repair Resurrecting thrift-store found Casio PT-20

[Etaoin Shrdlu]

Hi all,

 

I registered here in the hope I might find some help in repairing a lovely little Casio PT-20 I picked up for a couple of Euros in a second-hand shop. It's nice and tidy with no cracks or scratches, and all buttons & keys are present. Sadly there appears to be something wrong with the envelope signal on the melody, and the chord/bass doesn't make any sound at all. Here's a short MP3 which demonstrates what it sounds like, using the rock & bossa nova beats, and piano and organ melody sounds. Clearly a very sick puppy 😟 Some crappy phone photos:

 

 

 

 

The plastic is not nearly as yellow as the photos suggest! I would love to try and fix this lovely little keyboard, but haven't been able to locate a schematic other than for it's much bigger brother the PT-80. This does use the same CPU though, so there's probably a fair bit they have in common. I guess finding the envelope transistor for the melody might be a good place to start; on the PT-80 this hangs off pin #79 on the CPU...

[Etaoin Shrdlu]

I have started to painstakingly trace out the PCB, by continuity testing the physical board and looking at a photo overlay of the traces on the component side. It's hard going though, because the layout is quite messy.

 

 

 

I have used the D1868G pin designations from the PT-80, which is the closest relative I have been able to find a service manual for. Some of these will not match the PT-20, but I have to start somewhere... For example, the "white noise" output on pin #72 is n/c on the PT-20, which must mean it's using a different pin for this.

Edited December 23, 2023 by Etaoin Shrdlu
Updated schematic

[Etaoin Shrdlu]

Although it most likely won't have anything to do with the fault(s) on my PT-20, while working on the schematic I noticed there's very little in the way of documentation on the Hitachi HD 61914 SRAM chip, which seems to have been a custom jobbie made for Casio. It's an 8-kbit / 1kb static RAM in a 44-pin TQFP package, with a 4-bit bus width, so more than half of the pins are not needed. The thing is, sources differ on which pins are used and which are n/c. The only two sources I've found for pin-out information (the TRS-80 PC-4 service manual, and a Polish Casio FX-700P fan) show lots of pins as n/c which definitely are connected on the PT-20 and PT-30:

 

 

I'm curious to find out what's going on here, does anyone know? This confusion is a little annoying when I'm trying to produce an accurate schematic of the PT-20...

Edited December 22, 2023 by Etaoin Shrdlu
Added link to FX-700P site

[pianokeyjoe]

Oh man! SAME issue as mine! Look at your battery compartment for corrosion. Seems to be a common issue with all battery operated Casios that when the batteries are left to rot in them, the alkali eats circuits and causes havoc. I still can not fix mine. So yes, if anyone knows what is up and can offer a solution, we are all ears! Sad.. such a nice little keyboard. I have 2 of them actually, and one of them turns on but makes no sound. The other, makes the SAME sounds as yours.

[Etaoin Shrdlu]

Ah yes, I saw your earlier post and thought I'd ask if my MP3 sounded the same as your PT-20. Interesting. I suspect the fault we're having may have been caused by someone using an adapter with the wrong polarity, which has caused some components to fail. Hopefully that does not involve any microcontroller pins! At least the MCU is still running, so that gives me some hope. Still slowly working my way through the schematic (updated above), but it's a real slog 😒

Edited December 23, 2023 by Etaoin Shrdlu

[pianokeyjoe]

Ah yes, THAT old monster rearing it's ugly head again.. The old "using the wrong polarity on a Casio" bit. Yeah it is good at least no full damage. Though my other one does seem to have the memory reset stuck issue that @CYBERYOGI =CO=Windler mentioned to me but both my specimens did have battery compartment corrosion soooo.. Lets see where all this goes! LOL!

[CYBERYOGI =CO=Windler]

I wrote in this thread about the CPU pinout. PT-20 is a crippled PT-30 variant and barely related to PT-80.

 

https://www.casiomusicforums.com/index.php?/topic/21503-casio-pt20/#comment-79847

 

The RAM is special. I wrote down the pinout eons ago.

 

pinout HD61914

 

The "Hitachi HD61914 x" (44 pin SMD, pins count anticlockwise, x = a letter) is a 4 bit SRAM with built-in "self control circuit" that was used in Casio keyboards with D1868G CPU and programmable calculators (with CPU "Hitachi HD61913 xxx"). The "HD" and letter are printed above the "61914". Said letter can be absent, A, B or C; it is unknown whether it indicates the memory capacity or speed or other special functions (e.g. address decoding scheme); at least the version without letter has 1 KByte capacity (2048 words x 4 bit, seen in Radio Shack PC-4 service manual).

 

Most important is that the first 11 pins have the same pinout like a ROM-Pack and so version B was also used in the special RAM-Pack RA-1 for Casio PT-50. In PT-30 the HD61914B apparently uses 3 additional pins 14, 17, 19 multiplexed with keyboard matrix and datasette port (may be NC to ease PCB layout). My 2nd (likely older) PT-30 contains a HD61914A (not and has an additional IC "TC40H000P" (4x NAND).

 

In PT-50 both SRAMs are basically wired parallel. The only main difference seems pin 28 that is +Vs on the left IC but cpu pin 30 (VDD2) on the right. SRAM pins 1..12 go to the cpu, while 24..27 and 29..33 are on +Vs. Although pins 34..44 are connected with various pins from 1..11 (see here), I expect that they are NC and only used to simplify the PCB layout. But since VDD2 is only a standby signal, it may be that there is indeed a deeper meaning behind this (possibly decoded by some odd XOR tests). Pins 13..23 definitely do nothing and thus are very likely NC.

 

Also in PT-30 3 SRAM pin traces are additionally wired to other pins (6 to 19, 7 to 17, 11 to 14), those are likely NC to ease wiring. SRAM pin 6 (clock phase?) is wired through a diode to CPU pin 28. In my 2nd PT-30 SRAM pin 6 is additionally wired through another diode to the output of a driver (2 chained NANDs) from CPU pin 30 and some RC network circuitry to to the power switch "off" contact (likely to prevent data corruption), which hints that SRAM pin 6 is always input.

 

It is unknown what pins 24..33 exactly do; in the pocket computer Radio Shack PC-4 (HD61914 with no letter) they are all wired to pin 1 and the optional RAM-Pack (pinout like a ROM-Pack) is simply wired parallel to the internal SRAM. That RAM-Pack (only empty PCB layout shown in service manual) apparently contains a single "HD61914 x" with pin 1 wired only to pin 33, pins 2 additionally to 23, pin 3 additionally to 22, pin 4 additionally to 28, pin 27 only to 29, connector pin 1 to IC pin 24. These additional pins likely map the address range for the 2nd RAM at $800h directly above RAM 1, which in the PC-4 starts at $200h above the CPU internal RAM. Additionally the traces to pins 3..11 are wired through 21..13, and pins 34..44 are all interconnected (likely NC to ease wiring). But by their wiring at least pins 22 and 23 seem to be not NC (contrary to their naming in PC-4 schematics). Also in RA-1, PT-30 and PT-50 there are many such connections, of those many pins may be NC to simplify the PCB layout. In the PC-4 the cassette interface port (to an 28 pin DIL IC HD43110 for data compression, 300 baud "Kansas City standard", serial data on D4) is wired parallel to the RAMs and only uses a different ø2 clock line from the CPU, which shows that Casio used this 4 bit bus in many places.

 

This pinout is based on the service manual of the BASIC programmable pocket computer Radio Shack PC-4 (rebranded Casio PB-100) and the website of Piotr Piatek (thanks to him). The PB-100 was a crippled version of Casio FX-700P, which had an additional function button and the 2nd RAM already built-in (no RAM-Pack slot), marketed in 1982 as scientific calculator.

 

caution: The service manual indicates that this IC uses "negative logic", i.e. technically +5V is its GND while 0V is its -5V supply voltage. So the voltages are not was the pin names suggest. I use the positive voltage naming convention (from 0V to +5V, not -5V to 0V).
 

pin name purpose

1 GND supply voltage +5.5V

2 CE chip enable

3 VDD1 ground 0V

4 VDD2 auto-power-off voltage (hi when off)

5 ø1 clock1

6 ø2 clock2 (phase shifted against ø1)

7 OP data /address select

8 D1 address/data bus

9 D2 address/data bus

10 D3 address/data bus

11 D4 address/data bus

12 INT (interrupt?)

13 NC 

14 NC 

15 NC 

16 NC 

17 NC 

18 NC 

19 NC 

20 NC 

21 NC 

22 NC (device address select?)

23 NC (device address select?)

24 DC (alternative backup battery supply voltage?)

25 V4 device address select

26 V3 device address select

27 V2 device address select

28 V1 device address select A11

29 R/W (device address select?)

30 BL4 (device address select?)

31 BL3 (device address select?)

32 BL2 (device address select?)

33 BL1 (device address select?)

34 NC 

35 NC 

36 NC 

37 NC 

38 NC 

39 NC 

40 NC 

41 NC 

42 NC 

43 NC 

44 NC 

 

According to Piotr's website the write command is $4 (OP=0), followed by a 4 bit device address, followed by the 11 bit address within the device (MSB is ignored). After this any amount of data can be written (in an OP=1 cycle), which makes the address counter advance after every 4 bit word. The read command is $0 and works similar. A single $0 word regularly appears on the bus when the calculator is idle, which may be a kind of soft reset.

Edited December 26, 2023 by CYBERYOGI =CO=Windler

[dannzomatic]

 

 

Hello,

I welcome all long-standing members to this forum and am happy to be part of it now. Coincidentally, I also started reviving a Casio PT-20 around Christmas.

My PT-20 also makes similar sounds to yours, Etaoin. If necessary I can also upload an MP3.

After trying all sorts of things for the last 8 hours, I think it's time to join the thread. Maybe we can achieve something together to get the PT-20 fit again. The PT-20 means a lot to me and I would be happy if I can get it working again.

I replaced the electrolytic capacitors a long time ago, but that didn't bring any improvement. Today I took all sorts of measurements and unsoldered and measured some transistors, ceramic capacitors, diodes, etc., but unfortunately I couldn't locate the error.

For me too, the PT-20 behaves in such a way that as soon as the rhythm is on, the voltage collapses. For simple rhythms like the Waltz, only the LED gets a little darker at the low frequencies. With Ballad the PT-20 breaks down completely and the LED goes out completely. The PT-20 is still on because when you switch back to Waltz, the sound and the LED return.

The chords sound good. But the presets all have a small punch in the envelope. They do that “uuaaap” at the end. Exactly how pianokeyjoe described it.

OK. I'll leave it alone for today.

If I can help building a schematic or do some measurements, let me know.

Best regards,

Daniel

[dannzomatic]

Hello again,

I was successful, yippee!

Today I researched again whether it could be an APO problem, as CYBERYOGI suggested in another thread.

I also got to the bottom of the missing LED, my PT-20 no longer had a diode in the LN2G, apparently someone had already soldered it out before me. In the course of this, I set about examining the voltage regulator department in more detail because the LN2G is housed here. I had printed out various circuit diagrams from a service manual for the PT 1 and the SK 1, from which I tried to find clues.

It turned out that there are not quite as many transistors installed in the PT-20 as in the SK 1. The APO signal takes a different path in the PT-20 than in the SK 1.

Then I tried to check all transistors again, especially in the power supply area. When I soldered out the D545, it turned out that it no longer conducts in both directions (diode test, base to emitter and base to collector). The location on the mainboard is called D545 and a D734 F3B NPN transistor was installed. I then replaced it with another NPN transistor and it worked. There are no more breaks in voltage and the rhythms run smoothly.

By the way, I think the hi-hats caused the voltage dropouts, maybe a lot of amps are needed here, I don't know for sure.

I swapped the missing LED LN2G for another one from the spare parts box. It doesn't light up, but the PT-20 works.

I also played on it a bit afterwords, which is actually a lot of fun with the accompaniment.

Here are two more pictures of it.

Good luck to you too!

Best regards,

Daniel

P.S. I marked yellow the faulty transistor in second picture.

 

 

[pianokeyjoe]

When ever I can dig out my PT20s, I will try this fix. Oh OH!! I thinks we got it!! Thankyou @dannzomatic

[Etaoin Shrdlu]

On 12/27/2023 at 6:15 PM, dannzomatic said:

I was successful, yippee!

 

Fabulous! I read your post with great excitement, but was disappointed when I looked at my own PT-20; I had already replaced that transistor during a previous repair attempt, and it had not fixed my problem. Indeed mine too had failed open (I've kept the original part, so could double check). Then I had a brief moment where I thought I had found another dead transistor, but after desoldering it I realised it was a PNP and I had simply measured it the wrong way around I decided to spend some time looking at the signals from the CPU - here for example are the Melody Pitch (pin 80) and Melody Envelope (pin 79) signals:

 

 

 

That made me feel better again, since these are obviously correct, and the strange sounding instruments must be caused by something fixable. Based on what others had said about dim LEDs, and thinking mine too looked dim I took some voltage measurements (again) and this time I realised they were too low (-4.5V measure -3.7V for example), so I went back to the power supply section on the board and tested the caps for shorts, then the diodes, and... what? The diode between the DC barrel jack and the headphone jack measured 0.1V on diode test, in both directions Whoooa.

 

 

I desoldered it and confirmed: this diode was dead as a dodo and had a resistance of just 200-ish ohms in both directions. And it sits right across the power rails. Now I'm excited again! I grabbed whatever diode I had to hand (1N4148) and replaced it - and what do you know, IT FRIKKIN WORKED! I can't believe it, but melody and percussion now sound exactly as they should! I still have some other issues to fix; most importantly the chords section is totally mute, and some keys don't work on the keyboard (I suspect a good clean-up will fix those), but this is a huge step forward!

 

Big thanks to everyone who's contributed - and a Happy New Year to you all!

 

 

 

 

 

 

 

[pianokeyjoe]

@Etaoin Shrdlu ok, my original thought about that wrong polarity was the correct one at least for you. Ok, I have 2 specimens to work with so it will be interesting what I come up with for both.

[Etaoin Shrdlu]

Good news, and bad news. The good news is I've figured out why the chord section buttons (and the lowest note on the keyboard) don't work at all, and it's not due to an MCU failure, or any other component. As I was (very gently) washing it, I discovered that the keyboard PCB is cracked, just where the chord section sticks out from the rest 😫

 

 

 

I think the broken traces going to the chord section can be fairly easily repaired (after glueing a reinforcement on the back), but the contact pad for the lowest note on the keyboard is also cracked and that will be a lot harder to deal with... Any suggestions?

Edited January 2 by Etaoin Shrdlu

[Etaoin Shrdlu]

On 12/26/2023 at 5:21 AM, CYBERYOGI =CO=Windler said:

I wrote in this thread about the CPU pinout. PT-20 is a crippled PT-30 variant and barely related to PT-80.

 

https://www.casiomusicforums.com/index.php?/topic/21503-casio-pt20/#comment-79847

 

The RAM is special. I wrote down the pinout eons ago.

 

Thank you for this amazingly detailed information - you are a wizard! 🧙‍♂️ I do not have a service manual for the PT-30, or the PT-20, only the PT-80.

[Etaoin Shrdlu]

After sanding and degreasing the area I superglued a piece of FR-4 on top of the crack, which should make that part of the PCB much stronger than the rest:

 

 

But the damage was worse than I first thought... 😖 After removing the tape and cleaning up the adhesive residue it became clear that the crack had continued further and had broken another three copper traces, and the four graphite paint "jumpers" going across them:

 

 

I've fixed all the broken copper traces with solder bridges and bodge wires (I had no Kynar to hand, so had to make do with some thin regular wires), and now the top five chord buttons, and the leftmost one of the middle row, work as they should, but the rest remain silent. I have confirmed that these are all connected to the broken graphite paint bridges.

 

 

I guess I need to get some conductive paint before I can get any further, unless anyone has a better idea?

 

Edited January 2 by Etaoin Shrdlu

[CYBERYOGI =CO=Windler]

Do NOT use insulated cables - they are too thick and put even more strain on the PCB. I had badly messed up my VL-10 when I upgraded it with VL-1 functions and had to remove all that cable mess to use hair thin coil wires instead.

 

To patch PCB traces, simply use individual thin copper wires from a wick cable. Also enamelled coil wire can be used (remove insulation by soldering iron heat before pressing into place). The graphite/carbon traces can be patched with copper wire too (resistance is not critical here). Only those directly under the black rubber contacts may suffer of oxidization or mechanical wear when soldered directly underneath (but they will function too). You may buy conductive carbon paint instead, but it is not really necessary here.

 

Here I posted a photo of a patched PT-50 chord keypad PCB that got damaged much worse than yours.

https://www.casiomusicforums.com/index.php?/topic/22041-pt-31-wont-power-on-where-to-start/#comment-84290

 

Edited January 3 by CYBERYOGI =CO=Windler

[Etaoin Shrdlu]

6 hours ago, CYBERYOGI =CO=Windler said:

Do NOT use insulated cables - they are too thick and put even more strain on the PCB. I had badly messed up my VL-10 when I upgraded it with VL-1 functions and had to remove all that cable mess to use hair thin coil wires instead.

My go-to would be Kynar for PCB trace repairs, but I'm away from home and did not bring any with me. I also miss my inspection microscope, and soldering station (I have to do this repair with a Portasol instead) 😒 I agree plain single strand copper is the only way to go if you need to repair tracks under the rubber contact pads (or possibly conductive paint), but I've routed my wires outside these. I'll try repairing the four graphite bridges in a bit, and there I will use uninsulated copper wire, since it won't touch any other conductive areas. Also remains to be seen whether the white wires going around the back of the board will interfere with the keyboard keys - I may have to reroute these...

Edited January 3 by Etaoin Shrdlu

[Etaoin Shrdlu]

 

 

Mission accomplished!

 

(apart from the lowest note on the keyboard; I'll repair that sometime later, when I've got some graphite paint)

 

Thanks to everyone for their help!

 

 

Edited January 4 by Etaoin Shrdlu

[Etaoin Shrdlu]

Oh, and just one more thing...

 

 

Hopefully this will prevent the same thing from happening again

[pianokeyjoe]

18 hours ago, Etaoin Shrdlu said:

Oh, and just one more thing...

 

 

Hopefully this will prevent the same thing from happening again

LOL LOL! You would think that would help? But you would be wrong.. sadly.. And leaving the darn batteries in the Casio. that is the even worse problem since batteries leaking into the electronics causes much more damage than wrong polarity which blows on or 2 components and you can usually fix that but eaten components and traces? Not so easy to fix..

[CYBERYOGI =CO=Windler]

Nope, confusing polarity tends to destroy the CPU itself, so the thing is braindead (in Germany we call that now "vermariupolt" - as a pun of the word "verpolt" = "reversed polarity" and allusion to the name of the Ukraine city that was completely bombed away into ground). Battery leak vapour is mainly disastrous to LCD, the rest usually can be fixed.

Edited January 6 by CYBERYOGI =CO=Windler

[ericlimer]

On 12/27/2023 at 1:15 PM, dannzomatic said:

When I soldered out the D545, it turned out that it no longer conducts in both directions (diode test, base to emitter and base to collector).

 

I'm working on a PT-20 of my own, with similar issues to those described in this thread (more detail here). This thread inspired me to test this transistor on my unit. When I do this, I get a continuity reading from every combination of legs in both directions. This is indicative of a fault, yes? 

 

I'm a little unclear on how it _should_ test, but it is my general understanding that no transistor should have continuity from all directions at once. 

Edited February 8 by ericlimer
clarity

[CYBERYOGI =CO=Windler]

Desolder the transistor and test again. If any of its 3 pins have a low resistance short circuit, the thing is toast and needs to be replaced.

[Etaoin Shrdlu]

On 2/8/2024 at 9:04 PM, ericlimer said:

I'm a little unclear on how it _should_ test, but it is my general understanding that no transistor should have continuity from all directions at once. 

 

Ideally you would use the diode test mode of your multimeter, if it has this. From the meter's point of view, a transistor looks like two diodes connected from the base to the emitter & collector:

 

To test an NPN transistor:

•  With the meter set to diode test and with the positive lead connected to the base of the transistor you should see a forward voltage of ~0.5-0.8V when touching the negative lead to the emitter or collector.
• Staying in diode mode, and connecting the negative lead to the base of the transistor you should see no voltage ("OL") when touching the positive lead to the emitter or collector.
• Switching the meter to resistance mode, with the negative lead still connected to the base, you should see an open ("OL") when touching the positive lead to the emitter or collector.
• With the positive lead connected to the base you should see zero ohms (or close to it) when touching the negative lead to the emitter or collector.

 

For a PNP transistor everything is reversed:

• With the meter set to diode test and with the negative lead connected to the base of the transistor you should see a forward voltage of ~0.5-0.8V when touching the positive lead to the emitter or collector.
• Staying in diode mode, and connecting the positive lead to the base of the transistor you should see no voltage ("OL") when touching the negative lead to the emitter or collector.
• Switching the meter to resistance mode, with the positive lead still connected to the base, you should see an open ("OL") when touching the negative lead to the emitter or collector.
• With the negative lead connected to the base you should see zero ohms (or close to it) when touching the positive lead to the emitter or collector.

In both cases you should see an open ("OL") when measuring the resistance between the emitter and collector, regardless of the direction you measure in - no current can flow here since our imaginary diodes are pointing in the opposite direction to one another.

Edited February 28 by Etaoin Shrdlu