PPI a2150: FET driver blown? (waveforms attached)

[rth]

I have a circa 93 PPI art series 2150 (600W bridged) that sustained thermal damage due to my friend stacking this amp on top of another a2150

I just can't let this thing sit on the shelf any longer so I'm going to fix it but I hit a snag today. I replaced all 10 power supply FETs with Fairchild FDP20AN06A0 parts, which should be better than what most people are using to repair their amps since they are 60V/45A/17mohm/15nC... nicest thing I could find and money is no object since I sample them

Anyway, the damage consisted of a couple blown FETs and nothing else noticable so I ordered the 10 replacements and also went ahead and changed the gate drive resistors to 22 ohms like this thread suggests. I decided to make sure the sg3525a PWM driver was working right before I put the new FETs in so after I took the old ones out I hooked up the amp and watched the two gate drive channels on that chip. They looked great with 10.7 V amplitude and 2 us deadtime between pulses. I was happy so I changed the resistors and put the new FETs in and fired up the amp. About 2 seconds after I gave the remote lead 12v, the amp went from drawing 80mA to drawing 50 Amps! I looked at the gate drive and it looked UGLY. The pulses were only going to ~7 V and there was no real deadtime.

PPI drives the FETs directly from the 3525 so it seems that my problem must be the 3525, no? I guess I really need to pull the rectifiers to isolate the power supply because I was seeing some serious voltage drop across my supply wires that might've been affecting the operation of the driver.

[rth]

Here are the waveforms:
Gate drive with no FETs installed:


Gate drive with FETs installed:


Zoom-in of deadtime with no FETs installed:


Zoom-in with FETs installed... no deadtime, both outputs partially high during the transition:


Hopefully I can find a free sample for the sg3525, I would hate to actually have to spend money to fix this thing

[rth]

It appears that the crappy waveforms above were due to the voltage dropping very low due to high current through small wires, so to isolate the circuit I removed the transformer and here is what the gate drive signal looks like:


The amp is only pulling 97mA now so it's easier to work on than when it was pulling 50A Does that signal look like one would expect? The slew rate seems poor to me and the fact that one channel has instant 10V (10x probe) while the other has to build up to it seems very strange. In retrospect I should have pulled the secondary side of the transformer rather than the primary so that I can fully operate the power supply without a load... I'll put the primary side back on and lift the other side and see how it looks. I'm now using a uc3525 from TI and I stuck it in a socket to make things a little easier to test

[amplifierguru]

Ugh!!!

I'll stay analogue till my time's up!

[MikeB]

Did you really let this PWM-driver eat TEN of these mosfetsmonsters ?
Each of these have ~1nf inputcapacitance !
Looks like one or two of these would already do the job...
What mosfets were installed before ?
The waveforms look like the driver is completely unable to handle
the drain to gate capacitance.

Mike

[Eva]

These gate drive waveforms are defective, some of your gates are probably shorted to drain or source directly or indirecly through solder bridges. Check it carefully

[rth]

It should be able to handle those FETs because lots of folks are using the irfz44n to replace them and it has a 1470 pF input capacitance whereas the ones that I'm using are 950 pF. Maybe the 22 ohm gate resistors are too low and are causing too much power dissipation in the driver? That's doubtful since it doesn't even get warm and apparently PPI changed them to 22ohm in a later rev.

[rth]

Lots of progress!

It turns out that one of the new FETs had already blown due to the high current in the output stage, which shorted the gate to the drain and caused those nasty waveforms. I removed this FET and it the power supply works great with the rectifier legs lifted except only 4.7 V is seen at Vc, which means the FETs are only getting 4.7V at their gates, but otherwise the square wave is very square but does have some ringing (it happens). This low Vc seems to be due to a little TO92 PNP that sits between +12V and the Vc input on the sg3525... I don't know what the purpose of this thing is but I have a feeling the bad FET killed it too since it's only rated for 500 mA. I shorted that PNP's base to collector and then I had a nice 11V square wave going to the FETs

Why didn't PPI hook Vc directly to +12V like most other designs? What does this PNP do?

The source of the massive current draw turned out to be some toasted NPNs in the output stage... after I removed them the amplifier fired up fine with +/- 40V rails but I can't put a load on it till I get that Vgs up higher to avoid blowing up the FETs. I'll be back tomorrow with waveforms and (hopefully) good news

[rth]

Replacing that little PNP driving the 3525's Vc pin didn't change anything so instead of getting covered in thermal paste while tracing it to the other little NPN's in the circuit, I just hot wired Vc to the +12V line Now I have awesome looking gate waveforms and +/- 57V output rails

Everything is back together (with one missing power NPN, oh well, I'm running 4 ohms stereo) but now I see the -15V regulator is dead I reallllly don't see how PPI ever got the + and - 15 V regulators to work since they're tied directly to the output rails, which are 20 V above the max input rating (35V) of these 7915/7815 regulators! How the hell did PPI get away with this?

As soon as I get this regulator issue resolved, there will be one more art series rockin the block

[rth]

I don't have any 7915's so I decided to throw a 7812/7912 combo in there to power the op-amps. After 5 seconds of running like this the -12v capacitor exploded! I checked the voltage and it was at -49V! It seems the 7912 did NOT like the high input voltage, as I expected. The 7812 seems to have shut down instead of breaking down as it is only showing ~0.45V on the output. Again, I have to ask: How the hell did PPI make these voltage regulators live???