POST 709.
Hi again Colin. I've just dug out my summer receipt from Farnell, and see that my JLH amplifier psu capacitor of choice, the 10,000uF - 63V, has the part number 652179. These are Panasonic, and cost £6. 31 + vat each. Go to
www.farnellinone.co.uk
and enter this part number in their finder for details.
I've used these for years, and once charged two to 35V and left them to see how quickly the voltage would leak. After three weeks I gave up because they were both still at 27V.
Tschrama and Millwood, I wish you both the very best in finalising your Mosfet versions of the JLH-69.
However, can I clarify, as Geoff has already mentioned, that when the bipolar JLH is implemented using good devices, its open loop frequency response covers the whole audio spectrum.
There are any number of supposedly 'Hi-Fi' Mosfet (and bipolar) designs already in the market place that have much lower distortion specifications than the JLH-69, but when you examine their open loop characteristics their gains fall off within the audio frequency band due to path series inductance and parallel capacitance.
Now the Hi-Fi manufacturers suddenly appear to have discovered and have started making trendy claims for amplifier signal paths like those the JLH had over 30yrs ago.
As soon as you put Mosfets in the JLH-69 you are loading, within the closed nfb loop, the driver/splitter transistor with voltage (signal amplitude) variable (Mosfet) capacitors. To overcome the instability you then need a dominant pole capacitor within the overall nfb loop. The result is signal path delay, such that sibilants and percusive leading edge waveforms lose their initial coherence in a way that does not happen within the 'all bipolar' circuit. The sound is subliminally smooth, but it loses its original 'lifelikeness'.
This is not something that is easy to analyse using circuit simulators; the best you can do is observe first cycle distortion instead of 'conventional' second cycle output analysis in time delayed isolation.
Tschrama. Your Mosfet version includes circuit changes that benefit the original JLH, so can I suggest that it would be worthwhile updating your bipolar version so that you make comparisons on an openly fair ground. I am refering to the driver/splitter resistor values, also your nfb loop values, though the bipolar should not have the dominant pole capacitor. How about 2SC5200 outputs too; these are good for 25Wrms per pair in the JLH circuit with a 50V or +/-25V psu and 2.2A quiescent. (I've reliably run one pair at 2.4A quiescent.)
Cheers for now ............. Graham.
Hi again Colin. I've just dug out my summer receipt from Farnell, and see that my JLH amplifier psu capacitor of choice, the 10,000uF - 63V, has the part number 652179. These are Panasonic, and cost £6. 31 + vat each. Go to
www.farnellinone.co.uk
and enter this part number in their finder for details.
I've used these for years, and once charged two to 35V and left them to see how quickly the voltage would leak. After three weeks I gave up because they were both still at 27V.
Tschrama and Millwood, I wish you both the very best in finalising your Mosfet versions of the JLH-69.
However, can I clarify, as Geoff has already mentioned, that when the bipolar JLH is implemented using good devices, its open loop frequency response covers the whole audio spectrum.
There are any number of supposedly 'Hi-Fi' Mosfet (and bipolar) designs already in the market place that have much lower distortion specifications than the JLH-69, but when you examine their open loop characteristics their gains fall off within the audio frequency band due to path series inductance and parallel capacitance.
Now the Hi-Fi manufacturers suddenly appear to have discovered and have started making trendy claims for amplifier signal paths like those the JLH had over 30yrs ago.
As soon as you put Mosfets in the JLH-69 you are loading, within the closed nfb loop, the driver/splitter transistor with voltage (signal amplitude) variable (Mosfet) capacitors. To overcome the instability you then need a dominant pole capacitor within the overall nfb loop. The result is signal path delay, such that sibilants and percusive leading edge waveforms lose their initial coherence in a way that does not happen within the 'all bipolar' circuit. The sound is subliminally smooth, but it loses its original 'lifelikeness'.
This is not something that is easy to analyse using circuit simulators; the best you can do is observe first cycle distortion instead of 'conventional' second cycle output analysis in time delayed isolation.
Tschrama. Your Mosfet version includes circuit changes that benefit the original JLH, so can I suggest that it would be worthwhile updating your bipolar version so that you make comparisons on an openly fair ground. I am refering to the driver/splitter resistor values, also your nfb loop values, though the bipolar should not have the dominant pole capacitor. How about 2SC5200 outputs too; these are good for 25Wrms per pair in the JLH circuit with a 50V or +/-25V psu and 2.2A quiescent. (I've reliably run one pair at 2.4A quiescent.)
Cheers for now ............. Graham.
Graham Maynard said:
I believe this suggestion needs to be treated with a degree of caution. From feedback I have received from people who have tried the 2SC5200, and other high ft power transistors (2SC3281, MJL3281A and some Sanken devices whose numbers I cannot remember), the results have been mixed.
For some the change of output device worked, for others the result was oscillation and either a return to some slower transistors or the addition of a compensation capacitor (with the corresponding reduction in open-loop bandwidth).
The 1996 version appears to be more prone to the oscillation problem than the original 1969 version, possibly because the bootstrap capacitor in 1969 circuit loses its effectiveness at high frequencies thus increasing the collector load on the driver transistor and reducing the gain of this stage, though individual layouts could also have a bearing on the situation.
Hi Geoff and Graham,
open loop frequency response
Im very happy to tell you that if you value open loop frequency response, you will absolutely love the MOSFET version.
In my earlier post I have descibed the ringing and spontaneous oscilation of the MOSFET version. Later again, I have reported that the oscilation was due to output to input coupling. The(unshielded) wires were lying close to each other, no source connected, only 100K input resistor: this was asking for trouble.
I was cured very easly, but being carefull I installed shielded input wire, a 10K input resistor, and keep speaker wires away from input wires. It has never oscilated since. It is perfectly stable without compensation.
I also reported ringing on a 100KHz square wave. There must be a gain peak somewhere above 1MHz (my gerenator only goes to 1MHz). Equalizing the MOSFET drive impedance (by installing different gate-snoopers) and a 330pF in the good spot (thanks for the tips on that one!) (150pF also works) elliminates any overshoot giveing a very nice 100KHz square wave response.
My calculations show that this compensation cap has its -3dB at around 4MHz, so we assume that this is not the dominat pole. It just flattens the gain peak at some MHz. Like in the BJT version the output transistor form the dominant pole. With MOSFETs this the drive-impedance working in the gate-capacitance and Miller effect. If we assume 820R gate-resistance and 1200pF gate-capacitance, we get a dominant pole at around 161KHz.
My simulations so far (which have been very reliable so far) give an openloop freq-resp of 25KHz for the BJT and 68KHz for the MOSFET version. The 330pF compensation cap has no significant effect on this -3dB point.
I will have to measure this somehow, maybe someone could tell me a smart way to do this?
Further more my BJT doesn't come close to the MOSFET version when tested with a 100KHz square wave. The BJT has severe limmited up-going slewrate (3V/us max) making the square a triangle, it has servere overshoot at around 1MHz (?). The MOSFET completey outclasses the BJT in this case, having no overshoot and fast symmetrical slewrate (>10V/us).
I indeed should take another look (eg. rebuild ) at my BJT version, which still uses the 2N3055 (indeed slow and non-lineair as I have become to understand) and possibly sub-optimal driver-resistors.
I prommise I will find a way to measure openloop response, If one of you can explain how to calculate open-loop-fre-resp. OK?
Merry Christmas and best regards,
Thijs
open loop frequency response
Im very happy to tell you that if you value open loop frequency response, you will absolutely love the MOSFET version.
In my earlier post I have descibed the ringing and spontaneous oscilation of the MOSFET version. Later again, I have reported that the oscilation was due to output to input coupling. The(unshielded) wires were lying close to each other, no source connected, only 100K input resistor: this was asking for trouble.
I was cured very easly, but being carefull I installed shielded input wire, a 10K input resistor, and keep speaker wires away from input wires. It has never oscilated since. It is perfectly stable without compensation.
I also reported ringing on a 100KHz square wave. There must be a gain peak somewhere above 1MHz (my gerenator only goes to 1MHz). Equalizing the MOSFET drive impedance (by installing different gate-snoopers) and a 330pF in the good spot (thanks for the tips on that one!) (150pF also works) elliminates any overshoot giveing a very nice 100KHz square wave response.
My calculations show that this compensation cap has its -3dB at around 4MHz, so we assume that this is not the dominat pole. It just flattens the gain peak at some MHz. Like in the BJT version the output transistor form the dominant pole. With MOSFETs this the drive-impedance working in the gate-capacitance and Miller effect. If we assume 820R gate-resistance and 1200pF gate-capacitance, we get a dominant pole at around 161KHz.
My simulations so far (which have been very reliable so far) give an openloop freq-resp of 25KHz for the BJT and 68KHz for the MOSFET version. The 330pF compensation cap has no significant effect on this -3dB point.
I will have to measure this somehow, maybe someone could tell me a smart way to do this?
Further more my BJT doesn't come close to the MOSFET version when tested with a 100KHz square wave. The BJT has severe limmited up-going slewrate (3V/us max) making the square a triangle, it has servere overshoot at around 1MHz (?). The MOSFET completey outclasses the BJT in this case, having no overshoot and fast symmetrical slewrate (>10V/us).
I indeed should take another look (eg. rebuild
) at my BJT version, which still uses the 2N3055 (indeed slow and non-lineair as I have become to understand) and possibly sub-optimal driver-resistors. I prommise I will find a way to measure openloop response, If one of you can explain how to calculate open-loop-fre-resp. OK?
Merry Christmas and best regards,
Thijs
Hi,
I'm back again.
I first did some simulations on openloop charateristics. (as allways 1KHz, 1Watt into 8 ohm load)
original 1969 BJT version Iq 1.3A:
openloop gain : 57dB
-3dB: 30.5Khz
THD: -27dB
MOSFET version Iq 600mA :
openloop gain : 67.9dB
-3dB: 44Khz
THD: -45dB
MOSFET version with input transistor bootstrap collector Iq 600mA:
openloop gain : 79.9dB
-3dB: 26Khz
THD: -41dB
Measurements are offcourse what trully counts, but also must more difficlut to execute. I picked two 'crocodile snaps' and soldered a short wire from on to the other. Now it was easy to short the 100R feedback resistor, by clipping on the 'crocodile snaps'.
I made a 10K/1R voltage divider for my function-generator, giving 2mVpp output.
I connected a dummy load (8R2) and set Iq of my amp at 500mA.
I put the input signal and output signal on my scope. I noticed that gain increases with frequency as the feedback cap impedance decreases with frequency.
Maximum gain: 75dB (at about 10KHz)
-3dB :36.6KHz @ 45degree phase shift (=perfect 1st order)
Further more I notidec that increasing Iq to 900mA gave slightly higher the openloop (+30%) gain and going to 250mA reduced openloop gain by 30% ....phaseshift didn't change..0
Well that's it. looks good, not?
I think the reported overshoot is not due to a gain peak, but the result of crossconduction,which was due to the non-equal driver-output-impedance. However is think I lack the instruments to measure and confirm that.
Bye,
Thijs
PS made an error with alternative compensation cap arangement.. edtit: corrected... edit.. closed-loop bandwidth was measured as 625KHz...
I'm back again.
I first did some simulations on openloop charateristics. (as allways 1KHz, 1Watt into 8 ohm load)
original 1969 BJT version Iq 1.3A:
openloop gain : 57dB
-3dB: 30.5Khz
THD: -27dB
MOSFET version Iq 600mA :
openloop gain : 67.9dB
-3dB: 44Khz
THD: -45dB
MOSFET version with input transistor bootstrap collector Iq 600mA:
openloop gain : 79.9dB
-3dB: 26Khz
THD: -41dB
Measurements are offcourse what trully counts, but also must more difficlut to execute. I picked two 'crocodile snaps' and soldered a short wire from on to the other. Now it was easy to short the 100R feedback resistor, by clipping on the 'crocodile snaps'.
I made a 10K/1R voltage divider for my function-generator, giving 2mVpp output.
I connected a dummy load (8R2) and set Iq of my amp at 500mA.
I put the input signal and output signal on my scope. I noticed that gain increases with frequency as the feedback cap impedance decreases with frequency.
Maximum gain: 75dB (at about 10KHz)
-3dB :36.6KHz @ 45degree phase shift (=perfect 1st order)
Further more I notidec that increasing Iq to 900mA gave slightly higher the openloop (+30%) gain and going to 250mA reduced openloop gain by 30% ....phaseshift didn't change..0
Well that's it.
looks good, not?I think the reported overshoot is not due to a gain peak, but the result of crossconduction,which was due to the non-equal driver-output-impedance. However is think I lack the instruments to measure and confirm that.
Bye,
Thijs
PS made an error with alternative compensation cap arangement.. edtit: corrected... edit.. closed-loop bandwidth was measured as 625KHz...
tschrama, great numbers!
I did some thermal tests today. the whole amp module is put in an enclosure, and ran for 5 hours indoor (65-70F). the voltage drop on the 0.2ohm resistor rose to 260mv in about a couple of minutes, and steadily climbed upto 310mv - 320mv, that's about 1.6amp Iq, vs. 1.2amp Iq setting.
I think if you have a small heatsink, it is safer to set Iq at a lower figure (1amp is about as low as I would go). anything above 2amp seems to give marginal (non-audible?) gain but requires a much bigger heatsink.
For now, I am done with this thermal "instability" thing,
I did some thermal tests today. the whole amp module is put in an enclosure, and ran for 5 hours indoor (65-70F). the voltage drop on the 0.2ohm resistor rose to 260mv in about a couple of minutes, and steadily climbed upto 310mv - 320mv, that's about 1.6amp Iq, vs. 1.2amp Iq setting.
I think if you have a small heatsink, it is safer to set Iq at a lower figure (1amp is about as low as I would go). anything above 2amp seems to give marginal (non-audible?) gain but requires a much bigger heatsink.
For now, I am done with this thermal "instability" thing,
tschrama said:
Thijs
I am afraid that our simulations are at variance. My figures are:
JLH 1969 (BC560C/2N1771/2N3055) Iq 1.3A Vs +30V
Open-loop gain with 8ohm load: 63.5dB
-3dB point: 52.5kHz
Your MOSFET circuit (BC560C/2N1711/IRF540) Iq 0.6A Vs +30V
Open-loop gain with 8ohm load: 70.3dB
-3dB point: 33.5kHz
I have done any checks on open-loop distortion, but the simulations I ran yesterday, and confirmed again after your post, indicated that the closed-loop distortion at 1kHz for a 20Vp-p output into 8R was 0.0114% for the 1969 JLH and 0.0194% for your MOSFET circuit. Of course, these figures do not reflect reality since they are based on perfect matching of the output devices but they do indicate a reversal of your open-loop figures.
I don't know which simulation programme you use. I use SIMetrix 4.2. Perhaps someone else has some figures that might help to give an indication as to which set of results is (more nearly) correct (if either). Alternatively, we could discuss simulation methods etc by email to see if there is a variance in the models or the way we obtained the figures.
Geoff
Geoff,
I would be very interested in a discussion about simulation methods via email.
I actuyally find our findings pretty close for openloop bandwidth and openloop gain. I never expected simulation to be so close to each other, or so close to measurements..
But I used 2.8Vrms out not 20Vpp, this might be a factor that contributes to the differenes... Since Iq was 0.6A, driving 20Vpp (7.07Vrms) might some extra disproportional distortion from the MOSFET circuit..
BTW I not trying to say one or the other amplifier is better because of the simulation outcomes... That's up to listening test and measurements.. But the wide openloop gain might just motivate you for a MOSFET version...I hope it does..
Then again..it are just simulation, the most exciting thing was the measured openloop rersponse, actully very very close to you simulations! Isn't that cool ?
As to what figures are correct: only measurements can tell..
So there's one realy hard chanlence left: measure distortion..
Keep us posted or email me,
Kind regards,
Thijs
I would be very interested in a discussion about simulation methods via email.
I actuyally find our findings pretty close for openloop bandwidth and openloop gain. I never expected simulation to be so close to each other, or so close to measurements..
But I used 2.8Vrms out not 20Vpp, this might be a factor that contributes to the differenes... Since Iq was 0.6A, driving 20Vpp (7.07Vrms) might some extra disproportional distortion from the MOSFET circuit..
BTW I not trying to say one or the other amplifier is better because of the simulation outcomes... That's up to listening test and measurements.. But the wide openloop gain might just motivate you for a MOSFET version...I hope it does..
Then again..it are just simulation, the most exciting thing was the measured openloop rersponse, actully very very close to you simulations! Isn't that cool ?
As to what figures are correct: only measurements can tell..
So there's one realy hard chanlence left: measure distortion..
Keep us posted or email me,
Kind regards,
Thijs
my open loop frequency response, on protel dxp:
mosfet (mje15030 driven irf540s): about 100khz;
bjt (mje15030 driven mje15030s): about 10khz.
I didn't find an "automatic" way of doing this so I had to cut the feedback look and input a (DC) voltage source to maintain the same DC performance.
how did you guys do it on your simulators?
mosfet (mje15030 driven irf540s): about 100khz;
bjt (mje15030 driven mje15030s): about 10khz.
I didn't find an "automatic" way of doing this so I had to cut the feedback look and input a (DC) voltage source to maintain the same DC performance.
how did you guys do it on your simulators?
Millwood please don't blame me for being unable to follow what you think of as being an explanatory contextual flow. I also comment that I would be extremely embarrassed if I left a trail of unacknowledged, unretracted or uncorrected errors in this string.
Have you actually built or heard a 1969 BJT John Linsley-Hood 10W amplifier; the subject of this forum ?
In post 721 you informed us all that the BJT JLH-69 amplifier is running in one sided class-AB at 2Vp input. Not a hypothetical suggestion, not even an enquiry, but a bold statement; even though there is no such thing as one sided class-AB.
I suggested that you watch out for a bias shift that I observed 30 yrs ago using standard test bench equipment, in the hope that you will understand that what you had been simulating was not actually class-AB operation at all !
Then you come back with posts 741 and 742 as if I don't know how to read, and you smugly state that you could not get my prediction to work.
Because of your obvious inexperience I have taken the time to construct one of these amplifiers on my desktop, the 27V, 10W-8R version, and driven it with a 2Vp and 1kHz sinewave as you say you already have.
Hence my delay in reporting these my findings. It all takes time.
I have never simulated this before, but the bootstrap bias potential at the junction of R1, R2 and C1 on the original diagram did shift in the same way I remember, and
at time 0mS reads 25Volts
5mS 25.7
12mS 26.3
30mS 27.0
50mS 27.22
Due to the JLH-69 design having bootstrap and output capacitors, when it is driven beyond its class-A operating limit, a current averaging process gradually shifts the bootstrap bias potential as is illustrated in the above simulation results. As successive overloading sinewaves pass through their zero alternating potential, the bootstrap potential becomes slowly shifted; as is a bass loudspeaker cone when the amplifier is overdriven at a higher audio frequency.
Beyond class-A driving, the waveform becomes both time and amplitude shifted as the capacitors average out the available alternating current flow. This is not class-AB.
Further to your 721 post; there is no reason why the Mosfet equivalent of the JLH circuit should behave similarly, as with Tschrama's circuit, because driver currents are not being shared by output devices in the same way.
Cheers ......... Graham.
Have you actually built or heard a 1969 BJT John Linsley-Hood 10W amplifier; the subject of this forum ?
In post 721 you informed us all that the BJT JLH-69 amplifier is running in one sided class-AB at 2Vp input. Not a hypothetical suggestion, not even an enquiry, but a bold statement; even though there is no such thing as one sided class-AB.
I suggested that you watch out for a bias shift that I observed 30 yrs ago using standard test bench equipment, in the hope that you will understand that what you had been simulating was not actually class-AB operation at all !
Then you come back with posts 741 and 742 as if I don't know how to read, and you smugly state that you could not get my prediction to work.
Because of your obvious inexperience I have taken the time to construct one of these amplifiers on my desktop, the 27V, 10W-8R version, and driven it with a 2Vp and 1kHz sinewave as you say you already have.
Hence my delay in reporting these my findings. It all takes time.
I have never simulated this before, but the bootstrap bias potential at the junction of R1, R2 and C1 on the original diagram did shift in the same way I remember, and
at time 0mS reads 25Volts
5mS 25.7
12mS 26.3
30mS 27.0
50mS 27.22
Due to the JLH-69 design having bootstrap and output capacitors, when it is driven beyond its class-A operating limit, a current averaging process gradually shifts the bootstrap bias potential as is illustrated in the above simulation results. As successive overloading sinewaves pass through their zero alternating potential, the bootstrap potential becomes slowly shifted; as is a bass loudspeaker cone when the amplifier is overdriven at a higher audio frequency.
Beyond class-A driving, the waveform becomes both time and amplitude shifted as the capacitors average out the available alternating current flow. This is not class-AB.
Further to your 721 post; there is no reason why the Mosfet equivalent of the JLH circuit should behave similarly, as with Tschrama's circuit, because driver currents are not being shared by output devices in the same way.
Cheers ......... Graham.
tschrama said:
I tried that (upping the 2.7k resistor to like 2700k) and this thing didn't produce any signal for me (looks like random pulses).
Graham Maynard said:
don't worry about it, Graham. You have left plenty of them in your short stint here, so a couple more wouldn't really hurt,
Happy holidays, everyone.
HI YA MILLWOOD, I hace been following this thread with great interest as one of my projects [ for years] on the back-burner has been the jlh circuit in it's original form but now my interest has been sparked by your appearant success with a mosfet version of this circuit which I think I would like to emulate with your permission of course ... in an earlier post you said it might work with other mosfets than the ones you used .... so the question is do you think it would work with the IRF 740? THANKS Tomcat
Hi all,
I finally decided it was time top open up my original 1969 BTJ (bc560b, 2N1711, 2N3055, circuit as publiced in 1969) and do a similar openloop test as I reported a few post ago.. here comes...
openloop gain : 68.5dB
openloop -3dB: 31.7KHz
I think the trend is in nice agreement with my simulations..
I think it is time to prepare a THD measurement...
I need to arange my oscilator, Audigy 2 ZS soundcard etc etc..I don't promise anything, but that I'll do my best..
Graham Maynard and Geoff, I could be interesting to see how must variance there's is in 1969JLH openloop figures? Maybe you could find the time to do a similar test.
Best regards,
Thijs
PS
any comments (accept the obvious) on my openloop measurement methode?
I finally decided it was time top open up my original 1969 BTJ (bc560b, 2N1711, 2N3055, circuit as publiced in 1969) and do a similar openloop test as I reported a few post ago.. here comes...
openloop gain : 68.5dB
openloop -3dB: 31.7KHz
I think the trend is in nice agreement with my simulations..
I think it is time to prepare a THD measurement...
I need to arange my oscilator, Audigy 2 ZS soundcard etc etc..I don't promise anything, but that I'll do my best..
Graham Maynard and Geoff, I could be interesting to see how must variance there's is in 1969JLH openloop figures? Maybe you could find the time to do a similar test.
Best regards,
Thijs
PS
any comments (accept the obvious) on my openloop measurement methode?
Hi Millwood,
increasing the larger feedback resistor from 2K7 to 2700K can give someproblem. The input transistor bias current flows through this resistor so major offset can be introduced. I think a good method is to just delete the lower-value feedback resistor, or short it out like I do in practice.
Best regards,
Thijs
increasing the larger feedback resistor from 2K7 to 2700K can give someproblem. The input transistor bias current flows through this resistor so major offset can be introduced. I think a good method is to just delete the lower-value feedback resistor, or short it out like I do in practice.
Best regards,
Thijs
qwad said:do you think it would work with the IRF 740? THANKS Tomcat
Tomcat: I am no expert on this so please take whatever I have to offer with a big grain of salt,
Having said that, I see no reason why a 740 wouldn't work here. tschrama used the 510 which is slightly skinnier than my 540. there really is no science in my selecting the particular transistors in my amp: I happen to have lots of them around. so I would encourage you to try the 740 and let us know how it works. the more of us building this thing, the better we can understand how it compares with the bjt version.
tschrama has done extensive research and analysis around the circuitry so if you want to, you may look into his circuitry and see if that suits you better. All I can confirm for you is that mine works, and works nicely.
I do plan on breadboarding this thing during the holidays so I will report back how various variations of this (with different output transistors / drivers and complementory outputs too) work. Yes, I will try a BJT version of it as well to see what I am missin gout and how it baselines with the mosfet version.
I will report back to the forum when I get it going.
Hi Tomcat,
I actually know very little about MOSFETs, being more a BJT kind of guy... I decided against IRF510 because of lower transductance and power capabilities. The IRF540 seem very common since the small electricnic shop who usualy don't not sell what I need, this had did had those MOSFETs in stock .. they are cheap too (2,20 Euro).. If you can compare them by datasheets I'm sure you can find out if they are suitlble...
Goodluck,
Thijs
I actually know very little about MOSFETs, being more a BJT kind of guy... I decided against IRF510 because of lower transductance and power capabilities. The IRF540 seem very common since the small electricnic shop who usualy don't not sell what I need, this had did had those MOSFETs in stock .. they are cheap too (2,20 Euro).. If you can compare them by datasheets I'm sure you can find out if they are suitlble...
Goodluck,
Thijs
POST 745. Hi Geoff.
Yes I had heard these reports about oscillation with 2SC3281s etc., and they almost stopped me from trying them.
It is possible that the constructors who have had problems with these devices were ones who went to a lot of trouble making their amplifiers look pretty, and then found they didn't work, or they tried them as upgrades for poor layout versions that would not have oscillated with the 2N3055s anyway.
I recommend that constructors try these devices, and the only caution I would offer is that they be careful with their signal wiring and earthing layouts.
2SC3281/5200 types are much more gain linear, and provide a better hf accuracy.
The most significant fact about the two stage bipolar JLH-69 type circuit is its inherent stability; the 2SC5200 does not change this.
POST 752. Hi again Tschrama.
I'm following your Mosfet figures with interest.
You say you reduced the 100 ohm resistor to one micro-ohm and increased the capacitor to 2,200uF to establish the open loop gain.
You are actually boosting the first transistor gain way above what it runs at within the closed loop; your method will give a higher open loop figure than is actually realisable, and thus affect your nfb level calculation.
It is much better to insert a 100MH or 1TH inductor in series with the feedback resistor that comes from the output node. This removes negative feedback, and yet preserves the first stage gain and frequency response characteristics within the architecture.
Cheers .......... Graham.
Yes I had heard these reports about oscillation with 2SC3281s etc., and they almost stopped me from trying them.
It is possible that the constructors who have had problems with these devices were ones who went to a lot of trouble making their amplifiers look pretty, and then found they didn't work, or they tried them as upgrades for poor layout versions that would not have oscillated with the 2N3055s anyway.
I recommend that constructors try these devices, and the only caution I would offer is that they be careful with their signal wiring and earthing layouts.
2SC3281/5200 types are much more gain linear, and provide a better hf accuracy.
The most significant fact about the two stage bipolar JLH-69 type circuit is its inherent stability; the 2SC5200 does not change this.
POST 752. Hi again Tschrama.
I'm following your Mosfet figures with interest.
You say you reduced the 100 ohm resistor to one micro-ohm and increased the capacitor to 2,200uF to establish the open loop gain.
You are actually boosting the first transistor gain way above what it runs at within the closed loop; your method will give a higher open loop figure than is actually realisable, and thus affect your nfb level calculation.
It is much better to insert a 100MH or 1TH inductor in series with the feedback resistor that comes from the output node. This removes negative feedback, and yet preserves the first stage gain and frequency response characteristics within the architecture.
Cheers .......... Graham.