Discrete Opamp Open Design

[grhughes]

We need a list of transistors and diodes that Scott modeled with other than the j111. Did he use the latest Hitachi/Toshiba transistors and what are the output devices. The schematic is no longer mirror symetrical. Granted I'm not an EE and i trust Scott's design.I was told by another Phd EE that the 2n4403/2n4401 transistors have low base spreading resistance, the 03 particularly. Are there more modern transistor/s that are better. I will wait before starting to layout my board. It's too up in the air. Ray

[bcarso]

Originally Posted by grhughes

We need a list of transistors and diodes that Scott modeled with other than the j111. Did he use the latest Hitachi/Toshiba transistors and what are the output devices. The schematic is no longer mirror symetrical. Granted I'm not an EE and i trust Scott's design.I was told by another Phd EE that the 2n4403/2n4401 transistors have low base spreading resistance, the 03 particularly. Are there more modern transistor/s that are better. I will wait before starting to layout my board. It's too up in the air. Ray

At their locations in SW's circuit, low rbb' is not that critical.

Some very good parts that have high betas and breakdown voltages: 2SA970, 2SC2240, BL beta codes preferred. Although Toshiba does not mention rbb' explicitly for them, I backed out an estimate of 30 ohms for the 970, which is as good or better than the 4403. The low rbb' of the 4403, a fortuitous consequence of Motorola's trying to make a better switch, was a trade secret in places for a while. But that was long ago BTW not all 4403s are the same.

The output devices on the other hand I'm not as sure about. The input JFETs I believe are BF862 if you have the patience to deal with SMD and to match them.

[scott wurcer]

Originally Posted by EUVL

Should I use this for PCB layout ?


Patrick

I'll need one more post to show where some small R's and C's can go to help with lower closed loop gain applications. The input clamp diodes (back-to-back on the input like on the JE990 and many IC's) are also needed.

I have been using the Toshiba devices for the latest sims but the others will work. The bipolars figure very little into the noise budget, just trying to get the higher rails.

I hope someone will tinker with a breadboard, I don't trust the sims at the finest scale especially getting it unity gain stable.

[fitzfish]

It may be worth while to look at output protection as well. With a shorted output the sim suggests well more current than needed to slag any small signal output transistors.

Dave

[scott wurcer]

Originally Posted by fitzfish

It may be worth while to look at output protection as well. With a shorted output the sim suggests well more current than needed to slag any small signal output transistors.

Dave

Live dangerously for the best sound.

[kgrlee]

Originally Posted by scott wurcer

Latest schematic, I added input cascodes and tied the bias setting current surces together at the drains so only one cascode would be needed (for operation at high supply voltages).

As the whole left hand corner is now just 'constant current source' stuff, can't we get rid of J1, Q11, 9 and tie the bases of Q15, 35 to emitter & collector of Q7 respectively?

[fitzfish]

Originally Posted by scott wurcer

Live dangerously for the best sound.

I keep hearing that. Something about fearless... I tend to side with Stapp.

[scott wurcer]

If you are using a discrete op-amp it's crazy not to take advantage of the ability to custom compensate it for each application. After an afternoon's work it looks like the compensation capacitance needs to be circa 250pF at unity gain and can be as low as 75pF for 20dB of gain. Now even the first circuit has 50V/usec symmetrical slewrate at 20dB gain. I suggest here components that can be adjusted to optimize frequency response and assure stability. I realize that the details of optimizing this circuit are beyond the abilities of some of the folks here so let's help each other and include placeholders for the optional components in any layout and those that can develop a parts list for a specific application can inform the rest of us.

I have added a zero in series with the compensation capacitance as well as a pole zero network in both sides of the gain stage. These values depend on closed loop gain and matter less at G > 10.

EUVL - I consider that there is enough here to burn a prototype.

[scott wurcer]

Originally Posted by kgrlee

As the whole left hand corner is now just 'constant current source' stuff, can't we get rid of J1, Q11, 9 and tie the bases of Q15, 35 to emitter & collector of Q7 respectively?

Yes that could be, as long as the dynamic base current is captured it all works, time to Muntz away a little.

EDIT - It does not work out that easily, best left alone for now.

[fas42]

To keep the ball rolling, here's a quickie knock up of the schematic from #1568 for those who are averse to fiddling with the versions already posted. Appears to be OK, haven't checked everything, I'm sure people will point to any problems ...

The PSRR looks much, much better of course. One thing though, can't push the output standing current up to 8mA easily for some reason ...

Frank