Goldmund Mods, Improvements, Stability

Jump to Latest
ostripper

ostripper

Member
Joined 2008
keantoken said:
Krisfr, why not put 10n-100nF decoupling on the collectors of Q15 and Q22?

If you put the offset adjust at R6/R7 rather than the LTP sources, VAS matching won't matter but your FETs will have to be matched. My preference is to match the LTP and use anything for the VAS.

- keantoken
Click to expand...

Use "anything" for the VAS ??? that is what we hear.. man. the drivers/OPS just condition this for use with our primitive "atmosphere movers" (loudspeakers). ??:confused::confused: I listen to my drivers biased in class A with headphones even before I attach my OP devices for final biasing.

As far as Krisfr's adaptation of that circuit a few posts back , with the MPSA92/42 change , re-compensation would be in order. Nico's full wilson mirror and the beta enhanced differential will only "shine" with careful design and choice of operating points. I agree with the 1 pair of drivers ... my choice would be the 2sc4793/sa1837 in this application.
OS
 
Krisfr

Krisfr

Member
Joined 2007
IF and only IF we go to a thermal feedback for bias control, I would like to try a opto coupled one to minimize the induction into the feed back wires of the magnetic currents in the output area. Has anyone ever seen or tried that? Just get the information and eliminate the stray inductance.
 
jam

jam

Member
Joined 2001
One other thought.......

Since we are essentially buffering the VAS, could we not run the VAS at higher currents and since we have symmetrical drive due to the current mirror could we not dispense with the drivers and drive the mosfets directly. This would get rid of an extra junction and remove a potential problamatic pole. Just a thought.

Jam
 
Last edited:
Nico Ras

Nico Ras

Member
Joined 2005
jam said:
One other thought.......

Since we are essentially buffering the VAS, could we not run the VAS at higher currents and since we have symmetrical drive due to the current mirror could we not dispense with the drivers and drive the mosfets directly. This would get rid of an extra junction and remove a potential problamatic pole. Just a thought.

Jam
Click to expand...

Hi Jam, I suggested getting rid of the drivers early in the thread and you do not have to increase the VAS current either.

Regards

Nico
 
Krisfr

Krisfr

Member
Joined 2007
On the output devices heatsink there is a temperature that is linked to the output devices temperature at any given time based on output of the amplifier. Now take a diode, a 4148 or similar and place it on the heatsink. With wires, take the voltage drop across said junction and transfer it to the bias setting transistor of the amp. Normally the wire that transfer that information, also pickup the large magnetic fields of the output devices of the amplifier. This can start a loop between the output and the bias setting stage. By opto isolating this diode, you eliminate the induced voltage that put unwanted signals into the bias setting system. Also if feed to a op-amp you can tailor the action of the bias to reflect BETTER the control the thermal runaway system. I have done this before but I lost my notes from a FEW years ago. It was the best track of thermal info I had seen. And it could be tailored for optimization.
 
ostripper

ostripper

Member
Joined 2008
keantoken said:
OS, I meant I wouldn't bother matching the VAS. Not that I'd know how much it matters.

What do you mean about operating points? If I worried about operating points I would run nearly every transistor at 10-30mA where Hfe and Ft are best

- keantoken
Click to expand...
10-30mA for the to-92's ?? .. poor things. 1-2ma is where the 1845/992's develop the best Hfe , run cooool ,too. For the VAS , why not to-126's ?? then we can crank them up to 10+ mA and drive them laterals direct.
 
You must log in or register to reply here.
Top Bottom