Amp7 overcurrent on VNN

[goneXC]

OK, here's the admission - I did start up the test with the TA3020 installed but not powered up - don't ask how! Then got it right but V5 current seems a little high at 60mA but the VN10 current does what it is supposed to - i.e. small inrush then settles close to zero. Then connected the VNN with 50 ohm protection and it got very hot very quickly - it draws around 500mA (didn't bother with VPP).

Took the TA3020 out shorted the FETs gate to source and retested. All measurements correct. I cannot see or test any obvious shorts. I even took one FET out and compared it with a new one and all measurements seem OK but is this accurate? Can a FET be tested a) in situ and b)without a scope?

Have tested the TA3020 in another working Truepath and it is fine.

All suggestions welcome!

b.t.w. when searching for the data sheets for IRFP4321 Google kept directing me to IRFB4321 which is very similar but can handle higher current - interesting!

[V-bro]

As long as the gate charge is as low it could be a good alternative.

How was BBM timing set??

[goneXC]

BBM is set for 80nS i.e. 3-5 and 2-4. I have seen elsewhere that one way of finding a short is to allow it to get hot but this makes me nervous.

[V-bro]

It would make me nervous. Blowing up fets is one thing, but ruining a board is the last you want! It is common for the fets to draw monstrous current when the BBM timing is accidentally not set and zero by default. Happened to me once and was still lucky I pulled the plug in time and had no large capacitors fitted yet. Smelled like I warmed up the iron after a winters sleep...

[goneXC]

I really cannot see where else to put the blame apart from the MOSFETs. Unfortunately everything that I've read requires the FET to be unconnected. Rather than take everything out I'd prefer to nail the culprit/s first so my plan is this: short out the G-S and then progressively remove the problem children. If I do this with the TA3020 in situ will it be damaged? Testing without the TA3020 and the FETs shorted has not revealed any problems - am I looking in the wrong place?

[V-bro]

Well, the AMP7 doesn't have pull down resistors on the FET gates, whereas the Truepath does. Fets are latching by nature, which (when they are not switching, with no TA3020 connected) could put them in a state connected or unconnected (drain-source) at random (what they were last).... That could be your culprit perhaps?

[goneXC]

So your suggestion is to remove the whole lot? I'm hoping to avoid that!

[V-bro]

No, if they switch they're fine. Just measure if they do and if you feel like it needs pull down resistors it's not too hard to implement some 480K resistors to ground on the gates...

[goneXC]

Are these analagous to the ground reference resistors R218, R219, R118, R119 on the Truepath? I notice these are connected gate to source.

[audiophool]

OK, here's the admission - I did start up the test with the TA3020 installed but not powered up - don't ask how! Then got it right but V5 current seems a little high at 60mA but the VN10 current does what it is supposed to - i.e. small inrush then settles close to zero. Then connected the VNN with 50 ohm protection and it got very hot very quickly - it draws around 500mA (didn't bother with VPP).

Took the TA3020 out shorted the FETs gate to source and retested. All measurements correct. I cannot see or test any obvious shorts. I even took one FET out and compared it with a new one and all measurements seem OK but is this accurate? Can a FET be tested a) in situ and b)without a scope?

Have tested the TA3020 in another working Truepath and it is fine.

All suggestions welcome!

b.t.w. when searching for the data sheets for IRFP4321 Google kept directing me to IRFB4321 which is very similar but can handle higher current - interesting!

You can quickly test mosfets in-situ with a multimeter.

• Turn off power to the amp.
• Multimeter on diode test. Test DS and SD. DS should be open circuit. SD should be ~.4V
• Multimeter on ohmmeter. GD and DG should be open circuit. GS and SG should be equal to the RGS resistor (Open circuit for AMP7; 510k for Truepath)
• If you check all four, you will quickly see if one is defective.

Henceforth I'm assuming that AMP7 output section schematic is the same as Truepath (other members, please advise of any significant differences).

First consider this. Where is the INN coming from? Take a look at the schematic here: http://www.maedos.com/Guideline.html...ng_VNN_and_VPP
With no load connected, you can see that there is no INN fault current path to ground through the mosfets. The current has to be coming from VPP, but that can only happen if two mosfets are defective - one low side and one high side on the same channel.

The fact that the fault disappears when you remove the TA3020 and pull down the gates strongly suggests that there is nothing wrong with the mosfets. Either the TA3020 is pulling at least two gates high even when muted - you can check by checking VGS for each mosfet (but this is unlikely) or, the fault current is through OCSLP or OCSLN and through the TA3020 (which is also unlikely for a TA3020 that is not defective).

The fault current could also be from VN10 (because that's 10V above VNN) - but I doubt your VN10 can source 0.5A!

I suggest that rather than pulling mosfets off, it would be better to to trace the fault current to figure out where it is coming from. Some tips here: http://www.maedos.com/Guideline.html...on_trace_short

(Note, I updated the Maedos Guideline doc so you may have to clear your browser cache / refresh Maedos page to follow the links.)