I find that the open and DIY aspects of Mutable Instruments is a never ending source of inspiration and learning. For me there’s always new things to explore and discover, even for the, now quite old, pre-eurorack designs.
For the Ambika SMR4 voice card I have been thinking about increasing the resonance compensation capacitor to make the Juno/Jupiter like attenuation of lower frequencies at high resonance less pronounced. I’ve looked at the Shruthi-1 SDE modifications, where this capacitor was completely bypassed for a full bandwitdth resonance compensation, and was also intrigued by the possibilties of multiple filter slopes. The problem was that for some sounds I really like the shiny Juno/Jupiter quality, while for other bass like sounds I’d like the full bass end, so being able to select this programmatically would be important.
When Pharmasonic published the excellent SMR-4Pi design modifications I had the epiphany that if the filter slope selection was limited to “only” 4P and 2P low pass, it should be possible to use the other half of the 4052 to select between resonance compensation with, and without, the capacitor. While I was at it I also incorporated the filter tuning possibilities from Andres SMR4-Plus design.
The resulting untested works-in-progress is available at https://github.com/bjoeri/ambika_smr4x .
My main concern is that this is my very first attempt at any kind of circuit design and/or modification, so it’s highly likely that I’ve made some major rookie mistakes. For example, I wonder if the routing is ok with regards to noise etc. As you can see some analog traces pass, or are close to, digital signals. Compared to the Pharmasonic design I added decoupling capacitors for the 4052 but I have no idea how important this is. Any kind of feedback is highly appreciated before I go ahead and try these out in real life.
Is there anyone more skilled in electronics design than me that can tell whether the analog trace (for resonance feedback compensation) crossing (on the opposite side of the board) the digital SPI lines is a bad idea noise wise?
I’m not sure if it will have an effect at all but it helps to have them cross at 90 degrees… basically further away from parallel the better.
Like this maybe?
I purposefully stayed away from the VEE power line and then crossed the digital lines at a 90 degree angle and then routed over to the pin.
I think those are mostly programming lines anyway… maybe they aren’t used during normal operation?
ERC and DRC checks out with no new errors vs the original PCB.
Thanks for the feedback! I took the plunge and ordered a couple of boards a while ago that I’m now waiting for. I figure my routing should be ok, and I’m eager to know if the mods actually work as expected. I’ll know for sure when I have them built and tested.
Regarding the traffic on the digital lines: they’re used all the time in communication between Mobo and voicecard.
I see so they do make a bunch of noise all the time. You aren’t super close running parallel with the data lines at least so thats a good thing. Hopefully it won’t be an issue at all. The only other parallels I see are with a power line but that usually isn’t as big of an issue as data lines.
I’m just recently jumping into the Shruthi/Ambika stuff (super late to the party) so I’m a little new to these circuits. I finished a Shruthi XT yesterday and I’m making PCB’s for an Ambika now.
I will fix a couple things and export gerbers of your modified PCB too. I’ll probably have a few made when I get the control/motherboard made. Not sure I’ll beat you to testing though since I still have to get a BOM finished and build my control board first.
Any update on your PCB’s yet? I’m planning to have some made if they are working.
I haven’t gotten that far yet. I’m in the progress of building my 2nd Ambika, and still working on the mobo.
So I finished building one of these SMR-4x cards (together with my 2nd Ambika).
The cards work seems to work as an SMR-4 replacement in the terms of digital noise (none noticeable) and the 4P/2P modes works just fine BUT the cap mod is kind of a disappointment:
I cannot hear any difference whatsoever with/without the cap in terms of bass (low) end for high resonances. Either this means I have messed up as the electronics noob I am, or the high pass filtering effect of this cap in the resonance control described in section 4 of the SMR-4 description is much more subtle than I anticipated (https://mutable-instruments.net/archive/documents/smr4mkII_analysis.pdf).
Oh now I see, that blue thing on the left side is the actual trimmer! So I guess I should try tweaking those on my Ambika and see if it fixes or at least helps a bit with my keytracking issues.
But to have two trimmers so that we can match voicecards close together as far as filter tracking is concerned, yep, this is the dream
Bjarne, looks like when you’re done with your voicecard modification, I will probably want a set. Would you receive my Pharmasonic SMR4-Pi boards as a compensation (tough chance, I know)?
That said, if resonance compensation mod you tried to do doesn’t work, I’d suggest returning the 1-/3-pole jumper, as I find that a lot more useful. 1-pole sounds really great to me!
Yes, you need to calibrate your voicecards.
I’m currently in the progress of physically modding my already created PCB’s with the goal of achieving more resonance boost, see Tweaking resistors in SMR4 resonance compensation. If it works they will have 2P/4P with boosted resonance compensation, and a 4P mode similar to original SMR4. That’s what’s feasible without creating a new design and new PCB’s.
These PCB’s are really works-in-progress at this stage so I wouldn’t want to spread these for the sake of the added trimmer at this point.
No, I do understand your voicecard mods are WIP, I meant after they are completed I’d love to have a proper tuning across all the voicecards in my Ambika.
But boy, I would definitely miss that 1-pole sound There’s no way you could add it to yours, right?
@Bjarne Are there any news regarding your voicecard mod?
Unfortunately no. I’ve not had any time for music making and/or hacking lately.
I took some time to build and test a modified version of the SMR4x card posted above:
- Replaced 220 with 1k resistor in resonance compensation to avoid loss in low frequencies
- Cut trace between pin 4 and 5 in 4052
- Added 33k resistor between pin 4 and 5 in 4052
This gives the SMR4x the following modes:
- 4 pole low pass with boost’ed resonance compensation
- 2 pole low pass
- 4 pole low pass with original resonance compensation (as in SMR4)
Attached are some photos of the modified SMR4x.
That’s awesome!! Can you post some sound samples?
I was thinking of modding a 4P to be a multimode like the 4PM on the Shruthi.
I can post some samples later.
It’s funny you mention a 4P modification. I was just thinking about a modified Ambika 4P card with the following modes from Shruthi 4PM: LP4 liquid, LP2 liquid, LP2 ms. The odd thing is that on the original MI designs the resonance feedback is always from the 4th pole. I guess that’s why I like 2pole modes so much - all that bold resonance.
That’s the one thing I’d like to change with my modded SMR4x - now the 2P mode ‘correctly’ takes resonance feedback from 2nd pole.
Edit: I just might cut another trace and add a wire. Where will this end…
Here is an annoying bassline for comparing the original SMR4 with the three filter modes on the modified SMR4x.Note that the non-boosted LP4 of the SMR4x is equivalent to the original SMR4.
Edit: In each clip the resonance is gradually increased after the 1st repetition, and the attenuation of the passband (lower frequencies) can be heard on the non-boosted LP4’s.
SMR4_vs_SMR4x_Junos4.zip (3.3 MB)
And here is a pad sound with varying resonance comparing the original SMR4 and the three filter moeds on the modified SMR4x.
SMR4_vs_SRM4x_Junos5.zip (2.8 MB)