advertisement


SPDIF out from Naim CD3

Chops54

pfm Member
I'm sure this has been discussed before but has anyone actually done it? Spdif would come from pin 14 of the SSA7220 chip through a divider and I was going to use a 1:1 transformer as a buffer. Pin 14 needs to be floating but when I looked at the 7220 chip in my CD3 I find pin 14 goes to a 10K resistor which goes back to one of the LM317 regs. Can I simply lift that 10K resistor to float pin 14?
 
Yes, I expect that resistor is to tie the internal logic high for less noise when not used.

The DOBM pin on the 7220 doesn't need it though, and a basic get-you-going SPDIF output circuit is shown in the SA7220 circuit - two resistors as a divider. NB put this divider close to the 7220 and return the 0v end of this divider to the ground plane the 7220's 0v pin is tied to, then run coax out from here - with coax shield tied to the ov plane in the same place. Basically the coax output will connect across the 'lower' resistor in your voltage divider. Same connection if you use a signal transformer for isolation, but add a 0.1uF cap to stop the transformer being fed (saturated by) any DC from the DOBM pin output.
 
Thanks Martin. I have a Murata DA101C transformer which is quite small so will fit underneath the pcb ok. All the cdp circuit diagrams I looked at that had the 7220 chip and spdif had a 560R resistor coming from pin 14 to a 620R resistor which went to ground. The primary of the transformer is across the 620R resistor. I've not seen a cap in any of the diagrams but I'm happy to include it if it's beneficial. Would the cap come off pin 14 before the 560R resistor or be in the hot leg of the primary winding?
Whilst I was searching I found a reference to putting a 75R resistor across the secondary winding of the transformer to stop any chance of the signal reflecting in the 75 ohm line. Any thoughts?
 
The cap would be in the leg of the primary winding - something physically small would be good start (- ceramic or film, 10nF to 100nF not critical)

If you add 75R across the start of the line, then you are driving a 37ohm load and it's likely to be worse rather than better. The classic approach for video and other HF signals is to drive the coax centre through a series 'build-out' resistor (75ohm), which with the 75ohm load across the far end will kill reflections but also attenuate the signal by 50% too. This should be fine- the DOBM output on the 7220 is about 5v pk-pk i.e. supply voltage down to 0v - and SPDIF is only specced for 1v IIRC (so your initial divider, halved again is about right)

Like anything there are torrents of ideas out there on how 'best' to do SPDIF (mostly coming to the POV of 'it's a horrible standard'), but this will work.
 
Mission accomplished :)



It's playing through my dac as I write and it sounds great, a noticeable step up from audio out into the preamp.

Thanks for your help Martin.
 
Thanks trancera, I'll have a read of that tonight. The thing is I'm more electrical than electronic and don't have as deep an understanding of these things so I try to keep things simple. Before I even start wondering what effect a jfet will have on the sound as opposed to a transformer, I'm wondering what effect a different decoupling cap before the transformer would have?
 
I don't know much about it either (Martin ?).

However, from what I understand spdif aint great and then implementations of were perhaps based on not the best way of doing it (perhaps cheap to manufacture as per usual). Some transports perhaps doing a a better job being dedicated as such.

I'm sure you could build it without a board at all as long as you have the 12V ish

digifetishizator.jpg
 
I've read through the link and it looks simple enough. It would be easy to build it on perf board and fix it to the output socket. I must admit I'm very happy with the way it sounds right now.
 
That's just a buffer to drive the output instead of the chip direct. I'm not a fan of the schematic as drawn and think that the route you've chosen is probably more robust (and offers galvanic isolation, a good thing here). As you say, easy and cheap to play with (even maplin stocks BF245C jfets, its a very useful part)

If you wanted to add a buffer, you could use that circuit ahead of your transformer - but I suspect without a fast scope, willingness to drive yourself mad and knowing what you are looking for, it'll only change things - not necessarily improve at all. For example - since you are buffering a squarewave with a datarate of 2.8Mhz (and given is a square wave you want, that means lots of harmonics out beyond 10x even 20x that) the potential to make a bit of a mess at RF is definitely there. The phono plug itself becomes awkward at this bandwidth- and it is actually part of the spdif standard..!

tl;dr: Easy to get a different perceptual result, but non-trivial as casual DIY to get 'right' I suspect.

ETA:
If you do try the sketch - for heavens sake add a local decoupling cap e.g. physically-small 47-100nF ceramic or film cap from the drain (top terminal) of the jfet to the 0v connection (bottom of the 75ohm reistor, which should be the point yout signal output 0v returns to - and I'dd add a small RF choke on the supply side between the drain and th 9v -15v + ve supply (again, even maplin stocks these, 100uh or so should be fine, value uncritical)
 
Hi Martin, like I said I'm happy with the SQ from this simple spdif conversion. After reading your last post I'm inclined to leave it there for now. I know someone with a 'scope so maybe that should be my next step. Tbh I don't want this to turn in to a never ending upgrade thing. I just want to enjoy using my old CD3 with my new dac.
What did you think of his spdif tube buffer?
 
Yeah, not so sure of digi through tubes. But who knows if you don't try I suppose... the lampman is very enthusiastic that's for sure :)

Would a BNC be better than a phono, something about that in my grey matter...
 
I thought about using a bnc socket but my Musical Fidelity dac uses an rca phono socket as does my other cdp. I also have plenty of decent rca plugs and sockets in my spares box. There's a school of thought that says that with a bandwidth of around 12mhz and a wavelength around 3 metres that an spdif signal can't be affected by typical RCA connectors and cabling.
 


advertisement


Back
Top