Ovation e-Amp is Completed

[Nico Ras]

Originally Posted by Bonsai

Interesting Trannies Nico. Pd is 300mW so on 65V rails with 0.5 LTP current I am already dissipating 300mW, and on high output swing this will peak at over 600mW. (Another reason BTW for the cascodes was to limit the Pd in the diff amps)

Good explanation, thanks Andrew.

[wahab]

Originally Posted by Bonsai

The networks between the bases of Q19 and Q28 and their associated rails in the diagrams above are not Miller compensation networks. They are damper networks,

In fact , i was talking in the case where the drivers have
a resistor in serial with the collector , along with the said cap.

I guess this is to mitigate any inductance forming with their collector s
related tracks , provided the resistors are close enough of the drivers.

One of the byproduct is that the collector will have a slight inverted
signal compared to the base , hence a better efficency of the caps
whose value can be reduced.

[Bonsai]

Wahab,
yes, these resistors are to damp the lead inductances.

There is a small drop across the resistor, but I think for the most part its something you have to accept as part of making sure the amp is absolutely stable.

[Nico Ras]

Andrew,

Kindly try something in simulation and also in practice with your amp by placing a capacitor (experiment with 1 - 100 uF) between the emitters of each differential input pair, before the degeneration resistors.

I found that it creates an AC summing point and suppresses harmonic power density quite substantially.

[wahab]

This simply remove degeneration for frequencies at wich the cap
has low enough impedance in respect of the degeneration resistors.

[homemodder]

Wahab quite honestly Ive never seen a comercial design that had to use such "drastic" compensation methods to gain stability with triples. Neither did I have to in any of my designs using triples which is my preferred choice. At max a little miller cap on drivers do the trick. Instead of the caps Ive also made use of small inductor in parralel with the base resistors on ouput transistors for stability.

[homemodder]

Here is another interesting outputstage. Any comments on the connection of those cascodes ?? In spice its hardly stable but the schematic shown is a real commercial offering.

[Bonsai]

I prefer to tak about compensation and parasitic prevention as two separate things (which of course they are).

On the Ovation 250, I use a straight RC 'L pad' between the pre-driver emitter and the driver base with the R=27 Ohms and the C (1nF) connected straight to ground. The C can go to ground, or alternatively go to the rails if one assumes the supply is a short to AC. Of course, if you think about the parasitic inductances around the output stage then this is probably a dangerous assumption . . .

The Ovation 250 uses slow 21193/4 devices, BF471/BF472 pre-drivers and 15032/33 for the drivers. Without the 1nF cap to ground, the output stage was unstable and I had HF oscillation. The L pad cure BTW came after some research on the internet, where I picked up a good ap note from Microcap discussing emitter follower stability and how to cure it.

The e-Amp is another kettle of fish, because the output devices are an order of magnitide faster. Couple that to the fact that there are appreciable PCB inductances (similar to the Ovation 250) to contend with (from 2nF to 10nH depending on which output device you look at) and there is every opportunity for problems. I just did not want to take the risk. The amp will work with the removal of the the base stoppers for example, and you can lower the 1nF caps to 220pF and its stable. But, if you leave the caps out you will have problems, and if the caps and the stoppers are removed it can be provoked into HF ringing very easily. The output inductor and the Zobel also keep the operating environment of the output stage consistent at HF in my view in addition to the normal concerns about global stability when driving capacitive loads.

BTW, when I worked on the Ovation 250, one of the things I picked up on was that he speed up cap used across the driver emitter resistor was not good for HF stability and so removed it (I left the speed up cap on the pre-driver stage, but in retrospect do not think it adds any value). On the e-Amp I do not use a speed up cap, but instead rely on the fact that the driver stage current is quite a high, and the driver emitter resistor value low, providing fast turn off. In an exchange on this forum a year or two ago, Bob Cordell also remarked on this point and I was quite pleased to see he had arrived at a similar conclusion.

[Bonsai]

Ferrites in the base leads of the drivers work as well - I tried it on the Ovation 250!

[Dave Zan]

Originally Posted by Bonsai

On the e-Amp I do not use a speed up cap, but instead rely on the fact that the driver stage current is quite a high, and the driver emitter resistor value low, providing fast turn off. In an exchange on this forum a year or two ago, Bob Cordell also remarked on this point and I was quite pleased to see he had arrived at a similar conclusion.

Bob also recommended diode bias with ThermalTrak diodes as the driver emitter load. It looks nice to me, the dynamic resistance can be very low.
Have you looked at this?

Best wishes
David

BTW For a while I have meant to say thanks for the Middlebrook link on your website. Very "low entropy" information.