well - even if truce - back to sheeps 
over at Analog fora , in Deutsch playground , hb asked me to write few lines - how
and Shunty works ........ I do that , and then I thought that maybe someone here can be interested in that...
so - HOW PUMPKIN WORKS (don't ask me why it works ........
) :
main part - in fact - engine of Pumpkin is differential pair ( LTP - Long Tailed Pair) , made of Q1 and Q2 ; they are purposely chosen to be J309 or j310 - easy to find and with decent characteristics ;
toob electronic is my main passion ; I learned with Roehren that you must have enough current through stage to achieve decent driving characteristics , in other words - that stage is capable enough to drive next stage .... same story here - I choose J310 as replacement for 2SK389V , or pair of 2SK170V - most current capable sane jfets
I hope that I don't need to write how signal enters LTP and how's amplified ...... on both sides - at R7 and R8 as left and right loads ....
both mentioned jfets in LTP are cascoded with plain vanila BC critters - probably most often used BC critters after BC109 and similar ones ......... reason for cascoding is twofold - preserve tiny jfets from excessive voltage ( thus dissipation also) and to fight with Herr Miller's ugly head ( that's Herr connected with reflecting interelectrode capacitance of jfet via amplification factor ; when Uds is constant - we don't have variations in reflected (read- amplified) capacitance , also known as Miller's effect) .
LTP cascode is biased ( base of both Q3 and Q4 is placed at desired voltage potential) via R20-R21 string ; possible variations are filtered with C3.
current through entire LTP is set by CCS , formed of Q9 and Q10 ; I choose ~16mA , and that is ~ 7mA through left side of LTP,7mA through right side of LTP and ~1mA through cascode biasing string (look up to see what's that ) .
that CCS is biased ( his own DC conditions ) with D1 to D6 green led string ; each led have ~1,93V across it , at 5mA current , which is set with helper "micro" CCS , made of Q11+R24 . led's are enough silent for that purpose .
Q10 is cascode of main CCS part (Q9 ) , augmenting Q9's dynamic resistance ; it's biased (read - open full throttle) with D1-D5 string , which is standing on D6 shoulders.
so - small info regarding Q9 current - Ube = 0,65V typical ; voltage across D6 is 1,93V ; so - voltage across R22+WR2 is 1,93-0,65 = 1V28 ......... so - 1V28/17mA = 75ohms ( don't try to connect TV antenna there .... hehe)
same story goes for seting current through "micro" CCS - target is 5mA ....... resistor is 270E - so - calculate what's Ugs
now - look at Q6 and Q8 ; they are nothing more and nothing less than output CCS-es , left and right ,respectively;
they are veeeery cleverly ( ZM is also often known as Baldrick) biased along with Q10 - with same green LED string ;
so - 6 x Uled = 6x1V93 = 11V58 ; Ugs for these mosfets is in range of 4V ; now - we can calc what resistance we need in Q6 and Q8 sources to have ~ 30mA through each ..... 11V58-4V = 7V58/30mA = 252E , so I choose 470 and 680 ohms in parallel (think also about dissipation- so they are two) , giving ~277E ..... life is sometimes different than paper , and real mosfets are always different than paper ones
......... read - typical Ugs of 4V is often less than 4V .....
on top of each output CCS ( Q6 and Q8 ) are placed real working output mosfets - Q5 and Q7 ;
they are biased ( think again about magical number of 4V between G and S ) with currents flowing through R7 (for Q5 biasing) and R8 ( for Q7 biasing) ;
when your precious signal enter input LTP, current through these two resistors is going up and down , modulating gates of Q5 and Q7 ;
important thing is - blabla - you often read about "common source,common drain,common gate" stages of mosfets;
thing is simple - just forget that outputs are in common source connection ; look at that this way - for each output mosfet's
drain you have load - made of output CCS ; Q5's drain see Q6 CCS as load ; same- Q7's drain sees Q8 CCS as load ;
you must remember - when mosfet have load in drain, then that stage have gain , too ;
just to mention - R16 and R11 are source degeneration resistors - popular way of slightly decreasing gain of stage and achieving slightly better ....... everything
; also observe - role of WR1 - it's there for slight source degeneration of input jfets and also for setting DC symmetry - when needed .........
huh .......
further :
observe where is +input , where is -input, where is + output and where is - output ; look good where is each half gain contribution ........ and also look at few more things :
gain of entire stage is set with ratio of R1/R5 and R3/R6 ; you see that this approach is same as in so called - inverting Operational Amplifier stage .... one resistor inline with signal,one resistor taking signal back from output .....
specificum of this gadget is exactly where is taken drive for output stages - cross coupled !!
that's Super Symmetric Papa's thingie ...... think little about that ,and do not underestimate role of Tail in LTP ...... it's crucial for Super Symmetric effect , at least in Pumpkin .
I know that you'll have questions , so I will not even bother to write more , in this post ........
over at Analog fora , in Deutsch playground , hb asked me to write few lines - how
and Shunty works ........ I do that , and then I thought that maybe someone here can be interested in that...so - HOW PUMPKIN WORKS (don't ask me why it works ........
) :main part - in fact - engine of Pumpkin is differential pair ( LTP - Long Tailed Pair) , made of Q1 and Q2 ; they are purposely chosen to be J309 or j310 - easy to find and with decent characteristics ;
toob electronic is my main passion ; I learned with Roehren that you must have enough current through stage to achieve decent driving characteristics , in other words - that stage is capable enough to drive next stage .... same story here - I choose J310 as replacement for 2SK389V , or pair of 2SK170V - most current capable sane jfets
I hope that I don't need to write how signal enters LTP and how's amplified ...... on both sides - at R7 and R8 as left and right loads ....
both mentioned jfets in LTP are cascoded with plain vanila BC critters - probably most often used BC critters after BC109 and similar ones ......... reason for cascoding is twofold - preserve tiny jfets from excessive voltage ( thus dissipation also) and to fight with Herr Miller's ugly head ( that's Herr connected with reflecting interelectrode capacitance of jfet via amplification factor ; when Uds is constant - we don't have variations in reflected (read- amplified) capacitance , also known as Miller's effect) .
LTP cascode is biased ( base of both Q3 and Q4 is placed at desired voltage potential) via R20-R21 string ; possible variations are filtered with C3.
current through entire LTP is set by CCS , formed of Q9 and Q10 ; I choose ~16mA , and that is ~ 7mA through left side of LTP,7mA through right side of LTP and ~1mA through cascode biasing string (look up to see what's that
) .that CCS is biased ( his own DC conditions ) with D1 to D6 green led string ; each led have ~1,93V across it , at 5mA current , which is set with helper "micro" CCS , made of Q11+R24 . led's are enough silent for that purpose .
Q10 is cascode of main CCS part (Q9 ) , augmenting Q9's dynamic resistance ; it's biased (read - open full throttle) with D1-D5 string , which is standing on D6 shoulders.
so - small info regarding Q9 current - Ube = 0,65V typical ; voltage across D6 is 1,93V ; so - voltage across R22+WR2 is 1,93-0,65 = 1V28 ......... so - 1V28/17mA = 75ohms ( don't try to connect TV antenna there .... hehe)
same story goes for seting current through "micro" CCS - target is 5mA ....... resistor is 270E - so - calculate what's Ugs
now - look at Q6 and Q8 ; they are nothing more and nothing less than output CCS-es , left and right ,respectively;
they are veeeery cleverly ( ZM is also often known as Baldrick) biased along with Q10 - with same green LED string ;
so - 6 x Uled = 6x1V93 = 11V58 ; Ugs for these mosfets is in range of 4V ; now - we can calc what resistance we need in Q6 and Q8 sources to have ~ 30mA through each ..... 11V58-4V = 7V58/30mA = 252E , so I choose 470 and 680 ohms in parallel (think also about dissipation- so they are two) , giving ~277E ..... life is sometimes different than paper , and real mosfets are always different than paper ones
......... read - typical Ugs of 4V is often less than 4V .....on top of each output CCS ( Q6 and Q8 ) are placed real working output mosfets - Q5 and Q7 ;
they are biased ( think again about magical number of 4V between G and S ) with currents flowing through R7 (for Q5 biasing) and R8 ( for Q7 biasing) ;
when your precious signal enter input LTP, current through these two resistors is going up and down , modulating gates of Q5 and Q7 ;
important thing is - blabla - you often read about "common source,common drain,common gate" stages of mosfets;
thing is simple - just forget that outputs are in common source connection ; look at that this way - for each output mosfet's
drain you have load - made of output CCS ; Q5's drain see Q6 CCS as load ; same- Q7's drain sees Q8 CCS as load ;
you must remember - when mosfet have load in drain, then that stage have gain , too ;
just to mention - R16 and R11 are source degeneration resistors - popular way of slightly decreasing gain of stage and achieving slightly better ....... everything
; also observe - role of WR1 - it's there for slight source degeneration of input jfets and also for setting DC symmetry - when needed .........huh .......
further :
observe where is +input , where is -input, where is + output and where is - output ; look good where is each half gain contribution ........ and also look at few more things :
gain of entire stage is set with ratio of R1/R5 and R3/R6 ; you see that this approach is same as in so called - inverting Operational Amplifier stage .... one resistor inline with signal,one resistor taking signal back from output .....
specificum of this gadget is exactly where is taken drive for output stages - cross coupled !!
that's Super Symmetric Papa's thingie ...... think little about that ,and do not underestimate role of Tail in LTP ...... it's crucial for Super Symmetric effect , at least in Pumpkin .
I know that you'll have questions , so I will not even bother to write more , in this post ........
when I'm at it ........ Shunty :
HOW SHUNTY WORKS (if and when it works ) :
print that shematic - for start ........
I use to think about regs in "class" boundaries ...... series reg is B class and shunt reg is A class - same as in amp classes ; A class is always willing to release energy , and that is certainly more natural , than ....... blah blah ..
anyway - Shunty ; look at schematic , and just cover lower half with sheet of paper ; upper an lower halves are symmetric , and you really don't need to look at all this mess to understand how it works ..........
depending of Shunty's purpose and use - you'll have either xformer in front of it , or you'll have ready DC voltage (from main amp PSU perhaps) , but in any case you can use input Graetz , made of D6,,D7,D6a,D7a ; after Graetz - you see multiple CRCRC bank..... purpose of that is obvious ;
after that - you see plain vanilla series reg , made of Q1 and surrounding parts ; it's regulating voltage is set with D1-D5 zenner string .......... 5 x 10V - 4V (remember Ugs ? ) = 46 V
then you see CCS part of shunt reg , made with Q2,Q3 and few surrounding parts ; same as with main CCS in Pumpkin - current is set with Ube voltage of Q3/ R7 , and that's 0V65/4E7 = ~140mA ....... so - that CCS is trying all the time to push 140mA through everything after him ,and that's nothing else than Pumpkin's positive halve , in parallel with shunt part of shunt reg ;
in all shunt reg- load combos , most natural way of current sharing is when you have same currents through load and shunt element of reg ....
after Q2 CCS , you can see - going right - one little all bipolar Pumpkin ; so - that's again LTP, cascoded too , just because benefits of cascoding aren't limited on jfets/mosfets only , but on other active sorts too.....
CCS for LTP is made of Q8/Q9 ; current through it is set by R17 ; base current for Q8 is delivered via R10/R11 string , and C10 is there just for little filtering .cascode is , obviously , biased via R3/R14 string .
far right (sensing side - base of Q5 ) side of LTP is clamped to +5V via U1 , and that's voltage reference for 5V , eeny weeny 3-legged chip, which is like reference zenner in fact ;
current through U1 is held constant with "micro" CCS , made with Q10/R19 combo .
LTP is trying to always preserve same conditions on both sides ; so - when base of Q4 feels ( via voltage divider made of R8/WR1+R9 ) something different than 5V , it is trying to null that difference yoyo-ing current through halves .... yoyo-ing current through Q5 is routed via R12 , and that current is resulting in adequate amount for gentle opening and closing Q11 ;
that critter (Q11) is nothing else than good old BJT Darlington ; he needs ~1V3 of Ub1e2 voltage for full throttle opening ; just imagine it as rheostat , and you'll understand his role and work here ....
when LTP feels that voltage is greater than 36V ( set by tweedling with WR1) , it conducts more current through R12 , thus more opening Q11 , which then eats more current .......
simple - isn't it .
there are few tricks more , but I can't remember right now what they exactly are ...... I'm pretty sure that you'll ask what you must ask ...... so no simple stone will be left unturned
download Cook Book ; read it few times ...... and I'm sure that few things will be clearer ;
I can't even slightly remember how many times I re-read EVERY Papa's article ....... and not just Papa's ........
now I really know that I'm dumb
HOW SHUNTY WORKS (if and when it works
) :print that shematic - for start ........
I use to think about regs in "class" boundaries ...... series reg is B class and shunt reg is A class - same as in amp classes ; A class is always willing to release energy , and that is certainly more natural , than ....... blah blah ..
anyway - Shunty ; look at schematic , and just cover lower half with sheet of paper ; upper an lower halves are symmetric , and you really don't need to look at all this mess to understand how it works ..........
depending of Shunty's purpose and use - you'll have either xformer in front of it , or you'll have ready DC voltage (from main amp PSU perhaps) , but in any case you can use input Graetz , made of D6,,D7,D6a,D7a ; after Graetz - you see multiple CRCRC bank..... purpose of that is obvious ;
after that - you see plain vanilla series reg , made of Q1 and surrounding parts ; it's regulating voltage is set with D1-D5 zenner string .......... 5 x 10V - 4V (remember Ugs ? ) = 46 V
then you see CCS part of shunt reg , made with Q2,Q3 and few surrounding parts ; same as with main CCS in Pumpkin - current is set with Ube voltage of Q3/ R7 , and that's 0V65/4E7 = ~140mA ....... so - that CCS is trying all the time to push 140mA through everything after him ,and that's nothing else than Pumpkin's positive halve , in parallel with shunt part of shunt reg ;
in all shunt reg- load combos , most natural way of current sharing is when you have same currents through load and shunt element of reg ....
after Q2 CCS , you can see - going right - one little all bipolar Pumpkin ; so - that's again LTP, cascoded too , just because benefits of cascoding aren't limited on jfets/mosfets only , but on other active sorts too.....
CCS for LTP is made of Q8/Q9 ; current through it is set by R17 ; base current for Q8 is delivered via R10/R11 string , and C10 is there just for little filtering .cascode is , obviously , biased via R3/R14 string .
far right (sensing side - base of Q5 ) side of LTP is clamped to +5V via U1 , and that's voltage reference for 5V , eeny weeny 3-legged chip, which is like reference zenner in fact ;
current through U1 is held constant with "micro" CCS , made with Q10/R19 combo .
LTP is trying to always preserve same conditions on both sides ; so - when base of Q4 feels ( via voltage divider made of R8/WR1+R9 ) something different than 5V , it is trying to null that difference yoyo-ing current through halves .... yoyo-ing current through Q5 is routed via R12 , and that current is resulting in adequate amount for gentle opening and closing Q11 ;
that critter (Q11) is nothing else than good old BJT Darlington ; he needs ~1V3 of Ub1e2 voltage for full throttle opening ; just imagine it as rheostat , and you'll understand his role and work here ....
when LTP feels that voltage is greater than 36V ( set by tweedling with WR1) , it conducts more current through R12 , thus more opening Q11 , which then eats more current .......
simple - isn't it .
there are few tricks more , but I can't remember right now what they exactly are ...... I'm pretty sure that you'll ask what you must ask ...... so no simple stone will be left unturned
download Cook Book ; read it few times ...... and I'm sure that few things will be clearer ;
I can't even slightly remember how many times I re-read EVERY Papa's article ....... and not just Papa's ........
now I really know that I'm dumb
COOK BOOK
download it here :
http://czsr_bac.on.neobee.net/The Pumpkin & Shunty.pdf
there you'll find schematics , and few lines of txt
download it here :
http://czsr_bac.on.neobee.net/The Pumpkin & Shunty.pdf
there you'll find schematics , and few lines of txt
ZM,
Digikey or mouser does not have BDX54C. I hate to place a very small order with Newark just for these devices. Can TIP120 and its counter part TIP125 be an adequate replacement?
http://www.onsemi.com/pub/Collateral/TIP120-D.PDF
Thanks,
Dinesh
Digikey or mouser does not have BDX54C. I hate to place a very small order with Newark just for these devices. Can TIP120 and its counter part TIP125 be an adequate replacement?
http://www.onsemi.com/pub/Collateral/TIP120-D.PDF
Thanks,
Dinesh
All diymember's....
Just wanna wish you marry christmas
is on it's place...
lykkedk is sending a very marry christmas to all diy'er's out there... have a nice holly.diy
Re: Re: Re: when I'm at it ........ Shunty :
Certainly it doesn't stem from SHANT-av (very hard to properly translate, the closest one would be slang for someone that has "two left hands" Well... look at that now I am digging myself deeper and deeper - No offense to a left handed people, but that would be the description for someone who is technically challenged or who is just clumsy... The fact is- right handed people are majority, and they do sometimes joke about left handed people.
Let me correct myself euphemism for someone who is SHANTAV or with two left hands is someone with motor skills disorder. After this I will stop translating from Serbian...
Happy Holidays to all (left and right handed) fanatics.
spavleski said:
Now I understand the root cause for the name "SHUNTy"
Certainly it doesn't stem from SHANT-av (very hard to properly translate, the closest one would be slang for someone that has "two left hands" Well... look at that now I am digging myself deeper and deeper - No offense to a left handed people, but that would be the description for someone who is technically challenged or who is just clumsy... The fact is- right handed people are majority, and they do sometimes joke about left handed people.
Let me correct myself euphemism for someone who is SHANTAV or with two left hands is someone with motor skills disorder. After this I will stop translating from Serbian...
Happy Holidays to all (left and right handed) fanatics.
in any case - Shunty is mish-mash between implication to SHUNT and SHANTAV (fonetic written) , which on serbian language is verb for person with injured or (one ) shorter leg .
we have some other verbs for persons with both leg shorter
anyway - that can easily be implication for ZM ...... with my double discus-hernia, I'm often more than SHANTAV ............