Well, since it seems like you already have tried pretty much everything except total rebuild then maybe it's time for something far-fetched: Have you tried the amp in another room or apartment? Sometimes too long mains wiring, very long extension coords etc. causes a very irritating hum. Also, any motors or fluorescent lights nearby?

Anyway, that DC offset indeed seems too high: I'd suspect there's something breaking down and I'm sure Crate did not design an amp that would heat the voice coils with such a high DC figure. One possible source for unbalanced output is in the power amp input stage, which is very often a differential amp. I'm pretty sure this amp has a discrete one, (although a schematic would verify it). Any inbalance there will increase the DC offset - at least a bit. Maybe some components, solder joints or something have gone/are going bad in there. Besides a bad grounding, increased hum is as well a symptom of a stressed output stage. (The stage draws more current than it normally should). Are some of the transistors or resistors getting hotter than they normally should? Perhaps a busted CCS circuit somewhere in the PA is messing up the operation, causing inbalance and probably an excessive current draw - which then leads to supply sag, which causes an increase in ripple.

Can you test the supply voltages without the load of the power amp circuit to see whether you have sag even when idling? If this is the case then adding more capacitance would help in reducing the hum but would not cure the real problem. If the voltages still fluctuate notably (without the load of PA circuit) you most likely have either problems in the mains distribution or a faulty transformer.

Pin1 and 3 = Secondary ??! Secondary with such a small resistance to primary side - sounds very unreal. More likely the additional pin in the (assumed) primary is a tap for another voltage. i.e. 220 or 240 V. Makes more sense if you think: pins 3-2 = whole winding (9,5 ohms) = smaller turns ratio, pins 1-2 = little less winding (8,7 ohms) = higher turns ratio. After all 9.5 ohms is pretty close to 8.7 + 1.2 ohms.

Edit: So whatever you do, do not connect only pins 1 and 3 as primary (1,2 ohms): This would probably result into a very high voltage at secondary.

Those resistances seem pretty odd since they should be related to each other. Perhaps you did something that interfered with the probes. I know that sometimes measuring taps like that is pretty difficult. If the secondary is center tapped (I guess it is) resistance between pin 1 and 3 should be pin 1 and 2 + pin 2 and 3. Assuming that pin 2 is the CT. The resistances are pretty similar though.

How does a Zobel network reduce DC offset?

It's never good to assume anything when dealing with mains voltages. Plug it into a low voltage AC source and do some tests.

You could try to trace out a schematic of the output stage. As far as I´ve figured from Vox´s advertizements the output stage uses both halves of the 12AX7 to form a push-pull stage that is decoupled to SS power amp stage with a transformer. The ValveReactor pedals do not have a transformer, they simulate it. The power stage utilizes a feedback path after the speaker (not before as usual), which raises the output impedance and mimicks tube amp´s response to a speaker load. Vox calls this technology "reactive feedback" but it is better known as as a mixed mode feedback. This feedback path goes to either valve -or SS power stage. It even might go to both of them. There might be some DSP circuitry somewhere in the signal chain also - very likely there is - but you probably want to limit the examination to analog circuitry and signal path only. Although the circuit might seem and sound like a complex one it can be divided into simpler circuit parts. You do have a fairly good knowledge of what kind of circuitry you are expecting to find from both tube PP power stage and SS power stage since you are ready to modify it, right?

If you just want more loudness try more efficient speakers.

Quote from: Ge_Whiz on July 05, 2006, 05:21:59 PM
Symmetrical clipping does not introduce any DC component, and anyway, there is usually a big output capacitor to filter it out.

Well, unfortunately the bigger the capacitor the more lower frequencies it will pass. If the amplifier clips a signal with low enough frequency the clipped portion will essentially have the same effect for the speaker's voice coil as pure DC current - although the signal does not have a DC component in technical sense. The clipped portions will mainly just heat up the voice coil since DC can not act as a magnetizing motor. Naturally, a bass signal will more likely contain lower frequencies than a guitar signal. A lower frequency signal causes more supply voltage sag than a higher frequency, therefore resulting into heavier clipping. Bass amplifiers need that huge power and supply capacitance for reason: Clipped low frequency output signal is lethal for any speaker! Guitar speakers are quite sturdy but also fairly expensive and I wouldn't be too happy of stressing them too much with bass oriented signals. Another aspect is that the low frequency may force the speakers cone to travel too much and eventually break it.

The gain of LM3886 chip is adjustable just as in any operational amplifier - you just have to tweak some resistor values. The typical circuits circling around the internet seem to have a gain of approximately 20 - this requires something close to 840mV (RMS) for full output.

Negative terminal has to connect the ground otherwise the circuit is open. Consider the connection of the negative terminal equal to a connection of the positive terminal. No connection = no path to ground for current from the battery = open circuit = no power.

Don't know whether it makes a difference or not in this case but it is a common measure to "terminate" the unused opamp sections: Non-inverting input to ground, output and inverting input tied together. According to Texas Instruments whitepapers (http://focus.ti.com/lit/an/slyt166/slyt166.pdf) "Improper termination can result in greater power consumption, more heat, and more noise in op amps on the same physical IC." Shouldn´t be too tricky to re-route couple of traces. The opamp might need some decoupling between the rails too. A small cap in close approximity of the IC should do it. Otherwise oscillations might lurk out.

Quote from: lliw on July 02, 2006, 07:41:58 PM
Ok, maybe nobody understood me, ill attach a pik.

That´s a dual potentiometer, which is ideal for stereo circuits, and it has two identical potentiometers inside it. The center tap should be the wiper. At least it has been in every potentiometer i've seen so far. Don't solder pins together - you would connect the two resistances in parallel!

Quote from: lliw on July 02, 2006, 07:41:58 PM
I have read that you connect the tip to the input, the sleve to ground and ring to the negative battery terminal, how do you know what is what? one terminal is slightly golden (3) what one is this?

Do you connect the negative battery lead to ground?

If the polarity of the supply is positive (it is in this circuit) the negative battery terminal IS the ground. Mono plugs connect the sleeve and ring of the jack together so both of them are grounded if either one is connected to ground and a MONO plug is in the jack. If you connect the ring to the battery´s minus terminal and use the sleeve as a ground node the circuit will not be powered unless a mono jack is inserted (the circuit is open until this point) - thus you can make an "On/Off switch".

You can test which pin is which with multimeter's resistance tester: Connect one probe to tip and test which pin it is with the other probe. The resistance should be nearly zero. Same thing with ring and sleeve.

It seems that the output devices are still intact but working near their limits. My first suggestion would be to connect the amplifier in series with a light bulp in order to limit the current - this way it gets safer to test a live circuit for short periods of time. If the light bulp burns brghtly it indicates small loading in the amplifier, which could be i.e. a short circuit. Motorboating is usually a symptom of dried caps but if you replaced them this could be something else.

I assume that the motorboating goes away when the amp cools down? It could be cold solder joint but I'm more tempted to guess this problem might have something to do with a bias configuration that begins to drift when the amplifier heats up: After a while the huge current draw sags the supply voltages and you get an increased amount of ripple (motorboating). At the same time distortion rises since the rails have sagged. The transistors run very hot (usually they blow in seconds) but probably the heatsink is big enough to dissipate the heat - for a moment. Do these symptoms sound familiar? This is my first guess. Applying the cold spray at this moment might be too late to have any effect. You should rather check the amount of idle current. If it´s too high then you should adjust a lower bias.

They are pretty similar - meaning the substitutes should work without drastic rebiasing of the circuit. Little dissimilarity between specs of different jfet models is quite a small concern since the internal properties of jfets vary a lot and even two jfets of similar model could actually be very dissimilar. In Dr. Boogey circuit the trimmers should handle the fine tuning of bias anyway. I trust they'll work like they are supposed to.

As for the trimmers, try to find multiturn ones. Most trimmer resistors only turn less than 360 degrees (less than a full turn), while multiturn trimmers will turn more - usually 10 to 25 turns. Naturally multiturn ones are more accurate than ordinary trimmers. Carbon is the cheapest resistive material and will "degrade" as a trimmer - this is something you do not want. Prefer plastic, wirewound or cermet. Once again, one choosing criteria is whether the trimmer actually fits to it´s place. Someone suggested sealing the trimmer position with nail polish or similar substance. I think it´s a tip of the month.

The potentiometers marked with "A" (Audio) are logarithmic. The "L" stands for linear. Usually you should be able to guess which ones are logarithmic pots even when it´s not marked to the schematic since the human way to sense loudness is logarithmic. For example, volume or gain potentiometers are usually logarithmic. And again, a good rule has to have some confusing exceptions:
Old code: A = Linear, C = Logarithmic, F= Antilog
New code: B = Linear, A = Log, Antilogs are not produced
Alternate = Lin, Log - pretty self explanatory.

Looks like it. But with a single 12V supply the theoretical peak power to 4 ohms is only 9 watts and I guess something around 6 watts is more closer to the truth (assuming the amp can swing to +-5 volts of max +-6V). This is only about 3W RMS!

Finally found it from my archives: The amp I was talking about was National GA920P Tremolo Reverb. It has similar features plus it came equipped with a similar 2 x 8" cabinet. Both amps were made in Japan by Panasonic. I wonder if there´s more of these - at least I have heard about a similar Silvertone.