John Curl's Blowtorch preamplifier part II

[john curl]

Good topology, Joachim. First used by Harris Associates in some of their op amps. Your version is a little over the top, however. I generally use less parts, or parallel them.

[Joachim Gerhard]

Yes, this topology is not new. I only tryed to max it out. In simulation we also tryed current mirrors on the input and bootstrapped current mirrors in the VAS but settled on this one after much head scratching.
I usually design parallel symmetric but this one was designed to replace conventional Opamps so i neaded a diffential topology. Yes, i know, parallel symmetric, diffential can be done too but then the amount of active parts is even doubled and this one is already 9 x 6 cm in size.

[john curl]

Yes, I too, have needed an OP-AMP and used a single diff pair to make it. I have used this design for 3 master tape recorders, and two 8 channel boards. One helped record a number of direct disc masters for Crystal Clear Records. We always need 'building blocks'. Of course, at each juncture, I have tried to inject a quality IC, in order to simplify design even further. First in 1970, with the HA911, later in 1978 with the TDA1034, and just about everything that comes along, in general. So far, all discrete, especially with jfets is better.

[1audio]

The design looks familiar for the most part. It doesn't look as expensive as many of the circuits we see running around here (or as the circuit I used for the NuForce preamp which is not an expensive preamp). The folded cascode should make it easy to stabilize.
1) The cascode bias will limit the common mode range and the GD voltage (and leakage) will be affected by the common mode voltage adding a nonlinearity.
2) A trimpot across the 33 Ohm source resistors will allow offset trimming. If the input pair are coupled and pretty close the offset drift with voltage and time will be pretty small. I don't know if its a better solution than adjusting the drain currents but it has worked well for me.
3) I know John would rather see FETs for the current sources. He can explain why if he choses
4) You have a cascode on Q7 but not Q8. Why?
5) Are blue LEDs quiet? I had a problem in the past where red led's were photosensitive and allowed noise and hum into the circuit. I have avoided them since. Are these newer devices not photosensitive any more?

[PMA]

I would like to ask John Curl about the Parasound preamp flagship, Halo JC2. What is the reason of relatively poor CCIF IMD result (3rd order distortion component), as seen in

http://www.stereophile.com/images/ar...308Parfig6.jpg

(it is only 3Vpeak to 200kohm, i.e. very light load and quite low output amplitude)

and early distortion rise, see

http://www.stereophile.com/images/ar...308Parfig2.jpg

Is the issue in a poor dynamic linearity of the simple MOSFET output stage (extremely non-linear capacitances)? I would expect more from full balanced design with very high supply rail voltages, and high idle current.

Thanks and best regards,

[john curl]

The Parasound JC-2 preamp is designed to be a BUILDABLE preamp that would perform fairly well, be made 6000 mi away from the designer, and have a lot of 'features'. It is NOT a BLOWTORCH, and it never will be.
I would take care, if I were you, PMA, your input borders on professional rudeness.

[john curl]

I would also like to point out that PMA's measurement point would be 800W into 4 ohms, if it were used with a pair of JC-1 power amps, as it was designed to do. NOW, what about that 3'rd harmonic distortion?

[Joachim Gerhard]

Yes i have seen designs here that whould be nearly imposible to make commercially. The level of skill that some people show here in their DIY projects is amasing.
I am aware of the common mode range. We decided against a trimpot. One posibilty whould be to use a pot first and when the values are set, replace with fixed resistors.
The input transistors are matched doubles in one case anyway.
The constant current sources have been discussed at lenghth and my original design had red or green Leds with a lowpass filter to shunt the noise. The engineer that helped me with simulation was sure though that the blue Leds are quiet. Yes, a red Led has lower dynamic impedance ( around 5 Ohm i asume ) but only drops maybe 1.5V so you need a string for more voltage and then you end up with double or triple of that. I have made some good experiences with IR Leds and have some that are extremely "steep" so the voltage changes very little with current. They drop only 1.25V though. I have not experienced photosensitivity so far but i may have not looked carefull enough. There are active devices on the market for voltage reference with even lower dynamic impedance of less then 1 Ohm. I know that, but Leds are based on a quantum effect and they shine so nicely also giving me information if the circuit works. Maybe i am a bit romantic. Concerning the topology ( point 4 in your post ) i can not excactly remember how we arived at that. We tryed out more cascoding, mirroring, bootstrapping etc. but this circuit worked to our satisfaction. When i talk about the cost of making it, it is not only the amount of parts but the quality of parts that can be very expensive in the case of Vishey Z-Foils and the like. With standart parts the highest cost comes from the input tansistors. I buy them for around 10,-€ each here including shipping and tax. Maybe in quantity they are much cheeper or you have a better deal. And last not least, the layout is very generous ( 6 x 9cm ) so for example 6 of them to make a preamp with phono takes a lot of real estate and most of the cost is in the cabinet noverdays i am afraid.

[Joachim Gerhard]

P.S. Voltage Noise performance of this discrete Opamp is good but not outstanding. I whould not make an MC phonoamp with it. Parallel symmetric has a 6dB noise advantage there.

[john curl]

While I do not believe that PMA had this in mind, I would like to talk a bit about 3'rd harmonic distortion.
Many here fear 3rd harmonic distortion as if it were easily audible. It is not, and I can prove it.
I know this from decades of designing master tape recorders, from Ampex, to Mobile Fidelity, to Wilson Audio.
Do you know that originally, back in the 1950's with Scotch 111 mastering tape, 0 Vu was 1% 3'rd harmonic?
When I worked at Ampex, in 1968, the 3'rd harmonic distortion level at 0 Vu had dropped to about 0.7%. In the 1970's the 3'rd harmonic distortion at 0 Vu dropped, BUT many recording engineers decided to change the sensitivity of the Vu meter in order to get lower noise, and brought the 3'rd harmonic distortion back up to early levels!
So anyone who has heard a master tape, cassette tape, vinyl record or CD or even SACD, based on an analog master tape, can be promised at least 1% 3rd harmonic distortion virtually all the time when playing a selection. Ever notice?
Now, here is low level 3'rd harmonic distortion at 800W per channel into the loudspeaker system. Do you think this will be audible? Lots of luck!