My Amp Design

[dshortt9]

I would like comments on this design.
The driver and output transistors are sustained beta types which I think are better than the MJE 340, 350 - 2955, 3055. Outputs are Sanken devices roughly like the Toshiba A1302 C3281 but rated at 17A.
Will this thing even fly?


1. I am not an EE, just a hobbyist.
2. Don't know Spice, wish I did.

[hugobross]

here's the picture of the link above, works a little easier.
I think it's a weird design, but I'm too tired right now to put comments on it.
Maybe there will be a problem about the temperature compensation of the biasing voltage, but I'm not sure.

[peranders]

You have a potential problem here: high output impedance. Collector out gives you 1-10 kohm. You should have << 1 ohm output impedance in order to drive speaker without colouring the sound. I sucpect that your open-loop gain is too low. You can't feedback away the high impedance.

You must find a trick to reduce the output impedance (before you feedback the signal).

[Nelson Pass]

There is no trick available. Feedback creates the output
impedance.

[dshortt9]

Thanks for the input and help.

Perhaps I have not biased the stages high enough to give a good open loop gain. I wanted 20ma. at the input and 100 ma. for the output stage. This may be too low as I don't know what pspice uses for the threshold voltage of the fets. Maybe tweaking R3, R13 up from 20k would increase the bias of both stages and the OL gain? Say 100ma. for the input and .5 A at the output. R1, R2 may need to be tweaked to get the right bias. I also thought that the NTC of the input fets would aid the thermal stability if mounted on the heatsink.

[P.Lacombe]

dshortt9,

100 mA idle current for the output devices : your amp works in class AB. I think you will have problems with thermal stability, and therefore severe crossover distortion, depending on actual power output, voltage supply variations... Also, open loop gain seems to be very low.

Regards, P.Lacombe.

[dshortt9]

OK, After some more thought let's change R1, R2 to 6.6 ohms to bias the front end at 100ma and the output devices at 500ma. R3, R13 will also need to be tweaked to accomplish this. With these conditions I have an OL voltage gain of about 400 and a sensitivity of .1 V at 100W into an 8 ohm load. This only allows me about 20 db of feedback for a 1 V sensitivity, not enough to correct output impedance and distortion. Remembering what Mr. Pass said in his ZEN articles I don't think I can get more than a voltage gain of 10 regardless of what fets I would use in the front end so I'm stuck. Sure is hard to get something for nothing! Obviously I could add a driver stage but my mindset is to accomplish the goal with 2 stages. More mental excersize than reality at this point.

Thanks to all

[Circlotron]

Would bipolars be a better choice for the input devices so there would be a log/antilog = linear type thing happening from the first device input to the second device output?

GP.

[dshortt9]

Good Thought, There is more voltage gain ~ 1000 and the linearity is improved a lot. If we would then bias the outputs at 1 A both stages are very linear and we will have 40db of feedback for low distortion, but Im afraid the output impedance may still be higher than acceptable. Can anyone calculate or approximate this? Am I wrong in thinking that an output transistor running at 60 V and 1 A has about 60 ohms of resistance ( roughly = impedance ) and that 40 db would lower this to 0.6 ohm. I'm not technically proficient enough to know if this is close.

[peranders]

You must check once again the output stage (or the lack of it).

How many amps have purely collector out? Noone! Why? You can't get low output impedance at audio frequencies. If you wanted a bass amp you could have extremely high open-loop gain but that's no way to go for fullrange audio.

I suggest that you test your design with small signal transistrors and a aupply voltage of +-12-15 V, then you (maybe) will belive me ( I could be wrong but I don't think so). You can make some things better with higher bias currents but that's only aspirin for a brain tumour (right spelled?, I meen cancer).