Casio PT-7 instrument with tiny polyphonic touch sensor keypad & analogue rhythm
Casio MT-21

Casio PT-7

This oddity from 1984(?) is clearly one of the most bizarre and innovative instruments created by Casio, because this tiny thing has a detachable keyboard with 29 soft touch foil keys, and this is not just a monophonic toy tablehooter but a real 8 note(!) polyphonic instrument with analogue rhythm and a small but high quality loudspeaker that makes a very respectable organ sound.

(Old eBay picture, showing my specimen.)
 This strange Casio invention permits special play techniques with very rapid glissandos and was one of the forgotten milestones in the struggle of overcoming the clumsy piano key relic on electronic consumer instruments.

The original German retail price in 1985 was 199DM (seen on eBay).
The sound generator is a multipulse squarewave variant with up and down facing blocks.

main features:

eastereggs:

notes:

The main voice sound chip of this instrument is the same like in Casio MT-45. It plays high quality analogue timbres, those although not always natural, reproduce a warm and pleasant sound; it does not sound typically C64- like thin, but resembles rather fullsize home organs of that era. But like with Casio VL-1, the release phase of main voice envelopes is linear and thus sounds unrealistic since it fades silent too soon. By the lack of sustain, all sounds stop almost immediately after key release.

The 'elec. piano' sounds like a banjo, and also the normal piano resembles more a picked string. The 'harp' and 'organ' sounds pretend to add a bit of analogue distortion (genuinely a suboscillator with very short independent envelope) during attack phase. All sounds include a mild vibrato. The smooth touch sensor keypad responds quite sensitive and permits special play techniques with rapid note clusters and polyphonic glissandos (but no portamento - this is not a theremin!).

caution: Never play with sharp, spiky or rough objects (like finger nails) on the sensor keyboard surface - the foil may get damaged easily. At both keyboard sides the grooves for the 2 small metal hooks were filled with a smeary, tar-like substance, that sticks to the hooks and left black dirt stains everywhere. I suspect it was decomposed residues of wrongly mixed black rubber. (I found similar goo in Yamaha MP-1.) I used tissues, cotton swabs, a cable tie and a bit of isopropanol (do not soak the foil) to poke out and clean up most of it.

The percussion timbres resemble my Casio MT-70 and have the typical analogue home organ style. The rhythms are the same like with Casio MT-40 and their tempo can be set ridiculously high but not extremely low.
 

circuit bending details

The Casio PT-7 is build around the main voice CPU "Hitachi HD44140" with independent rhythm CPU "NEC D8048C 316".
The main unit of this instrument is crowded with 3 stacked PCBs those contain the rhythm CPU and a lot of analogue components. The PCBs are partly shielded by aluminiumized cardboard - likely to prevent interferences in the transistor noise generator of the analogue percussion. The detachable keyboard hangs on a very short, shielded 4 lead cable that only conducts supply voltage, 2x GND and the main voice audio signal. The keyboard unit contains a small SMD PCB with the "Hitachi HD44140" CPU, various discrete components, an LC clock oscillator (ferrite coil adjustable for tuning) and a bit compensation trimmer to tweak low DAC bits.

main unit

The rhythm is generated by the accompaniment CPU "NEC D8048C 316", which is the same IC like in my Casio MT-40 (see there), where it additionally generates a monophonic bass accompaniment. Thus theoretically it may be possible to add also a bass button keypad like in MT-40 to the PT-7. But I expect that this would also need various external components for implementing the analogue bass decay envelope circuit (I did a similar mod for the arpeggio upgrade of my Casio CT-410V). For the bass keypad matrix see here.

A lot of wild random crash sounds can be generated in my MT-40 by adding a shitshot button from pin 3 of the D8048C (for safety through a 1 kOhm resistor) against GND, which likely works also with the PT-7.

keyboard unit

Beside some other unused pins (anyway GND) wired to GND and the LC oscillator, the HD44140 is not wired that much differently than in Casio MT-45. A little strange is that the key matrix diodes were substituted with small SMD transistors (each used instead of 2 diodes) those have different types (pinouts) to minimize the use of bridges (the unmarked white resistors).

caution: The sensor keyboard case is held closed by 3 tiny screws and 4 plastic tabs; to open it it is important to pull the plastic cover backwards (in the direction of the cable hole); do not bend it upward because this would crack off the plastic tabs. Be generally extremely careful with dismantling this delicate component. When open, the PCB hangs only on 2 strips of adhesive film and the foil cable.

Danger!: The conductive glue under the foil cable is extremely brittle and will break by any small sideways motion or rotation of the PCB! My cable fell off without warning when I only tried to turn the ferrite core of the tuning coil (after 2 hours of successful measurement). It is impossible to remelt this glue by heat!
When I took the keyboard foil stack out for closer inspection, I noticed that the upper and lower contact foils are (instead of being made from one folded piece) connected in the upper right corner by another such glue joint, which also fell apart once I tried to take out and photograph the intermediate perforated insulation layer. Without success I tried to reconnect this end with a kitchen foil welder (did not even melt), SMD hot air soldering gun (up to 160°C, did rather warp a little without staying attached). These foils have copper paint traces (shiny from the transparent top) covered with conductive carbon or graphite paint. A careful test with SMD soldering iron revealed that despite apparent copper layer even at lowest possible heat molten solder destroys the paint and tends to melts the ancient foil. (Modern copper foil cables can be soldered.) So I simply aligned top and bottom foils with adhesive film, sharply rolled both contact ends together (held into place with adhesive film) and squeezed them under the sheet metal cover inside the case, which seems to work properly. I increased the pressure by installing short screws in the 2 unused screw holes at its sides. - This was the easier part. 

Unfortunately it turned out much more difficult to reattach the foil cable to the PCB. First of course hold it with adhesive tape into place, but the problem is that it needs firm permanent equal squeeze on all pins to make reliable contact, thus a strip of (window isolation) foam rubber on top has to stay pressed in place. Unfortunately in the PT-7 keyboard unit both case top and bottom are overly floppy and only bulge out instead of compressing the strip, and (unlike a glass LCD) also the PCB is rather flexible. So I tried to make several kinds of metal brackets (kind of safety pin or hair clip style) to clamp the PCB without warping the case, which still bent the PCB and didn't work too well. (I even broke a key matrix transistor by collision with pliers and replaced it with an SMD diode.) 

Against floppy parts help only  floppy disks, so I finally made 2 short vertical clamps from the sheet metal sliders of 2 broken 3.5'' diskettes. Bent around the entire foil cable (that I protected from the sharp metal rims by fabric adhesive tape) at the PCB rim, it applies much evener pressure. But even this still needs some metal reinforcement and a short foam strip on top that is squeezed by the case cover. Although it seems to work, it is still very prone to bad contacts. (I am not even sure if I damaged the foil harness elsewhere.) And be careful, there is another glue joint lurking between foil cable and keyboard foil stack, that is certainly of the same brittle stuff.

The following I found out by examining my MT-45. For the main keyboard matrix see here.

  • disable vibrato
    The permanent vibrato is likely hardwired by a transistor (used as diode) from CPU pin 4 to 23. To disable it, cut the trace to pin 23. You may install a locking switch instead.
     
  • sustain
    A sustain switch can be added by soldering a switch in series to a diode from CPU pin 3 to 23.
  • additional keys
The CPU has key matrix outputs from pin {6..12, 14..20, 22} and the corresponding inputs at pin {3..5}. Theoretically 8 additional keys can be added, but in practise this makes not much sense since it is mainly the sensor keypad that makes the special feel and the unique play techniques of this instrument possible and not just sound.
(I haven't upgraded my PT-7.)

Unfortunately the Casio PT-7 with its special glissando keypad seems to belong to the rarest of all Casio keyboards (I saw no other specimen on eBay yet) and I also never heard of any other polyphonic instrument with this kind of controller. I believe to remember that long ago I read in a magazine a Casio essay, that the PT-7 was originally planned as an electronic mouth-organ or blow-organ with foil keys on top (US patent 4619175 & 4566363, the latter even mentioning a solar cell), hence this tiny detachable keyboard unit with internal low-power sound IC. But the sound hardware was also used in cheap midsize Casio keyboards.

Casio MT-21

This very rare midsize keyboard from 1985 (embossed case date) has 8 preset sounds and only 4 analogue rhythms without accompaniment.

The sounds are like Casio PT-7 with additional sustain and vibrato switch. The rhythms have a stronger snare drum, but the main voice sounds cheaper (thinner and a bit hollower with less bass), which may be result of changed filters.

Despite normal midsize keys, the electronics has strong similarities with the PT-7, thus I only describe here the differences.

different main features:

eastereggs:

modifications:

notes:

The main voice is basically the same like in Casio PT-7 and MT-45, but here uses all functions of the main voice CPU and sounds a bit thinner than with PT-7. Despite lack of accompaniment it contains a special variant of the MT-40 accompaniment CPU. I first was sure that also the patterns had changed, but apparently this was all an illusion by different analogue percussion timbres (much stronger than the faint original MT-40 drums) those sound quite punchy. Particularly the better 'snare' sounds special (with a strange, somewhat handclap-like timbre), although not as realistic as my modified MT-40 snare. For only 4 rhythms without accompaniment the circuitry is rather complicated. May be it was meant to compete with Yamaha PS-1, which however came 5 years earlier, which makes the crippled section hard to understand and an obvious reason for its rarity because it certainly didn't sell well. The plastic case seems to be the first with embossed date stamp. Seeing the CPU, possibly Casio planned an MT-40 successor that never went into production.

The 'elec. piano' resembles a harp, the 'harp' rather a banjo and in high octave a (not ringing) mandolin. The 'violin' sounds a little hollow and also the 'clarinet' turns hollow in the bass range.
 

circuit bending details

The Casio MT-21 is built around the main voice CPU HD44140 and uses all its features (no matrix eastereggs). The rhythm is controlled by the unique accompaniment CPU "Toshiba TMP8048P 1104" (Intel MCS-48, variant of "NEC D8048C-316"). 
The analogue percussion section uses fairly much discrete components despite (by omitted "samba" rhythm) it lacks the 'clave' circuit. On the PCB are trimmers for main voice bit compensation (tweak low DAC bits), "main" and "noise".

Most unusual is that the rhythm uses the accompaniment CPU of Casio MT-40, so its entire bass accompaniment (including the famous Sleng Teng riddim pattern) is included, but unleashing it would need additional analogue circuits and buttons.

main voice keyboard matrix

The main voice is produced by the self-contained sound CPU HD44140, which has only few functions and supports no accompaniment. Very unusual for Casio is its keyboard matrix layout with 16 output and only 3 input lines. There are no eastereggs.
 
 6 KC1
7 KC2 
8 KC3
9 KC4
10 KC5
11 KC6
12 KC7
14 KC8
15 KC9
16 KC10
17 KC11
18 KC12
19 KC13
20 KC14
22 KC15
23 KC16
 
CPU pin
out 1
out 2
out 3
out 4
out 5
out 6
out 7
out 8
out 9
out 10
out 11
out 12
out 13
out 14
out 15
out 16
out / in
 
O.
clarinet
O.
organ
 O.
harp
o
F#0
o
A0
o
C1
o
D#1
o
F#1
o
A1
o
C2
o
D#2
o
F#2
o
A2
o
C3
o
D#3
sustain
in 1
3 KI1
 O.
violin
O.
pipe organ
 O.
piano
o
G0
o
A#0
o
C#1
o
E1
o
G1
o
A#1
o
C#2
o
E2
o
G2
o
A#2
o
C#3
o
E3
vibrato
in 2
4 KI2
 O.
accordion
O.
elec. piano
o
F0
o
G#0
o
B0
o
D1
o
F1
o
G#1
o
B1
o
D2
o
F2
o
G#2
o
B2
o
D3
o
F3
APO disable 
in 3
5 KI3

The input lines are active-low, i.e. react on GND. Any functions can be triggered by a non- locking switch in series to a diode from one "in" to one "out" pin.
 

legend:
"o"
 = keyboard key
underlined
 = function needs locking switch (i.e. stays active only so long the switch is closed)
O.
 = preset sound ('orchestra')
orange 
background		 = easteregg (unconnected feature)

  • APO disable
    Connecting a diode at pins 5->23 prevents auto-power-off (seen in Casio MT-45, thanks Traktor for info).

rhythm key matrix

The rhythm controls and bass key buttons are handled by the accompaniment CPU TMP8048P-1104. The matrix input lines are routed through the inverter TC4069BP ("Hitachi HD14969UBP", 14 pin DIL) next to it, so its input pins need to be used. 
 
5 /P15
none /P16
none /P17
<- inverter pin
accomp. CPU pin
in 1
in 2
in 3
in / out
 
R.
start/stop
C.
C1
C.
G#1
out 0
12 DB0
R.
synchro/ fill-in
C.
C#1
C.
A1
out 1
13 DB1
R.
waltz
C.
D1
C.
A#1
out 2
14 DB2
R.
samba
C.
D#1
C.
B1
out 3
15 DB3
R.
swing
C.
E1
C.
C2
out 4
16 DB4
R.
slow rock
C.
F1
C.
C#2
out 5
17 DB5
R.
pops
C.
F#1
C.
D2
out 6
18 DB6
R.
rock
G1
C.
D#2
out 7
19 DB7

The input lines are active-high, i.e. react on +Vs. Any functions can be triggered by a non- locking switch in series to a diode from one "out" to one "in" pin.
 

legend:
R.
 = preset rhythm
C.
 = chord
orange
background 		 = easteregg (unconnected feature)

  • 2 additional rhythms & fill-in
    The rhythms "samba", "pops" and a fill-in button can be added by wiring button switches through diodes from TC4069BP pin 5 to pin 15, 18 and 13 of the TMP8048P. Note that by the lack of a 'clave' circuit samba will play mutilated.
  • bass keypad & accompaniment
    Theoretically likely the whole bass accompaniment section of Casio MT-40 can be added, although it would be a lot of work.

pinout HD44140

The LSI "Hitachi HD44140" (54 pin SMD, pins count anticlockwise from below the left stub pin) was Casio's most simple polyphonic single chip main voice hardware of early 1980th keyboards (likely used first in MT-11 of 1982). Compared with the very versatile first generation predecessor D77xG, it was cut down to have as few functions as barely needed for a short beginners keyboard. So despite 8 note polyphony the keyboard matrix supports only 37 keys and 8 preset sounds with 1 sustain and 1 vibrato switch. Unlike D77xG the chip produces no feelable heat, so the transistor count was likely minimized for long battery life (possibly created for the mythical solar cell powered mouth organ that became the PT-7). A bit strange is the integrated DAC that outputs its lower bits (at increased level) only on a separate pin; also the highest bit is separately output in normal and inverted. All these have to be combined through an external voltage divider (small resistor network with ratio 1:64), which was likely done to reduce noise. The only 11 DAC bits also have separate output pins (not used in any instruments I know).

The sound generator is a very simplified variant of Consonant-Vowel synthesis. Each preset sound is made from 2 layered multipulse squarewaves (8 steps long?) crossfading by simple volume envelopes those do not change with note pitch. Each step can have only the height {-1, 0, +1}, i.e. protrude fully up, fully down or be zero. (I.e. unlike D77xG there are neither ramp nor curve sections.) The linear envelope and low bit resolution make it sound rather artificial, but the unusual sonorous waveforms are unique and the high internal clock rate (1.425 MHz?) prevents cold aliasing noise. The preset sounds 'clarinet' and 'piano' set the external fixed lowpass filter to 750Hz, else it stays 1.1kHz (seen in MT-11 service manual). Unlike D77xG there is no spread scale, i.e. holding the same notes in different octave causes no phasing (but phase changes among multiple key presses, thus there are no octave dividers involved).

caution: The pin numbering of this IC package is very bizarre. I expected it to start in the slanted lower left corner (text readable), but numbers on MT-45 and particularly MT-200 PCB (different IC in same package) indicate that it starts from below the stub pin to the left and the missing pin is skipped, i.e. this IC has only 54 (or 55??) instead of 56 pins. I first made this pinout from my own observation, but later got the MT-11 service manual for pin names.

caution: The MT-11 service manual indicates that this CPU 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). Apparently all "O#" pins are outputs, "I#" are inputs and "IO#" can be both.
 
pin name purpose
1 NC  
2 VDD1 ground 0V
3 KI1 key matrix in
4 KI2 key matrix in
5 KI3 key matrix in
6 KC1 key matrix out
7 KC2 key matrix out
8 KC3 key matrix out
9 KC4 key matrix out
10 KC5 key matrix out
11 KC6 key matrix out
12 KC7 key matrix out
13 NC  
14 KC8 key matrix out
15 KC9 key matrix out
16 KC10 key matrix out
17 KC11 key matrix out
18 KC12 key matrix out
19 KC13 key matrix out
20 KC14 key matrix out
21 NC  
22 KC15 key matrix out
23 KC16 key matrix out
24 NC  
25 TEST (wired to supply voltage GND)
26 NC  
27 GND supply voltage +5V
- (no pin)  
pin name purpose
28 PG1 clock in (5.752 MHz?)
29 PG2 clock out (inverted PG1)
30 MS (wired to supply voltage GND)
31 PGTO clock/4 out
32 PGTI clock/4 inverted? out
33 O1 audio out (upper bits)
34 O2 audio out (lower bits, weighted 1/64)
35 VDD4 ground 0V
36 IO12 dac bit out (LSB)
37 IO11 dac bit out
38 IO10 dac bit out
39 IO9 dac bit out
40 IO8 dac bit out
41 IO7 dac bit out
42 NC  
43 IO6 dac bit out
44 IO5 dac bit out
45 IO4 dac bit out
46 IO3 dac bit out
47 IO2 dac bit out (MSB)
48 NC  
49 IO1 bit compensation (inverted highest bit)
50 O3 bit compensation (highest bit, like pin 47)
51 VDD3 ground 0V
52 O5 /APO auto-power-off out (not used)
53 O4 filter control out (lo=duller)
54 VDD2 ground 0V
- (stub) (internally wired to supply voltage GND)

At the left side of the package is an unused short pin stub that is internally connected with +Vs (positive supply voltage) and apparently not counted as a pin. At the opposite side is no pin at all. The "NC" pins are unconnected, behave like inputs (high resistance) and have lo level. Pulling these to +Vs through 1K resistor seems to do nothing. It is unknown if some of the many pins wired to GND or +Vs are genuinely inputs or test pins.

I see no complex signals for communication with a host CPU; likely the HD44140 is too dumb for this or interesting pins are just grounded. When shitshot (touching clock pins) it does nothing spectacular (sometimes stuck notes or silence or plain notes decaying naturally, but no new timbres), which hints that is a rather banal LSI circuit of hardwired logic gates and counters (even simpler than D77xG) with next to no software inside. Also the strange keyboard matrix layout (16 x 3) was likely done to simplify internal wiring.

The PT-7 powers on with 'harp' preset sound, but MT-21 (slide switch in intermediate position) defaults to 'piano', thus I guess that in PT-7 external components (a capacitor?) set it to 'harp'.

An obviously related real sound IC by Casio (controlled by a CPU, with percussion generator and chord+bass channel support) is HD43720.
 

pinout TMP8048P-1104

The "Toshiba TMP8048P 1104" (40 pin DIL) is the rhythm CPU of Casio MT-21. This IC is almost identical with "NEC D8048C-316" (see there).

I first thought the rhythm patterns were more complex (automatic fill-ins), but apparently this was an illusion by different analogue percussion circuitry inside MT-21. The pinout and behaviour are identical; even the bass accompaniment is fully functional (verified by inserting into Casio MT-40 and vice versa) despite no instrument is known to make use of it.

Technically the "TMP8048P xxx" is a generic microcontroller of the well documented Intel MCS-48 family. Dumping its ROM with the eprommer Willem Pro4 isp however gave some strange results. When read as P8050AH, the 4KB output has 512 bytes followed by 14 repeats of the first 256 bytes. When read as other MCS-48 variants, the output repeats already after the first 256 bytes, so it is unknown whether the code is longer than the dumped 512 bytes. The trailing 287 bytes exactly match those of the 1KB ROM dump from "D8048C-316", which makes me expect that also the TMP8048P ROM is genuinely 1KB with the rear 512 bytes unreadable. It might even be that the software is identical with D8048C-316 but gets scrambled by some kind of copy protection circuit.

An MT-21 without rhythm hardware was released as Casio MT-20 (white case, seen on eBay). The same main voice hardware was also used in MT-45, and without additional rhythm hardware in Casio MT-11 (32 midsize keys, similar case style, service manual date 1982).
 

 removal of these screws voids warranty...    
WarrantyVoid
back to tablehooters collection
  
 
back