Utides II – Tides 2018 clone by Jakplugg, reworking Mutable Instruments’ updated Tides to 8hp format.
Turns out this was a headscratcher, for a number of reasons.
It’s a fairly densely packed board in places, and though the 0402 sized passives of the MI original have been worked out to less eyesight-challenging 0603s, there are a few potential pitfalls.
Firstly, at the top left of the board, you’ll see the text “Don’t short”. They’re not kidding, unless I’m missing something, the (conductive) metal frame of the top potentiometer will sit on the power header connections if you let it. I filed the offending part of the lugs on that pot away.
Flashing, ST links and ST Disco boards
Secondly, my (genuine, non-clone version) ST-Link 2, which happily flashed Cara, Plaits, etc, wanted nothing to do with it. I built the power section first before populating the rest of the board, wired the relevant pins to the SWD connector (there’s no mini JTAG on this), and couldn’t connect.
Normally that indicates a problem with the soldering on the STM, so I reflowed it anyway, but still nothing.
I suddenly remembered reading somewhere (Amazingsynth.com?) that people were having trouble flashing the newer MI modules with ST Links, so I dug out a ST discovery board and tried that, which worked immediately without a hitch. Hmm.
Calibration (thanks again, Forest)
The calibration procedure documented by Forest Caver worked perfectly, all I had to change was the decimal point resolution in Excel (select the Offset and Scale output cells in the spreadsheet, click Format Cells and the Number tab) and not power cycle the module after flashing it with the inital, pre-calibration firmware (in factory calibration mode the three mode LEDS flash to indicate it’s ready. Power cycing it boots it in normal mode). I built the firmware in the Mutable Development environment but used ST Utility to flash, as my Virtual Machine won’t seem to talk to my Windows laptop’s USB ports. Whatever. Here’s the SWD hookup (powering the module while flashing, so pin 1 on the disco board – VDD – not connected).
On my Windows 10 laptop, the Mutable development environment runs fine apart from the USB problem, but I can’t suspend or halt it, because it won’t restart – to run a new session (or use any USB components again) I have to use Vagrant Destroy and Vagrant Up from scratch, otherwise it locks into the “Can’t Connect – retrying” loop that a number of people seem to have problems with on Windows. I know nothing of Linux beyond a couple of basic commands, so I’m just living with this workaround.
The build commands for the bootloader and main hex file:
make -f tides2/bootloader/makefile hex make -f tides2/makefile
Files are in the Build directory…remember to flash the bootloader first.
The only other thing I have in build notes is the LED polarity for the SMD LEDs that sit under the transparent pots for the attenuators. Cathode down for the red LEDs, cathode up for the green LEDs (on the back of the SMD LEDs, arrow towards the bottom of the board for the reds and top for the greens). The two leg, bi-colour, through-hole output LEDs go long leg to the right.
Human error again
So, calibration done, and everything looking OK initially – nice slow LFO waves showing on the scope. Then I noticed that the Audio Rate mode didn’t seem to be working at all, and turning the frequency up in the middle rate mode dropped the amplitude of the wave until it disappeared. So, once again, a problem to track down…made more difficult because I wasn’t familiar with the module yet anyway, and the layout of this clone is different.
Having checked the manual and a couple of Youtube videos, I could see the rest of the module functioned perfectly. An audio probe showed the audio signal was there, just REALLY quiet. As the modes all share the basic hardware, the DAC was obviously working and there wasn’t a problem with the STM soldering, so it had to be something a long way downstream.
Thinking about it, it was behaving as if there was a low pass filter on all the outputs – the slow LFO-like cycles got through fine, but the faster ones are progressively filtered out.
I narrowed the probability down to this part of the schematic, where four identical circuits with the two dual op amps feed the four outputs.
The bias reference voltage VREF_BIAS checked out OK (1.087v according to the schematic). Checking C37, I got a reading of nearly 500nf. Much larger than the 47 picofarads it’s supposed to be. Same problem with the other three 47pfs – turns out I’d mixed up .47ufs, and the larger caps were indeed filtering out any audio rate output completely.
I need to slow down and pay more attention. Replacing them with what they were supposed to be lead to a fully working module. Which was nice.
EDIT March 2021: Built another one.