Buffer/Headphone Amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The buffer circuit described below can be found at http://www.digido.com/User/Assets/Active/PDF files/00478-buffer.pdf
My thanks to Bob Katz for putting it up.

A few spec’s and observations on the buffer/headphone amp circuit. Since we have a wide range of technical ability on this forum I am going to attempt to be as straightforward as I possibly can. The idea behind presenting this circuit is to give a “building block” for various applications.
The input impedance of the buffer alone (without U1, feedback, and U2) at the junction of the Cathode of D2 and Anode of D3 is 3 K ohms nominal.
The open loop output impedance of the buffer (without R3, feedback, U1 and U2) is 10 ohms nominal.
The open loop frequency response of the buffer alone (without the U1, feedback and U2) is DC to > 1Mhz.
The current through Q1 and Q4 is 11mA +/- 0.5 mA.
Peak output current is 250mA max.
Output DC offset: 20 to 40 micro volts typical

The general idea for this design was that I needed an amplifier to drive dynamic headphones as well as a line stage output. The other general idea was to use as many parts as possible that I had on hand. I have about 70 of the MPQ6600A1 transistor arrays (A 14 pin dip package containing two complimentary NPN and PNP transistors) left over from a previous job, so I decided to put them to good use. Please note that the output section, Q1, Q2, Q3, Q4, are all in one package. Having the output stage all thermally connected on one package is certainly not a bad idea, especially for bipolar transistors.. Q2 and Q3 are used as the current limit in conjunction with R1 and R2. This topology is a “semiconductor classic”, and is near bullet proof. At one point in test I changed R3 to 10 ohms and shorted the output to ground. I then put a sine wave at the input and drove it into hard clipping for 10 minutes. I observed no damage to the output transistors and no “hot” spots. There is one case where if the short circuit goes on long enough there could be damage. If I remove one of the supply voltages and short the output to ground, then the constant DC through the output could well damage it, if it were done for a long enough time.
The complimentarity of this buffer surprised me. At one point in test using the good old AD 711 at U1 and shorting the output of R3 to ground, I put a 10Volt peak signal across R3. I measured both the AC and DC voltages across R1 and R2 and found that they “mirrored” each other to within 1mV. I don’t think I’ll ever get to see that kind of performance in a power amp. I have also adjusted VR 1 for as close to 0 volts as I can get at pin 6 of U1. Expect some thermal drift here.
The other area of performance I was interested in was the output DC offset. I really don’t care for putting DC blocking caps in front of headphones (or any transducer for that matter) especially when the impedance of headphones can vary from 8 to 600 ohms. The performance of both U1 and U2 will have an effect on the DC output offset. U2 is a precision jfet input op amp and U1 is also a precision, low distortion jfet input op amp. At one point in testing I was using an AD711 at U1 and an OPA 227 at U2. The output DC offset went to 200 micro volts. Without any DC Servo at U2 the output DC offset went up to 600 micro volts. However; different applications may not require a very low DC offset, so I will be more than happy to let each of you be the judge of what is required for your particular application.
Capacitive loading appears not to be much of an issue. Square wave performance into a load of 10 k ohms in parallel with 0.01uf shows no overshoot or undershoot, and no hint of instability. Rise and fall sections of the square wave show no “glitches”. I have not taken any distortion or noise data as of yet. I am waiting for a friend of mine’s time and Audio Precision One to become available. I am, however, expecting to see noise and distortion similar to what is given in the Burr-Brown data sheets for the OPA 627.
From a strictly subjective point of view I have been field testing a prototype using a variation of this circuit (different gain, and switching to 10 ohms at R3 for headphones), and the field test is going extremely well. From where I sit this circuit is about as invisible as it gets.
For those who wish to experiment I have a few suggestions.
Changing the value of R4 and R6 will change the current through the output transistors Q1 and Q4. When these values are lowered more bias current will flow through Q1 and Q4. The drawback here is that the input impedance is also lowered. If the value of R4 and R6 are increased then less bias current will flow through Q1 and Q4 and the input impedance will increase. If the values of R1 and R2 are decreased than the possibility of thermal runaway may also increase. I would recommend not going below 5 ohms here, unless you thermally connect D1, D2, D3, & D4 to Q1 and Q4. I have not done any high gain amplifiers using this buffer. However I do suspect that the gain bandwidth will be primarily dictated by U1, and not the buffer circuit.
The MPQ6600A1 is now an obsolete part, although there are still a few thousand of these around. If you can get them, I would highly recommend it. Another alternative might be to use the BC550C and BC560C.
Good luck to all who build and use this, and remember the following:
When you miswire things and blow them up, try and remember that you now have joined the ranks of experienced professionals and have a good laugh.
d.b.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.