I want a bigger S/PDIF !

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am using a modified DCX2496 as my DAC, and I have found for some time that every now and again it will drop the high frequencies. It is exactly the symptom described in the CS8420 datasheet as follows:

Occasionally the CS8420 SRC will enter an invalid state. This can happen after the RUN bit has been set
when an AES3 stream is first plugged into the part or when a source device interrupts the SRC input stream.
When this happens, two symptoms may be noticeable - notches occurring in the frequency response, and
spurious tones being generated in response to some input frequencies.

After some messing about with my new pocket scope, I found that the digital output levels from both my CD player (Cyrus DAD7) and soundcard (M-Audio Fast Track Ultra) are below the minimum input for the CS8420! The minimum is 200mVpp. The CDP outputs about 100mV (under 75R load) and the soundcard around 170mV which still seems to play up.

I read that the spec for S/PDIF is 0.5Vpp - 1Vpp so why the feck are these supposedly professional business' making kit that doesn't meet spec?

Anyway, my main question is how do I up the voltage of each S/PDIF output? I have an old CA DiscMagic and that has a ~1V output on the AES/EBU connection and that never seems to 'loose lock' and drop the high frequencies. My thought is perhaps to replace the 1:1 ratio pulse transformer in the DCX2496 DAC with a higher ratio. OEP make some with 3:1 or even 8:1 ratio, though I'm not sure I can get hold of them. Spec in the File Here, are they even suitable for digital audio?

My other thought is that I have a digital switch box that uses an HD74HC04 gate device and I wonder if this can be configured to give 4x gain or something? I have traced the circuit and uploaded it below for you. I don't really know much about digital circuiets so please be gentle!

An externally hosted image should be here but it was not working when we last tested it.
 
Thanks for the replies.

You mention pulling R11. This co-insides with a question I had anyway - Is it only the reciving end of the line that needs to be terminated at 75R, or should the transmitting end also be done like this? I have seen digital outputs that have a 75R resistor between the pins of the pulse transformer, but I'd have thought this is only needed on the reciving end and the transmitting end should be low impedance?

If I pull R11, should I not also bypass R10? And should I put a 75R in the gap of R5, which is currently left blank.

I did actually remove a resistve diveider in my soundcard and it boosted the level a little. I can't really do this with my CD player as it has a strange digital output design that I don't really understand.

I'll try putting a pot in the feedback loop of the digital switch.

Thanks!!
 
The idea is to have a source impedance of 75 ohms. This might consist of a 5 ohm real world source in series with a 70 ohm resistor.

This source then feeds a line with a 75 ohm characteristic impedance (i.e. 75 ohm coax).

This is then terminated on the receiving end with another 75 ohm impedance. So, there are basically two 75 ohm resistors, one on each end of the 75 ohm "transmission line". The theory is that, at very high frequencies, the line transmits the most power to the load, and has the least amount of reflection, with identical 75 ohm resistors on each end.
 
Tenson said:
If I pull R11, should I not also bypass R10? And should I put a 75R in the gap of R5, which is currently left blank.

I would change one thing at a time and see how it sounds.

Pull out R11 and see what happens.

R10 should be in the circuit, and it should be 75R minus the output impedance of the 6 inverters.

As zigzagflux said, a 70R resistor in at R10 would be a good choice for starters.
 
Hi,

I removed R11 and the output is indeed higher, it has a gain of about 3x.

However, it doesn't work quite right like this. When connecting my CD player as the digital source, there is no sound, unless I touch the meter probe on the input terminal. I'm thinking this is maybe because the input impedance of the circuit is too high, (since they seem to have left out R5) and touching the probe on the input offers it a little input impedance to load the source? For some reason with R11 in place it never had this problem. My soundcard works fine though.

Also, with nothing connected there is about 2.3V on the input of the circuit, which I guess is generated by the output of the optical receiver. This drops to about 40mV if a coax source is connected.

Any suggestions? My thoughts are firstly to disconnect the optical receivers since I don't use them, and then add a 75R resistor in gap R5. Apparently the input impedance of the gate is very high so it can basically be ignored, although I don't know what happens to that when it has that 100K resistor in the feedback?

I think R10 should also be replaced with 75R because the output impedance of the parallel gates is also negligible.

Does all that sound sensible? :) Cheers!

P.S. this datasheet mentions input and output Z of the gates - http://www.datasheetcatalog.org/datasheet/HitachiSemiconductor/mXvyusq.pdf
 
Ex-Moderator
Joined 2003
I recently made an AES3 driver for my Marantz CD5400 to drive my DCX2496 for exactly the reasons you describe. Electrically, AES3 is balanced and 110 Ohm, rather than the unbalanced 75 Ohm of S/PDIF. There were some spare inverters in the player, so I used the paralleled inverter technique shown earlier but without feedback resistors. I used a transformer coupled by a 1uF capacitor and 95 Ohm (determined by experiment) of series resistance to get the output impedance right. Here's an eye pattern of the output data:
 

Attachments

  • aesebu00.png
    aesebu00.png
    18.2 KB · Views: 582
Thanks for all the replies!

I have got a little further with this now. It seems there were two reasons my DAC was loosing lock. Firstly as I said the S/PDIF output was too low. However, even when boosted with the switch-box it kept loosing it. It turned out that I needed to ground the switch-box because both my sources and my DAC had floating inputs/outputs. So I grounded the input of the DAC, that in turn grounds the switch-box.

Now, it does seem to be working reliably, but there are a few things I wonder if you can help clear up?

Firstly, removing the voltage divider on the output of the switch-box did boost the level, and I now get about 500mVpp on the output. However, something that I find puzzling is that changing the feedback resistors R8 and R6 do nothing to the output level. In fact even bypassing R7 does nothing. It seems that no matter what the level of the input signal, and no matter what the feedback resistors are, the output is always spot on 500mVpp. Why is this??! I'd like to make it about 1V if I could, then I can add a 75R series resistor after the gate to get the impedance right and still have a good level. At the moment there is no source resistor, it just connects the gate, via a cap to my DAC with a 75R input load.

Secondly, there seems to be some ringing when the signal is connected to the switch-box.

If I have the S/PDIF output from my CD player going through a coax wire unloaded or loaded with a simple 100R resistor it does ring a little, but interestingly this can be reduced with a change in cable from a coax to a twisted pair balanced cable. See the comparison below.

An externally hosted image should be here but it was not working when we last tested it.


Now the bit I don't get - when I connect it to the switch box, it rings no matter what cable I use, and it does so on the input, and the output! (p.s. I changed the scope range from 500mV to 1V when looking at the switch box output)

An externally hosted image should be here but it was not working when we last tested it.


So why does the cable suddenly make no difference, and how can I clean up that ringing?

Thanks for your help!
 
Ex-Moderator
Joined 2003
Re: AES3 driver

buble_corp said:
Could you share with us the circuit diagram of your AES3 driver? Did you use something similar to SPDIF to AES/EBU converter shown on http://www.epanorama.net/documents/audio/spdif.html ?

Close, but not quite. I picked up the signal inside the player where it came from the oversampling chip, so I didn't need the AC coupling and diodes in your reference. I did connect three inverters in parallel (like your reference) to drive the output. Then I added a series resistor and a series 1uF capacitor to drive the AES3 transformer. I'd draw it, but for various reasons, my scanner isn't working.
 
Ex-Moderator
Joined 2003
Tenson, you're going to need a better oscilloscope than that to see what's going on. As it is, your lack of analogue bandwidth or sample rate could be concealing all sorts of things. I think the best you can do is to set your source impedance to 110 Ohm, use 110 Ohm twisted pair and hope that everything is correct. Fortunately, you can still check your source impedance by seeing if adding a 110 Ohm halves the open-circuit output voltage.
 
Ex-Moderator
Joined 2003
The transformer was a 1:1 DIL type made by Newport Magnetics. I should warn you that although I've made this circuit a few times, the values of the series resistors may need to change depending on the individual inverters to get the right output resistance.
 

Attachments

  • cd5400 aes3.gif
    cd5400 aes3.gif
    7.5 KB · Views: 379
-> EC8010

Thanks for schematic, I see that you didn't use inverters to create balanced signal, you only used gates to buffer/amplify the signal.

I read the topic where Jocko Homo was teaching someone how to build correctly spdif output so I know about matching impedance, but still are there any datasheets telling you what the output impedance of chip's spdif digital output? Of course you can calculate it after measuring output voltage on different load, but to do it precisely you need a really good oscilloscope.
 
Ex-Moderator
Joined 2003
buble_corp said:
... but to do it precisely you need a really good oscilloscope.

Yes. :D

Actually, if you do the sums, you'll realise your oscilloscope doesn't need to be that good. Provided you compare open circuit with half amplitude (by adjusting a trimmer to give half amplitude), you can then just measure the trimmer with a DVM and the result will be good enough.

There's no way you can determine the inverter's output resistance from theory - they use FETs and FETs have enormous device to device tolerance. Also, the transformer adds resistance. It really is best to determine that series resistor by experiment.
 
gmarsh said:
Use a 74HCU04, not a 74HC04.

Putting feedback on a HC04 (non-U) is asking for spurious oscillation.


Hi,

The circuit in my first post seems to use a HD74HC04, and has some feedback resistors. Removing them seems to do nothing to the output level, so should I remove them? Do gates like these not have a near infinite open loop gain, like an op-amp?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.