Locating PCM sample chip

[tonewill]

Hello,
I have an old Technics K200 home keyboard from the early 80s. It was one of the first to use samples for the drum sounds. I don't have a service manual but am trying to locate the chip(s) that hold the samples.

Does anyone have any hints as to what to look for in a device of this age? Likely size of chip, number of pins, possible types or model numbers used in this era?

I'm just playing with the idea of reading the chip with a rom reader / writer (don't have one yet). I've seen examples of this with other synths and drum machines on the web, but I realise it's a long shot.

Anyway, if anyone can help me locate the chip(s) that would be great.

Many thanks,
Barry.

[sizone]

Sorry to crush your dreams bud, but the only place you're likely to find a replacement is out of another k200.
Getting your own samples into the thing falls somewhere between being a pipe dream and a MASSIVE expenditure of time and resources. The reason people flash their own samples for their Linndrums and stuff is because they were designed with that in mind.
To make new hardware to access the samples on the chip would require reverse engineering the entire sample handling scheme (compression, transposition, ect.) and you won't find details of that anywhere.
That being said, take the case off and look at the boards. There will probably be one large chip with pins on all four sides, that's the cpu. There will probably be 2 smaller chips (20-40 pins) with pins on two sides, those are the ram chips. Finally, there will be 1 or 2 chips that look more or less identical to the ram chips, those are the roms the samples are stored in (and all the cpu instructions and whatever other preset info the synth uses). Look up the character strings printed on the chips on google. The string closest to the top of the chip is usually useless manufacturer data (presumably things like date, faculty). The string of characters printed in the middle is usually the chip designation. That's what you want to search for. It's highly unlikely you'll find a data sheet for the exact chip, but by leaving sections of characters out you can usually find some info for similar chips in the family line. That way you'll at least have some idea of what chip is doing what.

[tonewill]

Hello sizone, thanks very much for taking the time to answer.
I know that there's very little chance of being able to do this but I must try or it will niggle away in the back of my mind.
Anyway, I opened up the case and there are 3 chips with pins on all sides (click to enlarge):



The most promising chips I think are these (click to enlarge):



NEC D80C49C 061
MSM83C55-17RS

The rest of the chips are small 8 or 16 pin types. Will do a google search. Could it be one of these?

Barry.

[sizone]

http://www.cpu-world.com/info/Pinouts/8049.html

1st of the dual's looks to be a microprocessor (a really common one at that)
(great fun to circuit bend if nothing more, ship it out to droffset if you don't want it)

2nd of the dual's looks also to be a microprocessor

http://www.utsource.net/ic-datasheet/83C55-38-603501.html

not finding anything on the quads, but the fact that the first two are identical leads me to believe they're your ram chips (by inference, the last on the right would stand to be the sample rom but I've been wrong on my speculation so far). Technic's ic choice is, like, the anti-Casio.

You may as well document (and research) the 8 (mostly opamps) and 16 (mostly logic) pin numbers too, it's kind of informative to figure out what's in a synth. Ones that don't turn up right away in searches should be regarded as potential roms.

If you decide to pray to Apollo for wisdom and insight, remember that blood sacrifices are generally well regarded.

Good luck. You'll probably learn a lot through your efforts even if you don't come anywhere near getting what you wanted out of it.

[tonewill]

Hello sizone,
I must just say that I know very little about electronics, though you've probably gathered that. Also, I decided against the blood sacrifice and had a nice cup of tea instead.

So, just to make sure I've got it: The large chips are CPUs and are not likely to hold the sample data.

The furthest to the right of the 3 chips could be a ROM chip holding the sample data. If it is, then I suppose it's unlikely that there will be an EPROM reader that will read a chip of that format (is that right?).

It is possible that one of the smaller 8/16 pin chips could hold the sample data. Again, unlikely to be able to be read by an EPROM reader I suppose.

If none of those types of chips can be read by a reader then I'll call it a day (unless you tell me otherwise). I knew it was a long shot but it was worth trying. Thanks again for all your help,

Barry.

[sizone]

Ahhh, you just want to dump the sample data (presumably something like what Rainer Butchy did with the SQ-80). What you'd end up with wouldn't be something directly usable like .wav format samples. The info would be samples, but encoded in whatever proprietary format technic used at the time, you'd basically need to write software emulation for the entire rest of the synth to do anything with them (similar to how you need a console emulator to do anything with the rom dump of a video game cartridge).
Please, for the sake of your own sanity, just connect it to your sound card and record the drum samples.
Your initial explorations were not totally in vain. This looks like it would make a great circuit bending project (common and fairly tough-to-destroy processor for starters), though maybe not so much a great first circuit bending project. You've got a potential gold mine of crazy sounds in front of you. Getting the case open is the first step and you figured that out.

[tonewill]

Hello sizone,

[sizone]

http://en.wikipedia.org/wiki/Audio_format

.aiff (1985)
.wav (1992 !)

Your keyboard existed before any modern encoding formats. Not that it really matters, ummm sorry, I just woke up and can't really explain this clearly right now, so for the time being take this as your mantra "it just doesn't work that way".

If the drums are analog, then they're being generated separately and it's just a matter of finding their outputs before they get mixed together. This is a good beginner project, the odds of damaging your synth or yourself in the process are -phenomenonly- low.
This is the cheap and easy way to do it:
Take a cheap mono audio cable (1/4'' or 1/8'' depending on what kind of jack is on your mixer or soundcard or whatever it is you're sampling into)
Cut off one of the plugs. There should be two wires (one "cool" colored one usually black and one "hot" colored one usually red) or one plastic coated wire and one bare wire. There may also be some foil shielding. Strip off a foot or two of the plastic insulation (the outer most plastic), then strip off a 1/2 cm or so of the plastic insulation around the two inner wires.
You can do the rest without a soldering iron, but having one makes it much easier.
With iron, tin both the red and black wires (to keep from damaging your audio gear it may also be a good idea to solder a 100k or so resistor to the red lead), solder the black wire to ground. Start a rhythm going (either use the internal speaker if it has one, or plug the normal line-out into your mixer, basically make sure that it's working). Then start probing with the red wire. Touch it to briefly to various pins and components until you find the signals you're looking for. On one hand, you want to turn your amp way down to keep from damaging things (like your ears), but on the other hand the unmixed signals are also going to be unamplified. Doing stuff like this isn't really teachable. You kind of just have to do it. Don't worry, you seem like a sharp guy and it's not that difficult. I think the biggest impediments are going to be doubt and fear.
Without iron, basically the same procedure but you'll need an alligator clip with a screw down contact.....I'll just assume you have an iron and that you know which end to hold.
Before you start the undertaking, maybe take a big, decent resolution photo of the board and post it so I can at least circle the areas of it you probably don't want to be probing.
Of coarse, if the drums are sampled then they're not going to be mixed down at a component level, they don't exist anywhere as separate signals.
Easy test to determine whether the rhythm section is analog or digital:
Trimmer pots, see any components on the board that look like they're meant to be adjusted with a small flat-head screw driver? Those are trimmers, pretty much every synth has at least one or two in it. If the drums are analog then there will be a lot more.

Again, good luck.

[tonewill]

Thanks a lot for the info sizone.

I can say for sure that the drums are samples, but the hi-hats are analog. They have that typical analog hi-hat sound. Also, on the you-tube video I mentioned, the guy messed about with the length of the hi-hat sound.

Soldering's not a problem so I'll try your experiments. As for the boards, there a several. One is marked 'main' and has trimmers for vibrato rate and speed. I already tried tweaking these and proved that this board is for the main poly voices. Attached to this is the small board with the 3 chips I posted earlier, and Technics own chip.

Then attached to the main board with stand-offs is a board marked 'solo' and it does control the solo voices. It has a trimmer for vibrato speed and also cut-off frequency, which would be good to replace with a proper pot perhaps.

Then there is an effects board which must control the chorus/celeste effect. It has a trimmer marked BBD bias which I assume is bucket-brigade device bias, though this doesn't do anything useful, rather the sound disappears either side of the sweet spot.

Then there is the board with the CPU chips I posted earlier.

I'd never heard of circuit bending until a few days ago. I must admit that some of the weird sounds that I've heard people make with their gear doesn't do it for me. But some of the more musical tweaks are useful.

Anyway, It's a bit late here now so I'll look at this again over the weekend. Thanks again,
Barry.

[droffset]

Hello !
I'll do a post about a really old keyboard I'm dissecting. It might help you determine where you can go with bending yours.

You probably won't have much luck extracting the samples themselves, BUT you might be able to interrupt or replace the command signals that trigger each instrument.

From the microprocessor to the soundchip there are likely 8 connections.
For preset rhythms each drum voice is triggered by a different dataline, if you leave a rhythm running and look at each of these through an oscilloscope or even straight to a simple audio amp you'll be able to observe the small blip that synchs with each instrument. Maybe connect those lines to on/off switches. Want to remove the hihat? Just knock that command line out.

Edit,
Oh yeah, as Sizone mentioned, out of the soundchip will be several audio outputs that will get mixed together by another section of the pcb.

You could leave a preset running and use/make a simple 386 audio amp (that you don't mind if it dies), connect it's ground to your keyboard's and try probing various connections of the keyboard to isolate/hear what the individual audio outs are.

Edit:
I posted without really reading the whole thread, if Sizone already said some of this stuff, sorry.

Hope that helps, your keyboard appears to be a lot more complex the one I'm playing with.

[tonewill]

First of all, here are some pics of the various boards.

Whole thing opened (click to enlarge):



Board with the CPU on (click to enlarge):



Main board with 'solo presets' (mono synth) board on top (click to enlarge):



Also attached on top of the main board is the little green board with the 3 chips posted earlier. Just below this you will see the word 'FRONT'. There is a Technics chip similar to the other 3 chips under this. At the top left of the picture is part of the effects board.

Barry.

[tonewill]

[sizone]

The "front" board is definitely the brain, the "technics" labeled chip is the main processor and I'm pretty sure now that the 3 chips on the other side of it are ram and rom. The other green board close to the keyboard is the other digital board, I assume it's responsible for parsing the keyboard and front panel inputs. Usually with Casios the rhythm and accompaniment is handled by a secondary processor section like that (while the main processor does the main voice). So it's, again, my guess that that board is also responsible for generating the sequence patterns. You've got a weird mix of analog and digital circuitry in front of you (similar to the Casio MT-400v). Looks like the white board next to the speaker is where the amplification, signal mixing and chorus effects are occurring. The other white board and the brown board underneath it are, I'm guessing, the synth sections.
Offset probably has the better idea, try to find the trigger signals for the various drum sounds. If you don't have a scope (don't feel bad, neither do I) you should be able to accomplish this with a good digital multimeter. I'd start with the green board closer to the keyboard. Get a note pad and a pen out. Start a preset playing and listen for the hihat. Set your meter to measure voltage (sensitivity should be such that pulses in the 1-5v range
are easy to see). Probe one pin at a time (for both of the microprocessors) and keep an eye on your meter. What you're looking for is a spike in voltage on a pin that corresponds with the hihat being triggered. If you don't find it, carefully flip your "front" board over and start probing the technics chip. Once you've found your hihat triggering pin, suppressing the signal should just be a matter of running a lead from that pin straight to ground (maybe wait for offset to chime in on this, he may have a more elegant way to do it). You can put a switch in to turn it on and off. You can likewise do the same for the rest of the drum sounds. The more thorough and clear you are in documenting your probings, the easier the work will be.

[droffset]

Good strategy there. Something I would add (on a bit of a tangent) is to look for the ICs nearest to the audio outputs, since those are likely to be opamps for mixing signals to go out, and are more likely to have datasheets available. If you can find the mixers you can find the inputs and stick audio probes in there to find the signal you want.

[tonewill]

I've tried probing the 2 CPU (large chips) as you suggested with my multi-meter set to DC voltage and the rhythm running but found nothing that would indicate a trigger of any of the drums.

However, I removed the small board with the Techincs chip on, switched the keyboard back on and everything works minus the PCM drum sounds! The hi-hat still works but then I know that the hi-hat is analog and that all other drum sounds are sampled. Then I noticed that the board has 'RHY' in brackets written on it —rhythm! The other boards also have similar braketed names (SOLO), (EFF), (SAS), (MAIN). So, unless I'm mistaken, this must be the chip with the samples in it.

Anyway, I tried probing the pins with a rhythm playing but there was no fluctuation in voltage. However, is it not the case that with sampled drum sounds a different memory address is read to access each sound rather than each sound being triggered on a different pin? Only guessing here.

Here is the chip:


Anyway, I'm a bit stuck now as to what to do.
Barry.

[sizone]

You tried working from the inside out, now it's time to try from the outside in. Get your audio probe set up, poke around where you suspect the hihat is being generated from (somewhere on the rhythm board?). Once you find find it, there are all sorts of things you can do to shut it off.
Generally you're right, it makes more sense for samples to be triggered by decoding a single signal, but there are a lot of reasons for designers (particularly around this time period) to use separate lines (simplicity, user programmability, hold-over from analog rhythm sections where every sounds comes from a bit of discreet circuitry). Don't give up hope just yet.
And good work on your progress so far.

[droffset]

Cool find!
Yup, time for the audio probe. Start at the keyboard's output and work your way back through the circuit to work out where the audio signals get mixed.
.
Sizone your multimeter approach is heaps more elegant than what i would have come up with, heh. Very methodical. I just break out the little amplifier and listen for blips. What can I say I like probing things....that didn't sound right...
.
re: supressing the command, my way wouldn't be all that elegant, short it to something else or cut it and hook up a toggle switch to either side maybe? Check out what Burnkit2600 does with yamaha pss-270 s.
.
Other ways to look at it:
Examine your rhythm chip there and work out which connections are coming in and which are going out. Use the audio probe on the outputs. Try shorting various inputs to each other while a rhythm is going.

Trace the inputs even further back to see where they come from, might be the 8049, maybe you'll find other lines that come from the same place but go somewhere else. Probe them. Listen for a blip that synchs with the hihat.
.
I suspect your pcbs are double sided, which could make understanding the paths a bit trickier, but not impossible.
.
Your keyboard is about 37 times more complex than the ones I'm playing with, so these are just ideas.
.
Fun, I'm learning a lot too, heh.

[tonewill]

Sorry, I forgot to reply.
In the end I managed to find a short that cut out the hi-hat and allowed me to get most of the samples individually. I didn't want to take out the boards and trace the circuit just for this. Besides, I'm convinced that the individual drum samples are combined in the digital domain rather than being separate anywhere where an audio tap can be made, so I'll just leave it at that for now.
Thanks for al the help,
Barry.

[CYBERYOGI =CO=Windler]

I dumped the ROMs of Technics SX-K250 and K350 (see attachment).

The K350 main eprom seems to be used also in K450 and related models. I remember that the internal ROM of the solo voice card CPU (an Intel 8048 (MCS48) variant) reads garbage and thus is undumpable, but I got that sample ROM (K250 and K350 solo board are (beside minor layout changes) identical). I think the drums are inside an SMD IC (K250 had 3 on a small daughterboard) which pinout is hard to investigate.

I now got a Technics SX-K200 but haven't looked inside yet.

---

Technics K250,K350 roms.zip
Description:	Technics SX-K250 and SX-K350 ROM dumps.	Download
Filename:	Technics K250,K350 roms.zip
Filesize:	36.81 KB
Downloaded:	390 Time(s)

---