S/PDIF 1x Cinch-Out to 2x BNC-Out extension/modification

Jump to Latest
W

wgh52

Member
Joined 2011
#1
Hi Folks,

I'm not an expert on SPDIF nor transmission lines, so bear with me please!

Background:
My good old well working Audio DSP has a digital output port with CS8420 feeding a AES3 (with TXP & TXN) and a Cinch (75Ohm) Output (with TXP) in parallel at Pin26.
The picture "original" shows the output schematic as far as I could analyse the (4-layer) board (pretty obviously with supply and GND layers). GND on the board is split between the output connector area and the circuit area (for unknown reasons) bridged with 0Ohm parts.

My Plan and request for confirmation or advice what to do:
The DSP is planned to drive 5m/15ft long koax stereo SPDIF BNC connections to each of my stereo speakers sporting digital 4-way "mono" x-overs. I.e. whether the speaker is "left" or "right" is selectable in the speaker. So, basically I need two parallel BNC Outs (would remove the cinch connector and disconnecting AES3 circuitry).

Please kindly have a look at the schematics and options I have come up with. I am sure my stuff needs changes to work properly, so I would be very grateful for recommendations "how to do this right".

SPDIF Out extension.PNG

Thanks a lot for reading all this and I hope you care to answer with any helpful recommendations.
Regards,
Winfried
 
#2
Hi again,

is there really nobody who would/could help to resolve this issue. This issue may even be an obvious non-issue, but I just don't know for sure and am eager to learn!

Thanks for any feedback!
Regards,
Winfried
 
#3
AES3 signals are at a higher voltage than SPDIF. See page 8 of CS8420 datasheet for Rtxp. Then the circuit for driving one 75R SPDIF coax is shown on Page 78, figure 31.

If you need to drive two SPDIF level coaxes at once, it depends if you will aways be driving two loads, or sometimes you may be only driving one load. It also depends how much output current TXP is able to supply, since two 75R loads in parallel would amount to about a 37R load that TXP would see (and I don't see a limit on output drive current specified in CS84200 datasheet).
Might be possible to change SC8420 output resistors to match to two cables, if it could handle the drive current. Maybe that could be helped by reducing the SPDIF voltage swing some. Might be workable that way.

I suppose you could also make an active splitter that uses a multi-channel clock buffer chip which is designed to drive 3.3v or 5v into a 50 R load. You could use resistive attenuator impedance matching at the outputs to match to 75R and the required voltage swing. Seems like a lot of trouble though.

Seems odd your speakers require one SPDIF cable each, since SPDIF usually carries both stereo channels. I would kind of expect you might run a SPDIF cable to one speaker, then from that speaker connect an analog audio cable to the other speaker. Something like that. Do you have a link to the manual for the speakers?
 
Last edited:
#4
Hello Mark,

Thanks sooo much for taking the time and effort to respond so detailed! I really appreciate it!

Let me give some more project insight to help focus the discussion:
  • The speakers are DIY fully active 4-way designs with DSP x-overs and DAC boards in each speaker.
  • Each speaker DSP has an SPDIF coax input (right or left can therefore be selected and I actually like it this way)
  • Due to the speakers standing 0.7-1m (or ~2-3ft.) from walls (and not least for positive WAF) I want to minimise cabling (i.e. also avoid daisy chaining the digital signal which would be an alternative, I guess)
  • The system driving the SPDIF output to the speakers is a DEQX HDP-3 with Dig. Output board. This board provides AES3 Output driven by CS8420 TXP/TXN through a transforemer AND an SPDIF Cinch Out in parallel from the CS8420 TXP (my reverse engineered schematic only covers the SPDIF Cinch circuit part, because I dont use AES3 anyway and have removed the transformers already, could remove the other AES Out related components as well if needed...)

So my guess and idea was (and I may be mistaken!):
  • If the CS8420 is capable of driving both AES and SPDIF coax in parallel anyway, it should also be capable of driving two 75Ohm SPDIF coax lines with BNC connections instead.
  • Yes, the two BNC SPDIF Outputs may not always both be used together, so a decoupling or separate line drivers may be needed (and I'd accept that).
  • The question is: How do I do all this right?

I hope I am able to clarify the concepts, hopefully helping to identify a viable solution.

Thanks and regards,
Winfried
 
#5
I would try two separate voltage dividers driven from a common TXP output, if there is space to fit the components. Preferably AC coupled to TXP to keep the current below the 21 mA specified on page 8 of this datasheet:

https://www.mouser.com/datasheet/2/76/CS8420_F4-1829.pdf

In pseudo-PSpice netlist syntax something like

C1 TXP node1 100 nF
R1a node1 outa 330 ohm
R1b node1 outb 330 ohm
R2a outa ground 91 ohm
R2b outb ground 91 ohm

The BNC connectors can either be connected straight to outa or outb and ground, or via 100 nF capacitors, or via S/PDIF transformers.

Edit: you can invert an S/PDIF signal without changing anything, so you could just build a second instantiation of the present circuit and connect it to TXN. TXP then drives one output and TXN the other. The series resistor then has to be about 350 ohm, I don't understand why it is only 56 ohm in post #1.
 
Last edited:
  • Like
Reactions: olsond3
#6
I do not have experience with SPDIF. But i have with similar signals. If you want to split A SPDIF signal, you are best served to use an actively distribution amp. SPDIF is pretty forgiving, you should be able to home brew a distribution amp with common op amps.

You can try your design using a copper splitter. If it does not work, give building an active splitter slash distribution amp a try. A cheap and quick trip down the experimental highway. SPDIF is a 75 Ohm standard, use 75 ohm BNC and 75 Ohm BNC cable. If you need to use a BNC to RCA adapter, put it at the RCA end.
 
#7
Not a DIY solution but this works very well for the purpose. It drives long cables and then you can convert the AES to BNC with a transformer at the end if needed. The key is to match the impedance or you get reflections and degradation of the signal. Some of the digital input speakers have the ability to daisy chain them so you do not need a splitter. The Hypex Fusion amps are an example.

https://www.bhphotovideo.com/a/new/1108543/henry_engineering_dd_14_digital_da_1x4.html

1744068397101.png
 
#9
This is a job for digital line drivers. Digital line drivers are electronic circuits that amplify and transmit digital signals over longer distances or through noisy environments, ensuring signal integrity and reliability. They act as an interface between logic circuits and transmission lines, converting signals into a format suitable for transmission
 
#10
Just build a second instantiation of the present circuit and connect it to TXN. TXP then drives one output and TXN the other, but an inversion has no effect on an S/PDIF signal. The series resistor has to be about 350 ohm, I don't understand why it is only 56 ohm in post #1.
 
#20
Hi folks,

a BIG thank you to all contributors! I'm overwhelmed by all the responses and discussion!

So let me try to boil all this down a bit as a few alternatives appear to exist for my case:

1. Duplicate the TXP Output circuitry (with Rtxp ~350Ohms) to drive the second BNC Output in parallel to the existing one
This seems viable to me because of the AES3 having been in parallel before and woule be disconnected, so TXP fan out would seem sufficient.

2. Duplicate the TXP Output circuitry (with Rtxn ~350Ohms) to drive the second BNC Output from TXN
This looks smart, but is neither recommended nor prohibited in the CS8420 Datascheet.

3. Implement dedicated Line drivers for each BNC Output
This seems to be the optimal technical solution, although, the existing system did not have that and uses the TXP/TXN outputs with their 25Ohm source impedance - so this option seems "not requiered" to me.

4. The external Splitterbox connecting to an AES3 output is not favourable to me in light of simpler system internal solutions providing the desired end to end 75Ohms BNC connections to begin with. But conceeded, this would be less DIY work...

So, I'll basically decide between option 1 and 2, probably try both.

This begs the question, though, which circuit to build in detail:

The reverse engineered schematic seems complex, but that may be for a (so far unknown) reason. Can anyone make sense of the original circuit?

Thanks again,
Regards,
Winfried
 
You must log in or register to reply here.
Top Bottom