We got our first Alesis Andromeda A6 last week, with a voice which passes it’s own internal calibrations but which is obviously out of tune afterwards, the client had pre-ordered 3 of the VCO ASIC’s in advance so we thought that all the bases were covered, a replacement chip and a couple of spares.
These chips were not cheap, even direct from Alesis so expectations were high of a satisfactory result from this. The IC’s themselves are surface mounted quad flat-packs which inherently carries it’s own difficulties. Lifting a dual sided SMD chip requires heating of one side of the chip until all the pins break free of their pads, then proceed to the other side to lift the chip out cleanly, but with quad packages this cannot be done without access to some very expensive tooling.
We actually carefully cut out all the pins of the VCO ASIC on the faulty voice (which is one option) and then carefully removed all the IC pins, the idea being that this would stress the PCB the least, and their delicate tracks, and this turned out to be the case, a nice clean removal and no pad damage. Lining up a QFPP with 64 pins is not a trivial matter either, the pins are only a millimetre apart and over four quadrants this isn’t easy, but the new chip went in, and after a long time under a magnifying light (which really wrecks your eyesight) we were happy to go for a test. This seemed to go well, the errant voice was now tuning every aspect and was staying there, all looked good apart from an adjacent voice which was always a semitone flat on Osc 1 and even more wayward on Osc 2 despite passing the Autotune routine. Repeated Autotuning showed this to be more than a transient abberation, this voice was faulty too in a similar way. This time we tried a different technique for chip removal, using very thin wire under the pins to lift out one side of the chip at a time while it was being heated with a temperature controlled Hot-Air rework tool. We did some rehearsals in advance on some old DVD drive and Hard drive circuit boards to get the temperatures, timing and techniques as perfect as possible for this form of IC, and then we went for it, and thankfully it went well, a clean chip lift and no damage to the board. Putting another one of the new chips in went well except that it didn’t work at all. no output whatsoever, the tuning sequence struggled with it and then gave up disabling the voice.
We checked and rechecked all of the connections and then gave up on it and next day installed the last of the three Osc chips into it. This seemed to work fine, and we regarded the synth as good to go, and indeed it was collected and went back to it’s owner, who after resetting the tuning tables started to run into a litany of problems with both of the new IC’s installed, and indeed a different voice has started to be problematic ( each oscillator ASIC handles two voices as does each Filter ASIC). Alesis UK have offered to send another two Osc ASIC’s free of charge to the client, but our main worry is about the number of times we can replace these chips before the pcb gets terminally damaged. This synth would seem to be a work in progress, and this is a shame, as what was done should have been enough to sort it completely, but the vagaries of the Osc chips make this a difficult problem.
The Tektronix Conundrum
As mentioned in a previous blog, we bought a Tektronix TDS420 scope for dealing with the Digital side of synth problem analysis, it has four channels, 100 MHz bandwidth, and lots of groovy DSP functions for Fourier analysis if necessary but also does funky things like reading out fairly accurately the frequency of the incoming signal, previously if we wanted to know if a CPU clock was correct, we would have to connect the Frequency counter but the Tek would onscreen give you a sufficient answer immediately.
Like the old Queen song Save Me says, “It started out so well” when we bought it, it passed all it’s boot diagnostics, which is why we bought it, but within a week it was starting to fail it’s front panel CPU diag’s and not much later on the Aquisition module started to randomly fail. Ironically perhaps we bought it to tide over a major rebuild of our Analog scope, a Fluke 3082 whose front panel controls had become totally shot, and indeed the Tek did tide us over this transition period while we rebuilt the Fluke.
Ironically the Fluke turned out to be a lot easier to repair than we expected, the switch contacts which looked like they should have been Gold plated turned out to have just been so dirty that they looked that way, scrubbing with Isopropanol brought them back to a nice silver colour. The weird surprise was that a lot of the Encoders on the front panel didn’t work either, and the reason was failed SMD diodes attached to them, about 9 if memory serves, and these were replaced with 4148 SMD equivalents which worked fine, the Fluke was now better than it had ever been, but it is a Digitally controlled Analog scope, not so dissimilar from most of the synths that pass through here, it is programmable, it has presets, and you can save your favourite settings. That being dealt with, and a great number of synths in the interim, we got back to the Tek for another try at the problem.
The Service Manual is available online but is largely hopeless as it doesn’t contain circuit diagrams and just recommends changing this board or that assuming that spares are going to be available for a 20 year old Scope, which they are not. We found the killer bug and it does have relevance to the modern synth world. as it seems to be all the Electrolytic SMD power supply capacitors on both of the failing boards, they leak something akin to battery acid across these tightly packed boards, and it does similar damage to the components and pcb tracks in their vicinity. This is Tektronix and they never knowingly used cheap components in their products, but here we are cleaning up noxious fluids fron the pcb’s and replacing about 50 capacitors and using Kynar wire to rebuild burnt out tracks on both of the failing pcb’s, and a few IC’s also, thankfully nothing that cannot be obtained locally.
About 50% of the capacitors measured were absolutely faulty and as for the rest we are dubious, the new replacements give far better readings. This Scope is from 1993 which is 20 years ago, how many synth time bombs are there waiting to expire for the same reason.