amplifier bias latchup ?

[Tekko]

My lateral amp have for some reason started latching the bias to 1.5-1.6A on startup, ignoring the bias pot, i have to turn it off, adjust bias to 0 and then it'll start up just fine.

It dident do this before i put it aside to built a dc/overheat protect.

I noticed the VAS is running way hot, could this be causing the problem ?

[Zero Cool]

is it oscillating?

[Tekko]

No oscillation what so ever, its rock stable right up to clipping and beyond.

The circuit is like thjis but with dual output pairs: http://i.imgur.com/kiwus.jpg

The rails are 50V and with that it gives 155 watts before it clips into 3.9 ohms.

[AndrewT]

I don't understand the operation of the circuit.

What is Q5 doing?
Why are there no source resistors on the 2pair output stage?
Why is the input LTP operating with unbalanced currents and voltages?
Why is R8 there?
How far does the maximum unclipped output voltage drop when changing from 8r0 load to 4r0 load?

[Tekko]

I don't have a 8 ohm dummyload so i cannot test that.

I've seen on this forum that with laterals you don't need the source resistors.

Q5 inverts the signal so that two N channel devices can be used.

This circuit is based on the citation 12, in cact the resistor values around the LTP are the same, 680 and 15k. Just slightly different values in the feedback and bootstrap, other than that it is the old citation 12 circuit.

R8 is what i use to null the dc offset.

[AndrewT]

Hi,
duplicate the drive to the other Nchannel. See how Quasi did his quasi FET amplifiers.
Both FETs need to see the same signal.
I am suspicious of the 1k5 as the load for Q5.

What are the currents in R10 and R17? They should be exactly 2:1 ratio.
You should not need to set output offset by adding such a large value resistor as R8.

Buy a job lot of 5W wirewound resistors so that you can assemble a selection of high dissipation 4r0 and 8r0 dummy loads.
Two 4r0 and one 8r0 allow 1r6, 2r0, 2r67, 4r0, 8r0, 10r0, 12r0, 16r0 dummy loads.
Or just make 4 identical 4r0 dummy loads using 10off 10r0 for each load. That also allows 1r0 and 1r33 loads as well.

[triode_al]

Originally Posted by Zero Cool

is it oscillating?

I support the 'does it oscillate' question.
It has been noted that differential input pairs will sometimes oscillate in the 100MHz-even GHz region (you won't ever see that on your scope) and the result is shifts in DC. It has even been reported (by the famous French writer Héphaïstos) that a differential pair will completely block an output transistor while this even isn't visible on the scope.

Even a small wiring problem can give this effect.

just by twopence.
albert

[Henry8]

Originally Posted by AndrewT

I don't understand the operation of the circuit.

What is Q5 doing?
Why are there no source resistors on the 2pair output stage?
Why is the input LTP operating with unbalanced currents and voltages?
Why is R8 there?
How far does the maximum unclipped output voltage drop when changing from 8r0 load to 4r0 load?

With advise like this...........

Why do you bother posting........

[Tekko]

I am on the lookout for nichrome wire to make a dummyload.

[ostripper]

Originally Posted by Zero Cool

is it oscillating?

I had the same issue of "bias lockup" on one of my new BX mongrels. One module had a misplaced 100R IPS degeneration resistor (10r instead - way too high UGP/low phase margin). Before I fixed it , I could turn it off , reduce the tail current with "CCS adj." The bias would be right. Music played , scope showed no oscillation until several hours later or if I increased the tail current. My oscillation was at about 2.4 mhz and intermittant ,I replaced the resistor ... all was well. The OPS survived but showed that "hairpulling" latch-up symptom corresponding to what is described. The bad voltage stage did the same thing on 4 other OPS's as well (warm zobel , "latchup"). One stupid 10R degeneration resistor after I changed every device in the whole board...

Give the poor amp at least a set of small drivers (mje340/350 or mpsa92/42) , and like Andrew suggested .. use quasi's method for the OPS , degenerate the VAS/IPS some (more). Too simple for a 120W+ amp ... good for a 20-40w student amp.

OS