Several years back, I got a brand new Crate GFX120. It's one of the older ones, not the ones with the red/black scheme, but blue/black color scheme. It shot sparks from a short last year. It is a 1x12 120w combo. I recently decided to fix it up. Replaced the fuse (3A 250V Slow Blow), and the thing powers up and plays fine. However, after maybe 10 minutes of playing, the sound starts to become extremely inconsistent. For instance, I will play a chord on the guitar, and the sound fades to a very low and quiet level. It will keep fading in and out according to the output volume of my guitar (I assume). If I cut the volume on my guitar (turn down the knob all the way), then the hum from the distortion channel from the amp slowly comes back and stays consistent. This leads me to believe my guitar is "overloading" something in the amp. A visual inspection leads me to believe nothing is wrong with the amp (no blown caps or ICs etc.). It is quite hard to explain.
Any clues? I can take pics of the amp, videos, sound clips etc.
I've confirmed it's not the speaker of the amp. My gut tells me it's in the poweramp stage.
Email me if you need the schematic. teemukyttala[removespam]@gmail.com
you have a cap. or a resistor breaking down when the amp gets hot. be careful of slow blow fuses, they are designed strictly for motors.
ger your amp to a good repair shop fast.
good luck
jimmy peters 8)
Quote from: jimmy peters on May 23, 2007, 12:28:14 PM
you have a cap. or a resistor breaking down when the amp gets hot. be careful of slow blow fuses, they are designed strictly for motors.
ger your amp to a good repair shop fast.
good luck
jimmy peters 8)
Slow blow fuses are used in amps all the time, in fact, Brian (chipamp.com) recommends using a slow blow fuse for any of his gainclone amps.
The fuse that came with the amp stock was a slow-blow fuse. My tube amp uses fast-blow fuses.
Is there any way to pinpoint what specific part is breaking down from heat? I'd assume it's after the preamp stage as it occurs for the clean and the gain channels. I can try rigging a fan above the amp to see if it stops/occurs later.
Slow blow fuses are common in amplifiers that have high inrush current.
"Is there any way to pinpoint what specific part is breaking down from heat?"
Yes. Cold spray, for example.
Anyway, plugging signal source into the "insert" jack should tell you whether the problem is in the preamp or in the power amp. (Be careful, there is no volume control for it in the amp.) You said the amp "shot sparks" and that doesn't sound too good. Sparks generally mean high current and high current melts things like solder joints etc. Was the short over speaker leads or something else? I second the opinion about thermal issues but you also have to narrow down the search area to locate the faulty part.
Just to make sure, I would first check that the quiescent current is right to find out if bias servo is still working correctly or if there's trouble with the power transistors. You should read 7mV across R72 with no signal and no load. Also, check the DC offset at the positive speaker terminal (use the negative terminal as reference instead of ground) - this has likely nothing to do with your problem but you want to be "safe" anyway. If it's anything higher than few millivolts then you should likely worry.
If it's the poweramp... Well, the concerned circuit in that amp is pretty basic so I would attack the section around the drivers/power transistors first to see that all those parts carrying high (or moderately high) current are OK and that there is no problem in their solder joints etc. Do you have the schematic for the amp?
I hope you do use a current limiter (i.e. light bulb) inline with mains since at current state it is reasonable to assume that the amplifier might just suddenly develop a major fault. Just because it seems to "work" now doesn't neccessarily mean it will do that for long. Do not measure the bias with current limiter inline though.
The short was under the amp, it wasn't the amp's fault, human fault for the shorts (some metal rod got stuck under the amp between the chassis and the board). The amp was doing the exact same thing before the short, so I'm just about positive the short had no effect on the amp besides blowing the fuse.
I do need the schematic. Email me at MetallicAman91 [at] gmail [dot] com
R72 was not reading 7mV, but after adjusting AP1 a bit, it is now 7mV exact according to my cheap DMM. This was without any input and with the speaker disconnected. That was how I was supposed to measure it, correct? With no signal/inputs and no load (speakers)?
I turned the amp on for all of a minute or two with the speaker connected but no input, and noticed the glue around the resistors towards the top-right of the PCB (R61, R88, R89, R90 and one other which I can't make out the name) was getting very warm. Maybe there's a problem there?
For the record, all of the solder joints on the board are fine from eye inspection. The input jack is missing. My Ext. Speaker jack broke, thus I replaced the Ext. Speaker jack with the Input jack. I made sure to solder the two points that weren't going to be connected with the jack gone (The jack was a switched jack, as in when a 1/4" cable was plugged in the switch was open, otherwise the switch is closed). I still have yet to test to see where the problem is, the poweramp or the preamp.
I don't have the greatest electrical knowledge, I know the basics. I don't have that many tools on hand either, I have a soldering iron and a cheap DMM...
Yes. According to schematic that is the way to measure the quiescent current. If it wasn't way off it's normal. What you really want to know as well is will the reading fluctuate once the amp has been on for some time. It should stay pretty steady once the amp has warmed up. BTW, do not adjust it "cold" (wait few minutes after turning the amp on so operating points have time to settle). Anyway, it seems like there is no problem with the bias.
According to schematic R88, R89 and R91 are series resistors for regulators. They are supposed to heat a little (thus the 5W rating) but if you think there might be a fault then check the voltages. There should be about 15-16V after R89 and R90 (+V and -V rails & zener diode regulation). Naturally, 5V after the DSP regulator - which is IC.
R61 is an emitter resistor. Basically, R72, R74 and R64 should heat as much as it does. If there is excessive heating in only one emitter resistor it is very likely an indication of a problem with the concerned transistor. At idle (no load, no signal) there should be little more than 1W wasted in the emitter resistors (again they have 5W rating) so they will heat a little.
However... Did the glue start getting warm after the adjustment? If so, recheck the adjustment. I don't know how Crate designs stuff but usually nothing should melt when the amp is on idle. Basically, less bias will do no harm. It will decrease the heat losses but in turn will also increase crossover distortion.
The schematic shows tables for voltage readings at certain test points (marked TPxx). You can check if they match. Naturally, they don't need to be exact, just reasonably close. I trust that you also know the basic transistor/diode measurement techniques. Your greatest problem will likely be trying to get the amp to exhibit the fault condition while you do the measuring (this is why intermittent problems and alike are really tricky). I already mentioned how to find out if the problem is in the power amp (the insert jack method).
The glue was getting warm before adjustment (the voltage across R72 was 2mV then). I haven't checked the resistors in suspicion since adjustment.
For voltages of the regulator resistors, I should not be measuring across the resistors, instead using the chassis or a ground as a common point, right?
The testpoints look as if they include the use of an oscilloscope, correct?
I can identify transistors and diodes, and know how to test diodes to see if they are still working. However, I don't know exactly what you mean by basic measurement techniques for each.
Quote from: SolidxSnake on May 28, 2007, 07:42:51 PM
The glue was getting warm before adjustment (the voltage across R72 was 2mV then). I haven't checked the resistors in suspicion since adjustment.
2 mV seems a bit low but now it should be up to the specs. The amp is likely running a bit hotter, though.
QuoteFor voltages of the regulator resistors, I should not be measuring across the resistors, instead using the chassis or a ground as a common point, right?
Right. You can see the zeners shunting the rails to ground after the 5W resistors. At those points you should read about 15-16V + or - according to the rail you measure. The 5V for the DSP circuitry is regulated with IC. I trust it should be pretty OK but you can measure that the circuit has an output of 5V. However, I don't think it helps you in the trouble you are having with the amp. If the voltage for digital circuitry would be way off you'd be having some other problems. I'm also pretty sure that the rail voltages are correct as well.
Quote
The testpoints look as if they include the use of an oscilloscope, correct?
Well... sure. But you see that one of the tables quotes also the DC. This you can measure with an ordinary DMM.
Quote
I can identify transistors and diodes, and know how to test diodes to see if they are still working. However, I don't know exactly what you mean by basic measurement techniques for each.
Well, I meant just that. You are familiar with concept that you should measure about 0,65V across base-emitter junction, likely more than 2V across collector-emitter. With fets about 1-4V gate-source and more than 2V drain-source. Basic silicon diodes should have voltage drop of about 0,65V etc. Since you know some voltage references you can test if semiconductors are OK.
However, I suspect that there is a failing part, intermittency introduced by heat or some other problem that is very difficult to locate by taking simple measurements. While the amp is functioning normally it will (very likely) also give readings that seem to indicate that everything is fine. I can't help you much further. All I can say is that...
1. Isolate the problem to certain parts in circuity as much as you can.
2. Think about the symptoms and what kind of problem in the concerned circuit might cause them.
3. Measure, inspect etc. to see if you are correct. Here is the major problem since you have to get the amplifier to exhibit the fault while you measure. If the fault is triggered thermally (at least it seems like it is) then a hair dryer and a can of cold spray will be quite useful tools. Naturally, they are a lot more effective if you can locate the fault to a certain part in the circuit. Heating or cooling the circuit randomly usually doesn't reveal anything concrete.
4. If the first educated guess was wrong think about other alternatives. Repeat 3 and 4 until you locate the problem.
Thanks for your help so far.
I brought the schematic to school and tried to explain to my electronics teacher what was going on, and she suggested it might be something from the power source. She said to try checking the diodes, and the test points around the filter capacitors (TP14 and 15).
For the test points, an oscilloscope is helpful for the AC readings. For the DC ones, a signal generator is required. Thankfully, my electronics teacher lent me a signal generator. The schematic says a 1KHz signal to the insert tip for the poweramp testpoints.
I'll post back with results as to what is happening if I can't figure anything out.
Once again, thank you for your time.
If you get it fixed, please inform us. I'm quite curious to find out what was going on.
Strangest thing.
Today, I was gonna work on my amp, measure test points and such. I first wanted to create a soundclip of what was going on to show my electronics teacher (as well as people on this forum) to see if that would help narrow down what was going on. I unattached everything from the chassis besides the transformer (all of the PCBs, the fuse holder and the power switch are not attached to anything), and then started playing to reproduce the problem. After 10, 20, 30 minutes of playing, I didn't get one single problem. I tried the caveman's testing for temperature by touching various components, and some resistors and transistors were BURNING. Resistors in question are R61, R64, R88, R89, R90. Transistors are Q12, Q13, Q15, Q16. The transistors I mentioned are attached to a metal plate which is normally attached to the chassis with some screws (some TIM in between it) for the chassis to act as a heatsink. However, even after the resistors and transistors were extremely hot, I had no problem.
This is very strange. The only difference is that the board is not enclosed or screwed into the chassis at all, and I have no problems. I can take a picture of my setup at the moment if you don't understand what I am saying.
Sounds like when it's in the chassis you are experiencing a short.
...Or the troublesome part of the circuit runs cooler when boards are not mounted to the chassis. (This usually results into better circulation of cool air). I would assume that a short circuit would exhibit some problem symptoms constantly - not after a delay. Delays are often sign of thermal problems.
If the chassis is a crucial part of the heatsink assembly DO NOT run the amplifier without it: Heat is the worst enemy of semi-conductors! Reducing the junction temperature by 10 degrees of Celcius will double the life of a transistor and vice versa: Running it 10 degrees of Celcius hotter will halve it's lifetime. The transistor operation is also very much affected by the temperature and reliability of most designs is based on assumption that the transistors are attached to at least a moderate size heatsink.
I would be interested in seeing couple of photos. Maybe they could tell something that the schematic fails to show.
The only part of the electronics that are cooled by the chassis are the four transistors (Q12, Q13, Q15, Q16).
I'm positive it's not a short in the chassis, there's not a single thing that could cause a short besides the two jumpers that contact the screws for the PCB.
I'll take some pics and host them.
Pics are here:
A pic of everything (chassis, PCB, speaker and everything else):
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0898.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0898.JPG)
A shot of the PCB:
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0899.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0899.JPG)
Shot of the fuse, power switch, transformer, and related areas (connections between transformer and PCB):
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0900.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0900.JPG)
Shot of the DSP and footswitch jacks:
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0901.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0901.JPG)
Shot of the resistors I suspect to have problems. The three towards the left from front to back are R89, R88, R90. The two to the left from front to back are R64, R61.
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0903.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0903.JPG)
Shot of capacitors C39 and C40, and another angle of suspicious resistors:
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0905.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0905.JPG)
Shot of R72 on the left and R74 on the right, AP1 to the far left:
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0906.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0906.JPG)
Shot of the four transistors Q12, Q13, Q15, Q16. The metal plate they are connected to had a thermal paste between it and the chassis when attached.
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0907.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0907.JPG)
Shot of the Insert and External Speaker jack. Note the Insert jack is missing (moved over to the external speaker jack which broke), but the only contact that is supposed to be shorted but wasn't is jumpered with a wire. I haven't bought a switching jack to replace the lack of jacks.
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0908.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0908.JPG)
Shot of the speaker:
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0909.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0909.JPG)
Shot of the chassis and transformer. Note the thermal paste in the BG, that is where the four transistors and their "heatsink" lay.
http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0910.JPG (http://bostonrinkrats.com/solidxsnake/GFX120/IMG_0910.JPG)
In the image of the suspicious resistors, the three on the left (R89, R88, R90) were getting extremely hot. The two to the left (R64, R61) were still very hot, but not as hot as the three to the left.
If you need any more pics or sound clips (I couldn't recreate the problem while the amp wasn't in the chassis), shoot me a message.
My first impressions:
I assume the photos did not show the real heatsink? Don't tell me your only heatsink is that attachment plate, which is mounted to chassis...
The resistors R88, R89, R90 in the power supply section look more burned than normal. Either Crate has done a poor design job or the preamplifier circuit is drawing excessive current. Do the resistors excessively heat that big capacitor beside them (C39?) ? If so, this is also a very bad thing since capacitors hate heat and their lifetime seriously decreases in elevated temperatures.
Q8 might not be tracking the heatsink's temperature accurately. It doesn't seem to be in much contact with it.
The only heatsink that amp has for the four transistors is the metal plate and the chassis. Nothing else.
The resistors do not heat up C39. C39 and C40 are both just lukewarm, nothing else.
Now that I look at those resistors, they do look a bit charred.
It does seem a large possibility that the preamp circuit is drawing too much current. Any ideas how that could have happened?
Quote from: SolidxSnake on June 04, 2007, 10:15:18 PM
The only heatsink that amp has for the four transistors is the metal plate and the chassis. Nothing else.
Wow! What the hell have the designers of this amp been thinking? Just take a look at these three amps and their heatsinks. They're all about the same power as your Crate. You likely will realize why your amp seems to run better in open air...
http://home.swipnet.se/malman/lowtim_circuitboard.jpg
http://www.littlefishbicycles.com/poweramp/monoblock_oblique.jpg
http://sound.westhost.com/p27-amp3.jpg
Quote
The resistors do not heat up C39. C39 and C40 are both just lukewarm, nothing else.
That's good.
Quote
It does seem a large possibility that the preamp circuit is drawing too much current. Any ideas how that could have happened?
Well, either Crate did bad poor on calculating the dissipation requirements of those resistors - which I seriously doubt - or there truly is a "soft" short in the preamp.
The simple Zener regulator is the greatest current hog of the preamplifier circuit since there is a constant shunt to ground over over it. During normal operation the circuit will draw about 90mA ((40V-15V)/270R) and the rail voltages should be around 15 - 16 V. Did you measure them? In idle state (and when the load impedance presented by the preamp is high) most of the current flows through the Zener (which dissipates about 1.4W). The series resistor constantly dissipates about 2.25 watts (25V*90mA) but since its rated for 5W it should be Ok. This is a pretty poor regulator circuit but that's another story.
If there is an excessive current draw (that is required to really cook up those resistors) the load circuit in parallel with the Zener becomes more dominant and begins to hog most of the current instead of the Zener. Even when the load is about 270 ohms (the size of series resistor) the regulator still works like described (drawing about 90mA). However, most of the current now flows to the load instead of the Zener.
If the load drops below 270 ohms the load and series resistor effectively form a resistor divider and the Zener begins to drop out of the equation. The current draw will be higher than 90mA and the amp will start to experience some supply voltage sag in the preamp rails. When the voltage has sagged below 15V or so the Zener can't conduct and its current draw drops to zero. The preamp load will now solely dictate how much current is drawn.
So, if the supply rails have lower voltage than about 15 -16 V it's a pretty good indication that there is for some reason excessive current draw. You can also measure rail-to-ground resistance (which is the "load" I'm talking about) but since this value is practically unknown it can provide only slight clues; it should likely be in the order of kilo-ohms but at least higher than few hundred ohms. A load of approximately 23 ohms would cause a current draw (and sag) that would result into full 5W over the series resistor.
Starting the amp cold, I measured across the Zener diodes, one has 16.15vDC, the other is 16.65vDC.
The amp is giving me a constant hum, starting cold or leaving on for a while. The hum doesn't go away. It did once, after I shook the chassis (and in turn shook the PCB). The hum I can record for you if you might want to hear it.
I rechecked the voltage across R72, and guess what, it dropped down to 5mV... I readjusted it. After letting the amp warm up a bit, still idle, I checked the Zener diodes again. The first was 16.7vDC across and the other dropped to 15.3vDC across. This is without any input, only the speaker connected. Also, this is with the amp back in the chassis.
Sorry for the double post, but..
Turns out all my measurements in my past post were wrong. Turns out the solder joints for C39 (very large capacitor) were cracked. The broken solder joints were also what was causing the hum, reheated the joints and let the solder reflow and now the amp works like it used to when cold. I measured the Zeners after playing for a bit. One was a little over 16v (I believe 16.34vDC), the other was nearly 17v (16.97vDC). My electronics teacher had told me that the Zeners should be extremely accurate to 16vDC, +- a few hundreths of a volt.
I haven't measured the rail voltages. Where would I go about measuring them? As you may have guessed, I am not the most electronically able person in the field of theory (such as, I don't know the basics about different kinds of components and how they should act, etc), but with some guidance I most of the time am able to troubleshoot problems if I know where to specifically look.
Edit: Looking at the schematic, I see four things which I assume to be rails. There is a +40 and a -40v rail, as well as a +V and -V rail. I assume you mean to measure the V+ and V- rails, and they should be ~16v?
If the +V and -V are the correct places to be measuring, then I have measured them. With the chassis as a reference point, +V (negative side of the Zener D25) is roughly 17V after a bit of playing, whereas the positive of D26 is around 16.3v.
Are D8 and D9 Zeners? Their schematic symbol don't look like Zener diodes, but they are reverse-biased, no?
Quote from: SolidxSnake on June 09, 2007, 01:32:31 PM
I am not the most electronically able person in the field of theory (such as, I don't know the basics about different kinds of components and how they should act, etc
I got to say what's on my mind: Are you sure you can fix this amp
mostly by yourself with the current level of technical knowledge you got? I mean, likely the people at this forum can only help you a little and the main amount of work is done by you. Is this a "learning project" or do you need it badly? In the latter case I would take it to a tech.
Helping to troubleshoot circuits in an internet forum is hard enough already - and becomes even harder if the guy (or girl) who's trying to fix the thing doesn't know a lot about the circuit in general. If there is anything but an obvious problem you might get plenty of different answers and most of them are speculations of what could be wrong - and people have to speculate because they can't have your amp and measure it. If the answer is obvious (like it was in the case of that bad solder joint you had: hum stops when you shake the amp etc.) people are usually too lazy to even bother answering.
Don't get me wrong, I don't mind answering basic questions but some of those answers... well, could be learned from theory books as well.
There is also a limit in how much I can help. I don't have the amp and I can't measure it. ...And at the current point I'm a bit confused what's wrong with it anyway: You fixed them hum - which was actually a very important fix because the concerned capacitor was pretty much the only effective filter for positive rail - but you also mentioned overheating and sound cutting in and out. Have you located if the latter problem is in the preamp or in the power amp yet?
What problems did fixing the solder joint fix? For the record, Crate is infamous for making amps with poor solder joints so you should check all of them, starting from the components under mechanical stress: Jacks, potentiometers, switches, heavy components such as big capacitors etc. Then proceed to components carrying high currents. Don't assume they are OK just because they look so: You really need a magnifier glass to inspect them properly so you might as well redo all soldering just to be sure.
Anyway, a fair warning is also in place: If you don't know the correct procedures of working with an amplifier you can easily ruin it to a point where even a certified tech wouldn't touch it. For example, it doesn't take much effort to ruin those "new type of" PC boards with extremely thin copper and no solder reinforcement. Even worse, you can shock yourself badly: That amplifier stores very high currents and the voltage across the rails is pretty high (although not considered lethal). ...And then there is mains circuitry.
Quote
Edit: Looking at the schematic, I see four things which I assume to be rails. There is a +40 and a -40v rail, as well as a +V and -V rail. I assume you mean to measure the V+ and V- rails, and they should be ~16v?
Yes. First few things about the higher rails: The supply is loaded by the varying resistance of output transistor and the speaker load. Together with the internal supply impedance they form a voltage divider. The high voltage rails (measurement points TP14 and TP15) are at the joining connection of the internal power supply impedance and transistors. This is why you measure about +-35V instead of +-40V when there is an 1kHz, 1Vpp sine wave at the insert tip. In short, the rail voltages drop/vary according to severity of loading. This is known as "sag".
The +-15V/16V supply rails are regulated, meaning they have (or should have) a constant voltage regardless of loading. The voltage is constant because Zeners shunt anything above 15V-16V to ground.
Quote
If the +V and -V are the correct places to be measuring, then I have measured them. With the chassis as a reference point, +V (negative side of the Zener D25) is roughly 17V after a bit of playing, whereas the positive of D26 is around 16.3v.
The Zeners are rated 16V so anything above this is not right. The voltage at these rails should stay constant under any normal operating condition. Check the Zeners and their solders.
Quote
Are D8 and D9 Zeners? Their schematic symbol don't look like Zener diodes, but they are reverse-biased, no?
I must be blind but I couldn't locate these. Where are they?
The amp at the moment is essentially a learning project, as this amp has been dead for a while and I got new one in the meantime.
Fixing the solder joint got rid of the hum. When it was humming, a guitar plugged into the input jack would not be heard, only the hum; changing anything (gain/clean channels) did nothing. Fixing the solder joints did allow the amp to "work" as i was previously.
I work with electronics quite a bit. I've got plenty of practical experience, but theory-wise I don't know so much. I've got knowledge about how to handle electronics and such, so I should be fine actually doing the physical work. However, I'm not very able alone to figure out the problem based on the measurements and other inspections I find. The amp however is basically a learning project, so fixing isn't on the top of my priority list.
I can't test TP14 or TP15 at the moment. I did have a function generator, but no dummy load to replace the speaker (I need a 4-ohm load, it is far too loud with the speaker connected). I will probably end up building a switchable 16-8-4ohm load soon enough however.
I'll check the Zeners and their solder joints.
For the past few days, I've been playing guitar for an hour or so and the amp gave me not a single problem. However, the voltage from +V is 17v still, and the 5w resistors are still extremely hot.
The resistors are supposed to be (at least) fairly hot. Currently there is (in the worst case scenario) a 23 V drop over the resistors: 40V - 17V = 23V. According to I=U/R, I = 85mA so P=UI means P = 1,96 W. Well within the rated 5W. Likely the heat is "normal condition". Each resistor still dissipates about 2 watts. That's the problem of shunt regulators in general; they are pretty inefficient.
If the solder joint in the mains filter capacitor was failing it could have caused some occasional volume drop but likely you would have heard the signal being modulated by hum as well - and some loud "pops" and "crackling" noises. (Up to the point the solder initially failed and you started hearing only hum). It's difficult to say anything certain about this. I still think it's a thermal problem, maybe a solder joint of a resistor failing somewhere.
Good luck on the project. I'll try to help if I can ...and for starters (on improving that theory part) you might want to take a look at this:
http://www.pacificrecone.com/JackDarrBook.html
It's not the best book to learn about solid-state power amplifiers and guitar circuits but has a strong focus on troubleshooting and fixing stuff. You might find it helpful.
Previously, my amp was giving me the symptoms you were describing (with the solder joints breaking). Sound cut out, hum appeared, pops, crackles etc. Haven't gotten any of those since I started working on the amp again (from the start of this thread).
You say the resistors are supposed to be fairly hot. I would assume they are too hot at the moment however. I can hardly touch them for more than a second or two without my fingertip nearing the scorching point.
I'll post back if the problem arises again, but at the moment the amp seems to be working fine.
Thank you very much for your time. It is highly appreciated.
And thanks for the link. I'll definitely check it out. I know the basics of the amp (preamp->poweramp, what those stages do, etc.).
Ok then. I guess it would be fairly safe to assume that the failing solder joint was causing those volume drop issues as well. Anyway, you'll know for sure if the problem appears (or doesn't appear) again.
The resistor issue is a bit more troublesome: According to your measurements (about 17V at the regulator side) they should not dissipate power thar exceeds their ratings. However, I would suggest that you measure the higher rail voltages just to check they are within about 40V - 35V. You don't need a dummy load for this, just check the voltages without signal applied - at this state the voltages are highest anyway. Anything exceptionally higher than +-40V would explain the very hot resistors - and indicate about more serious problems.
If those voltages are correct then it's likely just too much power for those resistors to handle. This isn't practically very ideal situation since too much heat can melt the solder joints (and that capacitor in close vicinity to resistors won't like it either). If this is indeed the normal operating condition - and if you want to solve this issue - you can substitute the resistors with something that can handle more power (i.e. two 5W 560 ohm resistors in parallel per each 5W 270 ohm resistor). Personally, I would rebuild the regulator circuit to be active type but that's another story.