion if this is a design fault or expected behavior. I have not seen eh heard this issue with my 2 hdsp9632, in 2 computers and different PSUs, 3 or 4. And I did use low impedance phones and am super critical of this noise. I had a digidesign Mbox that was having this. From all the computers I own it was only silent and then dead silent on one cheap laptop, all other desktops about 4 and laptops about 3 it was noisy.
There is no opinion hidden in what I typed just sharing.
Cheers and thanks for posting!

Vincent, Amsterdam
https://soundcloud.com/thesecretworld
Babyface pro fs, HDSP9652+ADI-8AE, HDSP9632

3 (edited by electdaw 2023-10-05 08:22:59)

Re: PCIe,PSU, Earth, common ground issue = noise!

The interference (windows open, mouse, render) noise does not change whether one moves the total mix fader up or down, or alter any of the control panel power settings, even when all faders are muted the noise is still present; so is being induced at a late stage.

To recap, this is found in the unbalanced section of the card.  with RCA the signals can be hot, so it is not as noticeable. 

it is more a concern with headphones, even when nothing else is connected: not even a display-monitor.

With Sensitive 'in-ear' 32 ohms, in a quiet room, this really lets us hear the noise well!

32 ohm 'cans/headphones' not so much as they are maybe a 2-3cm further from the ear drums.

64 ohm cans again far less noise, relative to the signal,  but still present in a quiet room.

The balanced outs are fine. 
the spdif is fine

The noise must be induced only in the unbalanced section,  I initially thought this was interference through  'the air' inside the noisy chassis from a graphics card,  until physically isolating the AIO from the chassis, 

When we disconnect the AIO from chassis Earth, even when monitoring via sensitive in-ear 32ohms headphones, the interference-noise becomes almost gone so tiny, just a light hiss. perfect, as expected for unbalanced,  and the signal is still as strong. so signal to noise is 'dramatically' improved.

so the cards unbalanced system works when it is not connected to the PC chassis/earth.

I do not understand the AIO design,

Q: why has the AIO not got a plastic PCIe bracket/slot  with strong non conductive plastic mounting holes to isolate from Earth?
Q:Why does it have to be connected to chassis/earth and the black ground wire?

The AIO surly needs a chassis disconnect switch or a simple plastic slot with plastic mounting holes!
This would solve the interference issue for so many people.

After all these years, why did the Pro version not get at least one of these or as an option?

Surly it is not dangerous to isolate the AIO from the PC chasiss (mains Earth) as it has the black ground from the PSU cabling.

We noticed the dramatic noise/interference reduction when bench testing out of a chassis. as the PSU chassis/Earth was not connected to the AIO or in fact the motherboard as they were all isolated on the bench, only the black ground was connected across the components and it all worked brilliantly!! So why has the AIO got a metal Pcie bracket/slot??

interesting note. whilst on the bench and the PSU chassis isolated we took a crocodile clip from the motherboard ground and attached the other end of the crocodile clip to the AIO bracket and it still worked. so the AIO bracket likes ground from the black ground wire, just not via chassis>PSU>Mains Earth.

I am using 1.4v
the motherboard  stand offs are connected to chassis very well  (checked with ohm meter)
The noise is more noticeable when total mix is open and the total mix window enlarged (dragging it out almost full screen. "bbbbbrrrrbrbrbrrrrr" sound.)

Do we need a special type of PSU? A PSU that is more balanced? So the earth is balanced to the black ground output is some way?  There seems to be a 'difference' that the AIO does not like.

Is there an RME ATX approved PSU we can buy? Or a way to modify the card by removing/ scratching off a circuit trace, instead of wrapping around lots of tape, to isolate it from the chassis?

Re: PCIe,PSU, Earth, common ground issue = noise!

If you can also measure volts you could measure the volts between the ground wire and the case when the PSU is bolted to the case.

Vincent, Amsterdam
https://soundcloud.com/thesecretworld
Babyface pro fs, HDSP9652+ADI-8AE, HDSP9632

5 (edited by electdaw 2023-10-05 23:19:39)

Re: PCIe,PSU, Earth, common ground issue = noise!

vinark wrote:

If you can also measure volts you could measure the volts between the ground wire and the case when the PSU is bolted to the case.

Thank you, very interesting idea.

I shall measure the black wire to the chassis bolt nearest to the PSU IEC power input.
I am limited on what i can do, only one z77 board to play with.

PSU's used for this test, I shall call PSUA and PSUB

1.PSU-A = quite expensive XFX TS PSU (solid state caps) about 100euro.
2.PSU-B = cheap PSU about 25 euro.

RESULTS:

PSUA whilst bolted to the chassis:

no AIO installed 3.3mv
With AIO installed = 3.8mv ( noise present in 32ohm earphones)

Unbolted to chassis (PSU floating/isolated in PC chassis same orientation position.)  = 0.7mv ( almost no noise in ear phones)

PSUB whilst bolted to the chassis:

no AIO installed 23.3mv
With AIO installed = 26mv ( noise present in 32 ohm earphones)

Unbolted to chassis (PSU floating/isolated in PC chassis same orientation position )  = 2.6mv (exceptable noise, in ear phones, very quite) 

I wonder can we tune a PSU to balance it?  or buy a specific type that has this already done?

Can RME team up with a manufacturer to build  an RME approved PSU?

This is a big struggle to get a pro card working. this should be an easy plug and play experience.

side note: These audio noise tests are with sensitive in-ear phones simply because it makes it easy to monitor the noise. I would normally use 64ohm reference headphones which are less sensitive and so that does reduce the noise but, i can still hear the noise with the 64hm reference headphones,The in-ear's are perfect for these tests.

Thank you, i hope this helps.

6 (edited by SSC. 2023-10-07 10:57:09)

Re: PCIe,PSU, Earth, common ground issue = noise!

Nice findings! Really apprciate you sharing these on here.

I have been having and have had the same problem on two totally different builds over 9 years, the AIO, now AIO Pro. Always a problem with the headphones out. When I sent my 1st AIO card to RME, all I got back as an answer on the return form was 'grounding issue' no real explanation and never a mention of what specs they ran the card on their end.

Not that it matters, but both my builds were built by a top DAW building firm here in the UK, years of good reputation, so not a random build my me ect. The company source the parts they feel and know to be great for audio work.

Both builds was different specs, but ASUS motherboards and BeQuiet PSU's. Quite popular parts I would imagine.

I have been told of trying tape or plastic to break connection between the card and case slot, but never have tried that as it seemed mad to think you should have to do that on quite pricey cards, plus I was never sure on the warranty side on the cards front, or my PC's when tampering. Me being an non-technical head, I would never go about un grounding any parts in my PC. Shouldn't have to be doing that.

Turning off all C-States was the only thing that helped a tiny bit. It cut back the high pitch whinning, but not mouse movements, processer noise, windows, apps ect.

I get that all builds/PC's are different, but from researching quite a few people have had a problem with AIO's over the years and now the Pro too. I also know there will be people, including RME themselves, who never have issues on their setups.

As I need to get working, I just took the easy bypass route of the headphones out as I am fortunate enough to have other headphones amps, DAC's ect, but it would be really annoying if I was stuck with the headphones out of the RME.

To add---
On the headphones front, I have various tried 250ohm and a few 68ohm pairs.

Thanks again for posting your findings, electdaw.

7 (edited by electdaw 2023-10-07 22:08:37)

Re: PCIe,PSU, Earth, common ground issue = noise!

SSC. wrote:

Nice findings! Really apprciate you sharing these on here.

I have been having and have had the same problem on two totally different builds over 9 years, the AIO, now AIO Pro. Always a problem with the headphones out. When I sent my 1st AIO card to RME, all I got back as an answer on the return form was 'grounding issue' no real explanation and never a mention of what specs they ran the card on their end.

Not that it matters, but both my builds were built by a top DAW building firm here in the UK, years of good reputation, so not a random build my me ect. The company source the parts they feel and know to be great for audio work.

Both builds was different specs, but ASUS motherboards and BeQuiet PSU's. Quite popular parts I would imagine.

I have been told of trying tape or plastic to break connection between the card and case slot, but never have tried that as it seemed mad to think you should have to do that on quite pricey cards, plus I was never sure on the warranty side on the cards front, or my PC's when tampering. Me being an non-technical head, I would never go about un grounding any parts in my PC. Shouldn't have to be doing that.

Turning off all C-States was the only thing that helped a tiny bit. It cut back the high pitch whinning, but not mouse movements, processer noise, windows, apps ect.

I get that all builds/PC's are different, but from researching quite a few people have had a problem with AIO's over the years and now the Pro too. I also know there will be people, including RME themselves, who never have issues on their setups.

As I need to get working, I just took the easy bypass route of the headphones out as I am fortunate enough to have other headphones amps, DAC's ect, but it would be really annoying if I was stuck with the headphones out of the RME.

To add---
On the headphones front, I have various tried 250ohm and a few 68ohm pairs.

Thanks again for posting your findings, electdaw.

HI SSC

thank you for your reply

with the card isolated I can turn the power saving features back on.  which is nice when the cost of energy is high!

Also being able to use the headphone out is nice too, as i do not need to turn my expensive amplifier on every day for 10 hours. Preserving the amp's value, longevity!! very important to me.. and using less energy.

The headphione out is a very nice feature that should work for all users. We all pay for it.

So I am trying my best. I think when other users realise the danger in selling this card later on. fear of refunds, They also may request a fix.

Even if it is a cheap insualation sticker or two stickers (front and rear of the slot) so that the slot is perfectky covered and a plastic screw also, i am sure this would be very cheap to do.

btw: Just one tiny bit of metal touching the PC chassis will not work ) So one has to cover the AIO slot perfeclty. Then everything is great!!

Lets hope RME see the many messages from this week There are solutions to this issue! And even if it costs a little bit in postage and for the new part to arrive  I think it would be very much worth it, dont you? to maintain the value and also get great soundf!

I see you have suffered allot buying more than one card and for many years  you deserve better!