Upon our inspection of this organ to verify the reported symptoms, Jim was surprised to hear that the problems were fewer and less severe than they had been previously. This was an indication that some or most of the trouble might be temperature related as, it was cold that day and, the organ had been either outside or in the back seat of the car for the last hour or so. As the unit warmed up in the shop the problems got worse - more to Jim's expectations.
Starting with the worst sounding note card, "C", I verified the errant signals with the scope. Most of the divider outputs were modulated by a signal containing dominately the lowest divider output from that card, but also seemed to contain some non-fundamental related noise. By varying the operating temperature (heat gun and freeze mist) the errant modulated sound would come and go. It appeared to be first one transistor, then another, then another - no - maybe it was that cap.
Twice I was convinced a particular transistor (first divider pair) was failing when warm but the circuit behaved the same with those transistors replaced. As the unit continued to warm up the outputs from other cards were showing signs of the same condition - leading to the conclusion that these transistors are all OK, but the bistable multi-vibrator divider circuits were just not so stable due to some other condition yet to be identified.
Starting with the worst sounding note card, "C", I verified the errant signals with the scope. Most of the divider outputs were modulated by a signal containing dominately the lowest divider output from that card, but also seemed to contain some non-fundamental related noise. By varying the operating temperature (heat gun and freeze mist) the errant modulated sound would come and go. It appeared to be first one transistor, then another, then another - no - maybe it was that cap.
Twice I was convinced a particular transistor (first divider pair) was failing when warm but the circuit behaved the same with those transistors replaced. As the unit continued to warm up the outputs from other cards were showing signs of the same condition - leading to the conclusion that these transistors are all OK, but the bistable multi-vibrator divider circuits were just not so stable due to some other condition yet to be identified.An 8V supply line to each card connects to the oscillator circuit via one of two coils it the tuning coil assembly. The same supply powers the divider circuits. This 8V line has the vibrato signal modulating the 8V DC. The coil connection results in frequency modulation of the oscillator and the modulated power supply for the dividers modulates the pulse width of the divider outputs. On closer inspection I found that, in addition to the vibrato signal of 3 or 4 hertz there is a lot of audio noise present at about 10dB below the LF component. I concluded that this was insignificant but upon reflection there may be an additional clue in that measurement. 
This supply line connection to the dividers is made from point "A" on the schematic. A table printed on the schematic shows a resistor value connecting point "A" to point "B". The value of this resistor varies from 330 ohms on the DO (C) card down to a direct connection on the lowest 4 cards. C - the highest pitched card, has one additional divide by output exclusive to the C card for providing the 16' tone for low C. This was the worst case note with the garbage modulation present on all octaves of the C pitch. I found noise and voltage drop across that resistor (330 ohms 10%) that was not present at that location on a good sounding card. That resistor checked out of tolerance at about 390 ohms. With the resistor replaced some of the noise went away - further reducing the value further reduced the noise until shorting points A and B, as specified for the lowest 4 cards, produced the best sounding tones. This process was repeated for the 7 other boards with resistors between Point A and point B, substituting values or shorting out A to B on most of those boards. Two of the 8 boards sounded best with the specified resistor value left in place - F# and G if I recall correctly - the two boards with the smallest specified resistor values.
This supply line connection to the dividers is made from point "A" on the schematic. A table printed on the schematic shows a resistor value connecting point "A" to point "B". The value of this resistor varies from 330 ohms on the DO (C) card down to a direct connection on the lowest 4 cards. C - the highest pitched card, has one additional divide by output exclusive to the C card for providing the 16' tone for low C. This was the worst case note with the garbage modulation present on all octaves of the C pitch. I found noise and voltage drop across that resistor (330 ohms 10%) that was not present at that location on a good sounding card. That resistor checked out of tolerance at about 390 ohms. With the resistor replaced some of the noise went away - further reducing the value further reduced the noise until shorting points A and B, as specified for the lowest 4 cards, produced the best sounding tones. This process was repeated for the 7 other boards with resistors between Point A and point B, substituting values or shorting out A to B on most of those boards. Two of the 8 boards sounded best with the specified resistor value left in place - F# and G if I recall correctly - the two boards with the smallest specified resistor values.I should confess that it did not occurr to me to check for pulse width modulation consistency while this unit was on the bench with scope probes in hand. I never did conclude the specific cause of the unstable multi-vibrator circuits, but there is no doubt that resistor value specified in the table affected the symptom. Also no dought that vibrato on or off, there was no effect on the symptom. I still do not understand the function of that resistor - perhaps to result in less pulse width modulation on the higher pitched cards (?) How can it work to attenuate the DC component proportionately with the vibrato LF? Seems like really questionable design criteria and leaves me thinking I've overlooked something. I'm anxious to get another Farfisa on the bench so I can check another theory. Jim has another one scheduled for next week. I'll update this article if I confirm an oversight. Any guesses as to what I may have failed to observe? The clue is in the shcematic.
My job is not to change the design, or even question the design under most circumstances. My job is to make the customer happy, and after this Deluxe Farfisa lost the glitchy sounding notes Jim was very happy. At least until he induced a problem on reassembly. Jim is one of my favorite customers as he wants to learn more about working on his own gear. Consequently he assisted and observed from start to what should have been the finish.
Five thumb screws hold the chassis inside the case. These screws thread up from underneath the case. Two long screws go one on each end, the other three shorter screws go one in front center and two towards the rear. I dumped out the screws and reminded Jim that the two long screws go on the ends. We start installing the screws, I, not paying much attention to Jim, but focusing instead on aligning the threads so I can get the first screw installed. A long screw in the end location. Before I realized what was going on, Jim had threaded a long screw so far into a short hole that the screw deflected the F# circuit board far enough to break the circuit board into.
Jim felt really bad about what had happened of course. I told him not to worry, that I could fix it and it would play OK again. He seemed skeptical so I reassured him that it would be OK but probably cost him another $50, and then sent him home for the night. Soooo - I get to write about repairing a broken printed circuit board (pcb).
The dark green circuit board color comes from a solder resist film applied to the pcb. My experience would suggest that this is done so that when the board goes through the solder flow operation, solder
only sticks where it is needed. The empty solder pads in the photo, with no solder resist and no solder, indicate that this board did not go through a solder flow operation and therefore must have been hand soldered. Hmmmm. (?)
The first step in repairing this damage is to remove the solder resist on all of the broken connection traces accross the fracture in the pcb. Use a sharp knife to scratch through the film as seen in this photo. In the middle of the photo you can see the exact point of contact of the guilty screw. I use a fine wire brush to clean up a little more, but it isn't really necessary. Once the solder resist is scored like this, solder will flow cleanly over this copper at 700 degrees - (next photo).
With the pcb prepaired with fresh solder flowed cleanly over the areas to repair, the wire is prepaired in the same way. This process of bonding solder to metal surfaces to be solder connected is called "tinning". Surfaces must be properly tinned prior to making the best possible solder connection. Use some stranded small gauge wire. Strip insulation and tin the wire. Embed the wire in the tinned traces bridging the cracked pcb using mostly the solder already on the trace and wire. Add additional solder as needed, and cover as much of the traces as possible with stranded wire embedded in solder. The technique is a little tricky - only apply enough heat to melt the solder in one small spot. Don't apply enough heat to melt all of the solder holding the wire as the
whole wire will move and smear solder - start over when this happens. Apply only enough heat to melt the solder in the immediate area of application. This allows for bending the wire to the exact radius needed to match the trace.
Common solder for electronic application is 60% tin and 40% lead. We were taught back at Tektronix that 63/47 solder is preferred for all industrial use. 63/37 has a liquid/solid vs temperature graph that results in significantly different and advantageous behavior through the
cooling cycle. The liquid / solid tranformation happens in a much narrower tempertature window resulting in minimal cold solder cooling. I recommend a fine guage, rosin core 63/37 solder as shown from Kester here.
My job is not to change the design, or even question the design under most circumstances. My job is to make the customer happy, and after this Deluxe Farfisa lost the glitchy sounding notes Jim was very happy. At least until he induced a problem on reassembly. Jim is one of my favorite customers as he wants to learn more about working on his own gear. Consequently he assisted and observed from start to what should have been the finish.
Five thumb screws hold the chassis inside the case. These screws thread up from underneath the case. Two long screws go one on each end, the other three shorter screws go one in front center and two towards the rear. I dumped out the screws and reminded Jim that the two long screws go on the ends. We start installing the screws, I, not paying much attention to Jim, but focusing instead on aligning the threads so I can get the first screw installed. A long screw in the end location. Before I realized what was going on, Jim had threaded a long screw so far into a short hole that the screw deflected the F# circuit board far enough to break the circuit board into.
Jim felt really bad about what had happened of course. I told him not to worry, that I could fix it and it would play OK again. He seemed skeptical so I reassured him that it would be OK but probably cost him another $50, and then sent him home for the night. Soooo - I get to write about repairing a broken printed circuit board (pcb).The dark green circuit board color comes from a solder resist film applied to the pcb. My experience would suggest that this is done so that when the board goes through the solder flow operation, solder
only sticks where it is needed. The empty solder pads in the photo, with no solder resist and no solder, indicate that this board did not go through a solder flow operation and therefore must have been hand soldered. Hmmmm. (?)The first step in repairing this damage is to remove the solder resist on all of the broken connection traces accross the fracture in the pcb. Use a sharp knife to scratch through the film as seen in this photo. In the middle of the photo you can see the exact point of contact of the guilty screw. I use a fine wire brush to clean up a little more, but it isn't really necessary. Once the solder resist is scored like this, solder will flow cleanly over this copper at 700 degrees - (next photo).
With the pcb prepaired with fresh solder flowed cleanly over the areas to repair, the wire is prepaired in the same way. This process of bonding solder to metal surfaces to be solder connected is called "tinning". Surfaces must be properly tinned prior to making the best possible solder connection. Use some stranded small gauge wire. Strip insulation and tin the wire. Embed the wire in the tinned traces bridging the cracked pcb using mostly the solder already on the trace and wire. Add additional solder as needed, and cover as much of the traces as possible with stranded wire embedded in solder. The technique is a little tricky - only apply enough heat to melt the solder in one small spot. Don't apply enough heat to melt all of the solder holding the wire as the
whole wire will move and smear solder - start over when this happens. Apply only enough heat to melt the solder in the immediate area of application. This allows for bending the wire to the exact radius needed to match the trace.Common solder for electronic application is 60% tin and 40% lead. We were taught back at Tektronix that 63/47 solder is preferred for all industrial use. 63/37 has a liquid/solid vs temperature graph that results in significantly different and advantageous behavior through the
cooling cycle. The liquid / solid tranformation happens in a much narrower tempertature window resulting in minimal cold solder cooling. I recommend a fine guage, rosin core 63/37 solder as shown from Kester here.
I know why the removal / changing resistor values has made the organ work.
ReplyDeleteThese transistor devices are GERMANIUM and are notoriously LEAKY when in the reverse bias condition.
This leakage increases with age and also when the transistors are on their last legs and one step away from failing.
By changing/ removing resistor values, you can force the transistor to operate on a VERY TEMPORARY BASIS.Having said that, the trnsistor tends to loose the low end frequency response as the transistor is now incorrectly BIASED and operating on the wrong part of the characteristic /load line curve.
The main danger of this is that by forcing the transistor to operate in this manner you are inctreasing the COLLECTOR current to an already weakened transistor.
This will inevitably lead to a condition called THERMAL RUNAWAY at which time the transistor will fail, usually short circuit and this can cause much damage.
I suspect that your TECH may be younger than me and simply may not have the experience of Germanium devices.
I mean no offence ,but it is extremely bad practice to replace resistors with links and increased values for the reasons stated above. Lets face it evey Duo was built with specific resistor vales and has worked that way for years; the only reason it will change is if the transistor is failing and, as I said ,with these types of devices , one can be easily fooled if using a modern day DVM to measure them.
The preceding comment was written in an email to me by a technician from England with experience repairing Farfisa organs and other vintage organs. We began a correspondence on this subject shortly after the work had been completed. As the customer who brought the organ in for repair, I'm kind of in the middle. Information on repairing these organs is in bits and pieces all over the internet but it is sometimes difficult to know what is correct and what is not. The work that Technician Larry did put the organ back into service and made it sound excellent. His expertise in electronics also enabled him to fix that badly cracked circuit board and get that back into service when it looked hopeless to me. While explanations about the work seemed reasonable to me, comments by the English technician also seemed reasonable and written with the conviction of experience. So like I said, as a customer who knows little about electronics I feel caught in the middle and slightly freaked! I'm hoping all this will be resolved in a friendly, good natured manner as everyone I'm sure is onboard with the ultimate objective: high quality service and repair that brings back and prolongs the life of these remarkable vintage organs.
ReplyDeleteThanks for those words Jim. Kenneth apparently did not assess the whole situation here as a couple of things he said are not applicable. What we changed did not bias the divider transistors to an excess load line. Given the available 8 volts and the 5K+ collector resistance, the maximum power dissipation is 2.8 milliwatts. Varying the limiting impedance by up to 330 ohms (the max value of the AB resistor) will not change the collector current enough to affect the transistor which is not even beginning to warm up under these circuit parameters.
ReplyDeleteThese transistors are functioning as switches, not amplifiers, and are therefore either on or off. Whether or not they are biased to the most linear portion of the load line has no significance in this circuit. Of course thermal conditions in transistors can certainly be a design concern - mostly when the transistor is functioning in a linear mode (amplifier) and not in a small signal circuit where available power is limited to a couple
of milliwatts. Power amplifiers generally have a thermal switch on the heat sink that interrupts the power when the temperature exceeds the designed level. This protects expensive output drivers from thermal
runaway. If one of these transistors shorts out it can take out other circuit components and the speakers with it - causing "much damage" as Kenneth reports.
Small signal switching transistor circuits are a different case. With a dead short C to E across the "switch" there is but a minuscule amount of power dissipated in the collector resistors. The circuit is designed to
function with the transistor turned full on for half the cycle. If a short between C and B occurs a BE junction failure might follow but at that point the transistor is already bad. The resistors in this circuit are 1/2 watt
rated. There could be 100 times the possible available power dissipation and there still would not be any smoke. A short from E to B would be equally benign by the same analysis. No additional damage can possibly
result from a transistor failure in this circuit even in the rare event of a
shorted junction.
I see no evidence that the transistors in your Farfisa are "on their last legs and one step away from failing" as Kenneth suggests. The reliability of the transistors on these cards is not too bad actually. Sometimes there
is a condition where the leads sheer off right at the case, but that is
mechanical - perhaps chemical (?) but cerainly not electronic. I usually find broken solder or wiring harness connections or other physical damage on the Farfisas that come in for service. Very rarely have I replaced one of these transistors due to a junction failure. For you to have to worry about turning on your Farfisa for fear of frying these transistors is absolutely unwarranted. Fear a lightning strike or a bad nasty power surge, or connecting to an outlet with 220VAC or plugging into an amplifier with a "hot chassis" instead. What I fear is that irrelevant or misinterpreted information will result in irrational and paranoid customers which I find much more difficult to deal with than a circuit I don't completely understand at first glance.
Back to the circuit analysis: The AB resistor works in conjunction with the 50 micro farad cap to filter out a portion of the vibrato modulation signal present on the emitters of the divider transistors. The four cards with the lowest master oscillator frequency have no filter and get the full vibrato signal modulating the reference voltage (see note 8 on schematic about omitting the condenser on cards with a Direct Connection at location A-B.). Modulating this reference voltage will alter the timing of the divider flip flops and thus modulate the pulse width of the output signal. This occurs as C A(1 thru 5) and C B (1 thru 5) discharge through the 100K resistors to ground. The binary state of the circuit toggles when this voltage reaches a certain level, and the level required to toggle is modulated by the vibrato voltage present on the emitters. The low pass filter created by the AB resistor and the 50 micro farad cap reduces the amount of modulation voltage on the emitters of the higher pitched cards. More resistance at AB means less pulse width modulation. As the frequency of the master oscillator increases the amount of timing modulation has to decrease in order to maintain a proportionate amount of pulse width modulation relative to the pitch of that card. When we had this unit on the bench I didn't see the whole picture. I added capacitance to the filter which caused no change in the abhorrent sound, but I did not remove and replace the 50 Microfarad cap or check the tolerance of the 100K discharge resistors. The cap on the cards where we replaced the AB resistor with a Direct Connection should be removed if we don't restore the specified AB resistor value. Another close look and the rest of the answers might become readily apparent as I now have an understanding of the AB resistor function. Let's put this back on the bench and answer those questions. I'll do it for no additional charge.
ReplyDeleteI was checking out your web site. Great information! I have one Farfisa tone oscillator that is is not functioning properly, bad tone. This information will help my troubleshooting thought process!
ReplyDeleteI am restoring a Farfisa Combo Compact with a Leslie Combo II Pre-Amp and a Leslie 825 speaker. This project is going well as I have two Farfisa Combo Organs - One for parts.
I was able to completely restore the Leslie PreAmp with new capacitors, transistors, power indicator lamps and switches. It is essentially functioning as new off the shelf!
You can find almost all the parts on Mouser.com. Also, you can purchase the Leslie Power supply capacitors and foot switches from ClassicHammonds. The caps they stock are a direct replacement and fit the PC board. Rather large electrolytics as they are 3,000 uf. Next will be replacement of the Leslie 825 power supply capacitors and main amplifier transistors.
I plan to refinish the the Farfisa Combo and Leslie 825 with new Tolex vinyl, replace all the capacitors, electron tubes (2), and clean all the connectors. I found a great product for removing contact corrosion, DeoxIT D5.
Works great - flushes out any buildup on the pots, contacts and connectors. The Farfisa Combo has a lot of contacts and pots! Some of the pots are hard to find, the issue is more of a replacement size problem. It is easy to find ones with correct resistance and/or wattage but the size or shape becomes problematic as sometimes the replacement will not fit the PC board location.
I have he Farfisa Combo Compact service manual and the manuals for the Leslie preamp and speaker, if anyone needs them.
Let me know if you want photos to post, and I can send a few.
I also have the tech spec sheets for the Combo's Triode tube ECC83. The GE 12AX7A electron tube is a replacement if anyone has a hard time finding the ECC83 tube.
I would rate my tech skills at Expert level, 10 years working as an electronics technician and another 15 years working as an electrical engineer. So hopefully, I will not need to borrow your hammer!
Regards, Joe
jodonnell18@yahoo.com
Hello Joe,
DeleteI tried emailing you, but the adres does no longer work. I would love to get a service manual for the Farfisa Compact (i already found the schematics).
I'm guessing this is a small chance...but who knows you drop by here every now and then.
Greets,
Bas
bastebraak [at] hotmail [dot] com
Hello ,
ReplyDeleteI have taken my Farfisa Cmpcat duo out of moth balls.
1). I remember that even in the 60's, till the 90's all C notes would become Fs, in cold weather, until it would warm up. Often times it took half the night. before all the Cs came back..
1). Well now this same issue is there still.
2). The sounds are good but wesk, acn be hearde well only when I turn the amp up to 10.
3). There is a lot of static, and hissing in the back ground. drlistro@gmail.com
Chuck
Hi, Great site! I dealt with this issue as well and have fixed it by replacing the caps that couple each individual divider to the next one below it... I think if those caps start to fail, and there isn't enough capacitance feeding from one divider to the next, the divider gets a weak or erratic input... and that causes this weird frequency modulated distortion. I detailed my experience on my own blog... http://fixingelectronics.blogspot.com/2012/10/farfisa-mini-compact-distorted-divider.html
ReplyDeleteLarry - I have a Farfisa Combo Compact that I played many years ago. Took it out to check and found a bad C note (sharp). Tried tuning but no change. Any ideas? Capacitors? Transistors? Could I send you the card for repair?
ReplyDeleteHow did you "try tuning"? If you use a metal screw driver, the pitch will change just by bringing the metal in contact with the tuning coil. Those coils have a ferrous plug that can break, or get stuck and be difficult to turn, which frequently causes them to break.
DeleteIf there are bad components you will either not get any audio, or there will be missing frequencies coming from the divider circuits. Try adjusting the tuning coil again and note whether or not the plug rotates and moves in or out as you turn it. Tell us what happens and I'll offer up additional opinions.
No, it is not a good idea to send me the card for repair. Maybe as a last resort down the line, but my mission is to provide assistance in your attempts to repair your own gear.
Great resource here! I'm 95% finished with fixing up a friend's Compact Deluxe and have run into a similar problem. Tonight I'll look at replacing that A-B resistor on the offending card (E). Basically I thought I had it fixed by replacing a few electrolytics on the boards and the transistor nearest the coil. It seems when starting up cold that the voice is several steps out of the tune, with some distorted harmonics and changing the coil doesn't affect the tune. After a while it warms up and starts to act normally. Would basic voltage readings on the legs of the transistors tell me if they are functioning or not? If the coil were bad what would the sympoms be? Thanks!
ReplyDelete