Midifying SSL EQ with Digital Potentiometers

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
PRR said:
You run a 10:1 (or 2:1 or 30:1) range on the pot, then the computer switches capacitors and runs the pot again from the top.
I started to do something like this once. The PCB layout should explain it mostly:

LcFilterAdg1414.png


It's just two 8 gate analog gate chips (ADG1414). One chip switched caps in or out to make a specific capacitance and one to switch inductor taps. Of course not all 2^7*6 combinations (or whatever it was - don't recall) were applicable for a specific Q value. But you could make 10 good frequency steps for a given Q per octave.

So this is just the LC resonant circuit that would be paired with one of the usual boost / cut filter circuits.

I never made the board though.
 
squarewave said:
I started to do something like this once. The PCB layout should explain it mostly:

LcFilterAdg1414.png


It's just two 8 gate analog gate chips (ADG1414). One chip switched caps in or out to make a specific capacitance and one to switch inductor taps. Of course not all 2^14 combinations (or whatever it was - don't recall) were applicable for a specific Q value. But you could make 10 good frequency steps for a given Q per octave.

So this is just the LC resonant circuit that would be paired with one of the usual boost / cut filter circuits.

I never made the board though.
I used analog gates (DG series) to switch caps in my PFL52
http://yahoo.aleado.com/lot?auctionID=k345769404
I wanted to have the best operational resolution, so I used slugged lin pots with only 1:10 range and 3 range multipliers.
 
Hi guys,

reviving the project and this thread after a long hiatus.
Thanks again for all your input, it has really helped me so far.
I have something to show for as well.

The first prototype is working!
I have switched to the AD5290 digipot in the meantime. It´s daisychainable, and it was easier for me to wrap my head around that. It´s 15V only so I´ve decided to drive the SSL EQ card with 15V as well and sacrifice some headroom.
Also I have replaced the SSL 82e132 with a G series EQ 82e292.
Anyway I have now wired up the card, got the HMF section to work with regular pots and now already replaced the frequency pots with two 50k digipots, and it´s working beautifully!
I don´t get any digital noise or zipper noise when adjusting the frequency. Also I have written a pseudo log conversion of the 128 midi values to the 256 pot values and must say I am really happy with the resolution this gives.
I have used a regular pot without centertap for the gain and generally this seems to be working well.

What´s still weird is that I only get around 15 dB of gain, instead of around 20 which the circuit should do. Negative gain I can go until -20dB. Dunno what´s up with that. Maybe because of the missing center tap, or because I have preliminarily removed the neighboring OP amps, to only use the HMF section. Will see.

The sound is great! Next steps are to replace the other 2 controls for the HMF section with digipots (already got these parts), and then scale up to the whole EQ.

Attaching a pic of my quick and dirty (very dirty) prototype. Will switch away from the breadboard for the next step for less chaos and more reliability :)
 

Attachments

  • IMG_1894.JPG
    IMG_1894.JPG
    1.9 MB · Views: 56
CoOl! Looks like fun. And AD5290 is a good part for this.

But the first thing you really need to do is see how low you can get the noise / distortion. And right now it's not going to be good with wires hanging out like that. You need to make a PCB that has the receptacle for that board so that you can just plug it in and then dodge all of those antennas.

Then break out the computer and audio interface and get some noise / distortion measurements / frequency plots. Otherwise you're pissing in the wind.
 
Thanks for your feedback and the advice.

Yeah, the prototype is really dirty, as I wrote. I wanted to see quickly if the chips worked as I envisioned, and how bad the zipper noise is (for the frequency pot it turned out to be zero). Update / correction: there is considerable switching noise when adjusting a parameter, see my post below.

I already have sort of a plan for the real board.
I plan to use a prototype PCB that already has leads for all the digital connections and power supplies printed on: +-15V, +5V, SCLK, CS, GND:
https://www.amazon.de/Prototyp-PCB-Universal-Matrix-Leiterplatten-Hongtian-St%C3%BCck/dp/B06XXPT7F1/ref=sr_1_6?__mk_de_DE=%C3%85M%C3%85%C5%BD%C3%95%C3%91&keywords=prototype+pcb&qid=1581458218&sr=8-6

Then I initially planned to wire the pots directly from the chips´ outputs on the board to the EQ. Do you recon this would still be too bad, antenna wise? I will work with sockets for the digipots, so I´m flexible on reusing them if something doesn´t work, but of course soldering the whole thing will be a lot of work and I´d like it to be as good as possible from the start.

Good point also about the measurements, will do these definitely.
 
Not sure what you mean. But it looks like that PCB has some kind of edge connector on it so if you can just interface with that somehow that would be important for trying to minimize the length of unshielded signal lines. Specifically it looks like you could make a long board that has the various digipots on it right next to the pins for the edge connector. But I'm just musing based on one photo so ...
 
Doesn't this card have some weird triple pots on it?

That may actually be easier in digital than finding an Analog one though.

Cool project!

S

 
Gertius said:
I plan to use a prototype PCB that already has leads for all the digital connections and power supplies printed on: +-15V, +5V, SCLK, CS, GND:
https://www.amazon.de/Prototyp-PCB-Universal-Matrix-Leiterplatten-Hongtian-St%C3%BCck/dp/B06XXPT7F1/ref=sr_1_6?__mk_de_DE=%C3%85M%C3%85%C5%BD%C3%95%C3%91&keywords=prototype+pcb&qid=1581458218&sr=8-6
I just see a typical prototype board. It's just one notch above a breadboard. Lots of wires, no ground plane. That should be ore or less ok for assessing basic operation, but certainly not for evaluating performance. For that you'll need a proper PCB.
 
abbey road d enfer said:
I just see a typical prototype board. It's just one notch above a breadboard. Lots of wires, no ground plane. That should be ore or less ok for assessing basic operation, but certainly not for evaluating performance. For that you'll need a proper PCB.

Thanks for your feedback. I´m a bit torn, because I understand what you´re saying, but to design a new PCBis out of my scope right now, as I have 0 experience in that. Are there some better readymade solutions that you know of?
Probably I´ll start building with the prototype PCB and evaluate every step (EQ stage) after it´s been built. When the noise gets too bad I then need to reevaluate. Actually I would already be happy if there is no digital noise when the pots are not being used. I could live with a little noise while adjusting parameters. It´s not meant to be a product, but just for my own mixes.
 
synthiaks said:
Doesn't this card have some weird triple pots on it?

That may actually be easier in digital than finding an Analog one though.

Cool project!

S

Thanks! Yeah that´s what I´m thinking too! The whole EQ with the filters still needs 26 digipots, so it´s not gonna be really cheap still. But I plan to use it across all channels with Waves QClone (along with digitally stored settings to go back and forth between channels for tweaking), so I hope it´s gonna be worth it!
 
I have i "black" in a drawer somewhere maybe i should dig it out and check the digi pot option!

As said really cool solution to this problem. One could use Digipots for the hard to find antilog and triple pots.

S
 
Gertius said:
so I hope it´s gonna be worth it!
Not only would I make a pcb for this, I would make several. I would make one little one for 2-4 digipots sharing an I2C bus so that you can just set some addressing pins (if you even need to do that for whatever digipot you're using) and then stack them in some way. You'll need to do something to get 26 of them in close proximity. You might consider using some simpler analog gate chips like ADG1414 (also an iCMOS chip) to make some of the controls stepped instead. With one chip you could replace up to 4 digipots (2 gates can make up to 4 steps. 3 gates makes 8, etc).

Then I would just build one of the "little" boards and test the performance of one eq band with and without digipots and compare the results. Use a decent quality audio interface and rig that has a noise floor low enough to see the difference.

You really need to follow the path-of-least resistance to verify performance before you become too emotionally invested in this.
 
squarewave said:
You might consider using some simpler analog gate chips like ADG1414 (also an iCMOS chip) to make some of the controls stepped instead. With one chip you could replace up to 4 digipots (2 gates can make up to 4 steps. 3 gates makes 8, etc).

This thread is turning into food for thought!

Like steped resistor values... man why haven't I thought of this before. I feel soooooo stupid. I need to dig out my E-series EQ and Channel Compressor cards right NOW!
 
I never had much interest in MDAC but admittedly only looked at them seriously several decades ago. 

Modern DPOTS are so much better than early parts, perhaps MDACs have likewise evolved.

JR
 
synthiaks said:
Like steped resistor values... man why haven't I thought of this before. I feel soooooo stupid. I need to dig out my E-series EQ and Channel Compressor cards right NOW!
Before you get too exited go back and read the beginning of this thread. Analog gates have significant on resistance and capacitance compared to conventional switch / relay contacts. The source impedance seen by the gate needs to be sufficiently low and the load impedance needs to be sufficiently high or you will get non-linearity. If you design the circuit specifically for analog gates then it would probably work great.
 
squarewave said:
Before you get too exited go back and read the beginning of this thread. Analog gates have significant on resistance and capacitance compared to conventional switch / relay contacts. The source impedance seen by the gate needs to be sufficiently low and the load impedance needs to be sufficiently high or you will get non-linearity. If you design the circuit specifically for analog gates then it would probably work great.
Transfer gates are the basic switching technology used inside DPOTs and even digitally controlled mic preamps to adjust gain digitally. The performance of these dedicated designs is more than adequate.

Inexpensive CMOS TGs can degrade performance if carelessly applied.  I switched tens of channels inside a console with one logic signal back in the 80s using inexpensive CMOS TGs but I was careful and distortion was below my bench test measurement floor.

JR 
 
Update / correction: there is considerable switching noise when adjusting a parameter.

I must have missed it, because I listened most of the time through a Waves QClone capture to see the freq response.
It´s not overly loud (-40dB, couldn´t hear it over music playing) and there is no noise at all when not adjusting a parameter.
I checked Serial clock ans MOSI lines with a scope, and there really is only activity when sending Midi.
For my project I´m good with that, but I think it probably wouldn´t be acceptable in a commercial project.

The noise floor when not adjusting a param (unbalanced connections, using 2 digipots for now) is at-90dB, which seems fine to me for now.
 
Gertius said:
Update / correction: there is considerable switching noise when adjusting a parameter.

I must have missed it, because I listened most of the time through a Waves QClone capture to see the freq response.
It´s not overly loud (-40dB, couldn´t hear it over music playing) and there is no noise at all when not adjusting a parameter.
That noise can be eliminated if you use a properly laid out pcb. There’s also noise and distortion in the analog output that will be better with a properly laid out pcb.

What your trying to do is more advanced than most projects. That’s why I think you should work your way through things incrementally.
 
After chipping away at it daily for the past month, I have now more or less finished the project.

The EQ is completely controllable by MIDI now, and it works very well.
In general I was surprised how straightforward everything worked.
I had one bug in my circuit, and one faulty Digipot, but other than that everything went smoothly.

I have one bug in the Midi plugin to iron out for the parameter update, but nothing major.
Noise level is at -80dB, which isn´t perfect, but I´m happy with it for now.
I can still hear clicks when switching settings, but that´s also no problem for my usecase.
Today I mixed my first complete song with the EQ ánd Waves QClone and I´m very happy with the result.

Thanks everybody for your input!
 

Attachments

  • IMG_1969_small.jpg
    IMG_1969_small.jpg
    356.7 KB · Views: 34

Latest posts

Back
Top