Hey Graeme,
These Jfets been talkin to me and they now sound different.
I am not sure what the ids of your Jfets are but I suspect I have a bit too much open loop gain in the circuit to to be carving off that much feedback for just 20 db of gain.
After burn in it now sounds a bit sterile with 20 db gain
From reading the X'd Zen articles I gather it works best with simple 2-3 gain stages where the stages are set up for maximum linearity before applying super symmetry.
High open loop gain and high linearity are opposing maxims if I understand what Nelson has written.
I am wondering if some local feedback in the form of source resisters might be beneficial for a 20 db gain JFet design.
I will look at this more closely over the weekend.
Perhaps Nelson can throw light on the subject.
Ian
These Jfets been talkin to me and they now sound different.
I am not sure what the ids of your Jfets are but I suspect I have a bit too much open loop gain in the circuit to to be carving off that much feedback for just 20 db of gain.
After burn in it now sounds a bit sterile with 20 db gain
From reading the X'd Zen articles I gather it works best with simple 2-3 gain stages where the stages are set up for maximum linearity before applying super symmetry.
High open loop gain and high linearity are opposing maxims if I understand what Nelson has written.
I am wondering if some local feedback in the form of source resisters might be beneficial for a 20 db gain JFet design.
I will look at this more closely over the weekend.
Perhaps Nelson can throw light on the subject.
Ian
Attachments
Hi Ian,
You bring up a number of design goals that do indeed interact. Many on this site have discussed (and continue to discuss) these all to death. Nelson has talked about using 10dB of SuSy feedback a number of times. I don't want to misinterpret his words but he seems to favor this as at least a nominal value or initial value. So take the rest of your design process from there.
Your idea of using Source degeneration resistors to reduce the sterile sound is certainly interesting. I have no opinion. So you'll need to try it and let us know how it sounds. My only thought is that the AX (XA) circuit is very interactive. It seems like varying any parts values at all in this circuit produces some degree of sonic change. There are so few parts in the core circuit that this seems inevitable because of interactions. So let your ears be your ultimate arbiter - not your scope.
Cheers,
Graeme
You bring up a number of design goals that do indeed interact. Many on this site have discussed (and continue to discuss) these all to death. Nelson has talked about using 10dB of SuSy feedback a number of times. I don't want to misinterpret his words but he seems to favor this as at least a nominal value or initial value. So take the rest of your design process from there.
Your idea of using Source degeneration resistors to reduce the sterile sound is certainly interesting. I have no opinion. So you'll need to try it and let us know how it sounds. My only thought is that the AX (XA) circuit is very interactive. It seems like varying any parts values at all in this circuit produces some degree of sonic change. There are so few parts in the core circuit that this seems inevitable because of interactions. So let your ears be your ultimate arbiter - not your scope.
Cheers,
Graeme
You can play with the open loop gain and feedback without
difficulty. If you simply want to reduce gain, reducing the high
value feedback loop resistor and increasing the input signal
resistor values is a start, and will improve the bandwidth
slightly. Usually acompanying this is adjustment of the resistors
to ground at the inputs, which can trim the amount of feedback
a bit.
Degeneration off the Sources works as advertised, but can
make the sound more clinical. There's no way of predicting this
effect, but it's easy enough to adjust against listening - all it
takes is good taste and lots of patience.
difficulty. If you simply want to reduce gain, reducing the high
value feedback loop resistor and increasing the input signal
resistor values is a start, and will improve the bandwidth
slightly. Usually acompanying this is adjustment of the resistors
to ground at the inputs, which can trim the amount of feedback
a bit.
Degeneration off the Sources works as advertised, but can
make the sound more clinical. There's no way of predicting this
effect, but it's easy enough to adjust against listening - all it
takes is good taste and lots of patience.
Hi Nelson,
Thanks for the tips. I will investigate this with a goof bottle of red over the weekend.
I noticed that since installing high speed rectifiers on each secondary I lost a few volts and the rails are now 22.9 volts.
The output source resisters are 0.75R using a total of 16 IRFP240.
Do you see any advantage in reducing them to 0.47R?
I have not measured it yet but the ouput stage obviously plays a part in the overall gain.
Ian
Thanks for the tips. I will investigate this with a goof bottle of red over the weekend.
I noticed that since installing high speed rectifiers on each secondary I lost a few volts and the rails are now 22.9 volts.
The output source resisters are 0.75R using a total of 16 IRFP240.
Do you see any advantage in reducing them to 0.47R?
I have not measured it yet but the ouput stage obviously plays a part in the overall gain.
Ian
gl said:
gl and tryonziess,
Thanks for the tips. My SJ74's and SK170's are on the way from ampslab.
Jeremy
PS. http://www.grandata.co.uk/ has SK170-BL in stock. No SJ74.
Well I'll be damned, 'I'll be damned.
This thing is judging my entire system..
Firstly over the past few days I have sensed an opening up of the sound, an air of transparency. The amps have been running continuously for 4 or 5 days now. Great room heaters!
However, until this morning I was not aware of its unnerving capacity to distinguish between my various signal sources, notably digital. This had made using auditioning various musical selections a bit pointless on say SACD or DVD audio with different machines.
At the moment the amps are telling more about what's right or wrong with my transports, dacs and universal players than what may or may not be the best set-up option.
This is not just subtle but chalk and cheeze, its brutal in some instances. Thing such as differences in dynamic contrast, bass definition that I felt were subtle really jump out in your face now.
I suppose that good news and as they say if its not broke don't fix it.
I guess I need to talk to Russ White about his Opus dac.
iMac
This thing is judging my entire system..
Firstly over the past few days I have sensed an opening up of the sound, an air of transparency. The amps have been running continuously for 4 or 5 days now. Great room heaters!
However, until this morning I was not aware of its unnerving capacity to distinguish between my various signal sources, notably digital. This had made using auditioning various musical selections a bit pointless on say SACD or DVD audio with different machines.
At the moment the amps are telling more about what's right or wrong with my transports, dacs and universal players than what may or may not be the best set-up option.
This is not just subtle but chalk and cheeze, its brutal in some instances. Thing such as differences in dynamic contrast, bass definition that I felt were subtle really jump out in your face now.
I suppose that good news and as they say if its not broke don't fix it.
I guess I need to talk to Russ White about his Opus dac.
iMac
I did some variations over a closed loop gain range last night.
I am incliend to think in the end its to degree system dependant and taste.
What I found is low feeback sounded open, light and relaxed with some what less clarity. This favoured simple musical passages like acoustic / vocals.
Increasing feedback increase apparent bandwidth. However the extension in bandwidth without sufficient damping may not suit all loudspeakers.
Therfore a bit more feedback (but not too much) can be batter in some instances
Imac
I am incliend to think in the end its to degree system dependant and taste.
What I found is low feeback sounded open, light and relaxed with some what less clarity. This favoured simple musical passages like acoustic / vocals.
Increasing feedback increase apparent bandwidth. However the extension in bandwidth without sufficient damping may not suit all loudspeakers.
Therfore a bit more feedback (but not too much) can be batter in some instances
Imac
Sorry I lost my interent connection earlier.
Too much feeback and the sound clamps up and become dry, sterile and lifeless.
So of my earlier X Aleph incarnations no doubt share similar behaviours but I never bothered to adjust them.
So I am carefully trying different closed loop gain settings because settings in such simple low open loop gain circuits I think there is a sweet spot that suits the individual.
I have done this on the basis of Nelson's earlier post. But at this stage have only adjusted the feedback and not the open loop gain wth source degeneration or adjusted the current source.
Graeme's JFT no douby hav different ids to mine, he uses a number of output device and source resisters, higher supply rails so I have not doubt I not getting an identical out come.
Bit this is what diy is all about
Too much feeback and the sound clamps up and become dry, sterile and lifeless.
So of my earlier X Aleph incarnations no doubt share similar behaviours but I never bothered to adjust them.
So I am carefully trying different closed loop gain settings because settings in such simple low open loop gain circuits I think there is a sweet spot that suits the individual.
I have done this on the basis of Nelson's earlier post. But at this stage have only adjusted the feedback and not the open loop gain wth source degeneration or adjusted the current source.
Graeme's JFT no douby hav different ids to mine, he uses a number of output device and source resisters, higher supply rails so I have not doubt I not getting an identical out come.
Bit this is what diy is all about
Hi Graeme,
To simplify assembly I moved the feedback network and drain resisters back on to the main pcb.
I also tried a different variety of capacitor in the current share loop and got quite different results. This has kind of blown away my previous impressions and have gone back to 26 db of gain for now.
My plan is to also try some small Zetex mosfet for the differential pair like those used on the Pass diy opamp article. Perhaps this will be best done in a new thread.
To simplify assembly I moved the feedback network and drain resisters back on to the main pcb.
I also tried a different variety of capacitor in the current share loop and got quite different results. This has kind of blown away my previous impressions and have gone back to 26 db of gain for now.
My plan is to also try some small Zetex mosfet for the differential pair like those used on the Pass diy opamp article. Perhaps this will be best done in a new thread.
Hi Graeme,
I will look at using the 3 parrellel Jfets this week and increase in bias current to 36 ma.
At this point I am quite pleased with the Jfet option and my small fet option is coming along.
From the adjustments I have done so far it would seem that any of the X Aleph diy incarnations my be working circuits but may be far from optimised to the get the best out of the Aleph cu/susy.
I would therefore encourage those who have built them and put them to one side to pull them out and have a look at the gain and feedback. Small adjustments can bring surprising results and I am certain susy works best with less rather than more feeback.
For those interested the Jfets can be purchased matched from Erno Borbely.
I plan to have a version of the ZVP2210 input stage working soon. This appears a much better option than the old 9610 part in lower voltage versions of the circuit.
I will look at using the 3 parrellel Jfets this week and increase in bias current to 36 ma.
At this point I am quite pleased with the Jfet option and my small fet option is coming along.
From the adjustments I have done so far it would seem that any of the X Aleph diy incarnations my be working circuits but may be far from optimised to the get the best out of the Aleph cu/susy.
I would therefore encourage those who have built them and put them to one side to pull them out and have a look at the gain and feedback. Small adjustments can bring surprising results and I am certain susy works best with less rather than more feeback.
For those interested the Jfets can be purchased matched from Erno Borbely.
I plan to have a version of the ZVP2210 input stage working soon. This appears a much better option than the old 9610 part in lower voltage versions of the circuit.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.