• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Matthias of RME on external power supplies!

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Don't own the RME and like KSTR believe it is well designed.

As such the leakage currents that may or may not be present may or may not show up in SE cable operation depending on the usage, used cables, used cable length and wire routing and equipment it is connected to.
In test conditions this is likely to NOT show up at all.

The concerns KSTR has with SMPS mirror mine.
In MOST cases this isn't an issue at all in audio but in some specific cases it might.
Also it really doesn't care if the PS in question is mV noise level or not.
The leakage currents are common mode and travel from mains via signal ground in other equipment and out back via mains.
Unless one specifically measures for this it won't show up.

The test from KSTR is very valid and usual in EMC testing (sometimes into the GHz range depending on the certification)
I rarely (if ever) see very relevant tests like this in the audiocommunity.
These kind of tests can ruthlessly show why some folks experience ground loop hums.

About power supplies in general.
Leakage currents can vary between very low and unacceptably high.
They can be mainly mains or HF only or combined.

For some equipment (low PSRR) a 'better' quality PS can give better results.
With high PSRR equipment it usually doesn't matter.
A wimpy power supply can be problematic when dynamic power is drawnsuch as in headphone amps driving low impedance inefficient headphones or speaker amps.

A higher amperage power supply in pretty power constant devices (DACs, pre-amps, smaal power headphone amps) usually is pointless unless the original one gets very hot and/or is not well suited for its task.
 
Last edited:

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,386
Likes
24,752
Location
Alfred, NY

A few. First, note that I used the (unbalanced) headphone outputs for most of my measurements and completely generic cables. I did not see any of the ground loop effects that you were able to create, but to be fair, I did not deliberately inject those currents as one would do in a test specifically aimed for EMC compliance and I strictly looked at the DUT output under a very typical user condition.

If I read your results correctly, worst case you were able to induce some noise at -110dB. I will admit that under any real-world use, I would be unable to hear that, and certainly not in the way described by the people on the RME forum linked before. My own opinion is that if this represents an exceptional case (and I think it does), why increase the costs for the 99.9% of other users by having a fancier power brick? The 0.1% who might experience the minuscule noise addition could either not use the RME or get themselves a different supply.
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
I do agree that for most any case of use the problem I illustrated shouldn't be relevant... except for one, clearly: The Adi-2 Pro is explicitly advertised as a measurement device. A less experienced engineer or hobbyist might fall into that trap, though. Those who are aware of this will use a better supply anyway (or an inserted isolator like I did), and they might also use an isolator for the USB connection (when TOSLINK isn't capable enough for the task).

The setup I used is actually a very common one, I did not trick in any way to get unrealistic high readings, though in a typical install the shield impedance of the RCA might be lower (then again, we know even unshielded interconnects with AWG28 internal wires are sold), plus there's two of them.

Personally, the mechanical issues of the stock supply is the really relevant topic. It most certainly is not "road-ready". The DC cable is crap and there is no easy way to fix this thing in a rack or something. For that reason alone RME should offer a more rugged version (solid cable, dogears for mounting), and while they're at it they could also change the electrical design. A low leakage design with (at least) medical leakage specs but without medical approvals doesn't cost a fortune, neither in development nor in production.
 

rwortman

Addicted to Fun and Learning
Forum Donor
Joined
Jan 29, 2019
Messages
740
Likes
683
[
RME builds gear to transport dozens of channels too and from mixing desks and that guy tries to convince them you need same magic cable to reliably transport 2 channel audio...
I have no problem getting 32 channels of bidirectional 48/24 audio from a mixing desk into my laptop with a cheap USB cable that came with something I bought. I don't even remember where I got it.
 

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,641
Likes
2,809
I might extend my measurements (like adding the exact cable specs, etc, testing different cable, test also against quasi-floating source which still can result in a lot of disturbance , provide a real-world example -- measuring noise floor of non-floating preamp with only unbalanced output, etc etc) and present this at the RME forum as well. RME is known to treat user input respectfully when one has data to back up. And as I said, some engineers at RME (actually IMM staff, which develop and build a lot of RME hardware) have confirmed they are aware about the leakage current issues of their stock supply. We can even find some posts by MC where he states in certain conditions he would suggest use of a linear supply with more benign leakage to solve some buzzing issues in the off state of the interface....

Hi KSTR, did you do more measurements?

And did you present your findings to RME? Any response from Matthias?
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
Not yet. More "hard data" is definitely needed before I will contact RME. The lacking mechanical ruggedness of the supply brick and its output cable is the only thing that is clear, without any doubt.
 

Perik

New Member
Joined
Jan 1, 2021
Messages
2
Likes
0
Not yet. More "hard data" is definitely needed before I will contact RME. The lacking mechanical ruggedness of the supply brick and its output cable is the only thing that is clear, without any doubt.

Hi @KSTR
I'm a ADI-2 Pro FS small scale audiophile user and curious if this went anywhere further?
I use high end balanced XLR output but the AC/DC input feels more like being to a clock radio. Or to be fair, I don't trust the wave and current from the wall.

Edit: I must add though, it's a fantastic little thing this DA/AD converter!
 

rwortman

Addicted to Fun and Learning
Forum Donor
Joined
Jan 29, 2019
Messages
740
Likes
683
"Or to be fair, I don't trust the wave and current from the wall."

Are you going to buy a generator or a solar power setup?
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
I use high end balanced XLR output but the AC/DC input feels more like being to a clock radio. Or to be fair, I don't trust the wave and current from the wall.
No need for worries here. With balanced I/O you're all set as this cancels out the (tiny) effect any shield current.
Remember that the "ground" path from the digital connection to the source is far more important when it comes to potential noise sources unless it is galvanically isolated. A 2-prong switch-mode supply of a typical notebook as playback source is dwarfing the noise current contribution of the RME stock supply by orders of magnitude which could likely be relevant with unbalanced cabling, but with XLR much less so.
As noted, the ADI-2 Pro itself is completely free of any power supply noise/hum effects to a level that truly is "instrumentation grade".
If you feel a whish to improve your system on this front, I'd suggest get the digital input side isolated. This will do much more good than a "better" power supply of the DAC.
 

Perik

New Member
Joined
Jan 1, 2021
Messages
2
Likes
0
No need for worries here. With balanced I/O you're all set as this cancels out the (tiny) effect any shield current.
Remember that the "ground" path from the digital connection to the source is far more important when it comes to potential noise sources unless it is galvanically isolated. A 2-prong switch-mode supply of a typical notebook as playback source is dwarfing the noise current contribution of the RME stock supply by orders of magnitude which could likely be relevant with unbalanced cabling, but with XLR much less so.
As noted, the ADI-2 Pro itself is completely free of any power supply noise/hum effects to a level that truly is "instrumentation grade".
If you feel a whish to improve your system on this front, I'd suggest get the digital input side isolated. This will do much more good than a "better" power supply of the DAC.

Thanks @KSTR for answering detailed and with short notice. Appreciated.
I'm running the USB on a stationary computer where the PSU has active PFC. And the USB-cable is flat and shielded like a SATA cable.
There is an awful amount of disturbance on the power lines within the house and beyond. Dimmers, devices sending signals back - people are even running network ethernet over the power net. Not optimal to build soundwaves from.
My hobby project is to create the equivalent of an online double conversion UPS with medical pure sine wave
(like this one for example POWERCOM (upspowercom.com) )
for my amp, 8k TV, XONE:96, CDJs and so on.

Would you recommend me to use cables like AKKUK on the bottom here RME: Accessories - Cables & Adapters (rme-audio.de) and take out 12V 2A DC directly to the ADI-2 Pro (protected from peaks, surge by 1 Farad backup or more if needed) or do you recommend me to connect the regular AC plug after I have transformed into a sine again? The last alternative seems like an extra step. Future Intel motherboards will handle 3.3V and 5V by themselves and will only need of 12V, so I will probably build a DC molex anyway to skip one fan in the computer.


"Or to be fair, I don't trust the wave and current from the wall."

Are you going to buy a generator or a solar power setup?

I will transform with something like 20 MHz or more.
There is a reason why people are hired to do only these things in high precision industries or healthcare for example. High Fidelity <=> No loss
Don't judge other peoples interests.
Edit: I might have misunderstood, sorry if that's the case.
 
Last edited:

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,641
Likes
2,809
As noted, the ADI-2 Pro itself is completely free of any power supply noise/hum effects to a level that truly is "instrumentation grade".

Hi, regarding the part quoted above, on previous page you wrote:

"The stock wallwart SMPS packed with the Adi-2 and Adi-2 Pro (and other RME units) is somewhat substandard, and RME knows this."

So just to clarify, are you saying ADI-2 Pro is or is not instrument grade, in terms of it's PSU noise ?

Also, what are you thoughts on this SMPS leakage current when ONLY using the ADI-2 headphone output? Assume USB source is a typical noisy laptop/PC which also use SMPS
 
Last edited:

Killingbeans

Major Contributor
Joined
Oct 23, 2018
Messages
4,089
Likes
7,552
Location
Bjerringbro, Denmark.
There is an awful amount of disturbance on the power lines within the house and beyond. Dimmers, devices sending signals back - people are even running network ethernet over the power net. Not optimal to build soundwaves from.

Luckily non of these devices "build soundwaves" from the mains voltage. Once these disturbances has been through the grinder of the SMPS its output filter makes short work of them. And if anything manages to piggyback ride on the output ripple, the post regulation and PSRR of the components in the ADI-2 Pro strangle it completely.

Like KSTR points out, there's no need to worry about the lousiness of the mains voltage. Keeping ground loops at bay is the only thing worth spending time on if you want to avoid something with a real risk of being audible :)

So just to clarify, are you saying ADI-2 Pro is or is not instrument grade, in terms of it's PSU noise ?

Not to put words in his mouth, but I think he's saying that the external power brick is crap while the internal regulation of the DAC/ADC itself is the instrument grade part. Meaning that even the most crappy PSU won't have an audible impact on the ADI-2 Pro.

(Assuming you don't stumble across a horror scenario of leakage currents and/or ground loops)
 
Last edited:

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
Thanks @KSTR for answering detailed and with short notice. Appreciated.
I'm running the USB on a stationary computer where the PSU has active PFC. And the USB-cable is flat and shielded like a SATA cable.
There is an awful amount of disturbance on the power lines within the house and beyond. Dimmers, devices sending signals back - people are even running network ethernet over the power net. Not optimal to build soundwaves from.
My hobby project is to create the equivalent of an online double conversion UPS with medical pure sine wave
(like this one for example POWERCOM (upspowercom.com) )
for my amp, 8k TV, XONE:96, CDJs and so on.

Would you recommend me to use cables like AKKUK on the bottom here RME: Accessories - Cables & Adapters (rme-audio.de) and take out 12V 2A DC directly to the ADI-2 Pro (protected from peaks, surge by 1 Farad backup or more if needed) or do you recommend me to connect the regular AC plug after I have transformed into a sine again? The last alternative seems like an extra step. Future Intel motherboards will handle 3.3V and 5V by themselves and will only need of 12V, so I will probably build a DC molex anyway to skip one fan in the computer.




I will transform with something like 20 MHz or more.
There is a reason why people are hired to do only these things in high precision industries or healthcare for example. High Fidelity <=> No loss
Don't judge other peoples interests.
Edit: I might have misunderstood, sorry if that's the case.
IMHO, you are attacking the problem (if there actually is any, that is) from the wrong end. You basically have a cluster of devices all interconnected in some way, some digital, some analog, some with SMPS, some with linear supplies, random connection to the outside world like ethernet and cable-tf/radio, antenna feed wall socket, etc. All powered from the same power strip, in the end. Replacing the input to the power strip with regenerated/rebuild mains from an UPS will do close to nothing to improve all the LF and RF balancing currents and RF pollution between gear. OK, a better sine, and hopefully well-filtered for RF both differential and common mode but that can be had with any decent mains filter as well.
Mains waveform is completely irrelevant to any SMPS-supplied piece of gear, and close to completely irrelevant for linear power supplies as well.

Only the analog section of you system is of relevance here, and the whole point is to keep the voltage potentials of the internal "ground" references as close together as possible. The connection of those common grounds (literally) between pieces of gear is the audio signal cabling and that's why it matters, especially with dreaded unbalanced connects where any shield current induced voltage drop becomes part of the signal. Therefore, the analog section should be treated as an island and all bridges to the world need their own dedicated inspection.

Top rule #1: make the shield/ground wiring of your audio interconnects as low resistance as possible, that is, use thick shields and shortest possible cables, and if possible, make additional connections between all available unused ground pin / shell contacts of analog inputs and outputs. Getting the resistance/impedance of that "ground grid" down is paramount.

Top rule #2: make all other connections to the world as high impedance (common mode) as you can. The idea is that the above reference ground grid develops lower error currents (and hence error voltages) from the same voltage differential when the total resistance it works upon is high (Ohm's law). This is especially relevant for earth-grounded pieces of equipment. We basically have a ground voltage divider here, the higher the ratio the better.

For the digital connection to the DAC we have USB isolators or TOSLINK which are close to ideal. The ideal for power supplies is floating battery, obviously not a real option in most cases. So what can we do with those linear and switch-mode supplies, some 3-prong (earth grounded), some 2-prong? Getting impedance up at RF is easy, just use tons of clamp-on ferrites on every(!) cable. To get at least some LF resistance increase, use long and thin gauge cables (as long as they are rated for the power rating!)
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
Also, what are you thoughts on this SMPS leakage current when ONLY using the ADI-2 headphone output? Assume USB source is a typical noisy laptop/PC which also use SMPS
Leakage current will be completely irrelevant as there is no analog audio connection to another piece of gear where leakage would induce shield currents and the associated error (RCA:high, XLR:neglegible).
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
So just to clarify, are you saying ADI-2 Pro is or is not instrument grade, in terms of it's PSU noise ?
The ADI-2 Pro doesn't have power supply noise. It's power supply does (like most any mains supply) but this is mostly irrelevant except for some specialist measurement setups and some edge cases. For example, when you want to test the reponse of phono amp and feed the amp with the DAC directly you don't want to inject any leakage current in the phono amp's input cable, given the high gain (~60dB at low freqs). Then again, this is incompetent technique and thus not a valid point in the end, one would correctly use a significant voltage divider at the output end of the signal cable to effectively kill the apparent error voltage...
 

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,641
Likes
2,809
The ADI-2 Pro doesn't have power supply noise. It's power supply does (like most any mains supply) but this is mostly irrelevant except for some specialist measurement setups and some edge cases. For example, when you want to test the reponse of phono amp and feed the amp with the DAC directly you don't want to inject any leakage current in the phono amp's input cable, given the high gain (~60dB at low freqs). Then again, this is incompetent technique and thus not a valid point in the end, one would correctly use a significant voltage divider at the output end of the signal cable to effectively kill the apparent error voltage...

Thanks!

Out of interest can you recommend a better SMPS to use than the stock RME supplies?

Something with world wide safety approvals, like MeanWell or others?
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
Comparable AC/DC SMPS with much lower leakage current and capacitance specs are close to nonexistent. Even the ones with approval for medical use still have 30% or so of the leakage of the RME stock brick.

Therefore, I went for an isolation DC/DC converter on the secondary side of the stock supply which offer my required specifications. Actual model I'm using is a Traco THM30 (~$70, so for a complete isolation adapter, with some filtering, metal case etc it's still below $100).
Adding the isolation to the floating secondary side of the stock supply has the nice option that this secondary now can be tied to any potential, so you can tie it to mains earth (or whatever reference potential needed) to reduce the remaining mains coupling even further because now it couples to a rather quiet potential.

EDIT: Just as a reminder, a better supply for the RME (in terms of leakage current) buys you absolutely nothing if you don't isolate the digital connection as well! For USB there is the Intona isolators (among others) but the simplest way (albeit not for the highest sample rates) is to use SPDIF optical input (TOSLINK).
 
Last edited:
Top Bottom