Method to protect a MOS-FET->BIPOLAR 200V powered amplifier during tests

[xavlaur]

Helo,guys !
Do anybody has different method to protect during trial tests a 200V powered amplifier - repaired on left channel ?
I mean everything seems fine on ± 20V, ± 30V etc. .... but on ±100V instead of burning 5A fuses (normally should be 10A fuses) , burned again final and prefinal transistors.
Idle current is normal (like in non-burned right channel) , all voltages seems the same (like in non-burned right channel), no DC on output.
I've marked on diagram "burned 1" and "burned 2" meaning "burned 1" = what found already burned when amplifier was received, "burned 2" = what keep on burning during trial tests.
On trial test I've had weaker final transistors (2SC5198 SI-N 140V 10A 100W 30MHZ - TO3PN / 2SA1941 SI-P 140V 10A 100W 30MHZ - TO3PN instead of : MJL3281A NPN TO-264 260 V, 15 A, 200 W / MJL1302A PNP TO-264 200 V, 15 A, 200 W and also Q409 = MJE350 SI-P 300V 0.5A 20W was substituted by weaker B861 PNP  -200 V   -2 A  max 30W) - and ONLY 2 of final transistors WERE USED , a PNP and a NPN - but based on datasheet should be working , at least for a while on ± 100V power supply.
Reason for burned2 is that potentiometers were so old that became noisy and during volume up it was a strong noise from potentiometer going to speaker - "overload" the stage - and after that R428 smoked by Q410,Q411 shorted etc .
Regarding fuses: Due amplifier is rated 2 x 600W I thing fuses should be around 6A not 10A as in diagram, no matter one MJL is max 15A.
One SC transistor is rated 10A but during trials both of them shorted altough 5A fuses not burned (to protect anything).
Burned2: Both 2SC5198 and 2SA1941 shorted between B-C not E-C
Burned2: IRF 630 shorted between G-D (10 ohms)
Burned2: IRF 9630 shorted between G-S (0 ohms)
Strange is R407 was burned1 altough Q401 ok.
Also, I think on 100V thermal heatsinks transistor silicon pads could be affected as electrical insulators.
My method was:
1. Tested idle current, DC current on ± 20V with 48V/ 20W  lamps instead of fuses - ok
2. Tested idle current, DC current on ± 80V with  48V/ 20W lamps instead of fuses - ok
3. Tested idle current, DC current on ± 100V with 220V / 100W lamps instead of fuses - ok
4. Tested idle current, DC current on ± 100V with 1 A fuses - skipped
5. Tested idle current, DC current on ± 100V with 5 A fuses - ok
6. Burned on step 5 while overloaded by noise. 5 A fuses remained ok.
I ordered today the properly spare parts but - meanwhile - due that 4 x MJL3281A + 4 x MJL1302A + IRF ... etc are quite expensive , I would like to receive any suggestion in order to avoid burning them again. Thanks in advance.

[Kleinstein]

The ouput transistors can see nearly 200 V. So 140 V transistors are not really sufficient it only take a transient excursion of the signal to more than 40 V to be beyound 140 V.

A burned resistor R407 is odd - this should not happen - though the current and power can be relatiey high under normal conditions, but should not change much with signal or load.

For pretecion to the power transistors one could try some 100 W light bulbs in sereis with the supplies and only a small capacitor (e.g. 100 µF) after the bulbs.

[xavlaur]

(Google translate, sorry , no time today to edit directly in English...)   
Today I've tested the problem channel with the new components:   
1b. load 100 + 8 ohms, ± 100V with 47 ohms instead of fuses - ok   
2b. load 100 + 8 ohms, ± 100V with 20 ohms instead of fuses - ok   
3b. 100 + 8 ohm load, ± 100V with 1.5 A fuses - ok   
5b. 24V / 50W + 8 ohm bulb load, ± 100V with 1.5 A fuses - ok   
I am going to put all the ends and I will go directly with 100W bulbs instead of fuses, then 5b, then 8A fuses directly with the 8 ohm load and that will be ...   
MJL21194G NPN 250V 16A 200W TO264   
MJL21193G PNP 250V 16A 200W TO264   
from Mivarom, I was warned that it might be off-specs but I am optimistic.   
   
My reservation is related to the following:   
R1. At 600W / channel the fuses should be 6-7 A instead of 10A as they are on the manufacturer's diagram, otherwise they are useless   
R2. SOA for MJL3281A and MJL1302A - on the original diagram - together with the protection on a single emitter leads to catastrophe in case of overload   
R3. Protection on a single emitter, on the resistance of 0.33 ohms (just like the others that are not protected) leads to another catastrophe because the transistors are expensive and you can't buy 20 and out of 20 to pair 8   
R4. Under the above conditions the scheme seems stable for ± 70V ... ± 80V, not for ± 100V as designed   
R5. At ± 100V it would be reasonable not to lower the 8 ohm sound. There are no manufacturer specifications.   
I'll be back after all the finals onboard.   

[t1d]

For pretecion to the power transistors one could try some 100 W light bulbs in sereis with the supplies and only a small capacitor (e.g. 100 µF) after the bulbs.

Kleinstein is the best! I have worked with him on other projects. If you don't know the trick that he suggests, it is called a "Dim Bulb Tester." If you search that term, you will find lots of DIY videos, schematics, etc.

If you need to test the mains power supply with an oscilloscope, you will also need an isolation transformer. Such a transformer separates the ground plane of the DUT from the scope, preventing explosion, etc. You can make one from two transformers connected "back to back." Meaning, the secondaries of the two transformers are connected. The mains input side of the second transformer becomes the output of the device. It is best to use two identical transformers. Search "DIY Isolation Transformer."