DIY Volume Control (Relay Attenuator)

[Ghoostknight]

Actually, i will scrap the idea of running both unbalanced and balanced, it makes the whole thing unnessecary complicated....
i might implement it with transformers like i initially planned

So i added:

• a output select relay to the relay attenuator, one way for a opamp buffer, the other passtrough
• increased audio traces to 2mm

and these are the three boards, nearly finished:

unless there are some problems i probably wont change much anymore before ordering 🙂

 

[Ghoostknight]

Hmm Question:

If i implement a "impedance balanced" resistor (maybe via relay) on the output of the relay attenuator wouldnt i be able to get all of these:

• unbalanced input
• unbalanced output
• balanced in
• balanced out

and can switch between these?

rca is unbalanced and xlr balanced

• rca in -> pseudo balance resistor -> xlr out
• rca in -> grounding cold wire -> rca out
• xlr in -> no resistor -> xlr out
• xlr in -> just taking hot wire -> rca out

pro:
completely passive

con:
volume can be quite jumpy

• xlr up to 6V on rca
• rca in with pseudo balancing resistor at only max 3V going into xlr

Did i miss something here?

 

[Ghoostknight]

actually might not need anything additional if i leave cold unconnected at the relay attenuator input, as the resistor deviders trough the relays should give the correct impedance/resistance for pseudo balance to work

So, i actually just have to connect XLR and Cinch at the outputs and both should work fine for both kind of inputs, leave cold unconnected for rca inputs

i hope im not thinking wrong here, but that seems by far the easiest solution

xlr input to rca output isnt ideal as you dont terminate the cold wire, so no noise rejection takes place, but i could look over this, just keep the corresponding xlr input cable short i guess

for the high xlr voltage towards rca i think its easy enough to implement a corresponding -3-6db reduction in software to not drive rca inputs into clipping

 

[tomchr]

Not sure what led to the decision to use 2 mm traces for audio. How many ampere of signal current are you expecting?

Tom

 

[tomchr]

suggesting the THAT1646 as DRV134 replacement, do you have expierence with this one?

As I mentioned earlier, I used the THAT1246 in my DIFF PRE, so, yes, I have experience with it. It's a fine chip albeit a bit noisy.

I have no experience with the DRV134. I'd use the LME49724 as the line driver. It's ~20 dB lower noise than the THAT1246.

The 10 Ω resistors in series with the inputs (R1 and R2) don't provide any ESD protection. If you want ESD protection, you need to go back to the clamping diodes and increase those resistors to 100-500 Ω. Higher would be better but that starts to wreck the noise performance.

Tom

 

[Ghoostknight]

Not sure what led to the decision to use 2 mm traces for audio. How many ampere of signal current are you expecting?

Tom

its really only my OCD wanting thicker wires lol, since i had the space i dont wanted the pcb traces to be a bottleneck compared to the relays

As I mentioned earlier, I used the THAT1246 in my DIFF PRE, so, yes, I have experience with it. It's a fine chip albeit a bit noisy.

I have no experience with the DRV134. I'd use the LME49724 as the line driver. It's ~20 dB lower noise than the THAT1246.

noise seems to be the big culprit of these unbalanced to balanced drivers

i guess you are recommending a "normal opamp" as balanced driver with the LME49724, would prefer it but the pseudo-balance solution seems like the easiest solution by far, you know, since balanced to balanced is still "perfectly" implemented i dont really care for subpar performance on other outputs/inputs combinations (and from what i read pseudo-balanced is still better than unbalanced), imo better easly implemented and having options than only balanced since i actually plan to try out some RCA devices in the future... unfortunaly you find balanced still only in the pro/high end segment

The 10 Ω resistors in series with the inputs (R1 and R2) don't provide any ESD protection. If you want ESD protection, you need to go back to the clamping diodes and increase those resistors to 100-500 Ω. Higher would be better but that starts to wreck the noise performance.

this is the current buffer board:

there are specific purpose ESD diodes

i think i read that its beneficial to match feedback and input impedances, tho the relay attenuator itself messes with the input impedance either way, so not really needed but i thought its better to have those resistor footprints to experiment since its a standalone module


-----

i was looking up how a opamp is powered by a single power supply, a easy solution seems to be a voltage devider + midpoint referenced to ground + buffer caps

im pretty sure some of you compared single vs double supply, is single supply that bad? it does seem like an easy solution, what could go wrong? the only negative i see is that you need 16+V for 6V output swing and that noise from different rails could influence eachother (tho im wondering if the resistance from the voltage devider helps here... or if i could add some passive filtering)

-----

need to rethink the output/input board, probably will just include both stereo channels on one board

 

[tomchr]

i dont wanted the pcb traces to be a bottleneck compared to the relays

How would thinner traces be a bottleneck?

Thinner traces would have less capacitance to whatever traces/planes they cross.

noise seems to be the big culprit of these unbalanced to balanced drivers

For some of them, yes.

i guess you are recommending a "normal opamp" as balanced driver with the LME49724

Not necessarily, though that would probably be the easiest. At least with an opamp driver in front of the LME49724 you don't have to worry about the input impedance of the differential driver. To me, adding an opamp is not a huge deal. It's just another part. Not much different than adding another transistor. Not much more expensive either in many cases.

i think i read that its beneficial to match feedback and input impedances

Impedance matching is a must for optimal power transfer in RF circuits. It doesn't matter for audio.

this is the current buffer board:

You have acid traps and your pad exits could be better. I'm also not a fan of via-in-pad. Those must be tiny holes in those vias. Are you willing to pay extra for those?

Are you sure you have the correct footprint selected for the IC? It looks much larger than an SOIC, but that could just be the scale of the board throwing things off visually.

Tom

 

[savan]

You should take an opamp to convert diff to single ended than you can implement tone control, volume control ...etc on single ended and finaly convert single ended to diff, for example : )