Mosfet Frontend Troubles

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Here is my Mosfet frontend. I am having trouble in testing. Q3 and Q4 get very hot! I measured 2.0 volts across R5 which means it is passing about 10mA. What is happening? Do i need to load the drives, and or tie the Fb in? I think it is possibly oscillating.

Please help with this!

DonS
 

Attachments

  • a40.jpg
    a40.jpg
    97.4 KB · Views: 1,301
The one and only
Joined 2001
Paid Member
Check the voltage across R16. If each LED is 4 volts,
then you will have something around 8 volts across
R16, meaning that about 80 mA is passing through the
output devices, and at 75V rails, that's about 6 watts
per device, which is quite a bit.

:cool:
 
Official Court Jester
Joined 2003
Paid Member
Don S said:
Here is my Mosfet frontend. I am having trouble in testing. Q3 and Q4 get very hot! I measured 2.0 volts across R5 which means it is passing about 10mA. What is happening? Do i need to load the drives, and or tie the Fb in? I think it is possibly oscillating.

Please help with this!

DonS


use just one led string for both Q3 and Q4

also - you can reference leds to +Ub (recalculated R6 or R14 to +75V, not gnd)
 
Official Court Jester
Joined 2003
Paid Member
Don S said:
R14 is now open, and Q4 appears to be shorted.

I think Q6 was full on!

DonS


first- make Iq through Q4 fixed , for start
then (besides already mentioned idea to share led string for both CCS biases) recheck value of R1 ( you need ~ 4V Ugs for open throttle of Q6......but you don't need much more) so - make R5 from one fixed resistor and trimpot in parallel- so you can tweak " openess" of Q6.

tweak that value looking at other DC voltages in circ

DC potentials of drive points are nothing more than result of transposed DC characteristics of Q5 thingie (some called it Vbe multiplier probably,but I'm dumb for that ) .

btw- sharing bias net for two CCSes on same side is benefitial from more than one reason, but I'll mention just one- less temp-co induced differences
 
ZenMod, if I just tie the gate resistors of Q3 and Q4 I should get the same effect as just one voltage source. Correct?

Thanks, jj. I will be even nicer when it works!

How can I calculate the best voltage on the gate of Q6. By my calculations with the 10mA current source I have 7.5 volts across R1. Is this right or have I been looking at this too long?

Someone please explain how to calculate for a balanced voltage across Q6 and Q4. Is my thinking wrong or should I not be looking for a small voltage above and below ground for drive+ and drive- respectively?

Mr. Pass, please help! After all you got me into this hobby with the A40! ;) ;)

DonS
 

Attachments

  • camera extact_4-30-05 012.jpg
    camera extact_4-30-05 012.jpg
    90.3 KB · Views: 842
Official Court Jester
Joined 2003
Paid Member
Don S said:
ZenMod, if I just tie the gate resistors of Q3 and Q4 I should get the same effect as just one voltage source. Correct?

Thanks, jj. I will be even nicer when it works!

How can I calculate the best voltage on the gate of Q6. By my calculations with the 10mA current source I have 7.5 volts across R1. Is this right or have I been looking at this too long?

Someone please explain how to calculate for a balanced voltage across Q6 and Q4. Is my thinking wrong or should I not be looking for a small voltage above and below ground for drive+ and drive- respectively?

Mr. Pass, please help! After all you got me into this hobby with the A40! ;) ;)

DonS

for starters......look at attached picture- so you can see bias arrangement

first thing first........


1. you shoot at desired current through driver stage adjusting CCS with Q4.
2. implementing pot with r5 you will adjust "openess" of Q6 , so desired current will flow through string consisted of Q4,Q5 and Q6;
as a rule of thumb- you need around 4V between gate and source of Q6........counting that you have current through LTP~9mA,so you'll have U across r1 ~6,78V..............
now- current through driver stage is presumed with :
((3x1V9)-4V)/100 =17mA

if you have 17mA flowing through R2 - that voltage is almost negligible for Q6's Ugs (voltage across R2) calculations , so - it's easy to see that you have too much Ugs voltage for Q6........ shoot for something as 1K for R1.

back and forth.......back and forth...........

to have drive potentials around 4V above and below ground ...........you must open enough Q5........so that DC voltage across it will be ~ 8V

exact position of that virtual ground (final DC offset) you can tweak later with mentioned pot in parallel with R5 ............proposal- use 390E as R5 and 1k pot across it..........

I hope that all this mumbling helps you
 

Attachments

  • a40.jpg
    a40.jpg
    71.3 KB · Views: 827
ZenMod, I wanted and now have 10mA through the longtail pair. If I go with 1k for R1 I will have ~5 volts on the gate of Q6.

This is what I was thinking!
Right now with R1 at 1k5 I will have ~7.5 volts on Q6 gate.

My question now is there a way to calculate R1 and or R2 to balance the VAS stage so that the voltages source to drain of Q6 and Q4 are close.

DonS

How do you include the post you are answering in your post?
 
Official Court Jester
Joined 2003
Paid Member
Don S said:
ZenMod, I wanted and now have 10mA through the longtail pair. If I go with 1k for R1 I will have ~5 volts on the gate of Q6.

This is what I was thinking!
Right now with R1 at 1k5 I will have ~7.5 volts on Q6 gate.

My question now is there a way to calculate R1 and or R2 to balance the VAS stage so that the voltages source to drain of Q6 and Q4 are close.

DonS

How do you include the post you are answering in your post?

I told you how you'll calc R1 ; R2 is nothing else than implementing sorta folded cascode , comprised of Q2 and Q6 , so leave it as is ;

now- what you wanna change ( and understand first) is role of Q5.....imagine it as variable resistor between Q4 and Q6 ;

you have current flowin' from Q6 to Q4 through that variable resistor. depending how big tahat resistor is in value , you'll have more or less or exactly needed 8V across it.

with variable R5 you will tweak current through LTP, so through R1 and with that - slightly varying where on voltage ladder Q6 will be , and with all that - you'll tweak DC offset on output of amp......

I hope that this overcomplicated and oversimplified explanation is good enough for you......
 
Zen Mod said:


I told you how you'll calc R1 ; R2 is nothing else than implementing sorta folded cascode , comprised of Q2 and Q6 , so leave it as is ;

now- what you wanna change ( and understand first) is role of Q5.....imagine it as variable resistor between Q4 and Q6 ;

you have current flowin' from Q6 to Q4 through that variable resistor. depending how big tahat resistor is in value , you'll have more or less or exactly needed 8V across it.

with variable R5 you will tweak current through LTP, so through R1 and with that - slightly varying where on voltage ladder Q6 will be , and with all that - you'll tweak DC offset on output of amp......

I hope that this overcomplicated and oversimplified explanation is good enough for you......


Thanks ZenMod!

Q5 is fully understood!

What I am asking is, we should be looking for equal voltage across drain to source of both Q6 and Q4, correct?

Or maybe better said is drain to + rail for Q6 and drain to - rail for Q4 should be equal and opposite, correct?

DonS
 
Official Court Jester
Joined 2003
Paid Member
Don S said:


...........

What I am asking is, we should be looking for equal voltage across drain to source of both Q6 and Q4, correct?

Or maybe better said is drain to + rail for Q6 and drain to - rail for Q4 should be equal and opposite, correct?

DonS


nope;

what you need is ............

look at this this way:

Q4 is CCS; it demands current (which you choose) .

then you must open enough faucet comprised of Q6 just enough to pass current demanded from Q4 . this is your starting point (around 4V Ugs) ;

then- you must set variable resistor (comprised of Q5 and surrounding trivia) to have ~ 8Vdc across it ( exactly 2x4V, needed to further open output mosfets) ;

then - slightly varying R5+trim combo you'll ballance virtual gnd pole , situated between + and - drive points


imagine Q4 as fixed CCS (and that is exactly) , Q5 as settable variable resistor and Q6 as "active" variable resistor,steered with current through left side of LTP.............
to answer on Q - for Q6 and Q4 Ugs-es will be same only if they are strictly matched.........and there is no need for that, and that isn't your goal at all

good enough?
 
Zen Mod said:



nope;

what you need is ............

look at this this way:

Q4 is CCS; it demands current (which you choose) .

then you must open enough faucet comprised of Q6 just enough to pass current demanded from Q4 . this is your starting point (around 4V) ;

then- you must set variable resistor (comprised of Q5 and surrounding trivia) to have ~ 8Vdc across it ( exactly 2x4V, needed to further open output mosfets) ;

then - slightly varying R5+trim combo you'll ballance virtual gnd pole , situated between + and - drive points


imagine Q4 as fixed CCS (and that is exactly) , Q5 as settable variable resistor and Q6 as "active" variable resistor,steered with current through left side of LTP.............

good enough?


ZenMod, we are on the same page. We are just telling the story a different way.

Please see my attached full schematic. No fets for drivers or outputs. Just good old bjt's!

My other concerns are my fb arrangement, taken from the drivers, not the outputs. Like Threshold. Also the A40 bias tracking from the output emitter resistor.

Have a look and let me know what you think!

It also includes your suggested changes. More or less:)
 

Attachments

  • untitled-1 copy.jpg
    untitled-1 copy.jpg
    96.6 KB · Views: 780
Official Court Jester
Joined 2003
Paid Member
Don S said:



ZenMod, we are on the same page. We are just telling the story a different way.

Please see my attached full schematic. No fets for drivers or outputs. Just good old bjt's!

My other concerns are my fb arrangement, taken from the drivers, not the outputs. Like Threshold. Also the A40 bias tracking from the output emitter resistor.

Have a look and let me know what you think!

It also includes your suggested changes. More or less:)


I like what I see.........but you must count on fact that my knowledge is limited ;)

anyway- my comment is- be it just a fashion or not........ replace input mosfets with J309 or J310 jfets ,cascoded;

I'm pretty sure that sound will be better.

you can see example (where else ? :devilr: ) than in Pumpkin thread........
http://www.diyaudio.com/forums/showthread.php?postid=1247745#post1247745
 
Zen Mod said:



I like what I see.........but you must count on fact that my knowledge is limited ;)

anyway- my comment is- be it just a fashion or not........ replace input mosfets with J309 or J310 jfets ,cascoded;

I'm pretty sure that sound will be better.

you can see example (where else ? :devilr: ) than in Pumpkin thread........
http://www.diyaudio.com/forums/showthread.php?postid=1247745#post1247745


ZenMod, thanks for the info. I went with the 610's so that I could keep it simple. Maybe next update will include the jfets and cascode. Unless you know of a jfet with 200 volt rating!

Thanks, DonS
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.