ARP Quadra service notes
Posted: Mon Mar 19, 2012 12:24 pm
In 1978 I attended factory training for ARP instruments. Back in those days ARP, SCI and Moog invested in training service technicians. In this case it was an all expenses paid trip to Denver for a week of classroom sessions. The Quadra was introduced that year and we were issued a preliminary set of schematics - 30 loose sheets of 11 X 17 schematics only. Over the next 34 years I worked on a total of about 3 of these instruments, and none of those repairs required any deep understanding of what made this design unique. I have one in the shop now that turned out to be quite a challenge, and I hope this report will save the next tech who goes there some time and aggravation.
***********
The ARP Quadra was a unique and innovative state of the art design for a very brief period in electronic music history. The schematics I have are dated 1978 - back when the use of microprocessors was in the infancy stages of circuit design. Oberheim and SCI were already on the verge of releasing their new microprocessor designs which would make the Quadra obsolete almost before it was released. One apparent consequence of these changing times was that ARP never published a service manual for the Quadra. If they ever did, I did not get my copy. 34 years later I am stuck fixing Quadras without the benefit of any circuit descriptions, completed block diagrams, error corrected schematics or PCB layout sheets. Fortunately the PCB artwork was pretty complete with key component designations silk screened on the component side and etched into the solder side of the boards.
The combination of old school analog circuits interfaced with primitive micro processor control created some confusion in the absence of any circuit descriptions. I have not seen any design quite like this on any other keyboard. The documentation I am writing here will serve as a circuit description for a complex gating arrangement dreamed up by ARP engineers. A design that faked me out of my sanity for a while.
When the Quadra being discussed came in for service it seemed to be mostly functional. Externally it appeared to be in excellent cosmetic condition except for the four sliders with sheared off handles. There were some straight forward logic problems causing some switching errors, the after touch was not responding hardly at all, the key contacts were in atrocious condition, and the gating was somewhat messed up. Putting the keyboard mode switch in the multi-trigger position made it appear that it mostly worked, but in single trigger mode not much worked as it should.
The switching logic was an easy fix. After touch was a matter of mechanical adjustment - again an easy fix. The key contacts and buss bar alignment was about as ugly as I've seen. After observing that each key switch functions as a single pole DOUBLE throw switch, and that the key up closure is as important as the key down closure, contact adjustment was pretty straight forward. Each key switch performs 2 functions. A digital signal is sent to the microprocessor and an analog gate circuit initiates the Envelop Generators and biases on the waveforms for strings and poly synth on the voicing boards. The processor responds to the digital signal and generates the CV for the bass and lead voices. It also produces trigger pulses when the keyboard control is set to multiple trigger mode or when the arpeggiator is on. The gate signals however originate from analog circuits on the Keyboard Electronics and Lower Voicing pcbs.
This keyboard has three separate gate buses designated lower Octave 1 Buss, 2nd Octave Buss, and Upper Buss. As I was probing around looking for the correct gate signals and reasons why they were not present, I discovered evidence that a previous technician had been chasing the same problems. I found related components had been replaced and also that the Octave 1 and 2nd Octave buses were bridged with a blob of solder. With enthusiasm I removed this "mod" thinking that now it will start to make more sense. Unfortunately, removing the bridge only led to more confusion. The circuit behavior became more unpredictable and confusing than before lifting the bridge. Now we get to the interesting part.
Each key switch (SPDT - single pole double throw) pole is connected to a CMOS non inverting buffer - a CD4050 located on the Keyboard Electronics pcb. These are hex buffer arrays so there is one chip for every 6 notes. Pin 1 (Vdd) of the buffer IC is connected to ground and pin 8 (Vss) is connected to -12V VIA a current sense circuit (see photo 1). R88 supplies voltage to buffer IC pin 8 with NO KEYS DEPRESSED. The key switch poles are at ground with key up, the buffer output is at 0.0 Volts, and there should be NO current load generated by the buffer IC. R88 will read 12.0 V on both sides. When a key is depressed the buffer drives a key voltage to the Voicing boards and then loads the Vss supply. The load creates a voltage drop across R88 which then forward biases CR 50. CR50 sources all necessary current to the buffer ICs and the voltage on the buss remains constant at a diode drop above -12V supply. Z 24D compares this voltage with a reference set by R87 and R89 and generates the Gate signal as long as the buss sees a load from any key depression connected to that bus.
The Gate circuits for 2nd Octave and Upper buses are located on the Lower Voicing board. These two comparator circuits are configured to output the opposite polarity from the Octave 1 circuit as some additional logic is employed before the Gate signal is generated. Other than that they function the same as the Octave 1 buss. 2nd Octave buss connects to Z3 and Z4 pin 8s. Upper buss is connected to Z5 thru Z11 pin 8s. Both of these buss comparator circuits were not behaving predictably. Strangely enough they sometimes worked when either a scope probe or meter leads were connected to one or the other comparator input, but sometimes they functioned briefly depending on some parameter I could not identify. I did find a "wrong" value resistor @ R162. The schematic designates 2.2K ohms but R162 was 2.2M ohms. I suspected a schematic error since the 2n Octave buss has 2.2M in this slot, but of course, not knowing what I know now, I had to check it out. That circuit does not even come close to working with 2.2K @ R162.
What I found was a 200 to 300mV drop across R147 (in the case of the Upper buss circuit) with no key depressed, and a similar but different condition on the 2nd Octave comparator. While all 12 of the CD4050 hex non inverting buffers were fully functional, 2 of them - one on each of the buses, were causing a current load on the buffer lines with no key depressed. There is no easy way to measure the current draw on each IC. To prove this I had three choices. #1: I could remove each IC one at a time. This board is no fun to work on and very time consuming to do a neat job of this, so forget that. #2: DIP (Dual Inline Package) pins can be carefully clipped such that they can be reliably reconnected with solder if you know how to do it. It requires good quality small cutters with the appropriate shaped nose. Mine have been misplaced and I use my backup cutters which don't do such a neat job. I opted for #3: Make a surgical incision on each pin 8 trace to isolate those ICs from the buss one at a time. This was the obvious choice dictated by the pcb layout making it easy to do so. The idea is to make one clean cut at an angle so one side of the trace can be lifted slightly creating a certain open circuit.
After the lifted trace is pressed back down there should be no gap and solder will flow over the cut assuring a reliable connection. If you damage the trace leaving a gap it is a good idea to bridge the gap with some wire.
The 2nd Octave buss has only Z3 and Z4. I cut the pin 8 trace to Z3 and presto - no more load on the bus. The Upper buss has 8 CD4050s. I was not so lucky and had to cut 5 traces before the errant load disappeared. Reconnecting them one at a time I tested each one to make sure there was only one CD4050 with this ailment. After removing these "rogue" ICs, the gate comparators worked reliably and I gained confidence in the design. Pretty innovative as long as you can count on CD4050s not drawing any current when connected in this way. Looking at the data sheet I don't necessarily see any guarantees.
All of this probing and testing made it necessary to remove the Keyboard Electronics board from the keyboard assembly multiple times. The connectors are supposed to be held snug to the Keyboard Electronics pcb by means of flimsy little plastic arms that clip over the edge of the fab. These are not adequate, and the pressure of inserting the pins when re-seating the pcb to the keyboard assembly causes these connectors to get pushed back stressing out the solder tabs and becoming miss aligned with the mating pins. Some hot melt glue was an easy and effective remedy.
***********
The ARP Quadra was a unique and innovative state of the art design for a very brief period in electronic music history. The schematics I have are dated 1978 - back when the use of microprocessors was in the infancy stages of circuit design. Oberheim and SCI were already on the verge of releasing their new microprocessor designs which would make the Quadra obsolete almost before it was released. One apparent consequence of these changing times was that ARP never published a service manual for the Quadra. If they ever did, I did not get my copy. 34 years later I am stuck fixing Quadras without the benefit of any circuit descriptions, completed block diagrams, error corrected schematics or PCB layout sheets. Fortunately the PCB artwork was pretty complete with key component designations silk screened on the component side and etched into the solder side of the boards.
The combination of old school analog circuits interfaced with primitive micro processor control created some confusion in the absence of any circuit descriptions. I have not seen any design quite like this on any other keyboard. The documentation I am writing here will serve as a circuit description for a complex gating arrangement dreamed up by ARP engineers. A design that faked me out of my sanity for a while.
When the Quadra being discussed came in for service it seemed to be mostly functional. Externally it appeared to be in excellent cosmetic condition except for the four sliders with sheared off handles. There were some straight forward logic problems causing some switching errors, the after touch was not responding hardly at all, the key contacts were in atrocious condition, and the gating was somewhat messed up. Putting the keyboard mode switch in the multi-trigger position made it appear that it mostly worked, but in single trigger mode not much worked as it should.
The switching logic was an easy fix. After touch was a matter of mechanical adjustment - again an easy fix. The key contacts and buss bar alignment was about as ugly as I've seen. After observing that each key switch functions as a single pole DOUBLE throw switch, and that the key up closure is as important as the key down closure, contact adjustment was pretty straight forward. Each key switch performs 2 functions. A digital signal is sent to the microprocessor and an analog gate circuit initiates the Envelop Generators and biases on the waveforms for strings and poly synth on the voicing boards. The processor responds to the digital signal and generates the CV for the bass and lead voices. It also produces trigger pulses when the keyboard control is set to multiple trigger mode or when the arpeggiator is on. The gate signals however originate from analog circuits on the Keyboard Electronics and Lower Voicing pcbs.
This keyboard has three separate gate buses designated lower Octave 1 Buss, 2nd Octave Buss, and Upper Buss. As I was probing around looking for the correct gate signals and reasons why they were not present, I discovered evidence that a previous technician had been chasing the same problems. I found related components had been replaced and also that the Octave 1 and 2nd Octave buses were bridged with a blob of solder. With enthusiasm I removed this "mod" thinking that now it will start to make more sense. Unfortunately, removing the bridge only led to more confusion. The circuit behavior became more unpredictable and confusing than before lifting the bridge. Now we get to the interesting part.
Each key switch (SPDT - single pole double throw) pole is connected to a CMOS non inverting buffer - a CD4050 located on the Keyboard Electronics pcb. These are hex buffer arrays so there is one chip for every 6 notes. Pin 1 (Vdd) of the buffer IC is connected to ground and pin 8 (Vss) is connected to -12V VIA a current sense circuit (see photo 1). R88 supplies voltage to buffer IC pin 8 with NO KEYS DEPRESSED. The key switch poles are at ground with key up, the buffer output is at 0.0 Volts, and there should be NO current load generated by the buffer IC. R88 will read 12.0 V on both sides. When a key is depressed the buffer drives a key voltage to the Voicing boards and then loads the Vss supply. The load creates a voltage drop across R88 which then forward biases CR 50. CR50 sources all necessary current to the buffer ICs and the voltage on the buss remains constant at a diode drop above -12V supply. Z 24D compares this voltage with a reference set by R87 and R89 and generates the Gate signal as long as the buss sees a load from any key depression connected to that bus.
The Gate circuits for 2nd Octave and Upper buses are located on the Lower Voicing board. These two comparator circuits are configured to output the opposite polarity from the Octave 1 circuit as some additional logic is employed before the Gate signal is generated. Other than that they function the same as the Octave 1 buss. 2nd Octave buss connects to Z3 and Z4 pin 8s. Upper buss is connected to Z5 thru Z11 pin 8s. Both of these buss comparator circuits were not behaving predictably. Strangely enough they sometimes worked when either a scope probe or meter leads were connected to one or the other comparator input, but sometimes they functioned briefly depending on some parameter I could not identify. I did find a "wrong" value resistor @ R162. The schematic designates 2.2K ohms but R162 was 2.2M ohms. I suspected a schematic error since the 2n Octave buss has 2.2M in this slot, but of course, not knowing what I know now, I had to check it out. That circuit does not even come close to working with 2.2K @ R162.
What I found was a 200 to 300mV drop across R147 (in the case of the Upper buss circuit) with no key depressed, and a similar but different condition on the 2nd Octave comparator. While all 12 of the CD4050 hex non inverting buffers were fully functional, 2 of them - one on each of the buses, were causing a current load on the buffer lines with no key depressed. There is no easy way to measure the current draw on each IC. To prove this I had three choices. #1: I could remove each IC one at a time. This board is no fun to work on and very time consuming to do a neat job of this, so forget that. #2: DIP (Dual Inline Package) pins can be carefully clipped such that they can be reliably reconnected with solder if you know how to do it. It requires good quality small cutters with the appropriate shaped nose. Mine have been misplaced and I use my backup cutters which don't do such a neat job. I opted for #3: Make a surgical incision on each pin 8 trace to isolate those ICs from the buss one at a time. This was the obvious choice dictated by the pcb layout making it easy to do so. The idea is to make one clean cut at an angle so one side of the trace can be lifted slightly creating a certain open circuit.
After the lifted trace is pressed back down there should be no gap and solder will flow over the cut assuring a reliable connection. If you damage the trace leaving a gap it is a good idea to bridge the gap with some wire.
The 2nd Octave buss has only Z3 and Z4. I cut the pin 8 trace to Z3 and presto - no more load on the bus. The Upper buss has 8 CD4050s. I was not so lucky and had to cut 5 traces before the errant load disappeared. Reconnecting them one at a time I tested each one to make sure there was only one CD4050 with this ailment. After removing these "rogue" ICs, the gate comparators worked reliably and I gained confidence in the design. Pretty innovative as long as you can count on CD4050s not drawing any current when connected in this way. Looking at the data sheet I don't necessarily see any guarantees.
All of this probing and testing made it necessary to remove the Keyboard Electronics board from the keyboard assembly multiple times. The connectors are supposed to be held snug to the Keyboard Electronics pcb by means of flimsy little plastic arms that clip over the edge of the fab. These are not adequate, and the pressure of inserting the pins when re-seating the pcb to the keyboard assembly causes these connectors to get pushed back stressing out the solder tabs and becoming miss aligned with the mating pins. Some hot melt glue was an easy and effective remedy.