AES/EBU to SPDIF coax input, how?

[gmarsh]

If you have gone to all the trouble of building in a SRC, then the difference may be small. But then, if you have gone to all that trouble, I can't imagine the extra dime breaking the budget. Even in Canadian money.......!

<snip>

Holy ****, I was listening to the exact King Crimson track when I read your post and glanced at your location... 😀

Building in a SRC isn't trouble, it saves us trouble since our product
has to accept a wide sample rate range. The alternative is to have our DSP chip measure the incoming sample rate somehow and recalculate all its filters... then find a way to dynamically change everything (FPGA's, etc) downstream from the DSP dynamically, too. I'll stick with the SRC thankyouverymuch.

Here's what I'm envisioning on this board:

TORX optical receiver -> CS8416
AES/EBU -> couple series C's (DC block) -> couple series R's -> transformer -> center tapped R termination tied to a bias voltage -> TL3016 -> CS8416.

I could probably tweak the series C's to cancel transformer leakage. I lack TDR tools, but an impedance vs frequency sweep of the 110 ohm input is doable.

I won't be using the CS8416's Rx, because there's no elegant way to use both a differential input and a single input on the input mux. This also means I don't have to do the stupid "reset after an unplug/replug" chip that Crystal can never seem to fix in their parts...

BTW.......if you do hook one up to a TDR......make sure that you show him all the **** the Crystal RX chip dumps back unto the line. He will really hate you then.

The chip has hysteresis, it's required for stability. Invariably this means dumping a bit of hysteresis current back onto the line...

<pulls Court out of the CD player, puts in Red>

 

[Jocko Homo]

Let's stick to the original problem, before we go off on a tangent. Although, it is encouraging that someone knows how a Schmitt trigger works. Not sure that it is necessary..........

OK...........you have a great big hole in your DAC to stick a XLR connector into. I don't like them, but you have an AES/EBU source. So........if you have something like a Crystal '841x RX chip, why do you need to convert to 75R? Those things only work well when you drive them differentially. All you would need to do is change the termnination. (And add a few picogates in front of that RX chip..........)

Unless you haven't told us the whole story about your DAC box............

Jocko

 

[tubenut]

I like Apogee names in general, wyde eye (their spelling) for the cable, big ben for their clock sync unit etc.
I already run SPDIF with BNC everywhere...... Noticable difference. Big diff was noticed when going from wyde eye with (temporarily but good quality) RCA plugs fitted using adaptors to my BNC DAC and Drive to crimped BNC on the cable.
The codec difference between SPDIF and AES3 has more to do with copy protection bits and track information I believe. Recorders may complain but DACs should interface OK. In any event I suspect the Theta does not output true AES3 but rather SPDIF codec on AES3 level and impedance...... (I see no extra cct appart from drivers for the transformer and all the CD logic is std Philips fare).

The DAC uses an 8412 as a receiver, but I think before the signal gets there it goes through a fancy PLL (Theta sold this cct seaprately later I think and called it the TLC). I would not want to loose that part of the DAC as I beleive it makes a substantial difference. Why do I think so? Because CDR recorded via the DAC and its digital out sound far better then direct from drive to CDR.....
Also input selection is done electronically from 3 inputs and I use one of the others for the DVD player (about 3 hours a month, that laser is gonna last forever). Picogates???? You mean inverters and a few caps???
Perhaps I should have a good look at the Theta CCT and see if they are not doing anything diffrential on the receiving end already.... Doubt it though.


Thanks
PS: My HiFi technician friend calls his company Repaircraft. He is a big Fripp fan....

 

[Jocko Homo]

Uh.........I think that the fancy PLL is after the RX chip.

Haven't worked on a Theta in a decade or so, but I recall that they did use a transformer on the one that I saw. And a good one, at that. Easiest way to get a differential signal, so there is a good chance yours has one.

Picogates are tiny little chips that have only one gate. Less stuff going through ground leads to induce jitter.

Jocko

 

[tubenut]

Only one txf on my Theta

and it is for the digital tape output........
I will try and see where the PLL slots into the whole thing over the weekend....

Complete tangent..... & I may be stupid........

Assuming that the input (receiver, buffer, whatever) has a 75 Ohm resistor across it (I have seen these 75 R things on the input...). And assuming an average guess of AES3 output of about 4 Volt P-P vs 0.5 V p-p for SPDIF...

What if I were to change the 75 Ohm resistor to 15 Ohm and connect the AES3 signal across this resistor via 47 Ohm resistors on either side of it.

_______________/\/\/\/\/\/\_______________
Added 47 Ohm | |
| Input buffer/receiver
AES3 >
<
15 Ohm < |
Now 75 < |
| |
______________/\/\/\/\/\/\____|____________|
Added 47 Ohm

Obviously once I scope the AES3 out from the drive (with a 110 ohm load across it) to see it's level I can taylor these resistors to make a devider that is at the right ratio for the application......


Any thoughts on the above anyone?

Thanks

 

[tubenut]

oops, diagram did not work....

hope my desrciption is understandable regardless.....

 

[analog_sa]

Hi G

I don't think it is important to scale down the voltage in front of the 8412 chip - it copes very well (better?) with higher levels. As long as your impedances are matched all should be well. Sorry if i didn't get your question right.

 

[tubenut]

Not sure if the inputs drive the 8412 directly though.... Do not think so....

Will check it out.........

 

[analog_sa]

Of course they don't. I just realised you mentioned having some digital input switching and digital out before the 8412. So, there must be a line input receiver in front. Looking at the ones i use they seem to be fine with input voltages up to 7v.

 

[tubenut]

Will mess around on the weekend I hope

I think it is a bunch of 14 pin dip packages on the input, I would guess 74 somethings but I need to have a look and see...

Anyways, I will check them out and either replace the 75 Ohm R with 110 Ohms or try my devider idea. If the devider is what is required it could be neat, just replace 1 TH resistor, cut 2 tracks and mount smd 47 ohm below the board... If I can get away with no attenuation I only need to replace one resistor....

I may try the devider story externally just to see if it works with a 18 ohm paralel R and the 2 47 ohms on a crib on one of the spare RCA single ended inputs...

Thanks

 

[gmarsh]

Let's stick to the original problem, before we go off on a tangent. Although, it is encouraging that someone knows how a Schmitt trigger works. Not sure that it is necessary..........

OK...........you have a great big hole in your DAC to stick a XLR connector into. I don't like them, but you have an AES/EBU source. So........if you have something like a Crystal '841x RX chip, why do you need to convert to 75R? Those things only work well when you drive them differentially. All you would need to do is change the termnination. (And add a few picogates in front of that RX chip..........)

Unless you haven't told us the whole story about your DAC box............

Jocko

CS8416 performs input selection (AES/toslink) and biphase decoding. I'm only using one of them, with RXN biased to 1.65V and a couple RXP 's driven with CMOS.

And this isn't strictly a DAC i'm working on. There is a 16-bit DAC in it, but it runs at >400 megasamples/sec and puts out FM radio. 😀

 

[tubenut]

Played a bit over the weekend

I did not get to open the DAC.... Just measured its input with a mm. 75 ohm resistor as expected so I made a crib with some crocodile clip leads and a few resistors soldered onto a socket. The divider idea should work as the crude crib locked without a problem.
I will get the bits I need during this weeek and implement it all over the coming weekend. Then we will do some testing with those flappy things on the side of my head and I will let you guys know what the verdict is......
Does anyone know the recommended connections for AES3 cable? Pin 2 and 3 are signal and 1 is screen / earth. On my friends fancy cable, the screen is not linked to the plug shell but internally on the Theta drive it is.... We listened to this Apogee cable I made up linking the the plug shells to screen and not and linked sounded nicer.... Just odd that the Analysis Plus which is still superior sounding does not link the cable screen to the plug shell but only to pin 1......

Thanks

 

[tubenut]

Progress and Theta Data Basic Mods

I just thought I would let you folks all know how I have progressed on this.
I tried my attenuation scheme but I was not convinced that this sounded any good, never mind bettering SPDIF.
I then wired up the transformer in a mirror of the Theta Data output. 37 Ohm resistor on each leg. This made no sense to me but I thought Theta would know what they were doing.
In subsequent correspondence with Theta I have established that they set the AES output up for 75 Ohms…. I thus redid my AES input on the DAC and rectified the transport output by changing resistor 17 and 18 to 55 Ohms. (Actually 56)
While the transport was open I decided to get stuck in there as well and the modifications done to it are:
Digital cables used from the main PCB to the output connectors. This looks real low rent stuff and I rewired using Apogee Wyde Eye AES and Co Ax cables. I changed the SPDIF RCA to a BNC some time ago already. Very worthwhile to run BNC instead of RCA…
Next up were the diodes. I used BYT53 all round, one of my faves as it is fast, soft recovery and not bulky. You will need 16 dides to replace 3 bridges and 4 separate diodes. There are diodes/bridges on the main PSU pcb, the main PCB as well as on the extra powersupply PCB (that I think is the main change from MK1 to II)
I also changed 2 of the 47 mfd electrolytic caps near the IF demodulator for some os-cons.
I also used some chunkier cable to wire up the power switch as my high quality mains lead was sure to be choked by the bell wire used to wire the switch to the PCB
All in all these changes have taken the units a few steps ahead and I am really enjoying the improvements, especially in the bass area.
I am not sure if AES has less “jitter” but perhaps a large reason for the benefits are its higher signal level 4 V instead of 0.5 V and the elimination of the ground links between DAC and drive.

Next up will be a clock upgrade somehow…

 

[tubenut]

Another thing

Also found on my DAC and my friends that having SPDIF connected between Drive and DAC while using AES EBU destroys the AES EBU sound....... Something to do with earths I guess.
I could not hear a difference with AES connected while using co ax but when you disconnect co ax while listening to AES things just open up big time!!!!

 

[Lourens]

Time Line Conditioner

The DAC uses an 8412 as a receiver, but I think before the signal gets there it goes through a fancy PLL (Theta sold this cct seaprately later I think and called it the TLC). I would not want to loose that part of the DAC as I beleive it makes a substantial difference.

Can confirm this TLC is built around the 8412 (and 74HC04, 74ACT74, 74HC157). I used it once as a jitter reduction device between a Philips CDD882 transport and a Meridian 203 DAC. After HifiNews/RR reported disappointing results (May 1996), compared to other jitter bugging devices, a fellow by the name of Mike Bearne proposed an effective mod: split the input and output grounds of the Vitec transformer (pin 4 directly to output, isolating it from the pcb) to reduce ground-borne interference.

If your Theta suffers from the same design flaw, could this explain the remarkable observation in your post #34?

BTW, I removed the TLC after reclocking the CDD882, modification of PSU (shunts), upgrading caps and diodes etc. etc. Even after building the thing into the Philips with it’s own dedicated psu, and replacing the DAC by a much better one, it appeared to deteriorate more than improve!

 

[Lourens]

Re: Another thing

Also found on my DAC and my friends that having SPDIF connected between Drive and DAC while using AES EBU destroys the AES EBU sound....... Something to do with earths I guess.
I could not hear a difference with AES connected while using co ax but when you disconnect co ax while listening to AES things just open up big time!!!!

Cannot confirm this 😉
My current transport (Studer A727 CDS) has both standard AES3 and S/PDIF (RCA) digital outputs. I fitted an extra BNC, directly from the Philips SAA7220 decoder (pin 14, DOBM) with Belden 1505A according to Jocko’s recipe http://www.diyaudio.com/forums/showthread.php?s=&postid=301131&highlight=#post301131
Sound wise it does not matter if they are connected simultaneously or one at a time, even if i add the CDD882 (see above) with it’s own direct BNC S/PDIF (yes the dCS has a lot of inputs ). I don’t know what my dCS DAC exactly does electrically with these three different sources when not selected, but my volt meter shows they all share the same chassis ground when the connected with the digital interlinks (I did not open up my dCS for this, but you can hear relays clicking when i switch inputs).

Otherwise these three inputs do sound different mutually, but that's another story and still under investigation 🙂

 

[tubenut]

TLC cct

Hi Lourens

It is not impossible that there is an earth problem on the TLC circuit inside the Theta DAC. In original form there is no transformer any where near this. The only transformer is on the digital tape output.
The "created" AES input is a transformer coupled one.
Interesting about the digital ouput recipe from Jocko. I may try it out. Would you rate it better then the original txf coupled outputs of your drive? Would using a transformer instead of the cap yield any benefit?

The better sound from unplugging co ax while using AES was observed on my DS Pro Basic II as well as my friends Proceed DDP (I think it is called that, the beige/grey DAC). We both use the same transport except his is a MKI with aftermarket upgrade to MKII. It may be worthwhile checking out the ground scheme on the transport itself rather then the DAC. I think the the aes and co ax transformers share ground on their input side but I doubt this can be avoided. The aes transformer outputs are not linked to earth obviously but chassis earth is linked to pin 1. I do not connect this earth on my "created " AES input, IE I ignore pin 1. The co ax output links the earth leg of the transformer to (same as input?) earth via a capacitor.

I remember the TLC modification talk but I thought it was the fact that they used PCB mount RCA sockets with a common earth connection to ground for both. I suppose it boils down to the same thing. Do you have any additional data or the original article for this TLC mod. A friend of mine owns one....

G

 

[Jocko Homo]

About my recipe........

That one works best if you use a zobel. You would put 75R in series with the right value cap (around 18-47 pF, depending on the transformer) across the output pins.

There are other versions..............

Jocko

 

[Lourens]

Re: TLC cct

Interesting about the digital ouput recipe from Jocko. I may try it out. Would you rate it better then the original txf coupled outputs of your drive?

Hi Guillaume

First impressions: not very much better but there are positive differences (low level detail, better depth and layering, more realistic horizontal en vertical proportions), but I am not sure yet to what extend, and at what price, still trying out. One thing is very clear though: when the Studer is slaved to the dCS (word clock that is) the differences are minor, although after changing the 330R series resistor from Vishay metal film to Welwyn carbon I noted a little less stress.
It did not test this in the traditional set up (transport as WCLK master) with the Studer, but I did so with the Philips CDD882. In that case the differences with the original output are much bigger, partly due to the inferior quality of its (now by passed) output transformer, I guess. The Studer is fitted with two high grade transformers for AES3 and S/PDIF (RCA) out.
More important in this respect is certainly the lacking word clock synchronisation, which makes your DAC more dependent of this cursed S/PDIF.
IMHO this means there is more to investigate, not only in the choice of components. Another example could be (in case of the Studer) to supply the AM26LS31C quad differential line driver with dedicated shunt regulated 5V, which I have not done yet. That driver splits AES3 and S/PDIF from DOBM. I attach a clip form the A727 dig out, maybe this helps.
Best option for you however is to slave the Theta to the word clock of your DAC 🙂

Would using a transformer instead of the cap yield any benefit?

I did not use a transformer, assuming I don't have to worry too much about galvanic problems with my dCS DAC completely in control. But as Jocko says there are many variations. I am certainly going to try out some of those.

It is not impossible that there is an earth problem on the TLC circuit inside the Theta DAC.
<snip>
The better sound from unplugging co ax while using AES was observed on my DS Pro Basic II as well as my friends Proceed DDP (I think it is called that, the beige/grey DAC). We both use the same transport except his is a MKI with aftermarket upgrade to MKII. It may be worthwhile checking out the ground scheme on the transport itself rather then the DAC.

First some questions:
What DAC did you use?
Does this Theta has 'Philips inside' or....?

As far as grounding is concerned, the Studer pcb’s are layed out very, very neatly. Should be of course, as a balanced pro machine, developed in a SOTA joint venture (CDS) between Studer and Philips. Messing around with shunt supplies did cause some ground (and clock!) related problems, varying from erratic behaviour to stressy (jittery?) sound. As the experts tell us, grounding deserves utmost care, so yes you are probably right to look at that again.

I remember the TLC modification talk but I thought it was the fact that they used PCB mount RCA sockets with a common earth connection to ground for both. I suppose it boils down to the same thing. Do you have any additional data or the original article for this TLC mod. A friend of mine owns one....

I do have the Paul Miller review and follow up by Mike Bearn, but no scanner. Will try to get it scanned for you one of these days. In the mean time -- if your friend hasn’t modified the thing yet, let him try the Toslink
Paul Miller’s summary table says about the TLC: “Jitter via coax: up (!) 16%. Jitter via optical: down 82%”...........
Best in this test: Monarchy Audio DIP (down 57% and 81% respectively).

Lourens

 

[Jocko Homo]

Yes, there are other variations, but all of those will add jitter.

The one that uses both ouptuts from a D flip-flop is the worst one around.

OK........not as bad as the stock output in Philips CDPs.

Jocko