sfcs ~ A-143-3 mods

I picked up a used A-143-3 to finish off my system with an array of simple manual modulators. I already have an A-143-4, but i will be letting it go - not only because of its hefty size, but primarily because the LFOs don't feature a sawtooth output. While it is possible to shape the triangle output into a sawtooth by feeding the square back to the unattenuated CV input, i don't want to be doing it each time while losing the square out. What intrigued me about the A-143-3 is that not only the sawtooth output is on board, it is also two times faster than the other outputs. This looks very similar to the working of a triangle-to-saw subcircuit i'm fairly familiar with and even used in my own VCO design.

The pecularity of the circuit is that the polarity of the resulting sawtooth can be switched by swapping out a single transistor. I found this out on my own while working on SFP47, because i prefer downwards sawtooths much more than the upwards ones, which i usually call 'ramps' instead. And while the A-143-3 comes with ramp outputs, i saw a chance to mod it so that two LFOs have ramps, and two have sawtoothes. Upon examining the board and retracing the schematic bit surrounding the transistor from it, it became apparent that it is indeed the same exact circuit, and swapping outt the NPN transistor BC547 for any general-purpose PNP tranaistor (2N3906 in my case) works like a charm.

I further discovered, however, that the resistors used in the tri-saw converter are far from being precision parts, and so, the resulting saw has one fin higher than the other. This can be helped by perfectly matching the two resistors - one going to the op-amp's inverting in from the triangle output source, and the other providing negative feedback around the said op-amp. Instead of matching them, though, i just replaced one with a trimmpot, and calibrated it so that the fins are the same exact size amplitude-wise.

Unfortunately, the LFOs themselves seem to provide somewhat asymmetric waveshapes, because every other fin of the sawtooth also ends up being just a tiny bit shorter, but this is not something that i can fix easily, it seems. Either way, a much more precise and downwards-going sawtooth is a huge improvement for this module, and below is an instruction about how you can do it yourself.

	On the bench is a freshly-received secondhand A-143-3. Today it is going to receive a modification that will make it even more useful than it already is!
	The module has a custom quad-header power ribbon. I'm going to be modding only two of the four LFOs, so i disconnect them from the ribbon. It's a bit tricky, but does not require excessive force. Be careful not to damage the ribbon, as making a new one will be a pain in the patookis.
	The pots come off with a bit of force and a lever using a plastic or wooden piece, if needed - not metal, as to not scratch the panel. A 10mm hex bit removes the nut holding the pot, while an 11mm one is fit for the jack nuts. The three jacks of each LFO are packed very closely and a garden variety 11mm bit does not fit, so the middle one has to be loosened with pliers and unscrewed manually. This opens up some space for the 11mm bit to take the other two.
	This is the transistor we need to take out, the BC547. I use my tweezers and stick them between the transistor's legs. Then i turn the board over, apply some flux to the three solder joints, take an additional solder blob to the tip of my iron and start heating up and liquifying the three pads. When all three are good, the transistor will be pulled out by the force constantly and gently applyied downwards by the hand to the tweezers.
	After the transistor is popped out, prepare the holes for the replacement, and the replacement itself. One can use a solder suction pump, but i'm a bum and a punk, so i just heat up the joint and then exhale a short firm blow onto it while facing a trash can; the excess solder just flies into the trashbag! 2N3906 is a fitting replacement, although any regular PNP transistor should be fine.
	The collector and emitter of the replacement transistor should land exactly where the collector and emitter of the original one were! In my case, this means violating the silkscreen and installing the 2n3906 as shown. Compare BC547 and your replacement before fitting the latter in! This concludes inverting the ramp's polarity into a sawtooth. You can test it by connecting one individual board to a eurorack system with a standard 16 to 10 pin power ribbon; red goes towards the near edge of the board!
	You may notice the LFO being heavily asymmetric when modulating things like VCO pitch with it. To remedy that, we're going to remove the resistor R10 as shown, and install a trimm potentiometer instead of it.
	R10 is 100K, hence the other resistor we have to match is also 100K. A fixed 82K resistor and a 50K trimmer in series would be ideal. But i'm not chasing ideals, so i prepare a 200K trimmer alone to allow going considerably above and below the ideal target value, matching the real resistor somewhere on the way. Note that the trimmer's wiper (middle leg) is bent and soldered to one of the edge legs. It doesn't matter which. Then the edge legs are bent so that they match the leg span of the freshly desoldered R10.
	R10 is replaced with a trimmer. Now it is time to plug the LFO's sawtooth output to some VCO and monitor the result. The trimmer should be tuned so that the fins of the sawtooth produce as identical a "peww" from a VCO as possible. This will completely remedy the difference in amplitude. However, minor difference in length may be preserved; i will look into ways of fixing that later.
	The original parts are stored away in a zip-lock in case de-modding will be required, e.g. if i sell the unit and the other party does not want the mod.
	Done! Small dots are placed next to the ramp drawings to signify that they are customized.