AES-3 via AES-3/RCA-RME-adapter to RME ADI-2 DAC FS causes sound dropouts

[weme]

In our setup there are sound dropouts when power consumers (LED lamps, neon lamps, ...) are switched on the same 230 Volt power circuit.
The problem is caused by the simple "AES-3/RCA-RME-adapter" - which has three problems:
- does not help with high frequency interference problems
- mismatch 110 to 75 ohms
- mismatch of voltage
RME should also give a better alternative in their manuals.

Reduced test setup in another 230 V power circuit:
AES source -> AES-3/RCA-adapter -> RME ADI-2-old
AES source -> AES-3/RCA-adapter -> RME ADI-2 DAC-FS

AES-3 source:
Behringer DEQ2496 or RME HDSP9632: both AES-3 outputs are galvanically isolated (transformer)

AES-3/RCA-adapter:
"AES-3/RCA-RME-adapter" with DMX CABLE 110 OHMS (approx. 1.5 m)
Schematic: e.g.: RME, User's Guide ADI-2 DAC v2.4, page 36.
https://www.rme-audio.de/downloads/adi2dacr_e.pdf


Sources of interference at 230 V:
Small transformer station with a switch that can be operated very quickly.

Results:
RME ADI-2-old (Hardware 1.4):
Audio dropouts are very short but audible

RME ADI-2 DAC FS (Firmware: 27/34 | 34/34):
Audio dropouts are much longer than with RME ADI-2 old. The very fast display "sync/lock" confirms the dropouts.

Workaround (everything has been tested):
Adapter "AES-3 Output to BMC-2 Input" with DMX CABLE 110 OHMS (approx. 1.5 m)
tc_electronic_bmc-2_manual_english.pdf, page 17
https://www.tcelectronic.com/product.html?modelCode=P0CIZ -> "Documentation"


A simple SPDIF toroidal transformer (1:1 with separate, insulated windings) before the ADI-2-DAC-FS | ADI-2-old

If you want to solder the "AES-3 Output to BMC-2 Input" adapter:



If you want to buy a professional adapter and you are living in the EU:
AES-3/RCA-adapter - with galvanically isolated (transformer)
http://www.funk-tonstudiotechnik.de/AES-EBU-Uebertragerkabel-192.pdf


SPDIF RCA connections and TOSLINK cable (of course) have no interference.

 

[NielsMayer]

I too had similar problems using a prewired and too-long Canare Rca->XLR AES/EBU cable i had laying around.

Switched back to a Sescom transformer ( https://www.amazon.com/gp/product/B00JZT65CE ) and a 1.5 ft male-male AES/EBU cable and it seems to work w/o dropouts.

And I thought it was working ok, till i tried higher rates -- 96k or 192k -- and saw it failing the RME bit test.

Testing using MPV and Linux, it recognizes only 44.1K rate files as "bit perfect" , whereas on the USB connection they are successful at all rates and depths. ( Unfortunately this is not the case for Strawberry music player: https://github.com/strawberrymusicplayer/strawberry/issues/730 )

On further investigation of the issue, I note that my current BNC->RCA solution for the sescom is suboptimal (it'll dropout if you jiggle it wrong due to bad contacts on the BNC->RCA adapter) so i'm looking for a decent direct BNC->RCA cable (or just solder my own) . It's very possible the potchky of four layers of connectors going from the Douk U2Pro ( https://www.amazon.com/Digital-Interface-Coaxial-Optical-Audio/dp/B08HN3VSF8/ ) to the ADI-2 has something to do with that problem....

More likely , however, the "6 Mhz bandwith" for the Sescom transformer is to blame... for example a calculation I stole off the internets suggests a bandwidth issue: "192 kHz * 24 bit * 2 channel = 9.216 MHz data stream. So would it require 2*9.216 Mhz = 18.4 MHz bandwidth?"

Essentially this transformer's bandwith would really only do 48k/32 bit files given this calculation:
"Signal bitrate is 3.1Mhz for sample rate 48kHz (I guess because 48 kHz * 16 bit * 2 channel = 1.536 MHz data stream)."

The original SPDIF spec was for 6Mhz BW -- https://www.epanorama.net/documents/audio/spdif.html -- from back in the 48k days.

So maybe i need to get one of the more expensive Neutriks?
https://www.amazon.com/Neutrik-NADITBNC-MX-Ohm-BNC-Adapter/dp/B005T7PSJW
https://www.amazon.com/Neutrik-NADITBNC-M-BNC-XLR-Converter/dp/B07GSHFKB9

Anybody know the bandwidth specs on those? I'm not coming up with anything.

Note the RME spec sheet indicates the AES and SPDIF inputs should accept 192k bitrates ( https://www.rme-usa.com/adi-2-pro-be.html )

• Lock Range: 28 kHz – 200 kHz

 

[NielsMayer]

Nope... the Neutriks also have a 6Mhz bandwidth too -- "SPDIF spec compliant" but useless for anything above 48k.

https://www.neutrik.com/en/product/naditbnc-mx

"Frequency range 0.1 - 6 MHz"

https://www.pdf-manuals.com/neutrik...pedance-transformer-naditbnc-mx-194583-manual

https://www.pdf-manuals.com/neutrik...pedance-transformer-naditbnc-fx-194579-manual

Given the bandwidth issue, I think i will want to revisit the method of RCA->XLR connection on the simple cables. It is possible they're leaving one half of the input transformer to the ADI-2 AES/EBU input "open" so it is picking up noise. Instead, perhaps ground one half and hope the signal is large enough on the other side. Alternately put the entire SPDIF signal across both arms of the ADI-2 input and leave the center tap floating (if possible). That would of course also result in a half-sized signal on the other side of the transformer, just like grounding the top and center taps in the former scenario and just driving half the balanced transformer. Note that per spec, the input is "transfomer balanced" suggesting the aformentioned circuit architecture of a center-tapped transformer on the input:

• 1 x XLR, transformer-balanced, galvanically isolated, according to AES3-1992

If above workarounds fail, some kind of active solution will be needed.

Of course I could just give up and use the problematic AES input as reason to return the ADI-2 (with much tears as it sounds phenomenal and i love how well supported it is in Linux). Alternately I can just give up on listening to higher rate audio through this particular input and just use the USB for higher rate audio.

PS: see also https://www.ranecommercial.com/legacy/note149.html "Interfacing AES3 and SPDIF"

 

[NielsMayer]

So i made my own transformerless cable which basically puts the coaxial signal to pin 2 (+) and the coaxial shield to pin 3 (-) while floating the ground (pin 1). I then got the same results with the shield and - connectected together and just Pin 2 driven.

In both cases it works and doesn't pick up noise.

However I still can't get any rates above 44.1K through Douk Audio U2PRO USB->SPDIF device or Topping D10 into this AES/EBU input on the RME.

At 44.1, through both strawberry music player and gstreamer directlly via

$ gst-launch-1.0 playbin uri=file:///home/npm/Music/RME-BitTest/441_24_adi2pro_bittest\ 4s.wav

I was able to get ADI-2 to pass the 16 bit and 24 bit tests.

The 32 bit test still indicated "bit test 24 bit passed" which at some level makes sense since the SPDIF spec only allows for 24 bits and apparently the remaining bits were just truncated so the DAC thought it was seeing the 24 bit test.
That of course means the 32 bit "bitrate" calculations in previous post, for the SPDIF->AES/EBU transformer, do not apply, just the 16 and 24 bit rates. However, since the transformer doesn't seem to change anything other than limiting the bandwidth, that whole issue is moot.

Just to make sure the USB->SPDIF wasn't the issue, I tried also with a Topping D10 and saw the same results. The Topping display indicated the bitrates not seen by the RME -- 96, 192. Whereas 44.1 worked as it did with the Douk Device. The Topping did seem to trigger the SPc mode on the ADI-2 at 192k whereas with the Douk, it didn't. Perhaps the D10 frequencies not as accurate?

As a final test, I connected the Douk SPDIF to the RME spdif in -- 192 & 96 were recognized in this configuration. Same as for the D10 which had always been working at 192k from NVIDIA Shield into the SPDIF input of the RME. Now I know why it displays "SPc" -- because of the D10!

Has anybody successfully used the AES/EBU input on the ADI-2 FS R BE at rates above 48k? (e.g. 88.2, 96k and192k)?