Discrete Opamp Open Design

[scott wurcer]

Originally Posted by john curl

Wangine in Taiwan did a jfet input version of the 990 more than 20 years ago for a Parasound preamp Called the P-Fet 900, I think.

It might be interesting to offer an update since the devices specified are listed as obsolete. I found it funny that Pete also discovered Dick Burwen's work, 46 years old. I think basicly the discrete op-amp ground was well covered 40 yr. ago.

[bcarso]

Originally Posted by grhughes

In one of Walt Jung's articles he expounds the virtues of virtually noisless mosfet constant current sources. Ray

They aren't noiseless, but they are not too bad. However the manufacturer of those depletion-mode devices offers no data, no guarantees, and they are sole-sourced. Not a bother for diy but the devil for production.

[bcarso]

Originally Posted by scott wurcer

Also as bcarso points out there might be an opportunity for power on latch-up so there might be a simple clamp needed somewhere. I don't think this is a big problem.

That apparent problem was I suspect just an artifact of initialization with an ideal buffer, one with unlimited output voltage swing. However, some worst-case rail sequencing and heavy overloading should probably be done before we go into production

[RNMarsh]

Often (not always) power on latch up can be handled by assuring both + and - supplies come up together. Often this complicates a dual reg supply design. This can be done many ways but the simplest way (thus my favorite) is to use a large value cap across the + and - rails.
As a byproduct, it cancels differential mode noise and aids in keeping the two rail voltages equal.

[bcarso]

Originally Posted by RNMarsh

Often (not always) power on latch up can be handled by assuring both + and - supplies come up together. Often this complicates a dual reg supply design. This can be done many ways but the simplest way (thus my favorite) is to use a large value cap across the + and - rails.
As a byproduct, it cancels differential mode noise and aids in keeping the two rail voltages equal.

For traditional regulated dual-rail supplies I usually have one reference voltage with lots of filtering determine the one rail, and a equal R voltage divider from that output to the other output be the complementary rail's output feedback, with common as the other error amp voltage. By the time the first reference comes up, both of the unregulated inputs have stabilized, so the overall effect is orderly symmetrical ramps on both outputs.

[RNMarsh]

I just love it when Scott puts up a circuit! This refinement on the first is a Very nice blend of volt and current feedback concepts. Still flexible re adjusting open/closed loop gain for DIY'ers to play with to learn about that issue. I was happy with the first design put up. Now this! Its getting dialed in... wont be long now (?) and we can build something. :-) Thx -RNM

[scott wurcer]

OK here is the Rush cascode version 200V/us into 600 Ohms not to shabby.

[Kindhornman]

Scott,
Being a complete newbie here and trying to follow what I can here I'm am lost as to the feedback circuit. Where does the r6 and r12 get in the circuit?

[john curl]

What seems to be lost here, is: WHAT IS IMPORTANT? Of course, many discrete op amp design ideas were developed more that 40 years ago, but not all! The OPTIMUM designs are usually elegant, as well as relatively easy to source parts, etc. Unfortunately, then the OPTIMUM design changes with source availability, as well as added circuit tricks.
What is important is how linear and/or how quiet the circuit is, how fast, etc.

[Chris Hornbeck]

A *real* opamp needs to start with a Telefunken ECC83 marked "GAP/R". The rest just follows! (Or a 6SL7 for traditionalists).

Thanks,
Chris