Ampex linedriver stage with SE/NE 540 (continued)

Having typed a reply to PRRs informative words on
"Looking for SE540 / NE540 info" last week I logged in to
post but things had changed... (and the rest we know).
Good everything's up & running again !

Wanted to add this:


Thanks PRR for your time & trouble to provide this info.
Wishing you many tasteful meals from your new kitchen :wink:


>> That +/-12V supply looks underwhelming
... but not really:

You're right, the healthy signal-levels can be there and fit the +/-12V supply. Just too focused on maxing out supply-voltages...

As I understood from the words above, for todays standards this linedriver-circuit is nothing special and its performance easily surpassed by more common components, although it has its uses.

There's of course still the fun of keeping an old PCB into service.

OK, I'll wire it up a bit further and use it with or without the output TXs - of which I now guess they're better 'spent' for another circuit. For the time being though could use them wityh the original driving stage.

The Ampex design uses an imprecise idle bias current setting. Thermal tracking is not an issue because the bias is set to 1.65 Vbe, and Q5 Q6 probably idle at very low current. Probably just enough to make crossover distortion very small, relying on massive feedback to eliminate the last bit of crud. It is a dirty trick that often does not sound bad, and is still popular because a high-bias Class AB stage is more troublesome in mass production.

Click to expand...

About the quiescent current of the output devices being
1.65*Vbe (instead of 2*Vbe as in the Walt Jung-circuit):
what I saw so far is that those output bipolars as used in
the Ampex-circuit are of about the same dissipation rating
as those in the Jung-circuit, so biasing the Ampex-version
a bit hotter seems possible without further mods, correct ?
Or is there something I'm overlooking for the '540 ?

Can check if there's a worthwhile THD-improvement.


Regards,

Peter

> About the quiescent current of the output devices being 1.65*Vbe

Don't change that without adding emitter resistors first!



2.7 ohms is an OK value for emitter resistors.

Make the Vbe multiplier adjustable from 1*Vbe to 2*Vbe by replacing the top bias-set 330 ohm (R48?) resistor with a 0 to 500 ohm resistor, possibly a 1K fixed and 1K trimmer.

Start with the trimmer set to zero ohms. Clip a battery powered DC milliVoltmeter from emitter to emitter (red marks), to read the sum of the voltages across the two emitter resistors. Slowly turn the trimmer. Bring it up to just 10mV-20mV and stop. Let it cook a while. The emitter voltage difference will probably drift up as Q5 Q6 heat up. Keep a finger on the transistors (one nice thing about 12V supplies!) and an eye on the meter. You want to sneak it up to 50mV-60mV across the two emitter resistors (10mA idle current), but no higher, and not even that much if it has any tendency to rise as it warms.

That gives 10mA idle current, meaninging it will run Class A up to full voltage in 500 ohms. It is also very nearly the optimum bias for a Class AB stage, so it will stay clean into much lower impedances (like those high step-up transformers).

Thanks PRR for the circuit-update,

Did some THD-measurements before modding.
Will do the mods tomorrow & measure again.

FWIW, unmodded was like this:



All with 500 Ohm load, no output-TX yet.
Measured idle current through output devices 2.44 mA.

That gives 10mA idle current, meaninging it will run Class A up to full voltage in 500 ohms.)

Click to expand...

10mA & 500 Ohms give a max. output voltage of 3.54 Vrms, correct ?
But I must be overlooking something here, since you're likely not considering 3.54 the full voltage.

** EDIT: I forgot a factor of 2; you likely meant biasing the output devices
at about half the max output current and still have class-A.
That brings the max output level indeed up to the full voltage. END **

It is also very nearly the optimum bias for a Class AB stage, so it will stay clean into much lower impedances (like those high step-up transformers).

Click to expand...

I remember hearing a now retired colleague
talking about that optimum 25 mV voltage drop years ago.
Thanks to your 'very nearly the optimum' remark I finally
looked for some more info and got the full story
( http://www.diyaudio.com/forums/showthread/t-21001.html )


Regards,

Peter

Some measurements on the now modded circuit:


Voltage across both emitter-resistors together was adjusted for
50 mV. Above this (~ 55mV) and the circuit indeed runs away.

At 50 mV both transistors get warm but not hot.
The SE540 driver-IC does get hot though (indep. of mod)


THD for modified circuit:



... which is about the same as before :!:

All with 500 Ohm load again, no output-TX.
Measuring filter 100 kHz, just like for unmodded figures.

The crossover-dist. is now gone - but that just didn't contribute that much to the THD-figures.


Just based on these THD-specs there really seems to be no reason to use this circuit as a linedriver instead of stacked 553X's or drivers like the SSM2142, DRV134 :wink: - like previous posts already suggested,
so just leaving the fun-factor of phones as a reason to use it.

I'll wire it up sometime and give it a listen.


Regards,

Peter

FWIW, I wanted to check if those THD-figures could be polished
a bit by reducing the closed loop gain of the circuit.
So far I had used the original gain setting resistors so gain
was 11.8 dB.

By using a larger input-resistor the closed loop gain is reduced
so in principle there's more 'gain-margin' to improve THD.
(R36 now 32k i.s.o. 10k, in the 'inverting input leg')

I didn't see much difference though, not more than 1 or 2 dB or so.

If it goes proportionally, THD should improve by some 10 dB, correct ?


So in addition to the various specs people kindly provided last week,
does anyone know what the open loop gain of the SE540 is ?


BYW, I guess it's OK to use this much feedback with the 540,
and otherwise the external compensation can be adapted,
just like it is done in the Jung-circuit as posted by PRR.


Thanks,

Peter

Just wondering, why could it be that even with more NFB (less closed loop gain), the THD won't decrease here ?

Maybe I'm overlooking something or making measuring errors...

Peter

Are you measuring THD or THD+N? The fact that it is higher at lower levels makes me think it is THD+N. You may be hitting the noise floor, so you can see any changes in the actual THD.

Thanks for the suggestion. I'll further check the residue on a scope
and do some noise-measurements.

Thanks,

Peter