Great looking circuit
Ricardo,
I really like what you have put together. Thinking about ordering the 2SK216/2SJ79 needed to build.
I have a number of questions, most come from my lack of knowledge.
1. The voltage source V2, can something like a LM329DZ be used here?
2. The gain resistor, I like either unity gain, or maybe a gain of two. My system gets all its gain from speaker efficiency. Too high a gain leaves it with a pot that only uses 25% of the rotation. Lookiing over the circuit, I do not understand how to calculate the gain, is it 400/330?
3. The servos are not defined, guess anywill be fine. Any idea of the range of actual offset and differential offset without? I have an unused pair of nice output iron. They are a great way of dealing with small amounts of offset. Plus allow super simple selection of SE or balanced output.
4. There are no gate stoppers on the mosfets. Do these lateral devices not require them?
George
Ricardo,
I really like what you have put together. Thinking about ordering the 2SK216/2SJ79 needed to build.
I have a number of questions, most come from my lack of knowledge.
1. The voltage source V2, can something like a LM329DZ be used here?
2. The gain resistor, I like either unity gain, or maybe a gain of two. My system gets all its gain from speaker efficiency. Too high a gain leaves it with a pot that only uses 25% of the rotation. Lookiing over the circuit, I do not understand how to calculate the gain, is it 400/330?
3. The servos are not defined, guess anywill be fine. Any idea of the range of actual offset and differential offset without? I have an unused pair of nice output iron. They are a great way of dealing with small amounts of offset. Plus allow super simple selection of SE or balanced output.
4. There are no gate stoppers on the mosfets. Do these lateral devices not require them?
George
Massimo,
They are the same devices except max Vds, but it’s just because I have not the ”pairs” spice models.
George,
This is just an attempt to a nice preamp, inspired by this thread, in any case don’t associated it with the Blowtorch, it could have similarities that’s all.
V2 is just here for simulation purposes, so you don’t need it, the gain is 2.5 or (8 dB) and is roughly set by the ratio R10/R24 , (R10/R22 for the other side) in fact a little less due to the limited fets transconductance).
The DC servo could look like the circuit illustrate here, it can be simpler without the differential servo, but I think it’s essential.
Gate stoppers are the compound R6//R4 or R1/R5, the same for the lower side.
I have not build this circuit yet, it’s only here for learning purposes with all of you, more experienced people could give some advises and improvements. So please don't hurry to build it, active devices are difficult to get, expensive and must be matched. As soon as I get some time (probably not before a few weeks) I will build it and test it (I must say that I don’t need it, it's just a way to learn). Of course if someone has the parts and really wants to carry it out meanwhile, his conclusions would be interesting and welcome…
They are the same devices except max Vds, but it’s just because I have not the ”pairs” spice models.
George,
This is just an attempt to a nice preamp, inspired by this thread, in any case don’t associated it with the Blowtorch, it could have similarities that’s all.
V2 is just here for simulation purposes, so you don’t need it, the gain is 2.5 or (8 dB) and is roughly set by the ratio R10/R24 , (R10/R22 for the other side) in fact a little less due to the limited fets transconductance).
The DC servo could look like the circuit illustrate here, it can be simpler without the differential servo, but I think it’s essential.
Gate stoppers are the compound R6//R4 or R1/R5, the same for the lower side.
I have not build this circuit yet, it’s only here for learning purposes with all of you, more experienced people could give some advises and improvements. So please don't hurry to build it, active devices are difficult to get, expensive and must be matched. As soon as I get some time (probably not before a few weeks) I will build it and test it (I must say that I don’t need it, it's just a way to learn). Of course if someone has the parts and really wants to carry it out meanwhile, his conclusions would be interesting and welcome…
Spice models
This should help somewhat...
*SRC=2SJ76;QSJ76;MOSFETs P;Gen. Purpose;140V 500mA
.MODEL 2SJ76 PMOS (LEVEL=1 VTO=-15 KP=122N GAMMA=18.6
+ PHI=.75 LAMBDA=1.48M RD=.84 RS=.84 IS=250F PB=.8 MJ=.46
+ CBD=444P CBS=533P CGSO=384N CGDO=320N CGBO=346N)
* -- Assumes default L=100U W=100U --
* 140 Volt .5 Amp 6 ohm Enh-Mode P-Channel MOSFET 07-28-1995
* 2SJ76, TOSHIBA, 1993 JAPANESE FET MANUAL, P.16
**********
*SRC=2SJ79;QSJ79;MOSFETs P;Gen. Purpose;200V 500mA
.MODEL 2SJ79 PMOS (LEVEL=1 VTO=-15 KP=122N GAMMA=18.6
+ PHI=.75 LAMBDA=1.04M RD=.84 RS=.84 IS=250F PB=.8 MJ=.46
+ CBD=862P CBS=1.03N CGSO=57.6N CGDO=48N CGBO=1.09U)
* -- Assumes default L=100U W=100U --
* 200 Volt .5 Amp 6 ohm Enh-Mode P-Channel MOSFET 07-28-1995
* 2SJ79, TOSHIBA, 1993 JAPANESE FET MANUAL, P.16
**********
*SRC=2SK213;QSK213;MOSFETs N;Gen. Purpose;140V 500mA
.MODEL 2SK213 NMOS (LEVEL=1 VTO=15 KP=.16 GAMMA=18.6
+ PHI=.75 LAMBDA=1.48M RD=.84 RS=.84 IS=250F PB=.8 MJ=.46
+ CBD=560P CBS=672P CGSO=26.4N CGDO=22N CGBO=852N)
* -- Assumes default L=100U W=100U --
* 140 Volt .5 Amp 6 ohm Enh-Mode N-Channel MOSFET 07-28-1995
* 2SK213, TOSHIBA, 1993 JAPANESE FET MANUAL, P.38
**********
*SRC=2SK216;QSK216;MOSFETs N;Gen. Purpose;200V 500mA
.MODEL 2SK216 NMOS (LEVEL=1 VTO=15 KP=80M GAMMA=18.6
+ PHI=.75 LAMBDA=521U RD=.84 RS=.84 IS=250F PB=.8 MJ=.46
+ CBD=658P CBS=790P CGSO=26.4N CGDO=22N CGBO=852N)
* -- Assumes default L=100U W=100U --
* 200 Volt .5 Amp 6 ohm Enh-Mode N-Channel MOSFET 07-28-1995
* 2SK216, TOSHIBA, 1993 JAPANESE FET MANUAL, P.38
**********
J. C once mentioned that servos should be kept away from input differentials, so could we control dc offset manualy if we put a trimmer in series with resistors R18(R1) and R20(R4) and tie them to output instead of ground. Output impedance will go up, so we could change folded cascode for 2SJ74/2SK170 and use 2SJ79/2SK216 for output buffer. Something like this...
http://www.diyaudio.com/forums/attachment.php?s=&postid=743611&stamp=1129238290
This should help somewhat...
*SRC=2SJ76;QSJ76;MOSFETs P;Gen. Purpose;140V 500mA
.MODEL 2SJ76 PMOS (LEVEL=1 VTO=-15 KP=122N GAMMA=18.6
+ PHI=.75 LAMBDA=1.48M RD=.84 RS=.84 IS=250F PB=.8 MJ=.46
+ CBD=444P CBS=533P CGSO=384N CGDO=320N CGBO=346N)
* -- Assumes default L=100U W=100U --
* 140 Volt .5 Amp 6 ohm Enh-Mode P-Channel MOSFET 07-28-1995
* 2SJ76, TOSHIBA, 1993 JAPANESE FET MANUAL, P.16
**********
*SRC=2SJ79;QSJ79;MOSFETs P;Gen. Purpose;200V 500mA
.MODEL 2SJ79 PMOS (LEVEL=1 VTO=-15 KP=122N GAMMA=18.6
+ PHI=.75 LAMBDA=1.04M RD=.84 RS=.84 IS=250F PB=.8 MJ=.46
+ CBD=862P CBS=1.03N CGSO=57.6N CGDO=48N CGBO=1.09U)
* -- Assumes default L=100U W=100U --
* 200 Volt .5 Amp 6 ohm Enh-Mode P-Channel MOSFET 07-28-1995
* 2SJ79, TOSHIBA, 1993 JAPANESE FET MANUAL, P.16
**********
*SRC=2SK213;QSK213;MOSFETs N;Gen. Purpose;140V 500mA
.MODEL 2SK213 NMOS (LEVEL=1 VTO=15 KP=.16 GAMMA=18.6
+ PHI=.75 LAMBDA=1.48M RD=.84 RS=.84 IS=250F PB=.8 MJ=.46
+ CBD=560P CBS=672P CGSO=26.4N CGDO=22N CGBO=852N)
* -- Assumes default L=100U W=100U --
* 140 Volt .5 Amp 6 ohm Enh-Mode N-Channel MOSFET 07-28-1995
* 2SK213, TOSHIBA, 1993 JAPANESE FET MANUAL, P.38
**********
*SRC=2SK216;QSK216;MOSFETs N;Gen. Purpose;200V 500mA
.MODEL 2SK216 NMOS (LEVEL=1 VTO=15 KP=80M GAMMA=18.6
+ PHI=.75 LAMBDA=521U RD=.84 RS=.84 IS=250F PB=.8 MJ=.46
+ CBD=658P CBS=790P CGSO=26.4N CGDO=22N CGBO=852N)
* -- Assumes default L=100U W=100U --
* 200 Volt .5 Amp 6 ohm Enh-Mode N-Channel MOSFET 07-28-1995
* 2SK216, TOSHIBA, 1993 JAPANESE FET MANUAL, P.38
**********
J. C once mentioned that servos should be kept away from input differentials, so could we control dc offset manualy if we put a trimmer in series with resistors R18(R1) and R20(R4) and tie them to output instead of ground. Output impedance will go up, so we could change folded cascode for 2SJ74/2SK170 and use 2SJ79/2SK216 for output buffer. Something like this...
http://www.diyaudio.com/forums/attachment.php?s=&postid=743611&stamp=1129238290
Richard's Proposed Schematics
Richard,
Why not use 5x 2SK170 in parallel to replace one 2SK216, and similarly 5x 2SJ74 for one 2SJ79 ?
Capacitance is more, but you have low resistor values anyway.
Heat dissipation should be OK at 1W max.
Bias current at 60 mA at least (at Idss) should also be sufficient.
Price would be comparable.
Advantage -- at least 2.5x higher Transconductance, lower noise, and self biasing ??
Patrick
Richard,
Why not use 5x 2SK170 in parallel to replace one 2SK216, and similarly 5x 2SJ74 for one 2SJ79 ?
Capacitance is more, but you have low resistor values anyway.
Heat dissipation should be OK at 1W max.
Bias current at 60 mA at least (at Idss) should also be sufficient.
Price would be comparable.
Advantage -- at least 2.5x higher Transconductance, lower noise, and self biasing ??
Patrick
May I jump back to earlier regulator circuits talked about in this thread, I would like to ask what is the difference if we would use a BJT instead of FET as a passive shunt voltage regulator shown in many circuits posted here?
Is it noise, or is a lower conductance wanted, or else?
Are there any real issues to not go with BJT's or are they just minor?
Also recently shown fully balanced circuit's, after one circuit was shown which happend to be a fully FET circuit all you guys are on FET's, can't we use BJT's?
Again what is the purpose, lower high order harmonics with FET's or else, or are we just "following blindly gurus" without own thinking outside the box and questioning why?
I just find the thread a bit peculiar how it get steared up by some initial circuit posted setting the topic line-up, no offense what so ever anyhow, it has been an educative thread never the less!
Cheers Michael
Is it noise, or is a lower conductance wanted, or else?
Are there any real issues to not go with BJT's or are they just minor?
Also recently shown fully balanced circuit's, after one circuit was shown which happend to be a fully FET circuit all you guys are on FET's, can't we use BJT's?
Again what is the purpose, lower high order harmonics with FET's or else, or are we just "following blindly gurus" without own thinking outside the box and questioning why?
I just find the thread a bit peculiar how it get steared up by some initial circuit posted setting the topic line-up, no offense what so ever anyhow, it has been an educative thread never the less!
Cheers Michael
Hi,
When I looked at their site in detail, I noticed that maybe only one (max.) on whole page of part # stated "Toshiba" alongside the device number.
This suggests to me that the remainder (i.e. the majority) are probably not original makers devices, and their prices are 'unusually' low compared with anyone else I know of.
I hope I am not doing them an injustice, but I decided not to buy anything from these people, as sometimes 'bargains' etc don't turn out to be so good.
Regards,
When I looked at their site in detail, I noticed that maybe only one (max.) on whole page of part # stated "Toshiba" alongside the device number.
This suggests to me that the remainder (i.e. the majority) are probably not original makers devices, and their prices are 'unusually' low compared with anyone else I know of.
I hope I am not doing them an injustice, but I decided not to buy anything from these people, as sometimes 'bargains' etc don't turn out to be so good.
Regards,
driver level power mosfets ...
To my knowledge, Hitachi/Renesas has been the only mfr of the 2SK216/2SJ79 (and their lower voltage siblings) TO220 packaged, lower powered lateral MOSFET.
If you get these parts labeled otherwise, I would be very suspicious.
Toshiba made some parts that some might have thought were similar, but I think many parameters, for example, the threshhold voltages, were very different. I'm not even sure they were lateral devices. Sorry, I don't remember the part numbers.
mlloyd1
To my knowledge, Hitachi/Renesas has been the only mfr of the 2SK216/2SJ79 (and their lower voltage siblings) TO220 packaged, lower powered lateral MOSFET.
If you get these parts labeled otherwise, I would be very suspicious.
Toshiba made some parts that some might have thought were similar, but I think many parameters, for example, the threshhold voltages, were very different. I'm not even sure they were lateral devices. Sorry, I don't remember the part numbers.
mlloyd1
Lobtron & Co.
I had 2SC2922 from Lobtron which came in a cheap look-alike package labed as SK 2SC2922. I had 2SA970 from Reichelt which came labeled CSA970. I had 2SK216 which did not even come with the correct pin assignment.
I can name you a few more. But I think if you want to discuss fake parts, we should have a new thread somewhere else. I think someone in the UK even got fakes from Farnell.
All I can say is if you can, buy parts from the Manufacturer direct (On Semi, Fairchild, etc.). Unfortunately Toshiba JFETs are just not available in Europe (I have asked their head office in Düsseldorf). Maybe we should organise group buys and purchase a few thousand from Japan direct.
Now let's go back to the circuit, please.
Patrick
I had 2SC2922 from Lobtron which came in a cheap look-alike package labed as SK 2SC2922. I had 2SA970 from Reichelt which came labeled CSA970. I had 2SK216 which did not even come with the correct pin assignment.
I can name you a few more. But I think if you want to discuss fake parts, we should have a new thread somewhere else. I think someone in the UK even got fakes from Farnell.
All I can say is if you can, buy parts from the Manufacturer direct (On Semi, Fairchild, etc.). Unfortunately Toshiba JFETs are just not available in Europe (I have asked their head office in Düsseldorf). Maybe we should organise group buys and purchase a few thousand from Japan direct.
Now let's go back to the circuit, please.
Patrick
Aparatusonitus, many thanks for the spice models
Hi Patrick, It’s something to try, why not, but in the case of a folded cascode configuration self bias isn’t possible, and the higher transconductance together with the bias resistors could amplify noise too, so the resistors values should be quite lower.
BJT could present a lower Zout for the PSU shunt device but a MOSFET is best suited in terms of noise and channel separation B-E vs. G-S, they are easier to use when filtering the reference, and show less ringing at high frequencies.
Fets are self biasing, are low noise, present low order harmonics (essentially 2) and high current vs. BJTs, moreover they sound especially good. This known I talk for this particular application (line preamp), in other type of audio circuits BJTs can be wonderfull.
Help John!
(No doubt you will tease on me, if I'm wrong about all that.)
EUVL said:
Hi Patrick, It’s something to try, why not, but in the case of a folded cascode configuration self bias isn’t possible, and the higher transconductance together with the bias resistors could amplify noise too, so the resistors values should be quite lower.
Ultima Thule said:
BJT could present a lower Zout for the PSU shunt device but a MOSFET is best suited in terms of noise and channel separation B-E vs. G-S, they are easier to use when filtering the reference, and show less ringing at high frequencies.
Fets are self biasing, are low noise, present low order harmonics (essentially 2) and high current vs. BJTs, moreover they sound especially good. This known I talk for this particular application (line preamp), in other type of audio circuits BJTs can be wonderfull.
Help John!
(No doubt you will tease on me, if I'm wrong about all that.)
Richard (or should I say Ricardo),
> but in the case of a folded cascode configuration self bias isn’t possible
If one were to use JFETs for the second stage, then there is not much more motivation to use folded cascode ?? Why not just drive the second stage at the gate, with the source connected to the supply rails ?
And Zetex is a good suggestion for substitute, especially in Europe.
Patrick
> but in the case of a folded cascode configuration self bias isn’t possible
If one were to use JFETs for the second stage, then there is not much more motivation to use folded cascode ?? Why not just drive the second stage at the gate, with the source connected to the supply rails ?
And Zetex is a good suggestion for substitute, especially in Europe.
Patrick
Output devices
I like the thought of using those 2SK216/2SJ79 mosfets. Low gate capacitance and decent current handling.
I am not sure about trying paralled jfets on the output. Even going with 5 may give it a wimpy sound. In my experiance using a way over speced output pair like IRF610/9610 or even IRF510/9510 gives better sonics. Bigger sound, more slam, deeper and tighter bass.
The folded cascode is more than capable of driving these devices. The suggested parts sure look good on paper though. Wish I had some.
George
I like the thought of using those 2SK216/2SJ79 mosfets. Low gate capacitance and decent current handling.
I am not sure about trying paralled jfets on the output. Even going with 5 may give it a wimpy sound. In my experiance using a way over speced output pair like IRF610/9610 or even IRF510/9510 gives better sonics. Bigger sound, more slam, deeper and tighter bass.
The folded cascode is more than capable of driving these devices. The suggested parts sure look good on paper though. Wish I had some.
George
Output devices
Assume that you have a balanced power amp, 200W class A. I suppose then +/- 4V ouput (rel to gnd) is a lot. Suppose you have total input resistance across +Vin / -Vin of 20k. This would draw a maximum dynamic current of 0.4mA. So I do not really understand why one would want 200mA bias current, which is the linear region for the Hitachi MOSFETs. To operate IRF610 / 9610 in their linear region, you need to bias to an amp. In which case you might as well use 2SK1058.
Why would I want to use JFET ?
I thought I were building preamp. So I wanted to use low noise devices, and I wanted to work them in the linear (and not parabolic) region, which is at Idss. And I would have chosen to use the V version. True, I would then have a dissipation problem at 24V. But I am sure there are solutions for that.
Just my 2 cents,
Patrick
Assume that you have a balanced power amp, 200W class A. I suppose then +/- 4V ouput (rel to gnd) is a lot. Suppose you have total input resistance across +Vin / -Vin of 20k. This would draw a maximum dynamic current of 0.4mA. So I do not really understand why one would want 200mA bias current, which is the linear region for the Hitachi MOSFETs. To operate IRF610 / 9610 in their linear region, you need to bias to an amp. In which case you might as well use 2SK1058.
Why would I want to use JFET ?
I thought I were building preamp. So I wanted to use low noise devices, and I wanted to work them in the linear (and not parabolic) region, which is at Idss. And I would have chosen to use the V version. True, I would then have a dissipation problem at 24V. But I am sure there are solutions for that.
Just my 2 cents,
Patrick
- Status
- Not open for further replies.