Lowering the output impedence of this dac's output

[georgehifi]

I have succesfully direct coupled the output by shorting the output cap and installing a dc offset nulling trimmer (indicated by the yellow highlighted components)
I now wish to short out the two resistors R119 and R219 47ohms to lower the output impedence to see how it affects the drive into a quite low input impedence of my next stage which is around 6kohms to see if it sounds any better, can everyone comment before I try it, will it be stable still? will it osscillate? it should drive better? I don't have much experience with darligtons BDX33's and BDX34's

Cheers George

[dddac]

The 47 Ohm are there to avoid shorting the output, so leave them out and do not short the output to avoid blowing up the end transistors.

Not sure 47 Ohm will make such a difference by the way for a 6kOhm input, but the proof is in the eating ...

[rfbrw]

Originally posted by georgehifi
I have succesfully direct coupled the output by shorting the output cap and installing a dc offset nulling trimmer (indicated by the yellow highlighted components)
I now wish to short out the two resistors R119 and R219 47ohms to lower the output impedence to see how it affects the drive into a quite low input impedence of my next stage which is around 6kohms to see if it sounds any better, can everyone comment before I try it, will it be stable still? will it osscillate? it should drive better? I don't have much experience with darligtons BDX33's and BDX34's

Cheers George

If the output transistors are adequately heatsinked, you might want to consider increasing the bias.

[georgehifi]

Originally posted by rfbrw


If the output transistors are adequately heatsinked, you might want to consider increasing the bias.

Heatsinks that can be done, what do you suggest to change to up the bias, I'm always in favour of more Class A , I have zero expirence in Darlingtons.
Just did the shorting of the R119 and R219 and had a quick listen, I had the inkling that the bottom end seamed stronger and more defined, not huge difference, but worth the trouble.

Cheesr George

[rfbrw]

Measure the voltage across emitter resistors R114/5 and R214/5. Dividing that by the value of those resistors, 10 ohms, will give you the current bias current, so to speak. Using that as a starting point you can then increase the bias by lowering value of the emitter resistors.

[georgehifi]

That's a simple way rfbrw. But as it is, with the 10ohm emiter resistors in, it should be quite stable, if I lower the emiter resistor are'nt we delving into making the output less stable and maybe prone to occilate if it sees some weird high interconect capacitance ?


Cheers George

[rfbrw]

Originally posted by georgehifi
That's a simple way rfbrw. But as it is, with the 10ohm emiter resistors in, it should be quite stable, if I lower the emiter resistor are'nt we delving into making the output less stable and maybe prone to occilate if it sees some weird high interconect capacitance ?


Cheers George

The emitter resistors are there for thermal stability and are often a lot smaller but you could always pose the question in the solid state forum for another view.

[banana]

R119, R219 would isolate the output amp from capacitive load. You might experiment with smaller value, but shorting them doesn't seems to be a wise idea.

If you need very low output resistor, the coil//resistor approach looks like a hit.

http://www.jensen-transformers.com/datashts/oli3.pdf

[georgehifi]

Originally posted by banana
R119, R219 would isolate the output amp from capacitive load. You might experiment with smaller value, but shorting them doesn't seems to be a wise idea.

If you need very low output resistor, the coil//resistor approach looks like a hit.

http://www.jensen-transformers.com/datashts/oli3.pdf

Very nice, first time I've seen these, I will definatly get some.

Thanks George

[jcx]

I could offer suggestions I believe will improve measurable performance if a higher technical quality dac I/V-buffer is what you are seeking

but by the evidence of that discrete transistor "op amp" with 4 MHz ft darlingtons in weak Clas AB bias I guess the designer was more interested in introducing some "sound effect" - I wouldn't want want to unknowingly mess with someone's fuzz box cricuit

assuming you want to technically improve the circuit my 1st mod would be to lose the discrete op amp, but you might want to start with Class A biasing all 4 of the I/V op amp outputs with ~ 5 mA from the negative rail - very specific to NE5532/4 where the pull down is very much slower than the internal NPN pull up half of the output stage - I don't recomend this generally but it really is applicable to the NE5532 in a I/V stage

deep Class A biasing the discrete op amp's output would help but I can't imagine tweaking this to within an order of magnitude of the gain accuracy and distortion performance of even another NE5532/4 per the datasheet circuit, at 6K load you have only 500uApp so it isn't even clear Class A bias of a replacement NE5532/4 would matter - I bet the datasheet circuit wins any measureable performance comparison for your load

your Vos tweak tirm pot may be upsetting the diff to single end conversion - ideally R103/RR105 = R104/R108 and any Vos should be added elsewhere - the fact that you need Vos adj is further evidence against the discrete op amp