Ampeg SVT 6 Pro repair assistance

[mxrshiver]

been repairing one of these, that was non-functioning after someone tried to power it by plugging it into their truck (don't ask). i know these are tough to work on, but i'm committed to it as a valuable learning experience, and the guy who wants it fixed got it for free, so there's no big loss for either of us in trying to fix it.

didn't try to start it up initially, but the 15A fuse was tripped and both the caps that see the direct AC high voltage (C16 and C31) were popped. so i decided to start with a complete recap, replaced all the FETs, and replaced the two finnicky bias pots with 20 turn pots.

started it up slow on a variac and dim bulb limiter, everything looked great at lower primary voltages (all DC voltages in the power supply normal at least) until about 90V input, when one of the FETs (Q12) shorted, and the gate resistor (R41) for that one started smoking somethin fierce. removing it and associated resistors from circuit, R41 still measured on spec, but i also found that the 0.47R source resistor (R35) measured about 1.1K! as i was stepping the primary voltage up, i was keeping an eye on the voltage across those source resistors, knowing that the target voltage is 25mV, but at the previous measurement spot (75VAC primary) i still didn't measure any voltage across any of the source resistors. i didn't get a chance to measure them at 90VAC primary before R41 started smoking and i had to shut it off.

my first question - i started the amp up with the bias pots turned to maximum possible resistance, as it seemed that should be the coldest bias setting. was i correct, or did i actually start it up at the hottest bias setting??

second question - if i was correct and did actually start it up at the coldest bias setting, do the details of this failure give anyone clues as to what went wrong? as mentioned, i'm pretty new to these kinds of circuits, but i can certainly see how R41 would be pulling way too much current with Q12 shorted (no resistance read between drain, source or gate in any combination), so it makes sense why that was burning. but i'm more confused as to why Q12 shorted in the first place, and what's up with R35 being way too high resistance... it was a while ago, but i'm almost positive i measured all the source resistors and they all measured correctly, so it seems that R35 somehow drifted up to that resistance value during the failure. but i've never seen a wirewound do that... in my experience, they tend to either keep their original resistance value, or fail completely shorted or open, not change resistance dramatically.

last question - does anyone have any references they'd recommend for understanding this power amp circuit? i'm a tube amp specialist, so this topology is pretty new to me. i've been recommended Douglas Self's Audio Power Amplifier Design and definitely plan to get it once i have the money, but in the meantime, i'd love to read something more in the order of an article or chapter on this particular circuit.

[The Dude]

If Q12 shorted, it makes perfect sense that resistors in it's path were damaged. Did you measure R35 in or out of circuit? My guess is that it's open and you are measuring parallel circuit resistance. Also, resistors don't short or go lower in resistance (though there have been EXTREMELY rare exceptions).

On bias: If you look at the circuit, in one bias pot wiper position the pot is shorted removing bias from Q2 and Q22 by effectively shorting B-E together so the transistor can't conduct. This would be the lowest bias setting. As you turn the pot and add resistance between E-B, the transistor starts to conduct as you reach it's on state. I'm not positive I understand your initial pot setting, but I believe you had it backwards.

You say you replaced the FET's. Did you get them from a reliable source? There are lots of Chinese fakes out there.

[mxrshiver]

thanks friend!! i removed the FET and both resistors from circuit to test them. and yeah, that's been my experience with most resistors, and especially wirewounds... very strange. i feel like it partially opened, and that just happened to produce a stable resistance when cool?

this tracks, thank you! i really appreciate the basic explanation, i'm familiar with the operating principles of transistors but am unused to understanding their context in a circuit. so if i have you right, setting the pot to maximum resistance would allow for the greatest possible conduction across Q2 / Q22, which leads to the hottest possible bias for the FETs? if that's the case, then yeah, i'm a dumbass hahaha at least i used the dimbulb limiter!! probably would have zippered the whole thing otherwise... if you could tell me any more about how the amount of conduction across Q2 / Q22 sets the bias for the FETs, i'd be super grateful.

do you think the other FETs are suspect? the rest of them still measure largely identical resistances across their terminals in circuit, in the order of around 12-16K, at least between the Gate and the other terminals; Drain to Source resistances take ages to stabilize. if they were power tubes and i knew the screens had been really badly overdissipated, for example, i'd consider them suspect until i put them on the tester and verified there's not increased screen current. but from what i've heard about transistors in general, it's more of an all or nothing thing - either they're totally fried, or they work fine. is that the case?

FETs are Vishay from Mouser. very tightly matched. i didn't use a fancy matching circuit, just a current limiting resistor for the Drain and measured the Gate voltage required to reach 53.2mA (the current they should be operating at in this circuit if the voltage across the Source resistors is 0.025V). bought 20 of each kind, and they all matched within 0.06V.

[nevetslab]

The Vishay IRFP240 N-Ch and IFRP9240 P-Ch MosFET's you bought from Mouser were bulk purchased, and from my experience with those parts, I've always had to set up a screening process, with a fixture for measuring the Vgs at 50mA at the nominal Vds supply voltage of 83VDC. I then used a somewhat limited range curve tracer (Hameg HM8042) coupled to my Tek scope in XY for measuring the Ids current and finally grouping them into reasonably matched sets. I was buying 50 or 100 pcs of each so I had a decent range to select matched parts from.

I did post this one thread back in 2018 to show what the results were from idle thru 200W output levels Ampeg SVT4-Pro No Output - Music Electronics Forum (music-electronics-forum.com)

Then, over time, I was accumulating a list of N-Ch and P-Ch parts measured having different lot codes and using common Vgs voltages (3.9VDC for the N-Ch and -3.6V for the P-Ch parts, I'd find the Id ranging from 35mA thru as high as 154mA on N-Ch parts, and as low as 4mA thru 103mA on the P-Ch parts.

Working on the SVT3, SVT6 amps, which have that additional Tube stage ahead of the MosFET-based output stage, that output stage is common to that of the SVT4-Pro that most of my testing was done on.

As I handle all of the backline gear maintenance for an equipment rental/rehearsal studio complex in Burbank, CA (CenterStaging, LLC) there's a large fleet of SVT3-Pro, SVT4-Pro, SVT5-Pro and SVT6-Pro MosFET-based bass amps to handle, along with a nearby company out in Alhambra where I see even more of the SVT Pro amps, I built some dedicated test fixtures to streamline the batching/matching process when I'd buy the two N-Ch and P-Ch MosFETs that are common to these amps. I did also have to repair the Hameg HM8042 to replace its rotary digital encoder which took additional effort to restore it back to service (now back up and running).

You may have lucked out if the result of the replacement MosFET's are nicely matched. From selecting sets of 5 parts each N-Ch and P-Ch parts from measured/marked lots, I'd end up with workable matched sets but have never found that sort of tight matching from just 20 pcs of each sex.

There is a wealth of service information on the Ampeg SVT Pro series amps using the Search function that opens up to a good fifteen or more years of posts from our members to read thru. I use that all the time to find specific issues on a product that's got me stumped at the bench. And a lot of info on the MosFET Matching and Batching for these amps, as Ampeg would be buying matched sets from the semiconductor mfgr (unless they too buy bulk and match in-house and then color code them in sets as you find in the new amps).