nar said:
Hi nar,
Thanks for your kind words!
No they do not need to be matched except the 2 CCS´s
down (BC549C) connected to R9, R10. All the others are
just cascode devices and the CCS´s on top (BC559C) are
adjustable.
You can swap BC549C to BC550C and BC559C to BC560C
if you want but IMHO you won´t hear the difference!
If you follow the circuit strictly, they run at 5mA (depends
on R9, R10)
If you do not want to match the 2SK369 (which can be a
hard timeconsuming matter) you can use one 2SK389
instead (slightly higher noise, slightly lower transconductance)
the pcb accepts both.
Uli
gain
hi Nar,
Calculation: for 2SK389 in brackets:
the transconductance of the 2SK369 (389) is typical 40 (20) mS.
this results in an openloopgain of 72 (66) dB without load except
R21, R22 thus showing an imaginary value of R17,18 of about
400k (200k). You must parallel this value to the real value to calculate gain.
As the imaginary value is a function not only of transconductance
but also of outputload its quite difficult to calculate.
It is much easier when you use external complimentary source-
followers at the output.Then you can assume the o.l. gain to be
loadindependant.
So if you use 10k input resistance (per half!) you loose ~10% at
R19, 20 you have to put in 152k (245k!)for R17,18 to achieve 20dB gain.
The unit has a typical gain of 53 (47) dB as drawn without
external load.
As like every inverting opamp its tricky to achieve high gain AND
high input impedance!
Uli
hi Nar,
Calculation: for 2SK389 in brackets:
the transconductance of the 2SK369 (389) is typical 40 (20) mS.
this results in an openloopgain of 72 (66) dB without load except
R21, R22 thus showing an imaginary value of R17,18 of about
400k (200k). You must parallel this value to the real value to calculate gain.
As the imaginary value is a function not only of transconductance
but also of outputload its quite difficult to calculate.
It is much easier when you use external complimentary source-
followers at the output.Then you can assume the o.l. gain to be
loadindependant.
So if you use 10k input resistance (per half!) you loose ~10% at
R19, 20 you have to put in 152k (245k!)for R17,18 to achieve 20dB gain.
The unit has a typical gain of 53 (47) dB as drawn without
external load.
As like every inverting opamp its tricky to achieve high gain AND
high input impedance!
Uli
Thanks for reply.
In my case achieving a high input impedance is not a real goal.
So I guess values to be adjusted to get proper gain.
I think this design cannot accept unbalanced input to get balanced out ? In that case how to achieve it ?
Thanks for replys to the " twenty questions" !!!
PS the outputs must be a few volts above ground ...
In my case achieving a high input impedance is not a real goal.
So I guess values to be adjusted to get proper gain.
I think this design cannot accept unbalanced input to get balanced out ? In that case how to achieve it ?
Thanks for replys to the " twenty questions" !!!
PS the outputs must be a few volts above ground ...
input issue
Hi Nar,
Yes of course it IS possible!!!
You can feed whatever you want!
Output stays symmetric all the time.
Only thing to be careful:
inputimpedance should be the same for both.
If you have 2x 10k as R15,16, and a source impedance
lets say 1k at R15 then put 1k between R16 and ground.
Uli
PS: this 1k isn´t absolutely necessary but you might get
some DC offset at both outputs.
PPS: no the outputs should be at zero to gnd.
This is the reason for R21, 22 and R5, 6
Hi Nar,
Yes of course it IS possible!!!
You can feed whatever you want!
Output stays symmetric all the time.
Only thing to be careful:
inputimpedance should be the same for both.
If you have 2x 10k as R15,16, and a source impedance
lets say 1k at R15 then put 1k between R16 and ground.
Uli
PS: this 1k isn´t absolutely necessary but you might get
some DC offset at both outputs.
PPS: no the outputs should be at zero to gnd.
This is the reason for R21, 22 and R5, 6
DC at the outputs ?If - input not terminated with right impedance when not used?
Damn.If you plan to use the X module as a preamp , how can you be sure that all your sources impedances has same value?
For my CD,1k perhaps but I don't know ... my DVD audio seems 200 ohms but I can't bet the farm on it ...other sources ? Honnestly I don't know....
Or to get rid of hazardous DC at the output ...why not capacitively couple the inputs ???
The question makes me extremely passionate .
If I want to drive an Aleph X with 25 k differential impedance,and have a 20 dB gain ...I am a little lost....how to calculate proper feedback resistor value ?
The thirteen devils
Best regards
Anael
Damn.If you plan to use the X module as a preamp , how can you be sure that all your sources impedances has same value?
For my CD,1k perhaps but I don't know ... my DVD audio seems 200 ohms but I can't bet the farm on it ...other sources ? Honnestly I don't know....
Or to get rid of hazardous DC at the output ...why not capacitively couple the inputs ???
The question makes me extremely passionate .
If I want to drive an Aleph X with 25 k differential impedance,and have a 20 dB gain ...I am a little lost....how to calculate proper feedback resistor value ?
The thirteen devils
Best regards
Anael
DC-issue
Hi Nar,
When you directly couple a pre you have this problem all the time.
Use either good caps (Mundorf or the like) or DC-servo.
PS: we are not talking about serious DC values. Every direct
coupled pre produces offset depending on inputimpedance.
Uli
To drive an X-Amp you better use the aforementioned source-
followers at the output to keep gain load independant!
U.
Hi Nar,
When you directly couple a pre you have this problem all the time.
Use either good caps (Mundorf or the like) or DC-servo.
PS: we are not talking about serious DC values. Every direct
coupled pre produces offset depending on inputimpedance.
Uli
To drive an X-Amp you better use the aforementioned source-
followers at the output to keep gain load independant!
U.
Hello Uli ,
I tried to understand the PCB version may 28 2003 but I am lost with In/Out connectors.
I suppose the one on upper part is for supply rails but I can't guess where V+ and V- are !!! At least the one in the middle is Ground ....
Second problem , the 2 connectors downside of the board . There is probably a mistake 
As no plot connection of the 6 plots downside is a ground
And the -in and + in are clear for me , in the middle of the board
But the outputs on the schematic are common point between R 11/13/17/21 ( or R 12/14 18/22 ) and on the bottom part of the layout , they are well alltogether but no link to output connectors and on the top plane neither
So where is the error ???
I want to try your module with output followers to have possibility to test it as a single preamp . So I propose to add Jfets output followers,conveniently self biasing ( in that case another 2sk389 per output will do the job . To add it here are the mods I propose :
R 19/20 go to 47 k instead of 100k
R 15/16 go to 10k
R 17/18 adjusted to get proper gain figure
Removal of R 21/22 or shall I let them in place ?
Output goes to the gate of 1/2 2sk389 ,its drain at V + , a small resistor (22 ohms) from its source pin to output of the final thing . Second half of 2sk 389 mounted as a current source between output and V-, gate at V- and source pin goes to V- through small resistor (22 ohms)
The DC at the outputs should be very small , i.e. 2sk389 is an already matched unit
The feedback resistor ( R 17/18 ) ..... crucial point , shall I leave it in place or shall I plug it real output from the follower and go to gates of the initial 2sk 389 ????
If it works.... this would be a total Jfet gain stage , as the cascodes are not considered to give voltage gain ... 
What do you think of the ideas ? Please answer I have to build the unit !!!!
Best regards
Anael
I tried to understand the PCB version may 28 2003 but I am lost with In/Out connectors.
I suppose the one on upper part is for supply rails but I can't guess where V+ and V- are !!! At least the one in the middle is Ground ....
Second problem , the 2 connectors downside of the board . There is probably a mistake
As no plot connection of the 6 plots downside is a ground
And the -in and + in are clear for me , in the middle of the board
But the outputs on the schematic are common point between R 11/13/17/21 ( or R 12/14 18/22 ) and on the bottom part of the layout , they are well alltogether but no link to output connectors and on the top plane neither
So where is the error ???
I want to try your module with output followers to have possibility to test it as a single preamp . So I propose to add Jfets output followers,conveniently self biasing ( in that case another 2sk389 per output will do the job . To add it here are the mods I propose :
R 19/20 go to 47 k instead of 100k
R 15/16 go to 10k
R 17/18 adjusted to get proper gain figure
Removal of R 21/22 or shall I let them in place ?
Output goes to the gate of 1/2 2sk389 ,its drain at V + , a small resistor (22 ohms) from its source pin to output of the final thing . Second half of 2sk 389 mounted as a current source between output and V-, gate at V- and source pin goes to V- through small resistor (22 ohms)
The DC at the outputs should be very small , i.e. 2sk389 is an already matched unit
The feedback resistor ( R 17/18 ) ..... crucial point , shall I leave it in place or shall I plug it real output from the follower and go to gates of the initial 2sk 389 ????
If it works....
this would be a total Jfet gain stage , as the cascodes are not considered to give voltage gain ... What do you think of the ideas ? Please answer I have to build the unit
!!!!Best regards
Anael
Dear Anael,nar said:
I suppose the one on upper part is for supply rails but I can't guess where V+ and V- are !!! At least the one in the middle is Ground ....
Second problem , the 2 connectors downside of the board . There is probably a mistake
As no plot connection of the 6 plots downside is a ground
And the -in and + in are clear for me , in the middle of the board
But the outputs on the schematic are common point between R 11/13/17/21 ( or R 12/14 18/22 ) and on the bottom part of the layout , they are well alltogether but no link to output connectors and on the top plane neither
So where is the error ???
I want to try your module with output followers to have possibility to test it as a single preamp . So I propose to add Jfets output followers,conveniently self biasing ( in that case another 2sk389 per output will do the job . To add it here are the mods I propose :
R 19/20 go to 47 k instead of 100k
R 15/16 go to 10k
R 17/18 adjusted to get proper gain figure
Removal of R 21/22 or shall I let them in place ?
Output goes to the gate of 1/2 2sk389 ,its drain at V + , a small resistor (22 ohms) from its source pin to output of the final thing . Second half of 2sk 389 mounted as a current source between output and V-, gate at V- and source pin goes to V- through small resistor (22 ohms)
The DC at the outputs should be very small , i.e. 2sk389 is an already matched unit
The feedback resistor ( R 17/18 ) ..... crucial point , shall I leave it in place or shall I plug it real output from the follower and go to gates of the initial 2sk 389 ????
If it works....
What do you think of the ideas ? Please answer I have to build the unit
Best regards
Anael
sorry for causing you any inconvenience!
Here we go:
CON3 pin1 = +rail, pin2 = gnd, pin3 = -rail
CON1 pin1 = +drive-, pin2 = +drive+, pin3 = -input
CON2 pin1 = +input, pin2 = -drive+, pin3 = -drive-
The drive pins are designed to be directcoupled to source-
followers gates.Therefore leave the R11, R12, R13, R14
untouched.
You are right there are no outputpins as the unit should be
coupled to sourcefollowers.
I would leave R17,18 untouched and put some FB from the
junction of the sourceresistors of the external sourcefollowers
back to the respective inputs.
I would NOT make single ended followers. Better is to use
1 2SK389 and his complementary part 2SJ109.
Lastly, do not put the fb resistor directly to the gate.
Leave the 100R (+100k) untouched and put an input-fb
network (eg 10k, 100k) outside the unit.
Uli
Hi Uli ,
Well integrated your thoughts ! Thank you very much indeed !
So leads me to a question never answered ( for me ! )
WHY THE USE OF SINGLE-ENDED FOLLOWERS FORBIDDEN ???
I have already seen and asked this question in the forums.
For a power amplifier , which is best designed to deliver voltage and current into a load that is never flat,for resistance and impedance,complementary followers seem to do the best job possible . In fact , if we look at the X series , we see that they go into class B after current bias level is reached .
But in the application here , the use of the module for a preamp is given to work on a load which is consistent across the audio band.Given the fact that a single ended follower always works in class A , why not use it ?
................................
And I remember a balanced single ended preamp by Borbely , which uses a differential 2sk389 cascoded coupled to single-ended output followers...and no global feedback , only local ( resistors on source pins of the 2sk389 , local degeneration ) .
Why ? 
Well integrated your thoughts ! Thank you very much indeed !
So leads me to a question never answered ( for me ! )
WHY THE USE OF SINGLE-ENDED FOLLOWERS FORBIDDEN ???
I have already seen and asked this question in the forums.
For a power amplifier , which is best designed to deliver voltage and current into a load that is never flat,for resistance and impedance,complementary followers seem to do the best job possible . In fact , if we look at the X series , we see that they go into class B after current bias level is reached .
But in the application here , the use of the module for a preamp is given to work on a load which is consistent across the audio band.Given the fact that a single ended follower always works in class A , why not use it ?
................................
And I remember a balanced single ended preamp by Borbely , which uses a differential 2sk389 cascoded coupled to single-ended output followers...and no global feedback , only local ( resistors on source pins of the 2sk389 , local degeneration ) .
Why ?
nar said:
Hi Anael,
Of course it is not forbidden!
I personnally prefer complementary followers in line stages
as the output can source and sink same current -> symmetry!
A SE follower normally can source much more current than sink.
Secondly the - half of a signal shows much more distortion than
the positive one because of the non linear behaviour of the
source resistor.
But IMO its absolutely up to your choice!
Uli
PS: I would not consider a feedback across one gainstage to be
global feedback.
Uli ,
Thanks for sharing your knowledge , that is GREAT !!!!
After reflections , I think I will go for the complementary followers .
A N channel positive and P channel negative .
As I have IRF 610s and 9610s in stock , I would be able to do it .
I think , accordingly to your efforts !!! to have a 0V DC output at the end of the module , if I match VGS for same value on 610 and 9610 , DC should be low .
I think I can trust the Alephono output stage !! which works with 1 610 and 1 9610 . I matched them , and output is very near to 0V
What current ? that's all the question , nice ...
I think for 24V rails , I should go for 12mA . Seems to be enough for a line transmission . So correct me if I am wrong ,
The 820 R on your schematic should give at 5mA /side a 4,1 V voltage drop,each side and positive /negative.
So , in the most perfect of all cases,using a 610 and a 9610 matched at 4V Vgs ( or less...) should do it . 221 gate stoppers,etc.
Now the source resistor, don't know how to choose its value but
at 12mA current, with a 4V Vgs should leave a 0,1V across this resistor ,so at 12mA gives a 8,3 ohms resistor on each side.And output should sit at 0V .
If i have problems , thermal runnaway ... what to do ?
Regards
Anael
PS Why not use some adjustable resistors ,small ppm for Temperature,in place of the 820 ohms . Should give control over bias pos/neg and drift
Thanks for sharing your knowledge , that is GREAT !!!!
After reflections , I think I will go for the complementary followers .
A N channel positive and P channel negative .
As I have IRF 610s and 9610s in stock , I would be able to do it .
I think , accordingly to your efforts !!! to have a 0V DC output at the end of the module , if I match VGS for same value on 610 and 9610 , DC should be low .
I think I can trust the Alephono output stage !! which works with 1 610 and 1 9610 . I matched them , and output is very near to 0V
What current ? that's all the question , nice ...
I think for 24V rails , I should go for 12mA . Seems to be enough for a line transmission . So correct me if I am wrong ,
The 820 R on your schematic should give at 5mA /side a 4,1 V voltage drop,each side and positive /negative.
So , in the most perfect of all cases,using a 610 and a 9610 matched at 4V Vgs ( or less...) should do it . 221 gate stoppers,etc.
Now the source resistor, don't know how to choose its value but
at 12mA current, with a 4V Vgs should leave a 0,1V across this resistor ,so at 12mA gives a 8,3 ohms resistor on each side.And output should sit at 0V .
If i have problems , thermal runnaway ... what to do ?
Regards
Anael
PS Why not use some adjustable resistors ,small ppm for Temperature,in place of the 820 ohms . Should give control over bias pos/neg and drift
adjust
Hi Nar,
As I am not a big fan of trimmers, I would go with 1k1 istead of
the 820R and use some bigger source resistors. The advantage
of this is the reduced thermal runaway caused by the local
feedback of the source resistors. Output impedance is not a real
issue in this case because it is determined much more by overall
feedback.
Why don´t you stay with 2SK389 (& 2SJ109) as outputfollowers?
In this case you can swap the 820R with e.g. 390R thus having
more undistorted voltage swing at the output?
Uli
Hi Nar,
As I am not a big fan of trimmers, I would go with 1k1 istead of
the 820R and use some bigger source resistors. The advantage
of this is the reduced thermal runaway caused by the local
feedback of the source resistors. Output impedance is not a real
issue in this case because it is determined much more by overall
feedback.
Why don´t you stay with 2SK389 (& 2SJ109) as outputfollowers?
In this case you can swap the 820R with e.g. 390R thus having
more undistorted voltage swing at the output?
Uli
OK Uli got it Now , I understand . And I am fond of extension to this design . Basically , with the X modul it is perfectly possible to make a real X amp , very close to the Pass product line except output power
But I look after a way to get the trick .
Possibly have power mosfets at the output , I think about IRFP 240 / 9240 as followers to make this real amp get it to real speakers . Possibly I will go for a test version .
In that case we will have to get the cascodes to run a greater current to drive the power followers . 5mA doesn't seem enough,I guess raise it to 10 or 15 mA and the thing would work better
Will have
4 amp total bias in the followers ( 2 A per s.e.)
8 mos each side, 4 240 and 4 9240 running at 500mA each /
25V supply , and 0,22 ohm source resistors.
If i understand the trick , we will try to raise gate voltage of the followers slowly as the amp heats to get 110 mV on those source resistors .
But I think thermal stability will not be optimal vs room temperature , if adjusted at 22 degrees , at 28 degrees in summer bias would be changed
in fact any time as room temp is never the same How to get rid of thermal runnaway ? Do you have any ideas ?
Moreover , as in the AlephX possibly some DC at the outputs.A resistor between outputs to sources of the 2sk389 differential will do the trick? I am not sure ....mmmmmhhhhh...possibly work....
or a DC servo .....as yours . Could you send me schematic on my e-mail ???
Last thing:How to calculate output power in class A of this type of X amp ?vs closed loop gain and output idle and power supply rails?
How to determine the precise point where the amp starts to go into class B ?
If you could give me the method to calculate it I would be so pleased
Well , the amp would be awsome...wouldn't it ?
Regards
response
Hi Moe,
actually I am running those X-es as bottom end amps in a
bi-amped configuration.
As I am on the way to finish my AX100 accompanied by kind
of MAXX clones which should be finished in about one month
I won´t run them bi-amped anymore as they will have a
serial second order 3-way XO. I will use the X200 for driving
the surround channels from that point on.
They are VERY powerful with fantastic lo end but I think I
like the AX-es better as fullrange amps.
A word to response:
Sometimes you have to be very patient, especially when you
try to put in something new in a place like this being a "newbie"
there
The X-modul thread died and months later somebody digged it
out of the dust, maybe it shall take this time that people
recognize the interesting points...
Uli
Hi Moe,
actually I am running those X-es as bottom end amps in a
bi-amped configuration.
As I am on the way to finish my AX100 accompanied by kind
of MAXX clones which should be finished in about one month
I won´t run them bi-amped anymore as they will have a
serial second order 3-way XO. I will use the X200 for driving
the surround channels from that point on.
They are VERY powerful with fantastic lo end but I think I
like the AX-es better as fullrange amps.
A word to response:
Sometimes you have to be very patient, especially when you
try to put in something new in a place like this being a "newbie"
there
The X-modul thread died and months later somebody digged it
out of the dust, maybe it shall take this time that people
recognize the interesting points...
Uli
Hi Uli ,
Your work on the x 200 is fantastic of course but a little too complicated for me for now . Thanks for the tips , and I perhaps will try it next year ...
For now I stick in mind using your X module to drive output followers as a test for a mini X .
First I will add 2 output followers , with the combination IRFP 240 / 9240 , with 0,22 ohm source resistors . Bias of 1 amp on 25 V for each . And see what happens ...
But later on adding some more mos parallel on the followers , I think I will get more output current to drive speakers better .
What i need is your advice to increase current through the cascode/out sect. of your X module . Maybe need to change some resistors , ie R 9 / R 10 ?And perhaps R 5 / R 6 and trimpots? What to do here ? Do the 549 / 559 stand a 20 mA current ? I think i require it to drive some power mos parallel as followers ....
I defenitely will use trimpots in place of R 11/12/13/14 , of 2k multiturn low temp drift . And perhaps . ... add some 47k5 resistors from output node of the followers to resistor gates of the 240/9240 I will use . Do you think it will help for bias stability ? Nelson used them in the Zen V5 ... ???Anyway,I want to use the output followers without feedback to the node R12/14/18 . As the existing X series ... I will leave your feedback resistor in place . I think I will go for 4 A bias per SE on a larger protoype , little by little ...
What I need for now is a little help to study a workable prototype , perfection will come later
Thanks for reply ,
Anael
Your work on the x 200 is fantastic of course but a little too complicated for me for now . Thanks for the tips , and I perhaps will try it next year ...
For now I stick in mind using your X module to drive output followers as a test for a mini X .
First I will add 2 output followers , with the combination IRFP 240 / 9240 , with 0,22 ohm source resistors . Bias of 1 amp on 25 V for each . And see what happens ...
But later on adding some more mos parallel on the followers , I think I will get more output current to drive speakers better .
What i need is your advice to increase current through the cascode/out sect. of your X module . Maybe need to change some resistors , ie R 9 / R 10 ?And perhaps R 5 / R 6 and trimpots? What to do here ? Do the 549 / 559 stand a 20 mA current ? I think i require it to drive some power mos parallel as followers ....
I defenitely will use trimpots in place of R 11/12/13/14 , of 2k multiturn low temp drift . And perhaps . ... add some 47k5 resistors from output node of the followers to resistor gates of the 240/9240 I will use . Do you think it will help for bias stability ? Nelson used them in the Zen V5 ... ???Anyway,I want to use the output followers without feedback to the node R12/14/18 . As the existing X series ... I will leave your feedback resistor in place . I think I will go for 4 A bias per SE on a larger protoype , little by little ...
What I need for now is a little help to study a workable prototype , perfection will come later
Thanks for reply ,
Anael
mods
Hi Anael,
1) I would go for IRFP240/ IRFP9140 because of better
resistance and transconductance matching.
2) You can alter the bias of the BCs by changing their
emitter resistors BUT take always the SAME factor!!
Eg: multiply the resistorvalues with 2 halves the current.
multiply the resistorvalues with 0.5 doubles the current.
3) The BCs can stand about 500mW until they reach the termal
limit. So going for not more than 200mW keeps you on the
safe side.
Uli
Hi Anael,
1) I would go for IRFP240/ IRFP9140 because of better
resistance and transconductance matching.
2) You can alter the bias of the BCs by changing their
emitter resistors BUT take always the SAME factor!!
Eg: multiply the resistorvalues with 2 halves the current.
multiply the resistorvalues with 0.5 doubles the current.
3) The BCs can stand about 500mW until they reach the termal
limit. So going for not more than 200mW keeps you on the
safe side.
Uli
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.