NE5534 with a new front end ...anybody?

[ashok]

Someone mentioned that the input pair of transistors of a NE5534 can be bypassed and an external pair can be used.
I was considering using a pair of 2SK170's. Has anyone tried this or with any other transistors ?
It's been done commercially long ago when they didn't have better opamps. However it will be interesting to know what the results will be.

Will this work in a Spice simulation ? I'm not sure if the bypassing scheme can be simulated !
Cheers.

[peranders]

Tie the inputs to V- and connect the new input pair to the offset trim inputs. But why do want to do this? This was a good idea 20 years ago. Now you can buy a good opamp with better results.

[ashok]

Just curious. I want to know how it tranforms the performance of the opamp audibly. Apart from electrical specs it must affect the sound of the standard chip. How ?
Cheers.

[janneman]

Originally posted by ashok
[snip]Apart from electrical specs it must affect the sound of the standard chip. How ?
Cheers.

It *must*? Why?

Jan Didden

[ashok]

Bipolar input changed to jfet.
Distortion spectrum could be different. I'm not sure if that should be audible but no harm checking it out. Maybe I shouldn't have said 'must'.
Cheers.

[lndm]

Originally posted by ashok
Distortion spectrum could be different.

This sounds possible. Under some (possibly less than ideal) circumstances I feel the single ended fet can lend an odd order harmonic influence more so than the bi-polar might.

[nuvistor]

http://www.diyaudio.com/forums/showt...907#post706907 has links that may answer your question and offer other information on 5534 hacks, the schematic link may not work now. The Siliconix appnote I referenced describes a cascoded JFET input pair with 2-BJT current sink but any reasonable input pair design could be used, I like using just matched high-beta NPN (BC550, 2N5210, 2N5087, MPSA18) with emitter degeneration resistors and a current sink resistor for low-gain inverting input stages.

The front-end bypass scheme can be simulated if the spice model is modified to remove the first stage NPN pair, I have attached such a model. Simulation with this model is OK for estimating DC biasing and feedback loop stability. Don't expect useful distortion analysis, the results will be insanely optimistic.

The front-end bypass allows frequency compensation by reducing DC open-loop gain, if input transistor gm is less than 2.5mmho no comp capacitor should be needed for unity gain stability, if higher I suggest adding emitter/source degeneration resistors if input noise is not critical.

Sure there are modern op amps that perform better than a hacked 5534, but if 5534s, matched 2SK170 and 1/4W resistors are all you can get your hands on and you have time to build then this hack is probably a better option than TL072 or 4558.

[AndrewT]

Hi Nuvistor,
can you post the schematics to go with the spice listings?

[nuvistor]

OK, here's a schematic.

[nuvistor]

This ZIP package contains LTSpice schematics and spice models. I further improved the normal 5534 model's open loop gain accuracy vs. datasheet curve in the 1 to 10MHz range, this is included. A text file with description of application and models and png files of the schematics are included.

To simulate, unzip to a directory and open the schematic(s) with LTSpice, no need to move the models and schematic symbol to your main LTSpice directories.

The simulation is only for the input hack, you will have to build a circuit to tinker with the output hack. The text file describes the output hack well enough.