dsi imaging


Human Connectome Project white matter fibers, via diffusion spectral imaging.

we had a neurosurgeon come in to talk to us about the brain. my take on neurosurgeons as lecturers is that they either are absolutely, irrefutably, incontrovertibly AWESOME-I-think-I-just-fell-in-love-please-marry-me-also-I-have-crippling-debt, or they are as dry as the purported wit I am known to shit out on a daily basis. or is it just the shit? either way is not completely true, actually, because my wit is hiLARious. and these neurosurg lecturers are very not. 

they do afford me the great opportunity to craft this skill I’ve been working on, which is falling asleep with my eyes wide open. hasn’t been working so well. although I HAVE managed to smoothly execute a ‘nodding off and violently pitching forward to oh-HAY-I’m-just-checking-my-shoelace/scratching my foot/adjusting my flats' manoeuvre.

fucking beautiful. 

as are these brain images (see what I did there? brought the post back 'round full-circle. genius). my alpha-waves-eyes-wide-open state was briefly interrupted when these babies appeared on our ridiculously large lecture screen. granted, resolution is kind of iffy, and there are things that definitely don’t belong in places that may or may not exist, but sooo puuurteh. and so many colours!

SCIENCE. making the world a more beautiful place, one tenuous-fibre-tract-imaging-study at a time. 

A quick tour of the Prophet 6 voice card

I’m in the process of designing a voice card for my polysynth project. It includes the oscillators, filter, amplifier and envelope generators. With the board laid out I faced the chore of figuring out how to connect it to the rest of the synth. I had originally planned to use a PCI or PCIe board edge connector as they are cheap and widely available thanks to their extensive use in PC expansion slots, but they seemed a little bulky and the edge ideally needed to be chamfered. I took a peek inside a Dave Smith Instruments Prophet 6 and found an ideal solution. The Prophet 6 has 6 discrete analogue voice cards in a 240 pin DDR3 DIMM format, another PC industry card edge standard with inexpensive and widely available sockets. Here’s a quick tour of the Prophet 6 voice card:

First a bit about the Prophet 6′s architecture: Each analog voice has 2 VCOs (one with a sub-oscillator) with continuously variable waveforms, a high pass filter, a low pass filter and a VCA. As with every Dave Smith design since the Prophet 600, it is almost certain that the envelope, LFO and other modulation control voltages are generated in software.

This is the ‘front’ of the card. It’s primarily populated with TL064 and TL062 op amps and 3 Cool Audio 2164 quad VCAs (12 total) plus a pair of PA381 (dual?) VCAs. The latter VCAs are a Curtis design not widely available. A through-hole version makes an appearance in the Prophet ‘08.  With the exception of the main VCA in the signal path, all of these VCAs are probably there for modulation depth control of the oscillators (pitch, pulse width) and filters (cutoff, resonance) as well as mixing the oscillator waveshapes.

The back of the PCB is where all the analog sound generation business takes place. On the left are the pair of VCOs, and on the right, the filters.

These are the oscillators.They look like a fairly standard saw core integrator with reset bolted to an exponential converter and waveshapers. I can’t say for sure without probing around but here’s my best guess:

 The first expo converter is on the bottom left. Q1 is a matched 2N3904 pair and just above it is a green 0402 3300ppm/°C tempco resistor.  Interestingly, the pair of op amps used in the expo converter are OPA2180 precision ‘zero’ drift (0.1 µV/°C). These guys are 4-5X the price of a TL064 and their inclusion here is likely key to the Prophet 6′s tuning stability over temperature and time.

Q2 is a MMBF4091 N-channel JFET. I suspect it’s the reset switch for the saw core integrator. C2 is probably the timing cap. It is noteworthy that the integrator appears to use a general purpose TL06x op amp, whereas other saw core designs often call for something more specialized like a CA3140. At first glance U4 looks like an LM311 comparator, which is standard fare in saw core VCOs. If it is, then the pin 7 output is grounded, which would be a deviation from canon.  It’s not clear to me how the sub-oscillator or low frequency mode of VCO2 is implemented.  The sync between VCO 2 and 1 is likely handled by the 74HCT4053 switch on the other side of the board. The 74HC4053 triple 2 channel multiplexer on the other side of the board is a bit of a mystery. It might be serving as a sync switch, but I’m suspicious there’s more going on, possibly related to the integrator reset and/or VCO2′s low frequency mode.

This is the filter section. Of special note here is the lack of an integrated Curtis filter like the one in the Prophet ‘08. Instead, the low pass filter looks to be implemented discretely with matched NPN and PNP transistor pairs, clearly visible as 4 stages. Is the high pass filter in there as well? And what is that LM13700 dual OTA doing? Perhaps it is handling voltage controlled feedback to implement resonance control for the two filters.

This voice card won’t reveal all of its secrets without a whole lot of probing around but it’s time for it to nuzzle up with its 5 brethren again and make some music…

P.S. Here are the higher resolution images.