A 2HP mix-attenuate-buffer module.
Initial features include:
- Open source hardware and software
- Two mix inputs
- Attenuation slider
- Two buffered outputs
- Works with audio and CV signals
While these modules can be hand-assembled, they are best ordered pre-assembled from your PCB house. BOM and component placement files are provided for the main board (2415-0201).
Full assembly instructions will be uploaded soon.
Feel free to ask questions by opening an issue!
The updated PCBs arrived:
Once assembled, they are still drawing more current than I was expecting:
- Without module attached: 41 mA (5V in to power supply)
- With module attached: 127 mA
- Difference: 86 ma (20 mA per op-amp)
As the previous design also draws 127 mA, this indicates that all the design changes I put in didn't make any difference at all with respect to current draw.
It's worth noting that these numbers include power supply conversion inefficeincies, and that the op-amps are running with +/- 12V rails, so at 24V, that's only 4 ma per op amp.
I re-ran my tests with the spectrum analyzer, and was able to track down the spurs to the power supply. So I don't have any evidence of internal oscillations.
Looking at the datasheet, Icc is 8 mA per IC with no load, so perhaps everything is operating correctly? It certainly functions as expected and passes all my tests.
I do prefer the new PCB layout, so I'll stick with it, but this sure ended up being quite the wild goose chase.
And I need to get or make a better power supply for testing purposes.
I re-did the routing for the analog board to eliminate the long traces that were likely adding significant capacitance, and also cleaned up the ground plane on the top side. I did have to move some of the resistors over to 0402, but didn't have to go to four layers.
I'll get this sent off to be manufactured in the next batch, and hopefully this will fix the power draw problem.
Despite the module working as expected, and not being able to see anything out of the ordinary on the oscilloscope, the NE5532 op-amps are oscillating internally. I broke out the spectrum analyzer and the H-field probe, and sure enough, there is a strong signature present above each op-amp IC:
Just look at those sprus!
I made a couple of improvements to the board layout for the 1.1 revision, namely a better ground plane, and moved the input mix resistors and the output protection resistors closer to the op-amps. Long traces can add capacitance, which may be what is causing the oscillations.
I also incorporated the design changes discussed yesterday to move from a unity follower to a gain of 2, which should also help with stability. Unfortunately, since this board is so cramped, I had to use some 0402 resistors to make everything fit, which does complicate things if I have to do any rework.
Here's the updated schematics for revision 1.1:
I'll have to get a new set of boards made, and re-test everything, but hopefully this will fix the problem.
The two NE5532 op-amps are each drawing around 40 ma, and get hot to the touch.
To check for stability, I applied a square wave to the input (blue trace), and measured the output (purple trace). The response looks well damped, and there aren't any oscillations other than what is present on the input.
I can't see any signs of stability problems on any of the output or input pins, so if there are any oscillations, they are internal to the op-amp.
Still working on figuring out where the excessive current draw is coming from.
The reason why the output trace is 50mV slower than the input trace is because the voltage divider formed by R5 and R10. If I increased the resistance of R5 to 10K, and reconfigured U2A and U2B to have a gain of 2, then the output would exactly match the input.
If I have to do a re-spin to fix the excessive current draw, I think I'll also add that change.
I noticed that while everything is working as expected, the op-amps are drawing way more power than I expected.
So I'd hold of on building any of these until I figure out what is going on.
Here's the schematics:
Got the assembled boards from the manufacturer:
The front panel is on the left, and the pannelized attenuator boards are on the right.
Once fully assembled, with the power and slider spacer boards added (not shown), the module looks like this:
And, after installing it into my Eurorack system, it works exactly as expected!
So this module is pretty much done!
Next steps are to order final boards, and to write up the assembly instructions.