Recording the MT-32's digital output

[NewRisingSUn]

We all know that the MT-32's LA and reverb chips produce a digital output that is then converted to analog with a D/A converter before it goes to the audio out jacks. Now, to create better-sounding audio files, it should be possible to sort of "grab" the digital data from the PCB before it reaches the MT-32's DAC; that way, one could bypass both the MT-32's DAC and the recording sound card's ADC, resulting in better sound quality as well as not having to worry about proper levels.

Unfortunately I have no electronic experience, so I would like to ask any of you who have if this is something that could be done by an amateur (using instructions given to him) with inexpensive equipment, and if so, how?

[Glottis]

I'd also love to know more about this matter. Supposedly it can't be simple as hardwiring the digital stream at the right point, to a say DIY cinch output at the back of the module?

[Laust]

It's not that simple unfortunately. The MT-32 only has one mono DAC and utilizes it in an ingenious way. Somewhat simplified, the DAC is driven at high speed, first converting a sample from the left channel to analog, then a sample from the right channel, then the processed left channel from the reverb chip, then the right, etc. The service guide refers to it as "timesharing" and there are eight "timeslots" in all, four direct from the LA32 chip (two left, two right) and two from the reverb chip (one L one R) as well as two more which seemingly aren't used. In each timeslot (apart from the unused ones), the DAC decodes a single sample either from the LA32 chip or the reverb chip.

After the DAC, the MT-32 has six "sample and hold" circuits, each holding one of the analog decoded samples (eg. voltages in other words) described above. These six samples are then essentially mixed down to two (stereo), all in the analog stage of the decoding process. So at no point are you actually dealing with a digital stereo signal, which is what would have made life easy...

To capture the audio, you would need to read the DAC at 256KHz (it's being driven at eight times the real speed of 32KHz), then mix the resulting samples down to two and then send it to something suitable. The S/PDIF standard has a 32KHz sample rate and that would do nicely as output. For reading and mixing, maybe a fast microprocessor (Atmel or similar) could be used.

All perfectly possible, but not trivial.

[NewRisingSUn]

Thanks for the information. Actually, I don't need a S/PDIF signal, I just need to get the samples somehow into my PC (I can do any processing in necessary from there in software; I don't need the signal in real-time). What method and what equipment would I need?