ampeg svt-3 pro bass problem

[Jazz P Bass]

Ampeg SVT3 Pro Schematic

Here is the SVT3 Pro schematic.

[J M Fahey]

I have to dig up the schematic unless somebody posts a link.

Dear Soundguruman, you don't need to look very far, it has been posted in this very same thread multiple times, such as Post#7 by Enzo, Post#34 by JazzPBass and again by him in post#47.
Maybe you rushed a little in posting your answer without enough research.
Yes, some other Ampegs (Crates too) have been successfully using optocoupler type limiters (as in output voltage limiters) for ages, to avoid buzzy/farty SS clipping, doubly annoying in a Bass Guitar.
You find one even in a tiny 15W, 8" speaker combo.
But *this* one, has not such a beast.
Having a full tube preamp and an "almost-tube" power amp, where a couple tubes do all the job, clipping and compressing included, and MosFets only buffer that high voltage high impedance waveform into the exact same waveform but at low enough impedance as to drive speakers , "extra" limiters are not needed (nor desirable).
Go ahead, study this schematic, you will find it very clever and at the same time, very minimalistic in basic concept.
It does not even have Negative Feedback !! Go figure.
Somebody worried earlier that it had no Zobel network at the speaker output.
Well, not having NFB means, among other things, that it does not need one.
Good luck.

PS: I forgot.
In those Ampeg/Crates who use them, the optocouplers work by lowering the value of the feedback resistor (thus lowering gain when needed) by paralleling the LDR with it.
Lower value=lower gain.
*If* you "bridge" them as you suggest, you kill the amp gain for good, effectively muting it.
jm2c

[hoagje1]

Greetings from Tacoma, Washington.
I have read this thread multiple times and noted the commentary by truly competent guys like Steve and Enzo. I have one of these amps with smoked 0.47 ohm resistors in the output (R25 and R27 too). Reading that some contributors have blown their MOSFETs multiple times is most discouraging. The preamp board is strange. I can't figure out which pins on the header bring in + and - 16V and the source of the filament voltage is mysterious too -- of course I can't turn it on in its current condition. Seems like the easiest thing for me to do given my limited intelligence is yank the mosfets and associated resistors, swap out the toroid for something light weight, and go right off of the White cathode follower into TUBES. A quartet of 6DQ6Bs or 1625s (very inexpensive) would give me about 150 watts. I'd be happy with that.

So here are my questions. Is the plan to take the fried output section out a good first step for fiddling with the preamp to see if that even works? Vis a vis comments above, could I just take out the gate resistors, or since the .47 ohm resistors are crumbling should I take them out too?

I am SURE that I am not smart enough to fix this thing, and it doesn't look like the toroid has the voltage for plate voltage, but I am open to suggestions if you have any. km6xz, St. Petersburg seems a bit far from US west coast, but if you are in the area with some rubles....

Thanks

John Hoagland-Scher

[J M Fahey]

Maybe you can repair them as-original (with MosFets) *and* have them perform more reliably in the harsh Touring/Stage World, if you correct what I see as a design flaw, and which has been scratching me the wrong way for a long time.
An absolutely guaranteed Mos killer is going beyond the 20V or so the Gate glass insulator stands.(Ängstroms thick, don't think it can be measured in Microns).
At the same time, the driving cascode 12AT7 can easily supply over a hundred volts.
No problem, you may say, since they are source followers.
Fine, while they may follow sources (duh).
Under certain circunstances, (transients, shorts) they can't do so.
OK, that's what those Zener diodes are there for, isn't it?
The problem is, Zeners must be driven from a generator or voltage source with a certain minimum internal impedance, other way of saying they must be *somewhat* current limited, for them to efortlessly doing their job.
In that schematic, the 12AT7 cascode drives those gates *directly* (ok, through a .68 capacitor, no big deal for a transient or pulse).
Also, those constant current drivers (Q3 and Q4 ) can under abnormal circunstances provide more current than those zeners can handle.
*So* I think some extra current limiting must be added.
I suggest adding a couple 4K7 resistors between those gates (and associated Zeners) and the outside world, measning the Tube driving circuit.
This will also allow VI current limiters Q5 and Q6 do their job as intended.
*Maybe* it works, sort of, without those resistors, but although being a Minimalist (less is always better), I *hate* undefined impedances.
The (easily) modded schematic would be:

[hoagje1]

Thanks J M!!!

Thanks JM

I found the + and - 16VDC on the header (I forgot that the TL074 uses pins 4 and 11 for power). Preamp tube filaments are still a mystery

OK -- maybe I will have a try at repairing this thing, but first I just want to keep the output stage from operating so that I can check some of the test points. Here are things I have thought of. Which is easiest?

A) Since I'm pulling out all the fried .47 5W (R32, R33 etc.) anyway, will this keep the MOSFETs (the whole power section) from drawing any current?
B) Pulling gate resistors (R28, R40) would be much more of a hassle since one set is crammed between the heat sink and the electrolytics.
C) How avout lifting R22 and R20??
D) Lifting C6 and C7??
E) Just pulling V2??

Any combination of the above.

PLEASE answer this question

The list of truly fried parts includes many of the .47s, R25, R27. The list of mildly baked parts (a little discolored) includes AP1 10K trimmer (looks to me like 1/4 watt resistors were substituted where there should be 1/2 watt). (Q6, R10 and R46 lightly toasted -- probably OK but near to the infamous R27). This information is provided with the delusional thinking that someone will say, "Yeah, mine did the exactly same thing and all I had to do was replace one...."

J M, the 4K7 should what wattage?

Finally, if this is just a big follower, is there any reason to install more than one pair of mosfets at a time to test my repairs initially?

Thanks

John Hoagland-Scher

[spy]

Did you check for dc on filaments? As I remember there is dc on filaments and some of them are connected in series. I'm not 100% sure but check it out anyway.

[J M Fahey]

OK, let's go step by step.

OK -- maybe I will have a try at repairing this thing, but first I just want to keep the output stage from operating so that I can check some of the test points. Here are things I have thought of. Which is easiest?

A) Since I'm pulling out all the fried .47 5W (R32, R33 etc.) anyway, will this keep the MOSFETs (the whole power section) from drawing any current?
B) Pulling gate resistors (R28, R40) would be much more of a hassle since one set is crammed between the heat sink and the electrolytics.

Maybe, but I prefer trying things functionally, meaning as close to normal as possible.
I would check and replace all ballast resistors, measure all Mosfets for shorts, in case of doubt pulling them out and later reinstalling the good or new ones.
That said, yes, you can use just one pair for initial testing.
Do not pull gate resistors, I guess that would be a death sentence.

C) How avout lifting R22 and R20??

No, they bias Q3 and Q4 which, together with the error signal detected by IC1 (1/4 TL074), form a servo which keeps the output node centered, i.e. with low DC offset (a few millivolts).
Let it work ; whether it can do its job or not will point to problems.

D) Lifting C6 and C7??

Yes, that's what I suggest, lift just one leg of each and you've separated the Tube and the SS stages, good for diagnosis.
There is no overall NFB loop, which allows you to do this without problems.

E) Just pulling V2??

Not necessary by lifting C6/7. Now you can test it on its own too.

The list of truly fried parts includes many of the .47s, R25, R27. The list of mildly baked parts (a little discolored) includes AP1 10K trimmer (looks to me like 1/4 watt resistors were substituted where there should be 1/2 watt). (Q6, R10 and R46 lightly toasted -- probably OK but near to the infamous R27). This information is provided with the delusional thinking that someone will say, "Yeah, mine did the exactly same thing and all I had to do was replace one...."

Change whatever looks toast, and whatever you measure to be more than 20% out of range.
Changing a 10K resistor because it measures 9K68 is not necessary.
Those 1/4W lookalikes probably are modern 1/2W rated .
Technology advances.

J M, the 4K7 should what wattage?

I think that 1/4W is enough.
Under normal use or even when Q5/6 are short protecting, it dissipates very little.

Finally, if this is just a big follower, is there any reason to install more than one pair of mosfets at a time to test my repairs initially?

One good pair is enough to start.

In any case, remember you will be testing the amplifier with a series lamp (lamp/bulb limiter) and no load speaker or resistor, until you have it working properly.

Start by replacing (known) bad parts, separating Tube from SS sides and measure voltages: *ideally* you should have both rails symmetrical and not much below rated voltage, no DC on the speaker out, you should be able to regulate bias voltage (C-E drop across Q2+drop across R10) from, say, around 2.5V to around 5V or even a little more.
Do this carefully because too high bias voltage will cause output transistors to pass too much current, we are simply testing the stage functionality.
If around the higher voltages the series lamp brightness (which should be dull red or orange) starts to increase, then step down.
Well, start measuring, I hope it works.
Do not make the added 4K7 mod yet, we want first to heve it working as original.
We want the least amount of unknowns possible.
Good luck.

[hoagje1]

Thanks for the advice. I've been on the web reading about testing mosfets. I'd like to know whether I should pull them before I test them. This is an academic question since I'm planning on using just one pair during the testing phase so most of them will get pulled regardless. Yes the tubes on the preamp board have filaments in series with a big fat hot resistor. Not my favorite design.

What is a series lamp? How do I hook it up and how many watts.

Thanks for your patience.

John Hoagland-Scher

[J M Fahey]

Since you are pulling all but two, go all the way and pull all, period.
Measure them outside and reinstall a certified good pair.
Before you worry about that, these power switching MosFets will seem to be "almost" shorted when reverse biased Drain-Source with the multimeter as diode tester.
That's normal, they have a built-in reverse polarity Drain/Source diode which works as an anti-inductive-kick protector.
Some datasheets show them, some do not, they are implied (I find that a confusing mistake) , but always there.
In fact it costs them nothing , because they are a byproduct of the manufacturing process.

As of the lamp bulb current limiter, search first inside this site and/or Google it.
A 60W lamp will be fine.

[tubby]

Maybe you can repair them as-original (with MosFets) *and* have them perform more reliably in the harsh Touring/Stage World, if you correct what I see as a design flaw, and which has been scratching me the wrong way for a long time.
An absolutely guaranteed Mos killer is going beyond the 20V or so the Gate glass insulator stands.(Ängstroms thick, don't think it can be measured in Microns).
At the same time, the driving cascode 12AT7 can easily supply over a hundred volts.
No problem, you may say, since they are source followers.
Fine, while they may follow sources (duh).
Under certain circunstances, (transients, shorts) they can't do so.
OK, that's what those Zener diodes are there for, isn't it?
The problem is, Zeners must be driven from a generator or voltage source with a certain minimum internal impedance, other way of saying they must be *somewhat* current limited, for them to efortlessly doing their job.
In that schematic, the 12AT7 cascode drives those gates *directly* (ok, through a .68 capacitor, no big deal for a transient or pulse).
Also, those constant current drivers (Q3 and Q4 ) can under abnormal circunstances provide more current than those zeners can handle.
*So* I think some extra current limiting must be added.
I suggest adding a couple 4K7 resistors between those gates (and associated Zeners) and the outside world, measning the Tube driving circuit.
This will also allow VI current limiters Q5 and Q6 do their job as intended.
*Maybe* it works, sort of, without those resistors, but although being a Minimalist (less is always better), I *hate* undefined impedances.
The (easily) modded schematic would be:
[ATTACH=CONFIG]14955[/ATTACH]

Hello JM,

could you reupload the mod schematic or specify the position of those 4k7 resistors? One each after the .68 coupling caps?

Thanks!

[J M Fahey]

Glad to help but I lost the schematic 2 or 3 hard disk crashes ago.

If somebody re-uploads it here, no problem.

Even MEF itself seems to have lost some old files, proof not only Humans suffer from Alzheimer