The XGC will be use with a preamp. I'm right now not fixed on gain I will apply to these circuit. It's based on Jfet 2SK389 diff paire. Nothing real new. For moment the first stage convert assym signal into sym signal to feed à double pot. After this there' a buffer (same stage with unit gain).
Don't take care on part value around current source, they are note fixed.
the board
May be some one has a recommandation for gain amout to apply to this scheme in case of association with XGC.
Marc
Don't take care on part value around current source, they are note fixed.
An externally hosted image should be here but it was not working when we last tested it.
the board
An externally hosted image should be here but it was not working when we last tested it.
May be some one has a recommandation for gain amout to apply to this scheme in case of association with XGC.
Marc
Pre-amp output Z
I have applied only a few minutes of thought to your design where clearly many hours have been spent, so do not take any criticism any more seriously than this quick a review deserves.
Your pre-amp has a fairly high output Z and the output is un-buffered. The pre-amp takes it's feedback directly from the output jack, this is effectively the same node where the XGC amplifier also applies it's feedback.
Unless this is an integrated pre and power amp, ideally each chassis should be stand alone and designed so that they work the same by themselves or interconnected, you may not have accomplished this. I'm not saying this won't work, but isolation between the pre and power amps could become an issue and might effect the sound.
Anything you do to increase the XGC input impedance and/or decrease the pre-amp output impedance will reduce the chances of this kind of difficulty.
Adding more parts or stages is rarely a good idea but maybe you would consider a single transistor unity gain buffer at the pre-amp output. You could always short stuff the board and delete it if it proved to be un-necessary. Either a source follower or emitter follower could reduce the pre output impedance by ten or twenty to one.
Another possible solution - currently the two pre-amp stages are effectivly identical; the first acts as a single to balanced converter and an impedance buffer. It should be possible to reduce the resitor values of the second stage by at least 2 to 1 and most probably by 5 to 1 (you'll need to recalculate the current source) this would significantly lower the output impedance. The capacitors might need to get bigger.
Also on the pre, a missing wire from R21 to R20?
I have applied only a few minutes of thought to your design where clearly many hours have been spent, so do not take any criticism any more seriously than this quick a review deserves.
Your pre-amp has a fairly high output Z and the output is un-buffered. The pre-amp takes it's feedback directly from the output jack, this is effectively the same node where the XGC amplifier also applies it's feedback.
Unless this is an integrated pre and power amp, ideally each chassis should be stand alone and designed so that they work the same by themselves or interconnected, you may not have accomplished this. I'm not saying this won't work, but isolation between the pre and power amps could become an issue and might effect the sound.
Anything you do to increase the XGC input impedance and/or decrease the pre-amp output impedance will reduce the chances of this kind of difficulty.
Adding more parts or stages is rarely a good idea but maybe you would consider a single transistor unity gain buffer at the pre-amp output. You could always short stuff the board and delete it if it proved to be un-necessary. Either a source follower or emitter follower could reduce the pre output impedance by ten or twenty to one.
Another possible solution - currently the two pre-amp stages are effectivly identical; the first acts as a single to balanced converter and an impedance buffer. It should be possible to reduce the resitor values of the second stage by at least 2 to 1 and most probably by 5 to 1 (you'll need to recalculate the current source) this would significantly lower the output impedance. The capacitors might need to get bigger.
Also on the pre, a missing wire from R21 to R20?
Re: Pre-amp output Z
Here is the correct scheme (with missing wire)
I rearrange the pcb and allow the possibility to feed it with sym signal.
Nice to read you. Marc
hermanv said:I have applied only a few minutes of thought to your design where clearly many hours have been spent, so do not take any criticism any more seriously than this quick a review deserves.
I posted here to discuss and learn
Your pre-amp has a fairly high output Z and the output is un-buffered. The pre-amp takes it's feedback directly from the output jack, this is effectively the same node where the XGC amplifier also applies it's feedback.
This two stage are built to have unit gain. The friend à discuss with said me i will have 100Ohm Z out. The first stage convert assym signal in symetric one to feed "X"GC structure.The stage after pot take place as buffer since it present unit gain.
Unless this is an integrated pre and power amp, ideally each chassis should be stand alone and designed so that they work the same by themselves or interconnected, you may not have accomplished this. I'm not saying this won't work, but isolation between the pre and power amps could become an issue and might effect the sound.
The pre will have his own box with symetric xlr link betwen XGC blocs.
Anything you do to increase the XGC input impedance and/or decrease the pre-amp output impedance will reduce the chances of this kind of difficulty.
Adding more parts or stages is rarely a good idea but maybe you would consider a single transistor unity gain buffer at the pre-amp output. You could always short stuff the board and delete it if it proved to be un-necessary. Either a source follower or emitter follower could reduce the pre output impedance by ten or twenty to one.
Another possible solution - currently the two pre-amp stages are effectivly identical; the first acts as a single to balanced converter and an impedance buffer. It should be possible to reduce the resitor values of the second stage by at least 2 to 1 and most probably by 5 to 1 (you'll need to recalculate the current source) this would significantly lower the output impedance. The capacitors might need to get bigger.
Gain is initial set at unit rate throw R12/R13 for example due to the X structure, but also i can modify gain with Trimmers near each 2SK389 (the unit lay at about 1k5 on trimmer. So in first issue this pre. as no gain : converts assym to symetric signal and buffer stage to suppress pot influance on Z output
Also on the pre, a missing wire from R21 to R20?
Here is the correct scheme (with missing wire)
An externally hosted image should be here but it was not working when we last tested it.
I rearrange the pcb and allow the possibility to feed it with sym signal.
An externally hosted image should be here but it was not working when we last tested it.
Nice to read you. Marc
Inter circuit impedance?
pre-amp out Z = ~1K
power amp in Z = ~3K
Both are one R from summing nodes for local feedback, just seems that interaction is quite possible.
The circuits only have near unity gain when you include the feedback loops. The actual circuit gains are much higher. So the worst case result might be an oscillation.
As you build this, test for a problem before hooking up anything expensive. Try a dummy load of 8 Oms paralleled by 1uF, if it's stable relax.
pre-amp out Z = ~1K
power amp in Z = ~3K
Both are one R from summing nodes for local feedback, just seems that interaction is quite possible.
The circuits only have near unity gain when you include the feedback loops. The actual circuit gains are much higher. So the worst case result might be an oscillation.
As you build this, test for a problem before hooking up anything expensive. Try a dummy load of 8 Oms paralleled by 1uF, if it's stable relax.
Re: Inter circuit impedance?
After simulation the scheme seems to have 500R Z output. We wiil try to lower these. The other option is to grow the XGC Zinput in changing R5/R7 and R6/R8 (10k/220K for exemple). Will these changing affect XGC sound caracterics and if in what suppose way?
Marc
hermanv said:
After simulation the scheme seems to have 500R Z output. We wiil try to lower these. The other option is to grow the XGC Zinput in changing R5/R7 and R6/R8 (10k/220K for exemple). Will these changing affect XGC sound caracterics and if in what suppose way?
Marc
Value ratios
Ah, here there be dragons.
As you lower the resistance in a given circuit stray capacitance effects, leakage current effects, EMI susceptability and resistor noise tend to decrease. All good things.
Now however, the reactive components need to be larger (capacitors in your design). On the whole, smaller capacitors tend to be better models of an ideal capacitor (the ratio of series R parallel R and series inductor gets better) or more simply, smaller capacitors tend towards higher Q. Dielectric absorbtion is also reduced in smaller capacitors.
Discrete semiconductors nearly always have a gain vs current curve and for most devices, bandwidth goes up with increasing current. So you often bias them to be near the middle of the current related gain peak. This of course ends up setting the R values for any given gain requirement.
My style is usually to design for the lowest practical R values keeping an eye on heat and loading of previous stages.
So, a quandary: lower R values result in more predictable performace and wider bandwidth (this helps feedback behave as planned) but loads the previous stage. If you add a buffer to the pre then suddenly the input Z of the amp is a non-issue for you but maybe not for others.
Not to duck the question, I would add a buffer (you did) before I'd make the R values a lot larger. Conversely many commercial amplifiers have input Z arounf 47K Ohms so you are nowhere near being unique if you increase the input R values. Maybe a compromise? Double the input divider R values?
Juggling umpteen variables makes designing simultaneously fun and tedious.
Ah, here there be dragons.
As you lower the resistance in a given circuit stray capacitance effects, leakage current effects, EMI susceptability and resistor noise tend to decrease. All good things.
Now however, the reactive components need to be larger (capacitors in your design). On the whole, smaller capacitors tend to be better models of an ideal capacitor (the ratio of series R parallel R and series inductor gets better) or more simply, smaller capacitors tend towards higher Q. Dielectric absorbtion is also reduced in smaller capacitors.
Discrete semiconductors nearly always have a gain vs current curve and for most devices, bandwidth goes up with increasing current. So you often bias them to be near the middle of the current related gain peak. This of course ends up setting the R values for any given gain requirement.
My style is usually to design for the lowest practical R values keeping an eye on heat and loading of previous stages.
So, a quandary: lower R values result in more predictable performace and wider bandwidth (this helps feedback behave as planned) but loads the previous stage. If you add a buffer to the pre then suddenly the input Z of the amp is a non-issue for you but maybe not for others.
Not to duck the question, I would add a buffer (you did) before I'd make the R values a lot larger. Conversely many commercial amplifiers have input Z arounf 47K Ohms so you are nowhere near being unique if you increase the input R values. Maybe a compromise? Double the input divider R values?
Juggling umpteen variables makes designing simultaneously fun and tedious.
Increasing XGC input impedance
One thing to keep in mind is that the supersymmetric feedback concept relies on the cross communication of the error current between the two halves of the input differential pair of a balanced circuit. Higher input impedance in this case result in reducing the error current. When Nelson Pass first suggested the XGC concept, he proposed an input Z around 1K.
There is no reason that the input impedance can't be higher, but the supersymmetry effect will be diminished. If you want a high input Z, id suggest changing the input differential to jjfets or mosfets, and go to voltage feedback rather than current feedback. Although Susy works best with current feedback, you can increase the input Z a lot higher with a voltage feedback scheme while keeping the error current level higher.
Something to keep in mind is that I developed the XGC circuit that is being used around having a preamp with a rediculously low output impedance. I didn't build it to be best for everybody but to be the best in my system. I encourage everybody to feel free to modify it to achieve the best results in their systems.
Cheers, Terry
One thing to keep in mind is that the supersymmetric feedback concept relies on the cross communication of the error current between the two halves of the input differential pair of a balanced circuit. Higher input impedance in this case result in reducing the error current. When Nelson Pass first suggested the XGC concept, he proposed an input Z around 1K.
There is no reason that the input impedance can't be higher, but the supersymmetry effect will be diminished. If you want a high input Z, id suggest changing the input differential to jjfets or mosfets, and go to voltage feedback rather than current feedback. Although Susy works best with current feedback, you can increase the input Z a lot higher with a voltage feedback scheme while keeping the error current level higher.
Something to keep in mind is that I developed the XGC circuit that is being used around having a preamp with a rediculously low output impedance. I didn't build it to be best for everybody but to be the best in my system. I encourage everybody to feel free to modify it to achieve the best results in their systems.
Cheers, Terry
Re: Increasing XGC input impedance
You well developpe the the XGC to be use with your CCS-X-Bosoz. As the Susy caracteristic need to have low impedance input i prefer to adapt preamp stage to this fact. So i first work on lowering the preamp Zoutput before changing XGC caracterics with consequences on its sound caracteristics.
Here the new version with the borbely buffer after pot and just before output with low Zoutput to preserve 3K XGC Zinput.
The related Board
Marc
metalman said:
You well developpe the the XGC to be use with your CCS-X-Bosoz. As the Susy caracteristic need to have low impedance input i prefer to adapt preamp stage to this fact. So i first work on lowering the preamp Zoutput before changing XGC caracterics with consequences on its sound caracteristics.
Here the new version with the borbely buffer after pot and just before output with low Zoutput to preserve 3K XGC Zinput.
An externally hosted image should be here but it was not working when we last tested it.
The related Board
An externally hosted image should be here but it was not working when we last tested it.
Marc
Here are the board
Marc
An externally hosted image should be here but it was not working when we last tested it.
Marc
The same with parts solderded on
Marc
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Marc
An another with the heatsink i will use.
Marc
An externally hosted image should be here but it was not working when we last tested it.
Marc
So amp board are complet :
The snubber
The 330nF cap between + and - voltage rail direct on chip pins
The input differentiel transitor pair
The Link cap modul (1µ MKP/220µ electrolytics/10nF MKP) between ztx diff par and LM3386
So Yet next is power supply
Marc
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
The snubber
An externally hosted image should be here but it was not working when we last tested it.
The 330nF cap between + and - voltage rail direct on chip pins
An externally hosted image should be here but it was not working when we last tested it.
The input differentiel transitor pair
An externally hosted image should be here but it was not working when we last tested it.
The Link cap modul (1µ MKP/220µ electrolytics/10nF MKP) between ztx diff par and LM3386
An externally hosted image should be here but it was not working when we last tested it.
So Yet next is power supply
Marc
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.