When a project doesn't work, the most powerful trouble shooting technique available is:
TAKE A BREAK.
Clear your mind. Find something else to do for a while, maybe put the project away for the night. When a newly assembled device doesn't work everyone's first impulse is to suspect that there's a bad part, but statistically this is rare. It is more likely that the problem is either a bad solder joint, a missing or broken wire or a component intalled in the wrong place or with improper polarity. Often, a rested eye and mind can spot errors that were completely invisible a few hours earlier.
Now, bad parts do happen - maybe a transistor was bad when you received it or an IC got too hot during installation or zapped by static somewhere along the line. If you feel that a part is bad you will find us very cooperative in replacing it as quickly as possible. To receive replacement parts, send email to firstname.lastname@example.org.
Scott Lee, PAiA Tech Services Director, has spent so many hours of head scratching figuring out what keeps kits from working that it's a wonder he has any hair left. Here are some of his notes on things he has seen go wrong with our more popular projects:
The 12V wall mount DC supply powers the board via the G and + wires. You can touch the solder lug which has one end of the G wire and the power supply (-) for the DC readings. The 12V going in to wiring point + goes to resistor R1. One side of the resistor should measure about 12V and the other end (the end attached to 8.2V Zener Diode, D1) should measure 8.2V. This voltage is labeled V+ on the schematic and you should find this 8.2V on the components on the board labeled with V+ on the schematic. This voltage is divided in half for 4.1V by the equal valued series resistors at Rs 22 and 23. This voltage is labeled Vr on the schematic and again, you should be sure that components on the board are getting the voltage as specified on the schematic. Since these voltages go to so many parts (ground/common too) they often go through jumper wires to get there. These are likely spots for an open circuit either as a result of the jumper not reaching all the way through to both ends of the printed circuit, or the tiny solder pad and connecting trace in the printed circuit for the jumper wires breaking.
IC2 gets its V+ power supply in on pin 7 (viewed from the top of the board, the pins count up around the part in a ccw direction from the notch). IC1 gets power on pin 3.
The legs of the transistors are other good spots to check for DC levels. Looking at the flat face of the part, with the legs down, they read l-r E for Emitter, B for Base, and C for Collector (on the schematic, the base is the center line "into" the part, the collector is the upward slanting line, and the emitter is the one with the arrow point, slanting downwards).
For transistors Q1-4, you should find about 2V on the emitter, 2-3V on the base, and 7V on the collector. Also, notice on the schematic that the circuits for the four oscillators are just about identical (two have panel controls, Volume Trim and Pitch Trim) while the others have a similar resistive circuit without the adjustment). DC voltage comparisons can be made between similar points in the circuits for oscillators 1 & 4 and oscillators 2 & 3. This might help isolate trouble in this area.
For transistors Q8 and Q9, you should find about 0.1V on the emitter, about 0.5V on the base, and about 2-5V on the collector. The transistors at Qs 8 and 9 work as amplifiers for the pitch and volume heterodyne (beat) frequencies and the appx. 1V p-p audio signal rides on the DC level on the collector circuit of these transistors.
The signal from the collectors of transistors 8 and 9 go to comparator stages which switch on (0V) and off (8.2V) at the frequency of the volume or pitch beat frequencies (there will only be beat frequencies when the oscillators are tuned to very near the same frequncy (null). A DC reading here (IC1:pin1 for volume, IC1: pin2 for pitch) should be in the range of 1 to 7 volts, but not 0 or 8.2 which would indicate trouble in this area, or more likely missing audio beat frequencies.
The voltage switching at the IC1 pins 1 and 2 goes to a circuit of diodes and a capacitor where its is averaged as Control Voltage and buffered for connection to the output jacks through transistors 5 and 6, for the Pitch and Volume CV outs.
The Pitch CV output could be missing and you'd probably never know, but the Volume CV output has to be there to turn "on" the VCA section comprised of transistors 10, 11 and 12. The Volume CV becomes a current flow into Q12 which allows current to flow through transistors 10 and 11 which allows the Pitch tone to get to IC2 which drives the audio output. IC2's output will measure appx. 4.1V if things are right in the VCA circuit.
Sometimes, even though the DC readings in the oscillator circuits are right, they may not be oscillating with sufficient strength or in the desired freqency range. An AM radio can be tuned to a quiet spot in the range from 6-10 on the dial to test the operation of any of the four oscillator circuits. Their frequency is adjusted with the slug in Ls 1-4 and when they match the radio tuning you hear the oscillator 'transmitting' to the radio. When operation is confirmed, adjust the oscillator 'off frequency' and try the next.
An open circuit (bad solder or broken printed circuit) in the oscillator transistors emitter resistor (56k, green-blue-orange-gold) to ground can be the cause of trouble where the DC looks OK, but the oscillation isn't. With power off, a resistance test from the transistor (Q1-4) emitter to ground should read about 56kOhms.
The IF transformers rank right up there with shielded cable and the power switch as being things that are particularly heat sensitive (the manual only mentions being careful with the diodes, transistors, and ICs, but they're really more tolerant than these 'plastic' parts. The pins on these transformers don't like being splayed too much during installation either. The wire used for the coil windings can detach with heat from soldering or movement of the terminal from bending too much in installation. A mulitester with a digital readout will indicate the resistance of the L1-4 (I.F. Transformer/oscillator coil) windings confirming there aren't open circuits in them.
Look on the bottom of the board for the five terminals of these parts and notice that pin 1 has a connection with the 680ohm (blue-grey-brown-gold) resistor to V+. Count from there around the part in a cw direction for pins 2-5. Here are the resistances of the wire comprising the windings and between these terminals (the side with three terminals has a tap towards one end of the winding): From one to two look for about 0.5 ohms; from 2 to three, 4.1-4.2 ohms; one to three 4.1 to 4.2 ohms; and, from four to five, 0.5-0.6 ohms
The first Theremax assembly manuals instructed calibration for "controller" mode (hand to volume antenna for volume increase) while now it is for "traditional" mode (hand to volume antenna for volume decrease) . As a controller, it seems more natural to move your hand towards the antenna for increasing volume CV and Gate/S-Trigger activation, but theremin enthusiasts recognize this as opposite the way the original instruments operated.
When calibrated in the "controller" mode, the volume pair is at null and hand motion about the volume antenna results in a sweep through audio frequency heterodyne signals just like the Pitch side of the circuit. This energy manages to find it's way into the audio output just above the noise floor or enough that it is noticeable. We have the sine wave getting to the VCA section fairly hot and the high frequency response in this section and the following op-amp output section fairly low to help mask the bleed of the audio frequency volume het.. This results in distortion of the intended Timbre waveforms (sine at ccw and square at cw).
Now, the calibration procedure is for "traditional" control response. The volume het. sits at the edge of audio frequency and increases with inward hand motion - the volume bleed isn't an issue and theremin buffs are happy because it's like the instrument they know. Here are some changes that get the sine and square back like they should be: Put an "attenuating" resistor (1k to 10k (6800 is often "right")) from lugs 3 to 2 on the Timbre control to bring the sine out of overdrive which sounds harsh or more squarelike. Select the resistor by listening to the sound with the Timbre ccw while touching resistors to lugs 3 and 2 and listening for the highest value that will cause a smooth tone (most noticeable with the following changes). Replace the capacitors at Cs 39 and 44 with 100pF to increase the high frequency response. This returns the "edge" to the square wave. You'll loose some of the immunity to the volume bleed if calibrated for "controller" response, but in this mode, you're usually not listening to the audio output but the device being controlled.
The audio bands each side of the volume null are what makes the two volume peaks that you encounter with inward adjustment of L3. The f-v circuit which makes the volume CV, has a frequency response which causes the CV to fall as the het increases above audio and C37's value affects this response. Sub'ing a 220pF (I've seen 100pF do good here too) for the 33pF here improves the volume cut with inward hand motion. If you find you have to go to close to the antenna for much change in volume, put the bigger capacitor at C37.
UPDATE. The C37 change also allows you to calibrate for a controller mode without the problem of vol. het. bleed too. Instead of going for null in this mode, go beyond the second volume peak til the sound just stops and stop with the inward adjustment of L3 - as the hand moves to the antenna, the volume goes up. The vol. het. is post audio frequency at this setting, so there is now controller mode without the bleed.
If the audio is present and the +5, +8, and -12V supplies are right, check to find out if the Pitch CV output changes when notes are played on the MIDI Controller. If not, it may be trouble with the MIDI or the processor not executing the program contained in the FatMop EPROM. There are some simple things that can keep the MIDI from being acknowledged and lots of things that can keep the processor from executing the program.
If the DIP switch used for selecting MIDI channel isn't doing what it's supposed to, the processor might think the MIDI is on the wrong channel. For instance, channel 1 is all four switches set to On/Closed and this shorts four of the processor port lines to 0V and if for some reason one of the port lines was still at +5, the processor would be looking for a different channel. A bad switch or cold solder joint at the processor or switch could be the trouble. Also, the channel setting is only input to the processor on power-up. Another possibility is that the controller is sending note numbers and velocity under running status and FatMan hasn't received a Note On Status byte yet. Turn the controller off and then on again to ensure the next note pressed has the Status byte for the FatMan, or, press a program change button, roll the pitch wheel or modulation wheel to change Running Status.
The most likely thing that would keep the processor from executing the program in the FatMop EPROM is a soldering problem such as a cold solder joint (insufficient heat to displace/boil away the resin in the solder) or solder bridge (two circuits shorting that shouldn't). Just about all the jumpers in the area of ICs 1,2,3,&4 are associated with the processor and rom or the data latches and these are likely spots for soldering problems. The solder pads for the jumpers are tiny and it can be difficult getting enough heat to both the jumper wire and the pad to make a good joint. They can break from their connecting traces easily too; either from force from the jumper on top of the board or from excess heat causing the pad to lift from the board, or both. They are close to each other too; look for bridges. If the solder looks like a smooth dome and there isn't evidence the excess wire needed to be clipped, it might be an indication that the jumper dropped back out of the board when it was turned up-side down to be soldered.
If you can't see any of the above-mentioned problems, a multitester might help. Look at the schematic for ICs 1,2,3,&4 and measure voltage on the pins of the processor IC and be sure the corresponding pins on ICs 2,3,&4 read the same. Be suspcious of shorts between adjacent pins that read identical voltages. Ground pins should measure 0V.
Here are some common troubles I've noticed people seem to have in the assembly EPFM projects:
Review Chapter Five, Connecting Power to the Projects pages 59-62 and Jumper Wires, page 62. Many people erroneously connect a single nine volt battery to power the circuit and end up wiring the jumper wire points to one of the 1/4" phone jacks (pages 210-212 cover DC adapters and obtaining a bipolar supply from a unipolar supply).
The 4739AK 5532 to 4739 adapter kit instructions regarding other than isolated solder pads for the printed circuit layout for the 4739 IC on the project (application) circuit board confuse many people.
Here's the deal: The 4739IC had no internal connection on pins 2, 3, 4, 10, 11, and 12 and because of this, some of the circuit board layouts have circuits running through these pins. The 4739 adapter kit utilizes all fourteen positions of the 14pin DIP pattern. So, once the long-legged, wire-wrap type fourteen pin IC socket has been inserted through the holes of the tiny 4739 adapter kit and all fourteen pins have been soldered, pins 2, 3, 4, 11, or 12 must be clipped (at the top of the joints on the tiny 4739ak pc board) if the application (EPFM project) board has circuits running through these pins. It is OK to simply clip these pins even if they don't have circuits running through them, this just increases the possibility of clipping one that you shouldn't.