Upgrading the output components in a prosumer dac

Status
Not open for further replies.
Hey guys,
I have a prosumer daw card here (a focusrite) which uses a CS42428 codec (http://www.cirrus.com/en/pubs/proDatasheet/CS42428_F1.pdf), and I would like to try to optimize the post-dac components for quality.

At present, the output from the dac goes to a JRC 4565 op amp. I have a feeling the JRC is a cheap component, and that perhaps a replacement along with either replaced or eliminated output capacitors might be an interesting upgrade.

Alternatively, I would be interested in bypassing the op amp completely and simply using the direct output from the dac though I am guessing there are going to be some issues (particular with regards to impedance and SNR). Anybody try this approach?

What would be a standard sort of mod for this sort of situation? Being that the 4565 seems to be a relatively common component, are there known drop-in replacements for it?

I'm approaching this with an open mind and have no preferences, so any recommendations or thoughts on the matter would be greatly appreciated. Thanks! 🙂
 
Maybe someone can help me with this:

The coupling cap for the opamp is 100uF electrolytic , followed by what appears to be a small 3 legged cap of unknown size. I can't make out what it says on the side, but it is small and orange/pinkish and is sort of shaped like an upside down version of this:
v\/v

I am presently using a jumper to bypass the coupling cap and whatever else that other thing is on one of the differential pairs on each channel. One side of each signal (of the differential output) has a good bit of DC, but the other doesn't seem to have any so I have the jumper on the the side of the signal with no dc and simply connect them. The output no longer measures and DC. I've done a quick signal check and there doesn't seem to be any detriment.

Any suggestions for the optimal output? Thanks to anyone who helps 😀
 

Attachments

  • outputstage_withjumper.jpg
    outputstage_withjumper.jpg
    60.2 KB · Views: 334
thanks for the reply Doug,

How large would such a transformer be? I don't have much room to work with in the enclosure, and I'm guessing snaking the wire a few inches and mounting the transformers outside the case wouldn't be such a good idea.

I am thinking back and the 4565 might be SMD which is beyond what tools I presently have. Any sort of attempt to preempt the 4565 would require me to cut the traces just past the dac, and then jump the dac output to either new transformers or some sort of external preamp is that correct?

If jumping directly is an option, how long a cable could I use or is that not even worth it with the noise and all?

Trying to get my head around dealing with small circuits and parts like this, oh how I wish this were 10x as large and used different boards! 😀
 
Billy,

This is a differential voltage output DAC with offset voltage of 2.7V. You do need the opamp, it does balanced/unbalanced with some gain. You can try a better opamp on a socket or a transformer. The 3-leg component is probably a muting transistor (Motorola?). You can also improve the stuff by replacing the output capacitor. A 10uF/63V polystyrene might be sufficient, like WIMA MKS-4 or better. In theory there should not be any DC voltage at the output of the opamp, but there is due to component tolerances. The capacitor is there just in case. Don't expect too much improvement, though, as long as you leave the power supply as is. I think there is a synergy in such devices. I mean if they use JRC4565, the regulators are very likely 78xx.

Laszlo
 
I vote for the transformer coupled output. It's simple and sounds very good. You can spend anywhere from $22 to $1200 for a pair of transformers.
(hint: the $22 pair will get you 90% of the way there)
 
Thanks for the replies,
oshifis- I figured out what the 3 legged component was, its a t-network capacitor. http://www.rapidonline.com/netalogue/specs/26-6122e.pdf and it looks like the one on the left in this picture: http://www.binbin.net/photos/murata/270/270pf-100v-t-network-capacitor-rc.jpg

Concerning the DAC outputs, I'm looking at the pcb again and its pretty damn complicated. All of the audio (with the exception of the final output caps) is on one side (the inaccessible side) while the power is on the other. There are just tons of those op amps, I'll have to try to redraw the circuit and get your opinions on it. There are 2 of the opamps per channel out, plus others before it. There is even an analog mixer IC in there before the final output.

The problem with using transformers, I see, is that the wires bridging the dac output legs to the transformer inputs would have to be atleast a few inches long. The card has two parallel PCBs configured like this:

|= the pcb

backside |top top| backside

The left PCB is the power board, and has jumpers going across between the two boards. The right PCB is where all of the signal conversion and analog output circuitry is located.

Because the boards face inwards, all of the smd audio pins are quite inaccessible. I could conceivably drill some holes in the PCB to pass some tiny wires through, but otherwise the minimum distance would have to be like 3".

Would I be able to directly connect the transformers to the dac outputs, or do I have to account for the output offset first?

Thanks for all assistance!
 
I may have to draw a diagram. The codec chip has 8 output channels, so I guess I would be willing to experiment on a set of them. If I manage to get the drawing right, would anybody here be willing to help take a look at it and suggest what sort of modifications might be advisable?
 
Alright, the pcb is too complicated to figure out. Its many layered and there are no schematics anywhere. Given the options, it seems like using a pair of transformers coupled to the dac outputs are the way to go.

Before I commit to this experiment, I want to be sure that there won't be a problem with the output levels going on. Attached is a diagram from the datasheet of the dac that shows the output levels. There is a 2.7v offset. With that issue, can I still use the transformers directly connected to it because the offset is identical and thus the transformer would effectively block the DC? Or does this mean that I need to correct the 2.7v offset first before I can use a transformer?

The other part I need help with is about the transformer itself. The dac datasheet says:

dac output impedance: 150 ohms
AC load resistance: 3K ohms
load capacitance max: 30 pF

So from that I know that I need a relatively high impedance input for the transformer, which is why the 10k input was suggested. Now the part I'm concerned about is that in the suggested 10k:10k. The second 10k listed is the load impedance correct? Does that mean that the output impedance is 10k and thus I would need a 100k input impedance on the receiving device, or does it simply mean that it expects a 10k or greater input impedance for that other device?

I think that about covers it. The edcors seem like they may be a decent budget way to try this out, anything to watch out for with them? Thanks.
 

Attachments

  • cs42428_outputdiagram.jpg
    cs42428_outputdiagram.jpg
    28.3 KB · Views: 146
Before I commit to this experiment, I want to be sure that there won't be a problem with the output levels going on. Attached is a diagram from the datasheet of the dac that shows the output levels. There is a 2.7v offset. With that issue, can I still use the transformers directly connected to it because the offset is identical and thus the transformer would effectively block the DC? Or does this mean that I need to correct the 2.7v offset first before I can use a transformer?
As long as you do not DC ground the center tap on the input side, the 2.7 volts will not cause a current and everything will work.
So from that I know that I need a relatively high impedance input for the transformer, which is why the 10k input was suggested. Now the part I'm concerned about is that in the suggested 10k:10k. The second 10k listed is the load impedance correct? Does that mean that the output impedance is 10k and thus I would need a 100k input impedance on the receiving device, or does it simply mean that it expects a 10k or greater input impedance for that other device?
Transformers do not have an impedance. On a 1:1 transformer, it reflects the output Z of the Dac, and reflects the load resistor.
I like to think of it this way, the output Z of the drivers should be less than 10K and the next stage should be 10K or higher.
 
Status
Not open for further replies.