Joachim Gerhard & I had some discussions here about a balanced to single ended converter :
http://www.diyaudio.com/forums/digi...minimalistic-iv-converter-21.html#post3097491
(up to post #1052)
I renamed it XCEN, as X (balanced or differential) is converted to C (single ended complementary).
The previous version (2C) still uses 2SK170/2SJ74 which are now difficult to get.
So I tried another version by just replacing the 2SK170/2SJ74s with 2SK246/2SJ103.
Performance is still very decent. So good news as the latter are still relatively easy to get.
PCB for this latest version will follow. To be continued.
Patrick
http://www.diyaudio.com/forums/digi...minimalistic-iv-converter-21.html#post3097491
(up to post #1052)
I renamed it XCEN, as X (balanced or differential) is converted to C (single ended complementary).
The previous version (2C) still uses 2SK170/2SJ74 which are now difficult to get.
So I tried another version by just replacing the 2SK170/2SJ74s with 2SK246/2SJ103.
Performance is still very decent. So good news as the latter are still relatively easy to get.
PCB for this latest version will follow. To be continued.
Patrick
Attachments
Last edited:
Good idea to have this new thread.
Rather than introduce some alternatives here I think I'll start another thread as well. I'm getting some extremely promising results in sim from an all-NJFET variant on the PNP Wilson-mirror-based current-differencing stage I showed in the Zen -> Cen -> thread.
That may also be the place to show some of the extensions to Sen that I studied several months ago. We shall see.
Brad
Rather than introduce some alternatives here I think I'll start another thread as well. I'm getting some extremely promising results in sim from an all-NJFET variant on the PNP Wilson-mirror-based current-differencing stage I showed in the Zen -> Cen -> thread.
That may also be the place to show some of the extensions to Sen that I studied several months ago. We shall see.
Brad
Brad, that will be very interesting.
But it would be good if you could reduce the number of voltage sources required in the circuit.
Also I wonder how you might deal with DC drifts.
JFETs are not known to be thermally stable current sources.
Regards,
Patrick
The new circuit will be just two supply rails, although relatively high voltage to get low noise and a large output signal swing. There will be two floating voltages, one of which can be a battery that will last the shelf life (only JFET gate current), another which could be derived from the rails but also could be a battery (stability and noise in that voltage will have a very weak effect on circuit performance).
If a d.c. servo is indicated I can provide one, although JFETs selected for operation at about 600mV away from pinchoff can be quite stable with temperature. But for the particular mirror, it will just need a matched pair (actually a matched cascode, although the upper devices are less critical) in order to do the job at hand*. One of the I-V (or I-I, current conveyor) derived from your circuits will drive a common load resistor for the single ended output. The output of the JFET mirror will be driven from the other polarity conveyor and directly sum as a current in that common load resistor. An additional voltage follower buffer will be optional.
BTW, I think your practice of keeping things balanced all the way to the loudspeaker is the best, for anyone who can manage it
Brad Wood
*I also have a whole bunch of 2SK389V and about 200 2SJ109BL. But I'm favoring the BF862 these days when used in a cascode with something like a process 51 part, after seeing the results that Wurcer and Groner have been getting, even though I need to sort and match them.
2N4391 -92, -93 and their plastic equivalents PN4391 etc., J111, 112, 113 I believe, also SM MMBF4391. See for example http://www.fairchildsemi.com/ds/MM/MMBF4391.pdf.
I like the 2N part because it can be used a fairly high dissipations when in a heatsink, some of which have a threaded stud and an insulating disc (the gate is common to the case). A few people still make them and someone found some in surplus recently.
Here's some data with typical curves: http://www.fairchildsemi.com/ds/J1/J112.pdf
I like the 2N part because it can be used a fairly high dissipations when in a heatsink, some of which have a threaded stud and an insulating disc (the gate is common to the case). A few people still make them and someone found some in surplus recently.
Here's some data with typical curves: http://www.fairchildsemi.com/ds/J1/J112.pdf
Last edited:
We had something like that here, but used J111 or J310 as cascode.
The reason that I prefer J111 is that there is a P-counterpart, namely J174.
http://www.diyaudio.com/forums/digi...ng-new-ess-vout-dac-es9022-7.html#post2867664
http://www.diyaudio.com/forums/atta...ss9022-euvl-filter-3.1-schematic.tsc-tina.pdf
But I would be interested in a 2Nxxxx part as an alternative to 2SK246/2SJ103.
Any suggestions for a good, truly complementary pair ?
Patrick
The reason that I prefer J111 is that there is a P-counterpart, namely J174.
http://www.diyaudio.com/forums/digi...ng-new-ess-vout-dac-es9022-7.html#post2867664
http://www.diyaudio.com/forums/atta...ss9022-euvl-filter-3.1-schematic.tsc-tina.pdf
But I would be interested in a 2Nxxxx part as an alternative to 2SK246/2SJ103.
Any suggestions for a good, truly complementary pair ?
Patrick
> Of course if you make it symmetric after the input Jfets it will also convert
single-ended to balanced.
Nelson, how about this one ?
http://www.diyaudio.com/forums/atta...mplifier-part-ii-folded-cascode-paraphase.jpg
Patrick
single-ended to balanced.
Nelson, how about this one ?
http://www.diyaudio.com/forums/atta...mplifier-part-ii-folded-cascode-paraphase.jpg
Patrick