Super Regulator, collecting the facts

[CheffDeGaar]

Originally posted by Fred Dieckmann
The use of a high Hfe pass transistor or even an additional follower between the op amp and the pass transistor is worth investigating ( big hint.......fet). The less output current the op amp has to deliver, the more linear the regulator is and lessit modulates its own supply voltage terminals

Sure I will say something stupid here, but what about a Mosfet as a pass transistor ? Sure the dropout will be higher, but the current delivered by the opamp will be very low.
But I surely miss important points here, such as fet's capacitances and so on... Tell me...

[peranders]

The original idea from Mr Jung I believe, was low noise AND low output impedance and a mosfet has higher output impedance than BJT. I'll guess that you also get a slower output stage so the benefit of a "electronic" capacitor will be less. An advantage though, it's easier to get more current out of the regulator.

[peranders]

A more original is from Kaneda 1977 and he used parts which were available at that time. 1977 the 709 was "hot" but he had also 741 to choose from.

If you compare Walt's design with this earlier you'll see small enhancements but important ones.

1 Better opamp

2 Better reference

3 Better output transistors

4 LP-filter for the reference

5 Current generator for the pass transistor driver

6 "Sense" connections

7 Pre-regulator

8 No current limitation (if this is better... lower output impedance anyway and that is good)

9 Better resistors

10 Better pcb(?)

[janneman]

Originally posted by peranders

Andy, you are almost right on target here. I get an oscillation from about 90 mA load at 6 MHz! A small LP-filter is necessary for the supply voltage of the AD797. I have tested 4.7 ohms + 100 nF. Without this filter I have a steady oscillation at no load except for the feedback and the reference.

With this filter AND without 100 nF at the output the regulator works very nicely.

The original series in AA by Jung/Didden/Galo said that the 797 is tricky to get stable in this application. Later follow-ups in letters etc emphasized this. I have always said that the 797 is NOT the best choice for a super regulator, and explained why. Yet, people still select it from the datasheet because it has low distortion . The 797 uses internal positive feedback to get the low THD. That spells "oscillations". But sure, go ahead.

Jan Didden

[peranders]

I used the AD797 because I felt for and with this little filter I get a stable regulator but then again it's a free world and you the use the opamp you want, even a 741! Jan, nothing stops you really from using any other opamp as long as it's good enough and with the right properties.

Andy, I do listen but this doesn't mean that I have to follow the advices.

[peranders]

Originally posted by janneman
The original series in AA by Jung/Didden/Galo said that the 797 is tricky to get stable in this application.

The SMD and the pcb may have influence on this, meaning more optimal environment if SMD is used.

BrianGT, how about your super regulator, did you succeed?
http://www.diyaudio.com/forums/showt...124#post246124

[peranders]

Originally posted by Fred Dieckmann
I think the Sulzer design is a much more solid foundation to build on for very good regulator design than Mr. Kaneda's implementation

Mr. Kaneda's circuit is exactly as Jung's excluding the improvements. I'll guess you are joking?

I think the main difference between those two is the output stage.

[janneman]

Originally posted by peranders

The SMD and the pcb may have influence on this, meaning more optimal environment if SMD is used.

BrianGT, how about your super regulator, did you succeed?
http://www.diyaudio.com/forums/showt...124#post246124

Indeed. That new Mercedes I say just a few minutes ago is basically *exact* the same concept as the T-ford. And per-anders, I get really tired from explaining that the 797 is internally really special, that that is the reason for its problematic behaviour in very high loop gain applications like this. I don't see that SMD or PCB changes the internal construction of this chip? You are certainly free to use *any* amp you want, but giving advice to people based on badly selected parts because you refuse to do your homework I think is not really helpful.

Jan Didden

[peranders]

Originally posted by janneman
You are certainly free to use *any* amp you want, but giving advice to people based on badly selected parts

ALW's pcb is highly regarded by Fred and others and from what I know he recommends AD797, so does aos. I may very likely test other opamps later but now I will concentrate on AD797. Personally I would have never chosen AD797, rather one "normal" general purpose oapmp, like OPA134, or AD8610 or somehting but I have decided to let AD797 have a go.

Jan, why don't you test my pcb and see if you dump into problems? It never hurts to lift up problems or potential problems but those potential problems may not be problems in real life.

I should also add that my pcb's have 50 um copper and everybody else have 35 um (1 oz), positive when it comes to high performance. I'm alone in this, am I?

[Fred Dieckmann]

The problematic nature of the AD 797 for this circuit has been discussed in print (Audio Electronics) and on the forum. Yes, it can be made to work but it is even more likely to cause problems, and problems of a nature that someone without a fast scope and experience in parasitic oscillations might find difficult to fix. Didn't you recommend a mod to your circuit to address just such a concern?

I am glad to hear about the increased PCB copper thickness. The correct routing and placement of the sense lines is designed to compensate for trace resistance in the load current traces. If this is done correctly, a change in resistance of the traces should be pretty well buried by the effect of the feedback. The impedance of the output transistor dominates the trace resistance by several orders of magnitude for the output (the one that is not ground) at any rate. Perhaps the thicker traces are beneficial to those soldering all those SMT parts by hand.