Very simple quasi complimentary MOSFET amplifier

Jump to Latest
#268
Guys, with NPN input pair, it's bases go slightly negative (very slightly, but still), so C3 is shown in good polarity, but C1 - I would also connect its negative side to the base (positive - to the input). Or - even better - use a non-polar one.
 
#275
For goodness sake...

If the circuit has a pair of NPNs for the input stage (upside-down LTP) then a polarised feedback shunt cap should be +ve to ground due to the (small) negative voltage across it. Or better yet use a bipolar cap as Hugh suggested.
 
#278
GEirin,

When you bias a long tailed pair amplifier, you support the very small current from ground. If the LTP is npn, then the current must pass INTO the base, typically around 5uA, and if you have an input bias resistor of 27k, this will pass 10uA and therefore the voltage dropped across this device is V = Rin x Ib = 27k x 0.01mA = 270mV. This voltage is impressed across R2, and since the upper point is ground (since the current base passes from ground, through R1 and to INPUT the base) it turns out that the voltage at the base of the input is -270mV, ie -0.27V.
A LTP operates by places identical voltage its both bases; this is the only way it can deliver differential operation by passing identical currents, in this case 0.4mA EACH, 0.8mA in total, the stage current. Since the voltage on the fb node, the T2 transistor, is identical, -0.27V (that is why the cap at the fb divider must have positive on the ground and negative on the fb shunt resistor (1K), and because the fb resistor from the output node (R7) is also 27k, if we start with 0mV on the ground, we finish up with 0mV on the output node. This assumes that the beta AND the Vbe voltage of T1 and T2 are the same - that is, they are MATCHED. We need 0V on the output node to deliver up and down voltage to the speaker; it is designed to operate correctly with just 0mV on the input. So this all works out very nicely, provided:

1. LTP transistors are matched (not essential but a great idea).
2. Identical currents through the transistors, T1 and T2.
3. Identical resistors R1 and the series feedback resistor R7 - both 27k here.

This describes all the wonderful qualities of a long tailed pair. They enable accurante offset control and marvellous fb point for the global feedback of the amp. They are reasonably linear within some limits.

There are many, many tricks in this design however. It looks simple but in fact the distortion is fairly high. But the distortions have been carefully profiled so it delivers sound quality consistent with the human ear. Think not of distortion but rather harmonics; the distortion of this amp might not suit Martians, or Fruit Bats, but for humans, this is a very good amplifier. Because it has relatively high distortion (around 0.05%) it might be criticised as poor performance. From an engineering specification POV, this is in fact a 'poor' amplifier.

However, when you build it, listen to it, and I will be very surprised if your friends will say this is the best sound they have ever heard.

I like explaining this stuff because it clarifies it well for ME and in fact I was originally trained as a science teacher. I might add that the electrical engineering in amp design is well understood and has been around for 60 years. Only the components have improved in modern times. The psychoacoustic issues, however, are still not very well understood. That is why audio designers are lauded because even today it is a black art, almost magic. Some really expensive, wonderfully measured amps sound like an ice pick in the ear. NP is one of the best magicians of all......... Happy Birthday, NP, I'm sure you see everthing in the forum!

Cheers,

Hugh
 
Last edited:
You must log in or register to reply here.
Top Bottom