Many years ago I used to compete in the annual Yamaha Electone Festivals here in Australia. I look back at that time with great fondness... it was great fun and I made life-long friends. It’s really no surprise then that some of my favourite Electone models are those that I spent slaving away on... fine tuning my next contest piece. One of those models was the E-75.

Dad sold the E-75 in the 80s to help fund an FX-20. Over the years I’ve regretted parting with the E-75, and recently I was fortunate enough to come across one in excellent cosmetic condition. A local church was giving it away because it had developed a fault and they had been quoted many hundreds of dollars to have it repaired. They were more than happy for a friend and I to pick up the E-75 and take it away.

I tracked down a service manual... a story in itself (it came from the current owner of my old E-75). The church had told me that the E-75 was working perfectly, and then one day it started to emit a “loud buzzing noise”. They were right... there was a loud buzz, and no sound output from any section of the E-75.

As always, the power supply is good place to start. I checked all the main supply rails, and they were within normal tolerances. I was interested in whether this was a tone generation issue or amplification issue, so I connected the E-75 stereo line outputs into a small keyboard amplifier, and the E-75 sprang into life. So, definitely not a tone generation issue. The power amplifiers seemed the likely culprits, but the E-75 has four separate amplifiers (left, centre, right and tremolo). It seemed very unlikely that all four would go belly-up. There are four separate supply rails associated with the power amplifiers (+38V, -38V, +27V, -27V). I doubled-checked these rails again to be sure.

I then disconnected the pre-amplifier signals from the power amplifier board (PA). Instant quiet. Confusing, as the pre-amplifiers also feed the stereo line outputs which I knew were working. I then examined the schematic for anything that may be common to all power amplifiers, and I came across the muting circuit. Power amplifier outputs are muted via two relays (RY-1 & RY-2 on PU). These relays are controlled by IC501 on PU. IC501 is a power amplifier protection circuit IC (TA7317P). This IC has internal over current detection, DC voltage detection, muting and relay driver circuits. One of the detection inputs is derived from the Tone Cabinet connector board (TC) at the rear of the E-75. There is a switch on TC used to remote control connected tone cabinets. There is a -20V supply rail associated with this circuit.

This supply rail was faulty. The smoothing capacitor (2.2uF/35V Tantalum) had failed. The photo above shows the failed capacitor. After replacement, the muting circuit sprang into life... the relays powered on after approximately 4 seconds, and the E-75 delivered all it’s sonic glory. It still fascinates me that a piece of equipment of such value (in it’s day) can be brought undone by a $0.78 component.

Many of early Yamaha electones (including the EX-42, E5-AR and YC-45D) had a Portamento strip above the UK or SK. Tones available on the strip included Trombone, Squawk, Birds and Astro (to name a few).

The Portamento strip on the GX-1 is a little different, as instead of being limited to a few preset tones it doubles as a pitch ribbon for whatever tone has been selected and modified for the SK. On my GX-1 the tone produced by the strip was unstable, whilst the corresponding output of the SK was fine.

Analysis of the Portamento Key On signal lead me to the SS board, which contains a Portamento Key Voltage Delay circuit. This circuit performs S/H on the key voltage to avoid a change in pitch during the sustain time following the release of the strip.

The S/H & Gate circuit is driven by a Clock Generator circuit which consists of SN7400N and SN7473N TTL ICs. Closer inspection of the SS board revealed that the SN7473N was showing signs of heat damage, as was TR24, a 2SC458 transistor associated with the power supply to the TTL ICs.

As I have mentioned earlier, the main supply rails for the GX-1 are +15V and -15V. The TTL ICs require +5V, so Yamaha generated a -10V supply from the -15V rail, and used the differential to supply +5V to the TTL ICs. Unfortunately it appears that when the -15V rail failed and drifted to around -23V it was enough to cook TR24 and the TTL ICs.

I replaced TR24 with a BC548 and replaced the TTL ICs with SN74LS00N and SN74LS73N equivalents I had on hand. This did not entirely correct the issue. Perplexed, I looked around the S/H circuit with an oscilloscope. The delay pulses generated from the clock generator were not stable. Noting that the first SN7400N is used to generate a clock for the SN7473N, I ordered some original SN7400N and SN7473N ICs from the US. These took a few weeks to arrive. I replaced the LS ICs with the new “originals” and re-checked the S/H. Success! It appears that whilst the LS ICs were logically equivalent, in this circuit they were not 100% electrically compatible.

The one remaining step was to tune the portamento strip to the SK. This was simply a matter of adjusting VR4 and VR5 on the KS (Solo Key Voltage board) whilst comparing the output of the portamento strip extremities with the output of the SK C3 and C6 keys respectively.

In an earlier blog entry, I commented on a mystery board that wasn’t documented in the service manual. If you google GX-1 photos, you can see that although Yamaha only made a limited number there are a few differences between various GX-1s. One difference is on panel PN4, to the left of the UK. Some GX-1s have Noise (VCF, VCO and Colour) controls (these effect the UK only) whilst other GX-1s have Random (VCA, VCF and VCO) controls instead which effect both the UK and LK.

The service manual I have documents the 1st configuration, but my GX-1 had the 2nd configuration. In testing, none of the controls had any effect on either the UK or LK, so it was time to investigate and reverse engineer the configuration. Tracing the wiring harness from PN4 lead me straight to the mystery board, which I’ll call XX. The harness continued from XX under the rack.

Upon a closer examination of PN4, things started to fall into place. It was clear that this GX-1 had been built with configuration 1 as the wiring harness for the configuration was still in place at PN4, with the unused wires terminated and bound.

In configuration 1, the Colour control adjusts the frequency of a random noise generator on the MOS board. The output of the noise generator is fed back through the VCO and VCF controls. These two signals are buffered through the PB board and are terminated respectively on V2 (13) and f3 (21) of all UK M boards. As configuration 1 only effects the UK, V2 and f3 on the LK M boards are grounded.

In configuration 2, the buffered signals above have been disconnected from the UK M boards, and the ground connections have been disconnected from the LK M boards. Instead, the Random VCO, VCF and VCA control signals are input into three 4558 op-amps on the XX board. The respective output signals are terminated on the rack backplane. To fully reverse engineer the configuration, I had to remove all the boards from the rack and flip it to examine the backplane. This was no easy task. The photo above shows underneath the rack with backplane exposed.

The VCO and VCF signals were connected to V2 (13) and f3 (21) of all the UK and LK M boards, although there were two wiring mistakes which meant that this area of the GX-1 probably never worked correctly from the time configuration 2 was installed. To implement the Random VCA, Yamaha linked LC3 (62) on the backplane for all UK and LK M boards, and hand wired LC3 on each M board with a resistor that terminated in the VCA circuit. Random VCA was achieved by using a different resistor value on each M board. Because there is no Colour control in configuration 2, Yamaha preset the “colour” via a 47K resistor hand soldered across C (75) and V (36) on the MOS board.

The more I thought about configuration 2, the more in felt like a “hack” by Yamaha. It would be interesting to know the history behind these variations, but I imagine that information has long since been lost within Yamaha. I decided that I preferred configuration 1 so I removed XX along with the associated wiring harness and LC3 links. I also removed the random VCA resistors from the M boards, and removed the 47K resistor from the MOS board. I re-attached the configuration 1 buffered signals to the backplane, along with the appropriate ground connections. Finally, I re-attached the PN4 wiring for configuration 1. Voila! Fully functional Noise (VCF, VCO and Colour) on the UK.

Over the past few weeks I’ve been looking at a few little faults that I observed during functional tests. The first was with the SK. All keyboards have a sub-oscillator which includes waveform function, waveform speed, VCO, VCF and VCA depth controls. In the case of the SK, the VCO and VCF controls were functional but the VCA control was not.

The VCA depth of the selected waveform function can be adjusted in two different ways, either by the VCA depth control, or if SK 2nd Touch sub-oscillator VCA is selected, via the pressure exerted on any of the SK keys. Based on the selection, the SR (Solo Touch Response) board outputs the final VCA depth signal from R4 to LC2 (56) of the MS (SK Master Generator) board.

Measuring LC2 yielded no signal, which let me to believe that SR may be faulty. A visual inspection of SR revealed that the wire meant to be connected to termination point R4 had broken away. A few seconds with the soldering iron was all that was needed to restore full sub-oscillator VCA depth control on the SK. The photo shows SR before the repair was made.

The broken brown wire can be seen floating below the middle electrolytic capacitor. The second fault was observed whilst I was checking functionality of various areas. The GX-1 had been turned on for a few hours, as I like to perform some tests after initial power-up, and again after the circuits have “settled”. I went to turn the GX-1 off, and it would not! This was somewhat disconcerting, and I quickly turned it off at the power point, not wanting to risk a PSU fault damaging other areas of the GX-1.

Re-visiting the PSU design, it seemed that the fault was going to be either the power switch itself, or the capacitor that protects the power switch from arcing. I disconnected the PSU from the main rack and turned it on. I thought I heard a “pop” but there was no visual tell-tale sign that the capacitor had failed. I removed the capacitor and measured it. It read 0.074uF, well over it’s value of 0.047uF. I put the power switch through a rigorous mechanical test, and installed a new 0.047uF/630V capacitor. There has been no re-occurrence of the issue.

During the functional test I observed that the there was no UK Portamento. The Portamento circuit is hosted by the UK Key Assigner board (KAU). Surprisingly, an inspection of KAU revealed that the entire Portamento circuit was missing! Comparison with the LK Key Assigner board (KAL) gave the answer... it appears that the two boards were transposed for no apparent reason. As the only difference between KAU and KAL is the Portamento function, I swapped the two boards back around and Portamento came to life on the UK.

During the functional test I also observed that notes played on the LK were “random” in tone, and did not follow the frequency of the note being played with any degree of predictability. Not surprisingly, after swapping KAU and KAL this issue presented itself on the UK instead. Now seemed a logical time to spend more time testing the Key Assigner boards.

The UK and LK Key Assigners essentially generate two outputs. A Key Voltage (K) representing the key being played, and a Trigger (TR) representing the channel assigned to the key. There are 7 or 8 channels per keyboard (depending upon the GX-1 configuration).

These two signals are responsible for telling each Master Generator board (M) when to play a note, and which note to play. There are 16 M boards for each keyboard (MU11..MU18, MU21..28, ML11..ML18, ML21..ML28). Hence, each key can play two tones simultaneously. For example, if a LK key is pressed, and assigned channel 7, K7 and TR7 are signaled to both ML17 and ML27. The tones generated by the M boards are dependent upon the tone selection panel, and if two tones are selected, they are mixed to taste via the keyboard channel mixer controls.

The K signal is generated using 3-bit Octave and 4-bit Note data which is converted to a Key Voltage (0.125V ~ 4.00V) representing the note being played. IC28 (LM310) is used to buffer the Octave voltage ladder output from the Note voltage ladder input. The output of IC28 (pin 6) did not follow the input (pin 3). I replaced the faulty LM310. I also replaced all the 4000-series CMOS ICs on both KAU and KAL as a precaution. The  photo above shows KAU reinstalled after replacement of IC28.

During the functional test I observed that the lower sustain was not working whilst upper and pedal sustain were working fine. The test point for lower sustain is ML11 SUS(72) and should measure +10V to 0V for sustain off to full on. Not surprisingly, the test point measured +10V regardless of the lower sustain panel control lever position. This suggested a fault with the control’s buffer amplifier.

The GX-1 designers loved buffer amplifiers. They are present for  many panel effect lever control signals as well as tone waveform parameter control signals. The board containing the suspect buffer is the PB board, which also hosts overtone preset selection and foot switch control circuits.

Each buffer is essentially just a complementary transistor pair, consisting of a 2SC458, 2SA561 and a few 1S2473 diodes. In the suspect buffer I replaced one 1S2473 diode (from the supplied box of spare parts) and both transistors (TR28, TR43) with BC548 and BC558 substitutes. Please note that these substitutes have different pinouts. There are other substitutes with identical pinouts, but they were not as easy to source.

A check of the test point confirmed the correct voltages present whilst adjusting the control lever, and a functional test confirmed that lower sustain was now present. The photo above shows the PB board after replacing the faulty components. Like the SM board, the wiring harness is soldered direct. Given the 60+ wires soldered to the board, I decided not to replace the board with the “new” spare I had. Speaking of which, I was surprised to note that the spare PB board I have has part number 15923, whilst the installed board has part number 15922. The layout of the two boards is different. It would seem that the design of the GX-1 was still evolving during the period of manufacture.

Further proof of this is the small additional circuit board in the photo above. There is no documentation for this board in either service manual that I have. According to the manuals it simply does not exist. When I have time I’ll trace the wiring harness to see where it terminates.

When testing the pedals I observed that D1 was “stuck”. The only way to cancel D1 was depress a pedal above D1, and once the higher pedal was released D1 sounded again.

The photo shows the pedal switches, which are wired in a simple low-to-high pedal priority arrangement. Although suspecting the pedal switch, I confirmed the switch was at fault before removing the pedal board, extracting the base board, and exposing the pedal switch assembly.

The pedal switch assembly connects to the KP board, which generates the pedal key voltage. With the GX-1 turned off and the KP board removed it was a simple procedure to determine that there was a closed circuit between pins E(3) and D1(10) of the KP edge connector socket, hence the pedal switch was at fault.

From the photo you can see the bent contact on the D1 switch. I’m curious as to how the pin became bent in the first place, but straightening the contact will likely be the easiest repair I’ll have during the restoration of the GX-1.

Tones are selected for each keyboard via the tone selector panel. For the Solo Keyboard, one tone can be selected from a bank of 10. For the Upper, Lower and Pedal Keyboards, two tones for each keyboard can be selected from a bank of 20. For these keyboards, the two tones can be mixed as desired.

In the GX-1 the PSM board is responsible for UK, LK and PK preset tone selection and memory. The SM board is responsible for the SK preset tone selection and memory. Tone selection was fully functional except for the SK.

When the GX-1 is turned on, Preset 1 is the default selection for each keyboard. In the case of the SK, Preset 9 was illuminating, and Preset 10 was not selectable. The selection logic on the SM board is implemented with a series of OR/NOR gates which feed into control and lamp driving circuits.

Given that the gates on the SM board are CMOS, and are known to deteriorate with age I decided to replace the board with a “new” spare supplied in a box of parts with the GX-1.

Replacement was not straight forward. Whilst Yamaha generally design their products well for servicing, there are a few curious design decisions in the GX-1. One is that edge connectors were used on most, but not all boards. The SM board is an example of the latter. To replace the board, 36 wires required unsoldering and re-soldering to the replacement board. The photo above shows the replacement SM board after the soldering was complete. Testing confirmed that the SK preset tone selection was now functioning as expected. I put the original SM board aside for repair at a later date.

A functional test of the GX-1 confirmed what I already knew... there were some challenges ahead. The only section producing audible output was the rhythm generator and it appeared fully functional. There was no audible tone generation from the solo, upper, lower, or pedal keyboards.

When faced with electronic repair, the power supply unit (PSU) is generally a good place to start. The previous owners purchased this GX-1 from the Tom Lee Piano Company in Hong Kong and imported the GX-1 into Australia. Hence this GX-1 has a General Specification PSU with Voltage Changer (according to the service manual the Australian PSU shipped without a Voltage Changer).

Inspection of the PSU confirmed that it had been correctly adjusted for 240V. Essentially, the GX-1 runs on three power supply rails; +15V and -15V for the main rack, and +10.6V for the control system. Measurement of these rails yielded +15V, -23V and +10.6V respectively. Each supply rail uses a uPC141A Voltage Regulator, and in the case of the -15V rail, one of the feedback voltage divider resistors (15K) was open circuit.

Replacement of this resistor was all that was required to correct the rail, however I replaced all the Electrolytics and Tantalums in the -15V supply whilst it was convenient. The photo above was taken after I’d replaced the resistor and capacitors in the -15V circuit and cleaned the PCB. A functional test after re-installing the PSU showed promise. All keyboards and pedals now had audible output. Not great output, but a step in the right direction.