MT-70 Analysis and (hopeful!) repair

[Ra226]

Hi, all.  New to the forum--I recently purchased an MT-70 which unfortunately is not completely working.  I was told it was being sold as is for parts, so fair warning... but I'd hoped it'd be easier to fix.  Turns out I have a bad LSI-2 chip.  I was hoping it would be any other part, but I'm afraid it's not.  So the short version of this post is: anyone got a spare LSI-2 chip from an MT-70 or ideas where I might get one?  My only thought is to wait for another to show up for parts (but then I'll be stuck with an extra pile of not-working MT-70).

 

The long version of this post is that I actually had a lot of fun analyzing this keyboard and hoped that my work might be useful to anyone else attempting to fix one of these.  My symptom is the first through fourth notes played sound like garbage.  The following three sound fine, then the pattern repeats.  Or, if in chord mode, the accompaniment sounds fine (bass and chords/arpeggio), and any other notes played do not.  This problem is similar to a few mentioned in the CT 701 thread in this same forum.

 

My initial guess was, due to the way it sounded, that the sample and hold circuit was bad, failing to to elliminate the noise.  Before taking a closer look, I figured this would be an electrolytic cap which can have issues.  As it turns out, those caps are tiny 0.01 uF ceramics and this was not the problem.  And also, the sound before the S&H circuit doesn't really sound all that different.  Working my way backward, I next had a look at the DAC.  I really doubted this would be the issue since it's just a resistor network--sure enough, it was not.  Having removed the DAC, that left the pins on the chip itself open, which turned out to be the problem.  Under oscilloscope, LSI-1 has nice, square outputs.  LSI-2 does not.

 

Attached to this post are several pictures of the analog board (where I originally thought the problem was).  I've mirrored, resized, and overlayed the back of the board on top of the front so you can easily trace signals.  I have not been able to find any kind of service manual or schematic for this keyboard.  There is a CT-701 "Textbook" document out there, which was very helpful, and the CZ-101 service manual does have several circuits in common with the MT-70.

 

Other discoveries I made about this keyboard as I went through analyzing it:

• LSI-1 is "Master."  It provides bass/chords when needed.
• LSI-1 and LSI-2 have different suffixes on my specimen.  LSI-2 is 43617 13, LSI-1 is 43617 23.  Maybe they're not identical?  (I'm guessing some ROM code?)
• DAS2 not used in MT-70
• DAS circuit much simplified compared to CT-701.  No discrete transistors, only NOR and bilateral switches.
• D/A Shift is really "Expander" (see CZ-1000 docs).  Increases dynamic range
• D/A Shift and S/H circuits somewhat combined.  One dual Op-Amp accomplishes both.
• White Noise source is quite different from CT-701.  Op-Amp-based source serves high-hats and snare
• Snare signal runs through 555 timer.  Presumably monostable, to lengthen pulse for noise portion
• Drum signals are different than CT-701.  No congas, and use different lines.  LSI's from CT-701 likely not compatible with MT-70.
• Voltage regulator is just plain old power transistor.  Warms up a bit, rather inefficient.
• Each chip has a 10-bit DAC despite being 12-bit, with lowest two bits made from discrete resistors
• DAC is typical R-2R, with R-25k.  The LSB does not have a 2R, though, so it was designed to have more bits added onto it.

 

The green arrows trace the raw LSI-2 analog audio signal from the digital board to the Expander, S&H, and into the audio mix.  LSI-1 audio (not shown) follows a similar track, but optionally gets routed through the accompaniment system (when in chord mode) so the accomp volume dial can control it along with the percussion.  Both are combined in Master Mix, then sent to the main volume dial, then to the power amp.

 

So I hope this is useful to someone out there--and if anyone can hook me up with an MT-70 LSI-2, I'd be grateful!

 

 

 

[Chas]

Sorry to hear about your poorly MT70, but thank you for your very useful information and analysis. 

 

As to where to find an LSi-2 chip, that's not going to be easy. Firstly, because it's a custom chip unique to Casiotones, and secondly because Casio only used this type of synthesis (Sine Wave based) in a few models compared with the much more common Consonant Vowel synthesis used in the majority of early to mid 80s Casios. 

 

The  Casiotone models that used Sine Wave synthesis are: CT701/ 601 and 501 (with the 501 being pretty much a full sized version of the MT70 with identical feature set), the MT70 and the CT1000P. To my knowledge, that's it.

 

From what I've observed, the MT70 is by far the most common model, followed by the CT701, then the CT1000P. The CT601 and 501 appear to be the rarest. 

 

It's probably best to look for a tatty/ physically damaged but working MT70 and make one good keyboard out of the two. An MT70 is likely to be found at affordable prices, plus the parts will be guaranteed to be compatible. The other models, with the exception of the CT501, may have differences that could lead to incompatibility with the MT70, though I'm not certain. 

 

Thanks again for the information you discovered and shared,  and I hope you manage to get your MT70 fixed and working fully. Keep us posted with any developments!

[CYBERYOGI =CO=Windler]

AFAIK the analogue sound output is NOT multiplexed at all, but one sound chip feeds its digital audio into the next one, so they get digitally summed and finally output by the last soundchip in that chain through a single resistor ladder DAC.

 

If only every n-th note is audible, likely one of the soundchips is dead or a pin disconnected. Their digital audio is daisychained from chip to chip instead of using separate DACs, which may complicate fault search.

 

See here:

https://www.casiomusicforums.com/index.php?/topic/18892-ct-701/page/3/#comment-74571

 

I never heard that Casio sinewave chips have software numbers (they are a bunch of gates summing sinewaves). Likely these were rather production dates. The main software is inside the CPU "NEC D7802G 038" (64 pin with strange zigzag layout).

Edited June 9, 2022 by CYBERYOGI =CO=Windler

[Macola]

I am sorry to piggyback your tread but I see that you did some research with this model so I have a question.

 

Every time I power it on it turns on different setup randomly. Everything is scrambled but every time in a different way and no keys are responding. Sometimes sound is played and some rhythm played for a second and different led lightened. 

 

Considering that it wasn't used for years, possibly decades I figure hard reset might help. As I saw on some page one guy has installed a hard reset button. Can you please tell me if you stumble upon this info which points on the mother board should be shortened for this hard reset.

Thanks.

[Ra226]

Hi @CYBERYOGI =CO=Windler

I was hoping you'd chime in here--I'm familiar with the thread you quoted, and their problem was very similar to mine.  I considered just adding my issue to it, but I don't believe the MT-70 works the same way.  On the MT-70, both LSI chips have a DAC, the outputs are sent to the dual-op amp (bottom center) then sent out the ribbon cable to the analog board.  Still, I'm pretty sure you're right about the chips being identical.  Keen On Keys (fantastic youtube channel) has torn down an MT-70 and his chips are identically labeled.  I could probably desolder both my chips and swap them to be absolutely sure, but that seems like a lot of work to still have a broken keyboard afterward!  I think Chas is right, I'll just have to wait for a suitably-priced broken one to show up.

 

Meanwhile, here are my photos of the digital board and my conceptual diagram of the Analog board.  Hope they're helpful to anyone trying to troubleshoot one of these.

 

 

 

 

[Ra226]

@Macola

I don't think I quite know enough about the digital board to answer that one.  There's a lot of good data on the CT-701, but it's different enough from the MT-70 that it's only so helpful.  I'd start with the CT-701 Textbook.  I found the CZ-101 service manual useful, too (it had several schematics), but mainly for the analog portions.  It might be the Schmitt circuit (section 7.4 in that document), but hard to say.  Also, feel free to piggy-back, I was hoping to just get more info on the MT-70 out there.

[IanB]

Fascinating and informative thread!

[CYBERYOGI =CO=Windler]

17 hours ago, Macola said:

I am sorry to piggyback your tread but I see that you did some research with this model so I have a question.

 

Every time I power it on it turns on different setup randomly. Everything is scrambled but every time in a different way and no keys are responding.

 

This sounds very much like a dead reset capacitor that needs to be replaced. But also lack of SRAM power (e.g. bad diode or corroded copper trace) may do such things. The keyboard contains no flash memory or such diseases, but persistent user memory (sequencer) is purely SRAM. I.e. removing batteries and power supply for a minute(?) should then make it reset itself.

Edited June 11, 2022 by CYBERYOGI =CO=Windler

[CYBERYOGI =CO=Windler]

8 hours ago, Ra226 said:

Hi @CYBERYOGI =CO=Windler

...

Meanwhile, here are my photos of the digital board and my conceptual diagram of the Analog board.  Hope they're helpful to anyone trying to troubleshoot one of these...

 

The 2 sound ICs play through different DACs during chord/accompaniment mode (so both can have different analogue filtering and volume control) but will use the same DAC to double main voice polyphony when chord is off.

[Ra226]

On 6/10/2022 at 11:05 PM, CYBERYOGI =CO=Windler said:

 

The 2 sound ICs play through different DACs during chord/accompaniment mode (so both can have different analogue filtering and volume control) but will use the same DAC to double main voice polyphony when chord is off.

If that were the case, wouldn't all the notes sound bad or all the notes sound good when in non-chord mode?  Also, I've definitely confirmed by listening to the outputs of both DACs that even in non-chord mode, both chips send their sounds analog, through their respective DACs.  I'm not sure why they didn't take advantage of that digital passthrough like the 701 does.

[CYBERYOGI =CO=Windler]

AFAIK one DAC output is not used (volume set zero through a 4066 or such?) so long chord mode is off. Each sound IC has to track in its registers where it currently is within a waveform (phase), which frequency is set etc., while the digital audio of the other is only passed through with less computation. So a broken register memory cell may not affect the rest.

 

According to the patent each chip is summing 5 sines (like drawbars) per waveform and also uses the sum of 2 sines to simulate multiplication to produce their volume envelopes. The concept looks very unintuitive and like Casio was already 75% on the way to phase distortion despite the resulting waveforms were still summed sines.

Edited June 13, 2022 by CYBERYOGI =CO=Windler

[Ra226]

Yes, the 4066 (at D4 on the Analog board) has two switches being used, always in opposite states (their inputs are TIM-0 or inverted TIM-0--line 7 on the ribbon cable).  They take in the audio from LSI-1 and route it to either the Op Amp right above in non-chord mode (where it's mixed with LSI-2's audio) or to the filter block, etc, where it's mixed with the percussion, sent to the Accomp volume control, etc...

 

I love what you mention about this chip--how it's almost like an alternative future that never was.  There's some advanced stuff going on and it really looks like they wanted this to be the future, but then they decided to go a different route.  What a great idea to look up the patent, I'll have to go find that, it sounds interesting.

[CYBERYOGI =CO=Windler]

The MT-70 was derived from the fullsize Casiotone 701, which has a 3rd HD43517 sound IC to support 8 note polyphonic play in chord mode, and a 3 digit 7 segment LED display instead of LCD. The general hardware architecture seems to be in US patent 4534257 (describing a CPU controlled keyboard that strongly resembles the 701, priority date 1981). The Casio barcode data format is described in the US patents 4422361 (general) and 4437378 (with key lighting). According to US patent 4464966 Casio also planned barcode programmable rhythm, which AFAIK was never released. The sound IC is explained in the Casiotone 1000P patent (US patent 4538495) and further details in US patent 4453440, which describes digital envelope control by adding phase shifted copies of the same sine wave to avoid multiplication, and how overtones are produced by a kind of phase distortion predecessor. In this patent Casio also describes how to mix the outputs of multiple sound ICs digitally, which was never done in any other 1980th Casio keyboards I am aware of. The sequencer is in US patent 4876938 (autoplay) which also mentions this sound IC. US patent 4622879 reveals the sequencer edit algorithm.

pinout HD43517

The Music LSI "Hitachi HD43517" (42 pin DIL) is a 4 note polyphonic sound IC based on additive synthesis, that became famous by the synthesizer Casiotone 1000p. Each polyphony channel is mixing 5 harmonics (sine waves) with different digital envelopes in a similar manner like a drawbar organ. The IC outputs 12 bit digital audio to an external DAC and (like the later D933) 3 additional highest DAC bits control an expander circuit (sort of fast switching VCA, US patent 4414878) to increase dynamic range. (Without expander the DAC waveform tip looks sunken in like a collapsed copula and sounds distorted.) A sample & hold circuit then removes high frequency components. Interesting is that the IC can route its output as digital audio into another HD43517 for digital mixing through thats DAC to simplify analogue wiring. Unfortunately this wastes another DAC bit, and with only about 25kHz output frequency (due to the additional summing of 5 overtones per channel) the specs and tone quality are inferior to the earlier D931C. Interesting is that the internal signal processing (US patent 4453440) already resembles phase distortion, using phase-shifted sine wave addition from a 12 bit lookup table for fast multiplication. Frequency- and phase data resolution for each overtone is 20 bit, with 7 bit envelope controls.

 

Multiple HD43517 (up to 3?, existing in Casiotone 701) can be controlled by one CPU on the same 8 bit data bus. Normally each sound IC outputs audio through its own resistor ladder DAC, but in master-slave mode they can output through the DAC of the master. For this the signals DAD (serial sound data), EVD (envelope data) and SYC (sync) are sent from slave to master sound IC. (The slave's DAC has to be muted externally -  e.g. by its sample & hold stage.) DAD and EVD are disconnected to select normal mode.

 

This pinout and description is based on the Casiotone 1000p service manual with its handwritten schematics and incomplete numberings. E.g. the order of DAD and EVD are impossible to see. Apparently all "O-#" pins are outputs, "I-#" are inputs and "IO-#" can be both.
 

pin	name	purpose
1	I-13	DB0 data bus
2	I-12	DB1 data bus
3	I-11	DB2 data bus
4	I-10	DB3 data bus
5	I-9	DB4 data bus
6	I-8	DB5 data bus
7	I-7	DB6 data bus
8	I-6	DB7 data bus
9	IO-1	DAD ?
10	IO-3	SYC (?)
11	IO-4	EVD ?
12	I-1	CS
13	I-2	/WE (write enable in)
14	I-3	/A/D (adress/data in)
15	O-19	(at 2nd sound IC wired to cpu INT1)
16	O-20	(at 2nd sound IC wired to cpu INT2)
17	I-4	(possibly chip id 2?)
18	I-5	chip id? (at 2nd sound IC wired to GND)
19	I-14	/reset
20	PG	clock in (4MHz)
21	Vss	supply voltage +5V
22	O-13
23	O-12	digital audio out (LSB)
24	O-11	digital audio out
25	O-10	digital audio out
26	O-9	digital audio out
27	O-8	digital audio out
28	O-7	digital audio out
29	O-6	digital audio out
30	O-5	digital audio out
31	O-4	digital audio out
32	O-3	digital audio out
33	O-2	digital audio out
34	O-1	digital audio out (MSB)
35	O-18	expander control out
36	O-17	expander control out
37	O-16	expander control out
38	IO-2	SH 25kHz out for sample & hold circuit
39	O-14
40	Vcc2	(wired to Vcc1)
41	Vcc1	GND
42	O-15

Edited June 14, 2022 by CYBERYOGI =CO=Windler

[Ra226]

According to CT-701 textbook, you are correct: 9, 10, and 11 are DAD, SYNC, and EVD, respectively.  That doc also defines pin 18 as MSO, Master/Slave select.  According to the CT-1000P Service Manual schematics, SYNC is directly connected between the chips, and DAD and EVD are connected through switches (4066 chip).  The diagram shows the Melody LSI as master and the Accomp (chords) LSI as slave, and data flows from slave to master.  The schematic shows MSO (pin 18) connected to Ground on the Accomp LSI, so I'm going to assume connecting MSO to ground is what determines slave (and therefore sets the data dirction on DAD/EVD).

 

On the MT-70, LSI-1 is the slave chip (MSO pin is connected to ground) but DAD/EVD are left open (see the photo of the digital board).  SYNC is still connected between the two.  On LSI-2, DAD/EVD are connected to Vcc via 47k resistors (probably just to keep noise from being treated as an input signal).  So for whatever reason, it looks like they elected not to use that digital channel on the MT-70.  Now I'm kind of curious what would happen if I connected DAD/EVD...

 

A workaround for my LSI-2 chip with bad outputs might be to swap the two chips, put the bad one in LSI-1's position, disconnect the LSI-1 DAC, then connect the DAD/EVD pins.  In theory, non-chord mode would work with 8 fully-functioning polyphonic notes instead of the 4 bad/3 good.  Chord mode would send the chords through the other chip and also work, but there'd be no way to control the volume on them.  I'm not sure I want to do that much soldering and de-soldering, but it's fun to think about.

[CYBERYOGI =CO=Windler]

I own 2 Casio MT-70 and thought about transplanting a 3rd HD43517 into one MT-70 to get 12 note polyphony. But yet I didn't try this (worry of breaking the LCD foil cable).

[TechnoSomnia]

I just got an MT-70 sold to me as working. However, the keys in the left side make no sound and just show the key on the digital display instead when pressed (normal?). And, the volume only works when just right, otherwise the sound is way too faint. If you're still looking for one to Frankenstein yours and mine sounds functional enough, you're welcome to get in touch. I don't know anything about these things and came across your post hoping to find an easy solution to mine's issues (no dice), and would like to see this go to somebody who can make it work. I didn't pay much for it and would pass it along for the same cost.

[Ra226]

Hi, all--I'm bringing this thread back because I've managed to fix the issue by doing exactly what I theorized earlier: removing LSI-1's DAC, swapping the chips, and connecting the serial lines.  The bad chip is now in LSI-1's position and does not output any audio via the audio out pins, sending it instead as data over the serial channel to LSI-2 (which is now the good chip).  The only drawback to this solution is the Accomp. Volume control no longer controls the chords/bass line (but does still control percussion).  I consider this to be pretty minor and am very happy to have a fully playable MT-70 now!

 

Problem

• If you have an MT-70 that's playing 3 or 4 good notes followed by 3 or 4 bad notes, it's possible you have a bad output stage on one of your LSI chips.  If chords sound bad, it's LSI-1 that's bad.  If chords sound good (which was my case), it's LSI-2 that's bad.

Solution

• Either way, remove the DAC from LSI-1 and disconnect the resistors at pins 23 and 24 on LSI-1.  Then:
  • If LSI-1 is bad: wire the two LSI Pin 9's to eachother, and the two Pin 11's to eachother
  • If LSI-2 is bad: swap the two LSI chips, then do the wires as above

See picture, and be careful of solder bridges.  I accidentally tied the CLOCK signal to 5V (pins 20 and 21) on one of my LSI's and the keyboard locked up on me.

 

[Zarquon]

New joiner here! Any ideas where these chips can be sourced, apart from the usual places (ebay/spares repairs etc)? I have a casio 1000P, with a distorted output on one of the IC sound chips (desoldered and swapped them over...), the other is fine. Are they all compatible - 601 /701 /MT-70 etc?

 

thankyou

[Chas]

The 701. 601, 501, MT70 and 1000P are all from the "Sine Wave" family of Casio models. They share similarities between them all, though they not 100% identical from what I've read.

 

The 701 and 601 are basically the same, with the 601 being a 701 minus the bar code reader and sequencer feature.  The 501 model is a full sized keys version of the MT70. The 1000P is unique in being semi programmable, and at a guess would share more in common  with the 601 and 701.

 

Unfortunately, the LSI/ CPUs used in these, as with most/ all Casios,  are proprietary and unique and were not available elsewhere. That means the only way you can find replacements is via another model being parted out, or by buying a faulty model sold as spares/ repairs.

 

The thing to look for is the number on the LSI- it will be something like "NEC D7802G". If the number of your LSI matches that in any of the other Sine Wave models mentioned above, it should be a 100% match. 

 

Hope this information helps, and I hope you can get your 1000P fixed. They are certainly a unique sounding keyboard. 

Edited January 22 by Chas

[Zarquon]

Hi Chas - thanks for the reply. My fault for not being clear! I'm after the Hitachi HD43517 IC's, rather than the CPU. I tried swapping the 2 chips over to check if the "good" one would work in the "upper octave keys" and vice versa - it is a bad chip, very distorted (quite nice sounding tbh, but I can get that with a fuzzbox!). Just wondering if they were used in any other keyboards that weren't Casio or maybe early computer boards etc?

[Chas]

The related models I mentioned all share very similar circuitry, so chances are they could also have the Hitachi HD chip that you are looking for. Outside of that, a Google search of the part number may come up with a suitable source. 

[JKG]

Were you able fix this? Bought an MT 70 which didn’t turn on to begin I opened it to see that it was missing the 2sd313 series pass transistor and replaced it with a tip41c and viola it turned on till it didn’t!!! It powers on in the sense that the power led glows and nothing else seems to happen…any help on how to go about figuring out what’s wrong and how do fix it will be helpful I’ve cleaned up the board with some wd40 so far 

Edited January 30 by JKG

[Ra226]

There are so many things you could try, it's hard to say where to start.  You might check the electrolytic capacitors, those can often die after many years and that can be a common problem in old electronics.  Specifically look to see if they're expanded and/or leaking.  Maybe take an oscilloscope to the CPU's clock pins, and then the other CPU pins to see if the CPU is at least doing anything?

[JKG]

Sure let me see if replacing electrolytic caps helps…sadly do not have an oscilloscope! Let me see what I can do to with a multimeter 😅

[CYBERYOGI =CO=Windler]

If something digital plays dead, always check the reset capacitor. If it is decomposed or shorted (e.g. by battery liquid) the thing won't wake up at all or crash. Also faulty quartz or clock oscillator can make strange trouble.