I had a quick play around with a Boss-Style discrete opamp using BJTs instead of JFETs.
I'm not making a project out of it and I wouldn't say it's complete.
I'm just putting it up as a stick in the sand.
As some simple goals I tried to match:
- maximum gain ; (which is something that doesn't match with the wrong JFETs)
See current thread:
https://www.diystompboxes.com/smfforum/index.php?topic=129418.0
- the opamp bandwidth, for no reason that than it will shape the response under clipping
- the overload behaviour
The downside of the BJT version is the gain pot can get a little scratchy as the DC bias has minute
shifts when the control is varied.
In the waveforms you will see a slight vertical offset. That's largely due to the slightly different DC
bias voltage at the output.
So the main thing I did to keep things the same was to tune the gain using emitter resistors on
the BJT differential pair. (Not stuffing around with the other stuff helped maintain a better match.)
Schematic and notes
(https://i.postimg.cc/qN76Cgvp/Discrete-Opamp-Boss-JFET-vs-BJT-sch-V1-0-initial.png) (https://postimg.cc/qN76Cgvp)
Frequency response
(https://i.postimg.cc/5jM3Qqd1/Discrete-Opamp-Boss-JFET-vs-BJT-Response.png) (https://postimg.cc/5jM3Qqd1)
Overload with Sine wave input +/-100mV
(https://i.postimg.cc/D4y4dzPQ/Discrete-Opamp-Boss-JFET-vs-BJT-100m-V-pk-sine.png) (https://postimg.cc/D4y4dzPQ)
Overload with Sine wave input +/-1V
(https://i.postimg.cc/FYJdx836/Discrete-Opamp-Boss-JFET-vs-BJT-1-V-pk-sine.png) (https://postimg.cc/FYJdx836)
Pulse response, input +/-10mV (not overloading)
(https://i.postimg.cc/K4v1WRtB/Discrete-Opamp-Boss-JFET-vs-BJT-10m-V-pulse.png) (https://postimg.cc/K4v1WRtB)
Pulse response, input +/-100mV (overloading)
(https://i.postimg.cc/sQfgbNGs/Discrete-Opamp-Boss-JFET-vs-BJT-100m-V-pulse.png) (https://postimg.cc/sQfgbNGs)
Pulse response, input +/-1V (overloading)
(https://i.postimg.cc/tnYC4nzc/Discrete-Opamp-Boss-JFET-vs-BJT-1-V-pulse.png) (https://postimg.cc/tnYC4nzc)
It occurred to me that reducing R13 to 3.9k might be "more correct" since the sources sit above VR in the JFET case and the emitter sit below VR in the BJT case. It's not going to affect things too greatly. With 3k9 the emitter resistors might need tweaking.
I check it out and the 4k7 as shown looks better than 3k9. Better waveform match on the +/-100mV pulsed overload test. In fact, in this case the waveform is a little closer with 5k6 + 82R emitter resistors.
R13 could be a current source such as a transistor collector or drain in the high-impedance part of the operating point so you can get the effect of 100K resistance without cutting the current or raising the voltage. There are current source diodes made from JFET's with the gate connected to the source internally. A high impedance at this point helps balance the input stage.
QuoteR13 could be a current source such as a transistor collector or drain in the high-impedance part of the operating point so you can get the effect of 100K resistance without cutting the current or raising the voltage. There are current source diodes made from JFET's with the gate connected to the source internally. A high impedance at this point helps balance the input stage.
I get what you are saying. It's also possible just to play with the R13 to tune the current and set the gain.
I didn't want to change the current since the circuit has an inherent imbalance of currents in the diff pair and that ties in with the R12 value (I specifically didn't want to change R12 in order to preserve the overload waveform). That imbalance is in the Boss JFET circuit as well. I noticed the imbalance when I looked at the circuit a few years back and it seems the motivation behind it is to push the open-loop (and closed loop) gain up.
The addition of the 100 ohm resistors (R18 and R19) actually changes the way the diff-pair overloads, the overload voltage is higher. That might make it a little closer to the JFET stage.
Basically I played around with a few things until I got a recipe which crudely duplicated the behaviour of the JFET version. There aren't that many parts in the circuit to play with. The results of the simulations certainly look close for a first-order approximation.
Check out how the Seymour Duncan Pickup Booster discrete BJT op amp does the gain.
https://web.archive.org/web/20110715171431/http://analogguru.an.ohost.de/193/schematics/SeymourDuncan_PickupBooster.gif