Ampeg SVT4-Pro N-Ch MosFETs Is current leakage?

[nickb]

My favorite tool for finding where large unwanted currents are going is a thermal imaging camera. Also handy for spotting mismatched power transistors. Do you have one? They are a huge time saver. Low res is fine and they are not so expensive nowadays. Failing that you'll have to pull non essential things off the board one by one.

[nevetslab]

Only device I have in that domain is a Fluke 80TK-IR infrared thermal probe. Thus far, I haven't allowed the amp to turn on so hard as to let failure to occur. Just got back to the shop yesterday after holidays. What I am seeing is near-equal current flow thru each half of the output devices, while still not finding 'bypass' path around P-Ch MosFETs.

[nevetslab]

Still trying to solve this Ch A Output Stage imbalance. I do still have oscillation....thought that had been tamed. During powering up the amp via Variac, until the DC Protect Relay closes (connecting the output of the amp to the load connectors), the output is shorted to ground.AC Mains voltage is about 90VAC when the relays close. Monitoring DCV across any of the Source resistors in the Output stage, the voltage drop is equal. I've had the bias adj pot AP101 set for 12mVDC across the 0.47 ohm Source resistors. Once the relays click in, the output is no longer grounded. If the amp hadn't been powered up for a little while, the DC balance remains equal, but then, the N-Ch side begins conducting more, with Vgs increasing, while the P-Ch side simultaneously conducts less, with it's Vgs decreasing. I'll also see an oscillation envelope develop. By the time the P-Ch side has dropped quite low, and the N-Ch side showing 30mV or more across the Source resistors, oscillation envelope increases. I shut it down at that point.

I've tried splitting up the compensation cap value of C115 (680pF nominal) to 100pF or larger at each MosFET. That's how Aguilar sets up their DB751 power amp...though those are BUZ901/BUZ906 Bilateral MosFET's. Their voltage gain stage is similar to that of the SVT4-Pro (apart from adding a bipolar emitter follower stage between the MosFET's and the Voltage gain stage), with no local compensation caps on the drive transistors feeding the gate busses. Splitting the comp caps up to each device here didn't change anything. Ch B, meanwhile, is perfectly stable, even though it's MosFET's aren't balanced as they once had been when new. But, no issues like the behavior I'm seeing on Ch A.

I'm about to tempt fate by un-soldering the replacement IC1 NE5532A with a socket so I can try alternate IC's, such as a MC33078 in place of the NE5532. I found my stash of 200V rated NPO Ceramic caps. The quality of the PCB on these amps is very poor, and traces/solder pads lift off with the greatest of ease, all having oversize ID holes to aid in that bonging failure! Sigh.........

[nevetslab]

I replaced IC1 with a new MC33078, mounted in gold machine pin socket, also replaced C104 100pF, after digging all of the RTV goop in the area preventing removal. I had restored the original compensation cap C115, though moved it to a new location so there wasn't this long snaking thin path to where the PCB has it. Got it nice and direct between Vdd buss and Vgs gate buss. When I powered up the amp again, bias still set for around 10mVDC across the source resistors I was monitoring, it remained stable for some time, AC mains at 110VAC. I did have the N-Ch Vgs buss being monitored, along with the op amp offset point at R111/R120/R121, and found after a while, the balance shifted, but this time, the P-Ch side increased while N-Ch side decreased. Opposite behavior as before. Made me wonder if I didn't have feedback connected. Shut down, found the path on the foil pattern drawing, then followed that on the PCB and verified continuity. Also, re-soldered the two jumpers that completed the path between the op amp circuit and the output buss, just to be sure.

Removed the DMM probes from the Op amp output and Vgs, as those do add lead capacitance to the amp circuit. I never saw oscillation break out again, nor did I see either N or P Ch stages change in conduction....Is current remained stable. I turned the variac back down to 90VAC where the relays open, and slipped my small screwdriver onto the bias pot AP101, brought the bias up to read around 18mVDC across the source resistors, and held the screwdriver in place, raising the variac back up to 110VAC. Tweaked bias for 20mVDC on both. Remained stable.

I then made a chart to record the readings of all 10 MosFETs, finding all ten very tight...DCV readings ranging from 20.8mV to 23.3mV (44.3mA to 49.6mA). Vgs N-Ch read 3.79VDC, Vgs P-Ch read -3.35VDC.

Turning off the amp, Ch A output swings as high as -14VDC (didn't have a fast reading DMM to actually watch it, nor had the scope attached on the output busses to see the transient), while that of Ch B never changed more than a few hundred mVDC. I don't know the significance of that, as I don't recall making those readings over the years of servicing SVT4-Pros or their partners in crime.

So, now, I just need to see if this is an amplifier...I'll have to wait until I come in tomorrow, as I'm heading home for the day. Maybe I finally DO have a working amp, though after so many days of defeat, hard not to have full confidence of it and willing to put the amp back together.

[nevetslab]

I took the weekend off to watch football (not much of a fan, but it was a distraction from this project and partially from the news media's wide range of views of the attempted coup Wednesday). I powered up the amp, no load attached from a cold start, shop sitting around 65 deg F. Both N-Ch and P-Ch MosFET's were still running around 25mVDC across their source resistors equally. I watched it for a while, and, after a few minutes, I again saw low level oscillation on the output (around 40mV RMS), and saw P-Ch stage begin conducting more, while N-Ch stage conducting less. Dropped the Variac to 90VAC where the output relay opened, shorting the output to ground. Both N and P ch stages back to around 20mVDC. Brought it back up to 110VAC, seeing about 25mV on both N & P-Ch source resistors. Let it sit there a few minutes, and this time, didn't see any oscillation start up.

I plugged in my sine wave signal...400Hz into the Power Amp input jack. Slowly brought up the level. I got up to around 8VAC when I suddenly saw the N-Ch stage conducting greater than 300mVDC and quickly turned off the AC switch. Turned the variac back to 0VAC. Then, with the oscillator muted, brought the variac back up to and past 90VAC, without seeing any DC offset on the output, and both N & P-ch MosFET's conducting normal....21mV range. Tried applying signal again, and, got the same results as I quickly switched power off. Turned back on with the variac from 0VAC, came up ok, brought it back up to 110VAC, output stage conducting equally (around 25mvDC across source resistors). After a minute or so, again see P-Ch stage conducting more, N-Ch conducting less, with low level oscillation oscillation on the output (around 40mVAC)

I did add a pair of 10k 3W Metal Oxide resistors across the main power supplies to ground for discharging them faster than the existing 47k 1/2W resistors in place (R35 & R36). Still takes 4 minutes to discharge the supplies to 1/10 the nominal supply voltage of around +/- 80VDC @ idle.

So, I don't yet have a stable amplifier. I still have the 'stock' circuit values on Ch A, apart from having changed to an MC33078 op amp for IC1 in place of the NE5532AN. I do at least have an IC socket in place, but, have been seeing the same behavior for the most part, though the trend of imbalance with no signal applied has switched from the N-Ch to the P-Ch conducting more at idle when the low level oscillation starts up. What to do?

[nickb]

What frequency is the oscillation?

[nevetslab]

Further findings. Using both an open-circuit T/S plug as well as a shorting T/S plug to insert into the Power Amp input jack to Ch A, when I see the oscillation come up, I can affect that by dead-patching the shorting plug, while dead-patching the open circuit plug there, does affect it, but doesn't always stop the oscillation. This, I think, suggests the oscillation is coming from the Preamp PCB assy, including the Graphic EQ PCB, as well as the rear panel I/O PCB...all of which have quad BiFET IC's TL074's, along with some TL072's. When I began the repairs on this amp, I did have failure of the -15V IC regulator LM7915T, with shorted output filter cap and had to dig all that out from under the vast amount of RTV gooped onto the supply PCB..

Looking at the PCB layout of the preamp PCB, Graphic EQ PCB & I/O PCB, all having IC's on them, the power supply bypassing of the IC's with 0.1uF caps are not so great. All are bypassing +15V to -15V, but at quite a distance from the IC's in the layout. None are using bypass caps AT the IC's supply terminals to ground. Ground plane isn't a highlight of this products' PCB layout. And, if you get oscillation in the system, ahead of the power amp, it does propagate along ground, or, on the preamp signal path thru the Preamp Output/Power Amp input jack to the power amp circuit. This doesn't yet answer why I only got so far directly injecting sine signal into the power amp INPUT jack before it broke out into oscillation and got high conduction on the N-Ch MosFET stage, and increase in the conduction of the P-Ch stage when that occurred. Does this seem like where I'm having the instability problem?

Bass Preamp PCB Schematics (53904H0).pdf

PWA Bass Preamp PCB Layout (53905PWA_1).pdf


I also wonder what happens if I mount the power amp PCB assy, and see if I get the same problem(s) mounted. I can still monitor the source voltages on both N-Ch and P-Ch stages, though more tedious due to the I/O PCB (if I mount it). I guess all good questions. I do feel like I'm only talking to myself in this pursuit. Has anyone else run into similar problems with the SVT4-Pro or SVT5-Pro amps like this? Both are nearly identical, dismissing the graphic EQ and one extra pair of MosFET's in the output stage.

[nevetslab]

This morning I disconnected the power cable to the preamp PCB assy, so there's just the power amp connected. I also installed the power amp assembly into the chassis. The replacement Cliff Male XLR's for the I/O PCB came in, but their mtg holes were not small to accept the small SMS screws...they were for using #6 hardware. So, had to install Clip Nuts onto the flanges, and had to nibble away part of the relief on the I/O board to clear the upper clip nuts. With the preamp powered down, the power amp worked fine, no oscillation or imbalance. Drove signal thru it using both 4 and 8 ohm loads, all working fine. I then powered down, dropped the AC mains to 88VAC, where the relays wouldn't pull in, and reconnected the preamp power connector. Dead-patched the power amp inputs, and turned it on, having power to the preamp, though a bit starved on the preamp tubes. Preamp working fine, and no oscillation found. So, I ramped up the variac to nominal line voltage, and have a working SVT4-Pro amp again. Not seeing any hint of oscillation, and the N-Ch and P-Ch MosFET's are looking equally matched. DC current levels driving load to full tilt are equal, and remain so all the way down thru lower levels.

The only thing that does puzzle me is when I power down the amp, Ch A, which is what I've restored...it's output goes negative at turn-off....haven't looked at it on the scope, but I've seen DMM readings as high as -20V at turn-off, which then drops down towards 0V after some time. Ch B doesn't have that behavior. The relays both open, so the outputs are shorted, so that large offset voltage doesn't hit the speaker. Any ideas on that one? I had gone thru my service notes on SVT4's and SVT5's, and don't find I had documented that turn-off behavior, so I'll have to look at other amps as they make their way thru the shop.

CenterStaging is just bringing the inventory back from their New York operation, so there should be work available again. And many SVT4-Pros to look at.

Connected a 410 GK cabinet to the amp, fetched the Fender Jazz bass to have a listen. Sounds like an SVT4-Pro. Still have to fetch a Bridge Mono cable for it to check it in that mode, but, plenty of time for that. I think the task is over for now.