New Amplifier Project

[slowhands]

Originally posted by sam9
I think the MPS8099s will limit you to 38V in the IS, at least if you want to use devices that are safe even if exposed to the total rail-to-rail potential. MPSA42/92 would take care of that problem. In fact you could then simplify things by having a single set of rails.

Since we don't know what the rail voltages are and we don't know exactly what current the input pair runs at, we must do a little analysis. I assume +-50v or so for the rails (100 watt output), and guess about 2ma for the pair (giving about a match for the 4 diode drop on the VAS emitter, assuming that should be equal to the voltage across the 3K0 load resistors).

For the input pair, you want the highest beta lowest noise parts you can use, which is not usually consistent with high breakdown voltage. What is really needed here? The input pair sees only half the rail to rail voltage (unless something is broken). The bases are tied to ground through low resistances, and with the currents I assumed they will be very close 0v.

For best performance I would try a 60v dual transistor. You can certainly leave the MPS8099s since their 80v rating is adequate and they have decent characteristics otherwise. The next step would be the 150v devices like the 2n5401/2N5551, with slightly worse parameters. The MPSA42/92 pair has low beta and I would not use them unless the rail voltages were much higher.

[fab]

Originally posted by The Peasant
Hello everyone,

I feel that a separate regulated power supply for the driver circuitry is very important and results in a more stable and better sounding amplifier. This may be why this type amplifier works well with the difficult impedances and current demands of the electrostatic speakers. How do others feel about this?
Doug

I would absolutely go for the regulated power supply for the driver. A simple RC filter will not be as efficient. I had an amplifier having already the RC (10 ohms and 100uF) along with a series diode but when I have added before it a series regulator, the improvement was quite noticeable. See my thread discussing this subject "DH-200/200 mods" for more details.

[sam9]

"The MPSA42/92 pair has low beta and I would not use them unless the rail voltages were much higher."

Good point. However the beta isn't quite as low as the minimum figures on the data sheet, I just check a batch I had and they were all about 70 which may or may not be sufficient. What I likew about them is their consistancy. At least from my supplier I can get usually get an exact beta match with in about the first four pieces. I just check my latest batch of ten, the range was approximately 68-73. My experience with 2n5551/2n5401 is that the range is much greater in a typical batch. Also the MPS's seem to be more consistent between batches.

Nonetheless, I use a lot of 2n5551/2n5401 as well.

[fab]

Originally posted by sam9
" My experience with 2n5551/2n5401 is that the range is much greater in a typical batch. Also the MPS's seem to be more consistent between batches.

Nonetheless, I use a lot of 2n5551/2n5401 as well.

So far the 2N5551 and 2N5401 I have found had almost all beta between 100 and 140. Of course, you may have to buy a few to get matched beta. I allow up to 20% difference between them for input pairs and I am still able to get less than 10 mv DC offset at amplifier output without any resistor adjustment or DC servo (in 3 amps).

[thanh]

value of input capacitor should be 470uF!

[lucpes]

Originally posted by thanh
value of input capacitor should be 470uF!

And why is that??? f0 (the input cap here acts as a RC hipass filter) is low enough with 8uf...

[thanh]

When output level is about 17volt,input capacitor can make distortion

[sam9]

"value of input capacitor should be 470uF!"

470uF will either be an electolytic or a film cap as big as the rest of the PCB. So I assume it is an electro. Now, I was aware that electros don't have infinite bandwidth but I didn't realize how significant that might be until the other day I re-read (browsed) an old JL Hood book. In it he presented the series resistance of a 470uF electro as an example. The series resistance starts rising sharply around 10kHz - thus using a 470uF will not be a high-pass but a band-pass filter!!

The highs dies!

If you insist on such a value you really need to parallel it with a small value cap. Oddly enough I've noticed that the electro models in LTspice don't account for this -- another case where Spice can lead you astray.

[fab]

"value of input capacitor should be 470uF!"

Other reasons to go with a big value here (assuming that it is in parallel with other lower value of better capacitor types):

1- several pre-amps have capacitors (unknown value) at their output (in series with their amp output). This capacitors appears in series in some respect with the input cap of the power amp. Series means lower effective value at input power amp, thus higher low frequency cut-off. With a value like 100uf or more, the power amp becomes less dependent of the pre-amp it is connected to (if pre-amp has really an output cap).

2- Having a very low low frequency cut-off at input of power amp prevents high phase shift of the signal at 20 Hz.

[sam9]

As I wrote in another thread I once discovered that a 70's (80's?) era reciever claimed an "intgrated rumble filter" (or maybe the adjective was "automatic") that was nothing but a 2uF cap on the input!!

Anyway one advantage of DIY is that you know and can choose what's in your equipment. I'm pretty suspicious about the practice of popping CD players or preamps open and willy-nilly swapping caps, but a bigger & better output cap is one thing that makes sense. The only thing is you need to locate it.