Triple emmiter follower

Nico Ras · 2008-12-17 9:13 pm

By the way my friend the phase response of this output stage is quite remarkable with 0,4 degrees shift at 20 kHz, I think this output stage will be almost characterless - which of course it good.

Nico Ras · 2008-12-17 9:15 pm

The only problem is how to control the thermal stability. If you solve this issue then I can assure you that it will be a remarkable and fast output stage.

Nico Ras · 2008-12-17 9:18 pm

lineup said:
Bonsai

Why settle for Tripple

When you can try the brand new

Lineup Quattro Output Stage
.. just invented

high gain super plus!!!!

Lineup, put values to it. That thing does not make sense to me, nothing is in balance. :bigeyes:

bogdan_borko · 2008-12-17 9:20 pm

Nico Ras said:
It would appear that the triple emitter follower has temperature stability issues. Lineup what is your opinion.

Nico

It`s thermaly stable there`s error in sim becouse of constant current sources...
This will do the trick

the attachment)

A word about oscilation issue?

bogdan_borko · 2008-12-17 9:27 pm

I used the pulse stimulus and I do not see the oscillations you do, in fact the wave has a small over and undershoot.

Yes, but look at the output waveform when you parallel 8 Ohm load with 100nF or more.

In practice, obviously you will be applying NFB when used in circuit.

No.

bogdan_borko · 2008-12-17 10:15 pm

NGFB amplifier
graph: full power 80W/8Ohm
p.s. distortion including gain stage.

wg_ski · 2008-12-18 12:26 am

Nico Ras said:
The only problem is how to control the thermal stability. If you solve this issue then I can assure you that it will be a remarkable and fast output stage.

It is a bit tricky with 3 Vbe pairs, running at 3 different currents and therefore 3 different temperature coefficients. On top of the transient thermal effects. But I thought Dr. Leach solved that problem (or at least, came up with a really good compromise) a long time ago with with a string of diodes in the base leg of the Vbe multiplier. That at least lets you track two slightly different tempcos at once.

Bonsai · 2008-12-18 12:46 am

Bogdan,

Some points on your design.

You are using a speed up cap (C3). Given the very low value resistor in parallel with it, you could probably drop the cap. I found that the speed up cap can make the situation worse in some cases so you need to check this in your final circuit.

Also, the peaking you see in your response will lead to loop stability problems - I don't think this is the cause of th e parasitics you are seeing on the output square wave test signal. My usual way to solve peaking is to adjust the inoput filter time constan on the amp - of course, you should only do this after the zobel and output inductor have been fitted and you have no other oscillation or stability problems.

I mentioned I saw 100kHz on my sine wave test signal in my earlier post - it should have read 1Mz - 3MHz.

Thermal stability. Yes, I agree it is a bit tricky with a triple. I dropped the idea of using diodes or a TO-126 style compensation scheme and instead went for an SMD transistor mounted physically very close to one of the output device collector leads (TO-3P style package). The collector is the die header and so tracks the chip temperature quite well. This scheme gives quite good thermal coupling with fast response because of the small sense transitor. I initially used thermal grease over th e sense transistor and around the power transistor collector lead to give really good coupling, but this in fact caused my Iq to decrease unacceptably as the amp warmed up. The best performance was with out thermal grease. I get very good stability now from cold to running very hot. With thermal compensation, you have to just try it out.

wojtek5001 · 2008-12-18 4:44 am

Hi Bogdan. You use low impedance source to drive triple EF. In most practical amps that circuit is driven from very high VAS output impedance, optionaly pararelled with resistor. Try to use source with about 100k output impedance. You can, for example set R58 and R59 to 200k, and use voltage driven current source at input. For me that helps with triple EF. With double EF there are still little overshots - I'am working on it.

KLe · 2008-12-18 7:17 am

lineup said:
Why settle for Tripple
When you can try the brand new

Lineup Quattro Output Stage in Post #8

.. just invented
high gain super plus!!!!

Hi Lineup
Very interesting circuit ...

I do not wish to high jack the thread but I thought that I would share his comments with Lineup ... sorry about that, in advance.

A good friend of mine decribed your circuit as follows ...

It falls into the category "triplet with current bypass" and is seldom seen in audio equipment. It is a circuit gimmick used quite often to extend the current sourcing capability of voltage regulators.

Basically, this is an intrinsic output stage (U3, U1, U4), which, as getting close to its designed current limit, cuts in U2 to take over the needed excess current. The first group enters a constant current mode, exciting the by-pass element U2.

This is practical when using power transistors with low hFE values, which need a rather large base driving current.
I wonder, how this performs at the upper end of the audio spectrum, when base charges cannot be depleted as fast as desired. Even though, R1 would be of a rather small ohmic value...
One problem can be the varying transonductance of the output stage, depending on the output current demand.

Thanks for sharing it with us ... interesting.

sajti · 2008-12-18 7:49 am

Hi Bogdan,

I used this bias setting network, but with the R34, and R35 connected to the emitter of the predrivers instead of their base. This solution cancel the temperature drift caused by Q15, and Q16. (Of course You have to redesign the values of the resistors)

Sajti

wojtek5001 · 2008-12-18 9:08 am

I have just fixed problem with double emiter follower - driving by 40k resistance. But that kills slew rate (about 3us rail to rail, versust 0.5us with triple follower driven by 100k).
When I look to schematics some 80s commercial amplifiers (yamaha, jvc, marantz, HK), i often
see resistors or capacitors (mostly both in pararell) connected from VAS output to ground. I was thought that is some way to compensation, but some of them have large miller cap, so they shuld be stable anyway. Also that decreases OLG, so increases didstortion (so important in these years). Now I think that I'am know.

Thanks Your problem I've noticed something that may be not noticed otherway. I saw some overshot in bode plot, but never suspected emiter follower. Thanks.

bogdan_borko · 2008-12-18 9:25 am

sajti said:
Hi Bogdan,

I used this bias setting network, but with the R34, and R35 connected to the emitter of the predrivers instead of their base. This solution cancel the temperature drift caused by Q15, and Q16. (Of course You have to redesign the values of the resistors)

Sajti

Thanks I`ll try that.

Nico Ras · 2008-12-18 7:16 pm

bogdan_borko said:
It`s thermaly stable there`s error in sim becouse of constant current sources...
This will do the trick the attachment)

A word about oscilation issue?

Yes you have solved the problem of imbalance.

sajti · 2008-12-19 8:47 am

Nico Ras said:
Yes you have solved the problem of imbalance.

There is another solution to get better DC balance. Use 2 resistors from the base of the predrivers to the ground. The resistors keeps the output close to the 0, in case of temperature increasing. The advantage is, that resistors can be higher value.

Sajti

sajti · 2008-12-19 8:51 am

This is the output stage of my tube+BJT amplifier.