pro: Yes i tested that with just a single pair of IRFP250N, it works fine.
(Up to 180 Watts in 8 Ohms with a 58V rail) Use IRF9640 for T14, it works nice and stable!
BTW: the folded cascode voltage amplifier has a funny property at power off it fades out slowly - exactly like a tube amplifier
(Up to 180 Watts in 8 Ohms with a 58V rail) Use IRF9640 for T14, it works nice and stable!
BTW: the folded cascode voltage amplifier has a funny property at power off it fades out slowly - exactly like a tube amplifier
Thanks Lars.
Can a couple of IRFP140N be enough for 100W or is better to use four?
I would like to push the bias current to about 1.2A for class A operation until 20W (supposing enough heatsink).
Does it work well with high bias, or it is just a loss of time?
The gain seems to be about 18dB. Is it possible to move a bit
(+-3dB) around this value or are there any contraindications? (I am gonna making a ampli for the bass section of my 3 way loadspeaker, so I have to match the gain of the two ampli).
Tank you.
Ciao.
Can a couple of IRFP140N be enough for 100W or is better to use four?
I would like to push the bias current to about 1.2A for class A operation until 20W (supposing enough heatsink).
Does it work well with high bias, or it is just a loss of time?
The gain seems to be about 18dB. Is it possible to move a bit
(+-3dB) around this value or are there any contraindications? (I am gonna making a ampli for the bass section of my 3 way loadspeaker, so I have to match the gain of the two ampli).
Tank you.
Ciao.
A very nice job Lars, it’s a good thing you pointed out that is a tested design and not just an idea. I’m really very interested to know how it sounds. When you update your web page, perhaps you could also post on your web site an explanation of how it works. It seems a little different then most of the basic topologies I’ve seen. I’ve never seen anyone use the other side of the duel differential pair like that.
If you don’t mind me asking, done you feel that this project will compete with the kits you are selling?
Leve
If you don’t mind me asking, done you feel that this project will compete with the kits you are selling?
Leve
Leve: It may compete to some extent with the kits we sell, but we don't have any high powered MOSFET amplifiers in our kit programme. So no direct competition.
And then i guess you can say that i am more a hifi freak than a businessman, so for me is more important to get a good design out there. We do well in the business anyway so i see no need to keep some designs secret, even if we don't make them into products. if DIY'ers could have fun with them, that's a good thing!
Pro: for 100 Watts IRFP140N will do fine! Without crossing safety limits you can get 239 Watts audio power out of a single pair of IRFP140N, (at 100 degrees C). If you have them in the drawer, use them
(And forget about the 0R33 resistors!)
If you're going out to buy, why not take the IRFP150N or IRFP250N instead, they are cheaper and higher power than IRFP140N.
Best regards
Lars
And then i guess you can say that i am more a hifi freak than a businessman, so for me is more important to get a good design out there. We do well in the business anyway so i see no need to keep some designs secret, even if we don't make them into products. if DIY'ers could have fun with them, that's a good thing!
Pro: for 100 Watts IRFP140N will do fine! Without crossing safety limits you can get 239 Watts audio power out of a single pair of IRFP140N, (at 100 degrees C). If you have them in the drawer, use them
(And forget about the 0R33 resistors!)
If you're going out to buy, why not take the IRFP150N or IRFP250N instead, they are cheaper and higher power than IRFP140N.
Best regards
Lars
Dear Lars!
I appreciate your helpful attitude, but I think you oversimplify the situation! To produce 239W/4ohm, you need over 48V stabilized Psupply for this circuit. If you drive it with half amplitude square wave (this is not a rare situation, some man may try to limit volume at the end, instead of the mixer, because mixer occupied by the DJ), then dissipation will be 144W for half period on a single device, wich is definitely not in the SOA. Some thing make things worst:
1.:high power woofer's impedance can be as low as 0,7*nominal at some freq, and phase is allmost never 0 degrees. It can be over 50 degrees (fortunately not at the freq, where impedance is min.), and all of the reactive power dissipates on output devices. This is part of normal usage.
2.:very few man use stabilised supply for power amp. If one want to achieve your rated power with sine wave and unregulated supply, then he need 15-30% more idle voltage.
3.: In my opinion, a power amp have to bear short circuit, but have to work well with 30 degrees out of phase impedance. This means that a 240W amp have to deliver 5,5A at short sircuit. 5,5A*48V is 264W, wich is far over the SOA.
IHMO!:
I like more when two amp can be run with shorted their outputs together accidentally. For this you need further more oversize. It's not theoretical situation, it has happened to me, that somebody used my amp this way, and the amp survived for hours, but heated up to 80-90 degrees (normally it runs at 45-50 degrees). This wouldn't be possible, if there wasn't 2 kW of output device in each 500W amp. This is somewhat strange, but I found it neccessary for safety. At partytime there can happen everything, but "the show must go on"!
I hope you won't take this as attack, just I couldn't hold it in me.
Best regards:
Peter Sass
I appreciate your helpful attitude, but I think you oversimplify the situation! To produce 239W/4ohm, you need over 48V stabilized Psupply for this circuit. If you drive it with half amplitude square wave (this is not a rare situation, some man may try to limit volume at the end, instead of the mixer, because mixer occupied by the DJ), then dissipation will be 144W for half period on a single device, wich is definitely not in the SOA. Some thing make things worst:
1.:high power woofer's impedance can be as low as 0,7*nominal at some freq, and phase is allmost never 0 degrees. It can be over 50 degrees (fortunately not at the freq, where impedance is min.), and all of the reactive power dissipates on output devices. This is part of normal usage.
2.:very few man use stabilised supply for power amp. If one want to achieve your rated power with sine wave and unregulated supply, then he need 15-30% more idle voltage.
3.: In my opinion, a power amp have to bear short circuit, but have to work well with 30 degrees out of phase impedance. This means that a 240W amp have to deliver 5,5A at short sircuit. 5,5A*48V is 264W, wich is far over the SOA.
IHMO!:
I like more when two amp can be run with shorted their outputs together accidentally. For this you need further more oversize. It's not theoretical situation, it has happened to me, that somebody used my amp this way, and the amp survived for hours, but heated up to 80-90 degrees (normally it runs at 45-50 degrees). This wouldn't be possible, if there wasn't 2 kW of output device in each 500W amp. This is somewhat strange, but I found it neccessary for safety. At partytime there can happen everything, but "the show must go on"!
I hope you won't take this as attack, just I couldn't hold it in me.
Best regards:
Peter Sass
Few Questions about variations and Supply Voltages
Hi Lars,
I been lerking about this fourm for few weeks as I research my options to meet some unique amplifier needs. I have seen the schematic you posted for teh 500W 8 ohm version. Just so you are aware I am interested in quality sounding amp, not one with "best" stats. I am aware you may not be able to comment yet on, when do what you expect and what you know will be just fine to know. That said I suspect THD to likely be =< 0.01% at any power/frequency level =< the "rated" highest wattage of design. As asked above by others, I am not in need of 500/1000W 8/4 ohm design. Just so you are aware, I am not even close to skills to be able to design amps or pre-amps, but I seem to do well in my "research" skills before I build and odd minor modification the odd time
Here are my questions:
1) Is this design stable and not too far out of a =< 0.01% THD at 2ohm load? Not reactive load as in electrostatic speakers, but for cone speakers, i.e. 4 ohm drivers or two 8 or 6 ohm units in parallel which implies 4 or 3 ohm effective load.
2) What power supply voltage/VA would be suggested for 60/120/240W, or 100/200/400W for 8/4/2 code driver load? I alos like one in the 15W range as well. I am fully aware ther eare lots of Class A designs in that range, but please humour me.
3) What changes would be made to limit -3db bandwidth to 3Hz - 75KHz, and formula in case I need to experiment with -3db points?
4) I am partial to a single pair or two pairs of power output devices. What ideal devices would you use for best sound and load handling stability? Device cost is not major factor, but feel free to give best choice and what you feel cost effective and any comments on differences.
5) Refining question 4, what would be the choices to using devices 100% or 200% derated to rated designed load/wattage? Example, for given device +-pair(s) where max would be 200/400/800 into 8/4/2 ohm I rather use and have PSU of V/VA suitable for 100/200/400, or 50/100/200 use
6) You state "To keep thermal contact as good as possible we recommend NOT using MICA or Silicone insulators. Simply mount the MOSFET's directly on a black anodized aluminium heatsink." does that mean the heatsink has to be sisulated form chassis and/or signal ground?
7) I happen to have IRF630's (10) and IRF9630's (10) here and not used yet. I still am most interested in answers to the above questions for system application I am trying to acheive, but this design you have may have a use for spare Amps or ones I can build for couple friends. Could either the IRF630 ir IRF9630 be used to build amps of this design using one output pair or two (as your design has, not the complementary pair I have spares of) of either or both (assume some flipping be required for the IRF9360) devices? What drive load and output power could be achieved? Any chance output would be x/2x/4x for 8/4/2 ohm output load? Any comments on expected sound quality and THD?
Of course if you have any comments or things that need to be asked or stated from the above questions feel free to do so Lars.
TIA, John L. Males
Hi Lars,
I been lerking about this fourm for few weeks as I research my options to meet some unique amplifier needs. I have seen the schematic you posted for teh 500W 8 ohm version. Just so you are aware I am interested in quality sounding amp, not one with "best" stats. I am aware you may not be able to comment yet on, when do what you expect and what you know will be just fine to know. That said I suspect THD to likely be =< 0.01% at any power/frequency level =< the "rated" highest wattage of design. As asked above by others, I am not in need of 500/1000W 8/4 ohm design. Just so you are aware, I am not even close to skills to be able to design amps or pre-amps, but I seem to do well in my "research" skills before I build and odd minor modification the odd time
Here are my questions:
1) Is this design stable and not too far out of a =< 0.01% THD at 2ohm load? Not reactive load as in electrostatic speakers, but for cone speakers, i.e. 4 ohm drivers or two 8 or 6 ohm units in parallel which implies 4 or 3 ohm effective load.
2) What power supply voltage/VA would be suggested for 60/120/240W, or 100/200/400W for 8/4/2 code driver load? I alos like one in the 15W range as well. I am fully aware ther eare lots of Class A designs in that range, but please humour me.
3) What changes would be made to limit -3db bandwidth to 3Hz - 75KHz, and formula in case I need to experiment with -3db points?
4) I am partial to a single pair or two pairs of power output devices. What ideal devices would you use for best sound and load handling stability? Device cost is not major factor, but feel free to give best choice and what you feel cost effective and any comments on differences.
5) Refining question 4, what would be the choices to using devices 100% or 200% derated to rated designed load/wattage? Example, for given device +-pair(s) where max would be 200/400/800 into 8/4/2 ohm I rather use and have PSU of V/VA suitable for 100/200/400, or 50/100/200 use
6) You state "To keep thermal contact as good as possible we recommend NOT using MICA or Silicone insulators. Simply mount the MOSFET's directly on a black anodized aluminium heatsink." does that mean the heatsink has to be sisulated form chassis and/or signal ground?
7) I happen to have IRF630's (10) and IRF9630's (10) here and not used yet. I still am most interested in answers to the above questions for system application I am trying to acheive, but this design you have may have a use for spare Amps or ones I can build for couple friends. Could either the IRF630 ir IRF9630 be used to build amps of this design using one output pair or two (as your design has, not the complementary pair I have spares of) of either or both (assume some flipping be required for the IRF9360) devices? What drive load and output power could be achieved? Any chance output would be x/2x/4x for 8/4/2 ohm output load? Any comments on expected sound quality and THD?
Of course if you have any comments or things that need to be asked or stated from the above questions feel free to do so Lars.
TIA, John L. Males
Hello,
I saw Mr Lars Clausen,s N-Channel amplifier schematic (500 Watt / 1000 Watt version) , Its quite different from N-channel designs I had seen eirlear . Mr Clausen has recomended 2pairs of output mosfets for lower power versions , instead it could be complementry output as there is already one nos of Irf -p9240 another could be added and on the other side another Irf-p250 thus two pairs of Irf-p250/ Irf-p9240 and the 10ohm resistor shorted.This will not make much differance in cost
For the higher wattage the N-channel version is most economical.
I saw Mr Lars Clausen,s N-Channel amplifier schematic (500 Watt / 1000 Watt version) , Its quite different from N-channel designs I had seen eirlear . Mr Clausen has recomended 2pairs of output mosfets for lower power versions , instead it could be complementry output as there is already one nos of Irf -p9240 another could be added and on the other side another Irf-p250 thus two pairs of Irf-p250/ Irf-p9240 and the 10ohm resistor shorted.This will not make much differance in cost
For the higher wattage the N-channel version is most economical.
Peter Sass: I agree with you, if you are going to make a PA amplifier for stage use, i would consider also the kind of things you mention, oversizing the amplifier and also building in some protection circuits. And it's definitely cool, that your amplifier could withstand the hard treatment you describe!
With regards to your calculations, they are correct except for one thing, you forget that even if you play a half Vrail squarewave, it is only dissipating power half of the time. Translating to the fact the output devices can share the dissipation load. (If you look at the SOA curve in the datasheet you can also see that the device can dissipate much larger amounts of power in short periods, than sustained DC).
For example IRPF250N can dissipate 214 Watts at 25C and derates 1.4 W/K, so at 100 degrees you are 75 x 1.4 watts down = 105 Watts.
This means you can dissipate average 214 - 105 Watts = 109 Watts (average) in each device at 100 degrees C. (Providing the good heat sinking). So 2 x 109 = 218 Watts of heat loss is the limit. In our example you need to dissipate "only" 144 Watts, with the nice regulated PSU of 48V, so it's no problem.
There are many ways to make short circuit protection, a good way is to shut down the amplifier all together in case of short circuit, and then maybe pump it back on after some seconds. This way you don't need to dissipate the whole power supply's max power in the output in case something is wrong. And the user is sure to know there is a problem Can be done with a few transistors and a NE555 chip.
Have fun : Lars)
With regards to your calculations, they are correct except for one thing, you forget that even if you play a half Vrail squarewave, it is only dissipating power half of the time. Translating to the fact the output devices can share the dissipation load. (If you look at the SOA curve in the datasheet you can also see that the device can dissipate much larger amounts of power in short periods, than sustained DC).
For example IRPF250N can dissipate 214 Watts at 25C and derates 1.4 W/K, so at 100 degrees you are 75 x 1.4 watts down = 105 Watts.
This means you can dissipate average 214 - 105 Watts = 109 Watts (average) in each device at 100 degrees C. (Providing the good heat sinking). So 2 x 109 = 218 Watts of heat loss is the limit. In our example you need to dissipate "only" 144 Watts, with the nice regulated PSU of 48V, so it's no problem.
There are many ways to make short circuit protection, a good way is to shut down the amplifier all together in case of short circuit, and then maybe pump it back on after some seconds. This way you don't need to dissipate the whole power supply's max power in the output in case something is wrong. And the user is sure to know there is a problem
Can be done with a few transistors and a NE555 chip.Have fun : Lars)
Lars!
Just quickly 'cos it's past midnight here:
I didn't forget. I wrote: "for half period". But for an IRFP140N ("Without crossing safety limits you can get 239 Watts audio power out of a single pair of IRFP140N, (at 100 degrees C)") even the half of it is too much at 100 C!
"If you look at the SOA curve in the datasheet you can also see that the device can dissipate much larger amounts of power in short periods, than sustained DC)"
These are single pulse ratings! Check it out!
"The user is sure to know there is a problem.."
I didn't meet any user, who wanted to know about this kind of problems (impedance curve, cos(fi), etc...)
Just quickly 'cos it's past midnight here:
I didn't forget. I wrote: "for half period". But for an IRFP140N ("Without crossing safety limits you can get 239 Watts audio power out of a single pair of IRFP140N, (at 100 degrees C)") even the half of it is too much at 100 C!
"If you look at the SOA curve in the datasheet you can also see that the device can dissipate much larger amounts of power in short periods, than sustained DC)"
These are single pulse ratings! Check it out!
"The user is sure to know there is a problem.."
I didn't meet any user, who wanted to know about this kind of problems (impedance curve, cos(fi), etc...)
Opps, seems being diyAudio Newbie Memeber has "delay"
Lars,
Post #49 by me yesterday I think you may have missed as it was hidden for over 12 hours. Seems as new diyAudio memner there is a initial screen process of postings. Fair and ok, though I was not aware of that. Net of this is the posting was not allowed to be viewed for over 12 hours. I think this may have caused you to miss the posting as well as others. Hence this posting to go look at posting 49 by me is you wish and if you decide you have any comments.
TIA, John L. Males
08 September 2004 23:22 EDT
Lars,
Post #49 by me yesterday I think you may have missed as it was hidden for over 12 hours. Seems as new diyAudio memner there is a initial screen process of postings. Fair and ok, though I was not aware of that. Net of this is the posting was not allowed to be viewed for over 12 hours. I think this may have caused you to miss the posting as well as others. Hence this posting to go look at posting 49 by me is you wish and if you decide you have any comments.
TIA, John L. Males
08 September 2004 23:22 EDT
((( Charles: Yes exactly.
The schematic of the amplifier (with this safety circuit) is also to be found on the website, so you can take it from there, if you dont want to design your own version.
Cheers Lars )))
Lars
could you please give the link for the safety circuit as I could not find it
Thanks
Rajeev
The schematic of the amplifier (with this safety circuit) is also to be found on the website, so you can take it from there, if you dont want to design your own version.
Cheers Lars )))
Lars
could you please give the link for the safety circuit as I could not find it
Thanks
Rajeev
Lars
A little OT. I saw your Millennium schematic on your web site. Just curious - without NFB, what sets the gain?
I see an initial gain from the 2k diffamp emitter resistors and the 4k R18/R20. But there is further gain from T10/T12, with 1k emitter resistors, and ???? collector load ????. So what exactly sets the gain?
(Nice job, by the way!)
Rgrds
Alan
A little OT. I saw your Millennium schematic on your web site. Just curious - without NFB, what sets the gain?
I see an initial gain from the 2k diffamp emitter resistors and the 4k R18/R20. But there is further gain from T10/T12, with 1k emitter resistors, and ???? collector load ????. So what exactly sets the gain?
(Nice job, by the way!)
Rgrds
Alan
Hi Lars.
I tried to simulate your schematic, and he seems to work well, I only have 2 questions:
1) How shoud be the value of the offset trimmer R8 for the DC offset? I can't go down 80 mV with it. Meybe a DC servo could be better?
2) I have seen that the overall gain of the amp is about 46 dB and you apply about 6dB of negative feedback. Because I need a gain of about 20dB, can I increase the feedback to 26dB or it is better to reduce the gain elsewhere, and if so, How. I want to use this amp on a 12" woofer cutted at 700 Hz 12 dB/octave in sealed box.
Thank you.
Ciao.
I tried to simulate your schematic, and he seems to work well, I only have 2 questions:
1) How shoud be the value of the offset trimmer R8 for the DC offset? I can't go down 80 mV with it. Meybe a DC servo could be better?
2) I have seen that the overall gain of the amp is about 46 dB and you apply about 6dB of negative feedback. Because I need a gain of about 20dB, can I increase the feedback to 26dB or it is better to reduce the gain elsewhere, and if so, How. I want to use this amp on a 12" woofer cutted at 700 Hz 12 dB/octave in sealed box.
Thank you.
Ciao.
pro: Hi
1..If you use a 10k you will get a nice and quiet adjustment on the middle, and more aggressive at the ends. You can change the voltage +/- 100 mV on the input, so it should be enough. I never really trusted simulators anyway
2..I think you can reduce the feedback resistor, to get 20dB gain, didnt try it though. Worst case you would have to connect a small capacitor from the gain stage to the feedback input. Maybe 10pF. I doubt it's nessescary.
Alan: The gain is controlled by converting voltages into currents, and back again to voltages with a higher resistance, and so higher voltage (with same current). It can be boiled down to (R4+(1/2xR29)) / (1/2xR17).
NFB can be various things, and thus have lead to many unnecessary discussions over the years.
By Non-Feedback we mean: No feedback LOOPS.
Theoretically a resistor in the emitter of a transistor is feedback, (but not a feedback LOOP) so many people have during the cause of time claimed that we actually have feedback, because of the resistors. With their interpretation (as taken from the general analog school) of the word feedback, they are absolutely right. However in case of Audio, the interpretation is different. An audio amplifier completely without feedback (in sense from the engineering world) would be worthless, and so during the years the term non-feedback has been adapted to something slightly different, but more meaningful for audio amplifiers. And that is: No feedback LOOPS. The harmful thing with feedback loops is the time delay in the loop. To say it in a popular language:
1..You correct your amplifier's output signal in 2 microseconds by comparing the input signal, with the output signal 2 microseconds ago. In other words the input and the output of the amplifier are not on the same place of the signal curve.
2..You cannot correct an error on the output before it is present on the output. And then it is kind of too late. If you understand
Thanks for your comments
Lars
1..If you use a 10k you will get a nice and quiet adjustment on the middle, and more aggressive at the ends. You can change the voltage +/- 100 mV on the input, so it should be enough. I never really trusted simulators anyway
2..I think you can reduce the feedback resistor, to get 20dB gain, didnt try it though. Worst case you would have to connect a small capacitor from the gain stage to the feedback input. Maybe 10pF. I doubt it's nessescary.
Alan: The gain is controlled by converting voltages into currents, and back again to voltages with a higher resistance, and so higher voltage (with same current). It can be boiled down to (R4+(1/2xR29)) / (1/2xR17).
NFB can be various things, and thus have lead to many unnecessary discussions over the years.
By Non-Feedback we mean: No feedback LOOPS.
Theoretically a resistor in the emitter of a transistor is feedback, (but not a feedback LOOP) so many people have during the cause of time claimed that we actually have feedback, because of the resistors. With their interpretation (as taken from the general analog school) of the word feedback, they are absolutely right. However in case of Audio, the interpretation is different. An audio amplifier completely without feedback (in sense from the engineering world) would be worthless, and so during the years the term non-feedback has been adapted to something slightly different, but more meaningful for audio amplifiers. And that is: No feedback LOOPS. The harmful thing with feedback loops is the time delay in the loop. To say it in a popular language:
1..You correct your amplifier's output signal in 2 microseconds by comparing the input signal, with the output signal 2 microseconds ago. In other words the input and the output of the amplifier are not on the same place of the signal curve.
2..You cannot correct an error on the output before it is present on the output. And then it is kind of too late. If you understand
Thanks for your comments
Lars
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.