Discrete Opamp Open Design

Jump to Latest
#203
For transistors the Toshiba 2SK2145 dual is still available 50 Volt, 300 mW 15 mS and noise figure under .5 dB. Bipolars like the KSC945 and KSA773 are readily available, complimentary, 300 mHZ and low noise. Dual P Fets are history though. The 2SK2145 has the disadvantage of sources tied though.
Maybe a nice tube for voltage gain. I will defer this to Wavebourn.
 
#204
Wavebourn said:
I am going to try today in the field Behringer 2x300W per channel amp, class D with SMPS. I paid $185 for it including shipment. No heatsinks (almost), just a fan.
Click to expand...

We tried something similar 2 years ago. It was not Behringer, but similar brand, same output power, class D.
It was an open-air garden listening session, with JBL 250TI LE speakers.

http://www.uwe-neidhardt.de/assets/images/JBL_250Ti_Jubilee2.png

The class D was unable to compete with complementary-differential power amplifier, also 2x300W, 2SK170/2SJ74 input, cascoded, MOSFET VAS and 4 pairs MJL1302/3281 output. The class D sound was more harsh, less fluent, less full-bodied.
 
#206
scott wurcer said:
The one I was thinking of is well known (Bernie Oliver?), just the right drop on the emitter degeneration resistors (output devices) will do some first order distortion cancellation. I don't remember the exact numbers.
Click to expand...
The paper was in the Hewlett-Packard Journal and is reproduced in The Selected Papers of Bernard Oliver, pp. 311-317. I have that book handy for a change 🙂

An excerpt: "However, the drop across this resistance produced by the operating current is only Io/2go = 1/2 kT/q to Io/go = kT/q or from 13 to 26 millivolts. Over the temperature ranges from 0 C to 100 C, the junction drop of a silicon transistor will change typically by 250 mV, Thus, unless the biasing diodes ... track this change within a few percent, Io will be very unstable ... At present the most practical solution to the temperature stability problem appears to be to make Rc [the emitter resistors] many times larger than 1/go, and to rely on negative feedback to reduce the resulting distortion."

Also in the collection is his great paper Thermal and Quantum Noise, from Proc. IEEE, May 1965.

What a wig he was.
 
#208
The matching of the 2SK2145 seems pretty good on spot testing. We buy them on reels of 3,000 and they are pretty consistent on the reel. Nice dual part for all but the lowest noise apps. I mostly current source them in circuits using the Y type.
 
#209
If you close local loop before buffer, and add resistor from buffer's input to output, you will get a version of "current dumping" amp. Try and compare. However you can't get high gain from such an amp, but low distortions on a wide band for sure.

Speaking of transistors, why you don't consider Zetex devices? I would call them "Toshiba of modern days", in terms of innovations.
 
#210
bcarso said:
The paper was in the Hewlett-Packard Journal and is reproduced in The Selected Papers of Bernard Oliver, pp. 311-317. ... At present the most practical solution to the temperature stability problem appears to be to make Rc [the emitter resistors] many times larger than 1/go, and to rely on negative feedback to reduce the resulting distortion."
Click to expand...

Bingo. See my output stage for headphone amp. The whole package doesnt drift at all. very temp stable circuit. Surprised no one commented before as to why circuit was so temp stable.
 
#212
fitzfish said:
Thanks Brad, Wayne.

Are we talking adjacent die? perhaps 5% or less matching?

Nice, small package for an smt discrete opamp LTP. Beats the SO8 from LSK.

Dave
Click to expand...
Yes, they mention 2% for some of the parameters. A great idea; formerly they just sliced and diced and dumped into a big hopper before packaging, I guess. Having things off the same wafer was better than nothing, but this is a lot better.
 
#220
jneutron said:
Dumped into a big hopper??? Shirley you jest..😱

I'da thunk either an expanded wafer, or waffle pack..

jn
Click to expand...

Typically they're scribed on a sheet of heat activated sticky plastic and stretched. Now they would probably use a vacuum pencil pick and place after that.

The same die dual issue would be what is the substrate, I thought it might be the back gate so isolated duals would have to be two die. We always bought them that way.

EDIT Sorry EUVL the FET's were drawn funny, and some symbols show the arrow at the drain.
 
Last edited:
You must log in or register to reply here.
Top Bottom