Scott: Please forgive my non familiarity with electronics and audio design but if TO126 or TO220 (mje171/181 or mje243/253 output devices are used then how much heat sinking is required? How close to Class A can it be biased? Can this signal dependent bias be pushed into class A? What do you think the quiescent or collector current in the output devices would amount to? Thanks! Ray
Most device datasheets will have a ambient temp and a case temp spec on power dissipation. Or, at least a thermal resistance spec for chip to case and chip to ambient. With the maximum operating chip temp also given, one can then calculate the permissible dissipation.
The circuits published (for example 1568) only have a weak signal-dependent bias, as Scott commented. The bias for the output stage current sources is just convenient, not essential to the circuit operation.
I left the bias values at the ones in the JE990, there are several descriptions floating around describing the thermal issues for the module. If you want very low impedance drive at higher power I would put the bias diodes and output devices on a piece of metal either case or heat-sink.
I just noticed that I left off the output bias trim and instead matched the Is of the diodes to the outputs.
I run a small recording studio and do location recording of bands. I need 16 stereo headphone amplifiers (32 channels) for performer reinforcement. NO BATTERIES! They have to drive mostly Sennheiser HD380Pro ear buds (54 ohms). If the output stage can't handle that I'm not interested. Thanks, Ray
Headphone driver no problem I assume you are not putting 10V rms on earbuds. I suspect this circuit is also fine in Dick's 1V rms on 30 requirement. With real musical crest factors the average power dissipated while well blasting your ears is fairly small.
Looks good, all SMT right? How small can can you get it?
EDIT - Those complimentary pairs are a great idea, and I love those prices. How low do you think the BOM could get?
Last edited:
Careful. BF545 has a minimum pinch voltage of -0.4V. If by chance you run into the worst case combination of BF862 and BF545, the BF862 may not saturate (but stay in the linear region).
Better find a n jfet with higher pinch voltage, like this. Only 25V Vds but dual, and -2V minimum pinch voltage.
Good advice, the data sheet is a little unclear but from my read the BF545C is fine. That last line of Vp at 1uA looks like it covers all the grades, not sure why it's there.
Hmmm ... in my sim I have very good agreement on the +ve PSRR figures, but the -ve PSRR is dramatically superior at lower frequencies for stinius. I do note that the -ve PSRR is highly variable, depending strongly on the choice of BJTs, so I'll try replicating stinius's selections ...
Frank
Frank
AC or DC? You can scope out the DC component by doing a DC sweep and poking around.
AC. I also note that PSRR is highly dependent on the output DC operating point; if I trim the circuit to minimise DC offset it actually worsens PSRR behaviour - I'm not sure whether this is a sim artifact so far, and I don't have anywhere near enough knowledge about amplifier tolopogies to understand behaviours in an intuitive way ...
Frank
Frank
This thread has highlighted some very interesting small JFET dual parts. Thanks Scott, Brad, and everybody else who has made parts suggestions along the way.
Dave
Mystery solved - he's added two more cascodes, Q7 and Q8. Using J111s there I'm now able to get 125dB -ve PSRR with one BJT combo, still 100dB at 20kHz ...
Have to say the sim distortion figures are truly excellent, even at gains of 100 ...
Frank
Just noticed this morning another AM radio FET but dual. Someone should evaluate the matching and audio frequency noise. This could be another gem like the BF862. But the user will be punished by an even smaller package.
If you refer to the PMBFJ620, then it has much higher noise and much lower Yfs than the BF862.
Idss is also way too high.
The package is a small problem. We have soldered even 0.5mm pitch 6-pin.
It is humanly possible, even by hand.
Or you are talking about another device ?
Patrick
Yes, then it should stay in cascode duty. I do like the contrast between the parts we are discussing and the usual almost obsolete ones. I think one of the npn/pnp duals had 5 million and 3 million "ready to ship" at the first two distributors.