Eurorack Synth Master Output Module

[PermO]

While enjoying a BBQ a friend told me he needed to buy a master output module for his modular synth.

So we decided, let's make one, a really good one.

There's 10Vtt level in a modular system, so we don't need to add gain, but the impedance is rather high so I decided to do another version of the Proto Preamp instrument amp with a different gain structure and a headphone output instead of the FX output.
It's 2ch balanced output on mini XLR + phones.

Last night we did a test run to figure out the resistors for the 6 pos stepped gain switch, it now has a nice range so it can drive anything.
Step 1 -10dB
Step 2 -6dB
Step 3 -3dB
Step 4 0dB
Step 5 +4dB
Step 6 +8dB

It sounds really good, low end is incredible, tight, punchy, every little nuance in timbre comes trough nicely, I don't think I ever heard a synthesizer sound this good, I really love this chipset combination, it sounds amazing.

Here's what we did;


Cheers !

 

[Brian Roth]

I am surprised it works without a "bias" resistor from pin 5 of the 2134 to ground. Don't understand the output FET design with the switch which would seem to deactivate that FET??

Shrug...if it all works, good news!

Bri

 

[PermO]

It works quite well... and it sounds really good.

The switch takes out the mosfet, it is a phones mute switch, so it will draw less current when phones not in use.
It does not really need that, but the first headphone amp I designed ran in Class A, and it was much needed.

That class A amp was too much of a load for the shared PSU these eurorack systems have.

But the switch was allready on the front so I just kept it there.

 

[PermO]

I am surprised it works without a "bias" resistor from pin 5 of the 2134 to ground. Don't understand the output FET design with the switch which would seem to deactivate that FET??

Shrug...if it all works, good news!

Bri

On second thought, I guess you are right !

So far I've done some testruns with the main outs but have not tested the phones output yet as I still had to solder some bits together.
It's sitting on my desk now...

I think you just saved the day !

Thanks for this !


100K to ground ?

 

[Brian Roth]

IMHO, I would increase those two 0.22 uF caps. If I ran the calculations correctly, the cap and level pot combo has a - 3dB point of 15 Hz. The second 0.22 and "bias" resistor combo will add additional LF rolloff.

EDIT: Since the 50K level pot is log taper, the "bias" resistor needs to be quite a bit higher than 50K to keep the taper of the pot reasonable....maybe at least 5x higher than 50K

Bri

 

[PermO]

I just checked and noticed I'd allready increased those 0.22uF to 1uF caps.

 

[PermO]

So we did another testrun with the unit, without the before mentioned missing resistors and it worked just fine.
I did put 470K to ground at pin 5 now, just for good measure.

That headphone output kinda surprised me, it sounds really good ! and yet so simple, I might apply this again somewhere.

There was a problem with the stepped gain switch, it's a 2P 6step Elma, so there's two rows of 6 resistors switched to the centerpins A and B.
I tested both loops and every step equalled the resistor as in the schematic, I had two multimeters one on each loop both showed the same thing. So I connected the loops over the fixed resistor and it ended up in a distorteed mess that was L R not equal, mess level depending on setting.... so wtf ?

With the resistor ladder disconnected found that centre A sometimes connects to centre B at one position with 0 ohm and other positions with several K.. huh ?

I can imagine that happening when the pins are miscounted, but that would show, that would not give me two equal ladders on two multimeters at the same time ?

Weird, I'm going to remove all resitors from the rotary switch and see if P1 and P2 still connect, it looks like a faulty switch.

 

[PermO]

So I found the culprit, it was a little solderblob that sometimes shorted centre pin B to resistor ladder A.

So here's the as build schematic;


I picked the 50K stereo log pot as that was the only small stereo pot in my drawer, I think 25K would be a more sensible option. I added Brian's bias resistor R2.

This is a build for a friend of mine, I don't have a eurorack synth, but if you have you might want to try this as both master outputs and headphone outputs sound really good, fast and punchy.
If you run a really beafy PSU in your system you can decrease R12 to 270 Ohm so the IRF's will run a bit hotter (apply heatsink) and you might enjoy the sound even more.


Cheers !

 

[Brian Roth]

Glad it's all working out well!

Something I noticed earlier in this thread (but forgot to mention) is the attenuator at the input of the 1646. The 1646 datasheet says:

2. Both devices must be driven from a low impedance source, preferably directly from opamp
outputs, to maintain the specified performance.

("Both devices" being the 1606 and 1646.)

Discussion on this topic is included in this thread, particularly on page 2:

https://groupdiy.com/threads/drivin...ith-audio-interface-line-outs-line-ins.85873/
Perhaps Wayne (@KA-Electonics.com) will chime in on this topic.

Bri

 

[EthanSpaulding]

For simplicity, you could get rid of C9 and R2 and get your headphone driver ground reference from the pot. Or keep R2 in, if you dont like your bias coming from a mechanical connection.

 

[Brian Roth]

I definitely suggest keeping something in place for R2. @PermO indicated he was using a random pot from his inventory. If the track or wiper of that was a bit wonky, the opamp's output might slam randomly to one rail or the other as the pot was adjusted.

Still unsure how it actually worked with C9 in place but without R2......Something like a 553x would go batsh!t without the "bias" resistor in that circuit. Must be something I'm not perceiving with a FET input opamp.

Bri

 

[Brian Roth]

@PermO... was your build through-hole or surface mount? Asking because Mouser sez the PDIP 2134 is EOL.

Bri

 

[PermO]

@PermO... was your build through-hole or surface mount? Asking because Mouser sez the PDIP 2134 is EOL.

Bri

Mine is all true hole on prefboard, it was a one-off build for a friend of mine.
They are gone @mouser, but I did stock up a bit, I have 25 pieces OPA2134 trough hole in my drawer, as I really like how these things sound.

RS still has some stock left... Texas Instruments, 2-Channel, 8-Pin PDIP OPA2134PA | RS

 

[PermO]

Glad it's all working out well!

Something I noticed earlier in this thread (but forgot to mention) is the attenuator at the input of the 1646. The 1646 datasheet says:

2. Both devices must be driven from a low impedance source, preferably directly from opamp
outputs, to maintain the specified performance.

("Both devices" being the 1606 and 1646.)

Discussion on this topic is included in this thread, particularly on page 2:

https://groupdiy.com/threads/drivin...ith-audio-interface-line-outs-line-ins.85873/
Perhaps Wayne (@KA-Electonics.com) will chime in on this topic.

Bri

Yes that happens when a novice starts designing things

The circuit is based on this instrument amplifier circuit.

Here the output of U1.1 directly connects to input U1.2 and input U2 and you'll notice there's also no bias resistors at the inputs.

As I did not need any gain with 10V comming from the synth I decided to move the gain structure between the first input stage and the balanced line driver. Then I decided the headphone output would need a separate control to keep the master output at a fixed level.
Then I noticed turning down the headphone would effect the master output as it runs the signal to ground, so I decoupled with a cap. A friend told me to put a cap after the pot also, to prevent "scratch noises" when turning the pot.
So that is how this came to be.

OPA2134 has a 10M input, I don't see any trouble there.

Judging by the sound of this thing, THAT seems to have no trouble being driven from a higher impedance, it might not be ideal, but it all sounds great.

 

[Brian Roth]

I haven't done a "deep dive" about what 1646 parameters degrade with a "non-zero Ohm" driving source, but it seems to be the "quality" of the output CMRR is reduced. I hope Wayne will see this and comment.

In my few 1646 applications, I just took THAT's spec sheet at face value vs. trying to second guess. <g>

Bri

 

[PermO]

I did check the datasheet when I first started with the instrument amplifier.

So a better design would be to have a 6 step L-pad after the 1646 on the outputs ?
That means the 1646 would get the full 10V at it's inputs.

As it is a one off build, and it's allready build I think I will keep it like this as it seems to work fine judging the by the sound of it. But I would like to know this, to improve my future designs.

I'm having a lot of fun doing this, GDIY is a great place !

 

[Brian Roth]

It's a puzzle...do you require totally independent level adjustments for the main and headphone outputs?

Bri

 

[PermO]

Yes I think so...

Thing with these things is, it's such a mishmash of patchcables and controls, that when you end up with something really cool you have to record that right away, as you might not get the same sound back the next day.

So I think playing late at night on headphones should not effect the recording outputs.

I think it's fine as it is now, might not be ideal from the datasheets point of view but in the end it's all about the sound.

And I never heard any synthesizer sound as good as this thing does, so we're probably doing something right here

 

[Brian Roth]

OK...I'm just curious about the workflow. In post #14 you had the gain adjustment around the input stage for that particular design. Obviously that arrangement would also impact the level to the "branch" for headphones.

The "correct" (AKA "over designed?" lol) solution based on the circuit in post #8 is to add an extra active stage with variable gain in place of the passive attenuator.

Hell....."if it ain't broke, don't fix it." <G!>

Bri

 

[PermO]

The instrument line amp has a different gain structure and no headphone outut, in this case the output is used to drive FX pedals, so here it has a contraption to increase output impedance and a guitar tonestack can be switched in to mimic the ouptut of the instrument impedance or increase the output impedance of a line source (keyboard / synth) to better match the pedals.

This schematic is still "under construction"

Thinking now, adding the switched L Pad after the 1646 will increase the overall output impedance of the eurorack module, that's not what I want...
Yes an extra gainstage / buffer would help out, but that would also need a complete redesign.

Ah, nothing is perfect, but it all sounds good !