Sound Quality Vs. Measurements

[dvv]

Nige, it could be so much more than that.

Suppose you arranged two of them in a bridged configuration. I reckon each could effectively deliver up to 18V peak, which is some 40 Watts into 4 ohms. Theoretically, in a bridged configuration, you should have twice the voltage and twice the current, or 4 times the power - this is rarely quite so, but say 3 times the power should be real enough. That's still say 120 W into 4 Ohms.

At that voltage, the BD 249/250 C can deliver way over 10 Amps no problemo, using low voltage high capacity capacitors, say 22,000 uF/40V.

You might consider running it off higher voltages, say +/- 25V, and use zener diodes to regulate to say +/- 22V for the op amp, but compensating for voltage drops across the output stage.

And so forth - the point being that many opportunities arise, some of them being very interesting.

[nigel pearson]

Lets say 1 R also and make sure it can drive it . Still will be a simple amp . On my old Ferrograph distortion analyzer it might say zero distortion ! That is if I can work out how to use it ?

[dvv]

Originally Posted by nigel pearson

Lets say 1 R also and make sure it can drive it . Still will be a simple amp . On my old Ferrograph distortion analyzer it might say zero distortion ! That is if I can work out how to use it ?

How to use what, the Ferrograph analyzer, or the bridged amp?

As for 1 Ohm, given proper design, that should not be too much of a problem, although beside our old buddy Wayne, I am not aware of anyone else being a Member of the Sub 1 Ohm League. Normally, bridged amps have problems with very low impedance loads, I am told, but in this case, given the low voltage rails and the prodigious currents the BD pair can handle with safety, this may well not apply to you.

[nigel pearson]

Good question . Ferrograph mostly .
1 R at 15 Vrms , nice .

[a.wayne]

What about feedback, when doing low -z , does the feedback amt change, is more necessary ...?

[nigel pearson]

I always say I am too lazy to do simulations . Usually I rush out to my shed and solder up a few things . With this one I would love Dvv to see what he comes up with . Even 7 Vrms into 1 R would be interesting . I think I would follow the advice of Douglas Self and use 0R1 emitter resistors . I would start with 2N3055/2955 and BD139/140 . From my tests the complimentary feedback pair is the only option . Ideally 3 A of standing current would be used . That would be about 200 W of heat to get rid of . In bridged mode isn't worth talking about .

Casting aside that option a conversational class AB might be interesting . I would base it on the Quad 303 . I would use the 2N3055 . The standing current is 10 mA , we can be generous and use 20 sets in bridge . 100 mA and will only be 2W of waste heat . I will now say something which is seldom said . Regardless of transistor type or any other consideration this circuit has seldom been beaten for how easily it works . It would laugh at 1 R . I suspect even better to use 2N3055/2955 are driver ( 303 is a triple ) . Gogny made an amp like this in 1967 with just 2 x 2N3055 diving the load , it was 50 W into 0.8R ( 0R8 ) . +/-18 V rails and 10A fuses as emitter resistors . This would be a loop feedabck amp . It's crossover distortion is very low and predictable . I would expect problems if " better " transistor were used .

Off to the garage now and I am late , forgive any typo's .

[dvv]

Of course, others among us might wonder why does Nige always insist on the one output stage known to be difficult to get right without oscillating, when he could have used a nice triple Darlington emiiter follower configuration, well known for proving the best possible control of the load an amp will do even with reasonably low global NFB figures, and, MOST important, offers the lowest output impedance. This alone would suggest its use in a balanced cofiguration as the most recommended, given that whatever you have will halve in a balaced setup.

Nige, it's better to think it all through at least one scenario before you're off to solder somethig.

[nigel pearson]

Cfbp like most old wives tales is exactly that . Op-amps oscillating when used with boosters is common regardless of output stage . OPA 604 seems very obliging . The 303 circuit is very stable . Another old wives tale was the 303 has many tweaks to get it stable , I never saw where they were . It has output tipples which I feel have never been creditor with how they completely they eliminate most of the problems . Not least the use of all NPN outputs which makes for rugged designs . BTW , The bottom half of Quasi comp is Cfbp . People dream up all sorts of reason not to use this proven arrangement , old fashioned is not valid I feel .

[dvv]

Blimey Nige, whyever would you want to use a quasi-complementary output stage in this day and age?

40 years ago it may have been necessary, as there was a shortage of reliable high power transistors, but today?

[nigel pearson]

Because it is better ? This debate raged in Wireless World for years . Reluctantly it was agreed that an NPN with 100 % local negative feedback becomes a PNP to all intents and purposes . The addition of a cheap diode solves most of the asymmetry . Even to this day the way a PNP works is more difficult in terms of transition speed and durability . The fact the PNP's look so good is testament to the hard work done. Some say all NPN sound more alive and punchy . In the 303 and Crimson amps they show how good it can be . The Crimson was especially good .