Discrete Opamp Open Design

Has anyone assembled a working Wurcer opamp with real parts or are we still simulating?

Scott has built working SWOPAs, for sure - but I don't know about any others. I have built a 1st-gen FET990 variant with real parts and may build a 2nd-gen variant with a Diamond Cascode output stage later - it depends on the interest from the Gainwire/ODNF folks, who have some interest in using a discrete opamp for the error-correcting servo.

I'm still simulating a number of alternative topologies, but none comes close in THD20 to either the FET990 or SWOPA.
 
Thanks Linuxguru. So Scott has not changed any parts since his build (2 board sandwich) using 2n4401/03. I guess I will follow his lead. The JE990 people are still using the mje171/181 as output devices. Hardy and Jensen are still using them??? Is there a better pair of output BJTs. I just want the latest info before I lay out a board. They are still using pn/mmbt2484 as current sinks,why? Thanks, Ray
 
Thanks Linuxguru. So Scott has not changed any parts since his build (2 board sandwich) using 2n4401/03. I guess I will follow his lead. The JE990 people are still using the mje171/181 as output devices. Hardy and Jensen are still using them??? Is there a better pair of output BJTs. I just want the latest info before I lay out a board. They are still using pn/mmbt2484 as current sinks,why?

No idea, it's possible they want to stay as close as possible to the proven design and BoM.

Output BJTs - certainly there's a huge choice in through-hole parts. In SMD, the choices include BCX53/56, BCP53/56 and 2sc2873/2sa1213, among others. If you're going to use lower rails (say +/- 15v max) and lower output-stage current, some generic SOT23s may suffice - e.g. MMBT5401/5551.
 
Disabled Account
Joined 2012
SIM this --

What can you guys do with this --- SIM it and change output stage and then bend it into a compl. push-pull. I find the cancellation methods very interesting toplogies.

Thx-RNMarsh



output distortion cancellation ckt 1.png
 
What can you guys do with this --- SIM it and change output stage and then bend it into a compl. push-pull. I find the cancellation methods very interesting toplogies.

The output stage base-current equalization idea is not at all novel - that particular complementary push-pull buffer Vbe-biased by a dual-diode and a current source has been used for maybe 50+ years. There's no way that sub-circuit can be defended against patent infringement.

However, the odd-looking loads Q6 and Q4 on the LTP collectors look novel to me, and I'll take a look later in simulation to see if they provide any benefit in distortion cancellation.

My preliminary guess is that the benefit is insignificant, since the bases of Q6 and Q4 are low-impedance nodes with negligible voltage swing, hence emitter of Q4 also sees negligible voltage swing (it's also low impedance), so it cannot help much in linearizing the voltage-follower Q37. But intuition can be misleading, and Spice may throw up some insights - will take a look.
 
Scott has built working SWOPAs, for sure - but I don't know about any others. I have built a 1st-gen FET990 variant with real parts and may build a 2nd-gen variant with a Diamond Cascode output stage later - it depends on the interest from the Gainwire/ODNF folks, who have some interest in using a discrete opamp for the error-correcting servo.

I'm still simulating a number of alternative topologies, but none comes close in THD20 to either the FET990 or SWOPA.

I will try to have universal GaiWire PCB to accommodate op amp or your discrete op amp, but first I have to finish it and I am slow, hot summer here.
BR Damir
 
I will try to have universal GaiWire PCB to accommodate op amp or your discrete op amp, but first I have to finish it and I am slow, hot summer here.

No hurry - the 1st-gen FET990 derivative is ready and should work, and the 2nd-gen FET990 will also eventually have a PCB done. I'll use the intervening time to include some tweaks to the layout, if needed.
 
Disabled Account
Joined 2012
However, the odd-looking loads Q6 and Q4 on the LTP collectors look novel to me, and I'll take a look later in simulation to see if they provide any benefit in distortion cancellation.

In doing so, could this lead to another idea that cancels distortion. With low voltage IC's there has been a lot of recent (<10years) interest and patents in cancellation rather than cascoding, high nfb etc. A lot of new and interesting designs. Many of them.

The issue of ' is it audible' isnt always of primary interest. but new ideas are because they lead to other spin-off ideas and solutions to new applications. And, learning new concepts is fun in itself.


Thx-RNMarsh
 
jcx,
The Boneyard! It isn't even that some of these things are small as much as some of us are getting blind as we age! I know I will have to put on my magnifying lenses to work on a smd or else forget about it. I just haven't had to deal with any so no experience. Not really afraid to start just haven't had the opportunity or the need up to now.
 
My first job out of college was at Motorola, it must have been one of the first SMT production lines in N/A, ~1985, had a FR4 and a ceramic substrate hybrid line, running in parallel.
One day a green ME did a test around lunch time, put an FR4 pcb through the hybrid line, smoked to whole place out, people sick/fire dept etc, got to go home early that day. :)
Yes magnification is the answer.
All for SMT, for many assy's, it really is a mix of SM/TH. Paste/glue/reflow/wave/back-load etc.
P.S. still got parts I picked off the floor while fixing the pick/place machine. Now have a LCR meter, so can finally sort them MLC caps out.
 
All you young guys with perfect vision go for SMT. My eyes are not what they used to be. Personally I don't relish chasing a SMT part around a 1 inch square board with while holding a magnifying glass or wearing uncompfortable visors or peering through a microscope. I haven't built anything since about 1980 and I have a drawer full of matched 2n4401/03 in TO92 and I don't want to go through that pain of matching again. I would rather have holes to anchor the part before I solder it. Call me a dinosaur or whatever. I'll probably only build four of Scott's preamps and that will be it, so why invest in all the paraphenalia I would need for SMT. So please consider a thru hole board for us aging dinosaurs, PLEASE. Ray
 
diyAudio Member RIP
Joined 2005
The advantages of SM are manifold. It's possible to get performance at high frequencies that simply isn't accessible with radial/axial components. But they are difficult to work with. You can however sometimes get subassemblies with SM on little boards that convert, say an SO-8 to an 8-pin DIP. But you lose performance advantages with the added parasitic inductances (in particular).

I designed an elaborate hybrid analog and digital board for Epson America, for a home theater system, which went into the motorized screen and was both a motor controller and IR receiver/processor. I was rather pleased with how I managed to infer the motor condition, including its internal and otherwise inaccessible limit switches, looking at phase and transmitting signals across a galvanic barrier. I brought in an early version to show them, and the thing that they were impressed with is how I'd managed to put it together, as there were a lot of surface mount parts :)
 
If you do the math, I ain't young (54) but god did bless me with being very myopic, which in the case of small, is very good. Must realize that SMT is geared for automated assembly, period.
I have done some big pcb's, where I designed, gave to fab, got an assembled pcb back, with some BGA pkg's used, so it really is the economics of scale. If everything worked, as it should, I did not even have to touch a soldering iron.
Yes as DIY, THT is the best for all concerned, but I want to use a BF862 or one of those duals in a SOT-363? and NXP don't want to bother with a THT pkg, so it is, use what ya got and most possibly limit yourself, or get with the times.
Actually those clowns, finally standardized on the pin out for a bjt/fet in a SOT-23, where as, for say a T0-92, it is scattered about and thus is a PITA.
Awful shame that this SMT issue limits many a man's horizons, so that ends the story for me.
Rick
 
Last edited:
diyAudio Member RIP
Joined 2005
I'm retired now. All my kids have grown and out of the house and I have time to upgrade my gear that I built in 1980; replace all the electrolytics, etc. Is that enough of an answer? Ray
At least most electrolytics will be large enough to be seen.

I finally had to replace one in an NAD3020. Considering I'd used the amp since circa 1980 as well, I hardly was surprised. And it wasn't even one of the big bulk caps, rather one of the auxiliary higher-voltage rail ones. Of course I changed both aux rail ones together. I looked at the ripple on the main rails and it was still just fine, so I decided not to bother with those. Of course the switches and the volume pot have long ago gone south, so I simply run into the power amp input on the back.