Patrick's measurements are very intresting, thank you for the detailed presentation.
From the whole thing, it seems to me that the quite substantial lower transconductance of the P channels MOSFET's will not have major drawbacks because they have also a much higher Vgs threshold.
Therefore, the required higher JFET drain resistor of the upper halve does compensate for this.
Also shown clearly in Patrick's spreadsheet.
My thoughts on better complements here:
http://www.diyaudio.com/forums/showthread.php?postid=1494636#post1494636
As OT, what would be the consequence in a F4 when using MOSFET's with these transconductance differences? Very high K2?
Tino
From the whole thing, it seems to me that the quite substantial lower transconductance of the P channels MOSFET's will not have major drawbacks because they have also a much higher Vgs threshold.
Therefore, the required higher JFET drain resistor of the upper halve does compensate for this.
Also shown clearly in Patrick's spreadsheet.
My thoughts on better complements here:
http://www.diyaudio.com/forums/showthread.php?postid=1494636#post1494636
As OT, what would be the consequence in a F4 when using MOSFET's with these transconductance differences? Very high K2?
Tino
> My thoughts on better complements here:
Tino,
You cannot get better complementaries than the Toshibas, IMHO !!
http://www.diyaudio.com/forums/showthread.php?s=&postid=1507044&highlight=#post1507044
Also the spreadsheet I posted only shows gain balancing, but says nothing about 2nd and 3rd. I have another spreadsheet for that, which is 1.8MB and too complicated to explain. But there, it shows that the Toshibas give half the distortion as the Fairchilds when trimmed properly (to account for the Transconductance differences of the JFETs.)
Not that it is important, whether at 0.003% or 0.006%
Patrick
Tino,
You cannot get better complementaries than the Toshibas, IMHO !!
http://www.diyaudio.com/forums/showthread.php?s=&postid=1507044&highlight=#post1507044
Also the spreadsheet I posted only shows gain balancing, but says nothing about 2nd and 3rd. I have another spreadsheet for that, which is 1.8MB and too complicated to explain. But there, it shows that the Toshibas give half the distortion as the Fairchilds when trimmed properly (to account for the Transconductance differences of the JFETs.)
Not that it is important, whether at 0.003% or 0.006%
Patrick
May I come back to my question a couple of pages before ?
http://www.diyaudio.com/forums/showthread.php?postid=1567732#post1567732
>> What would you suggest to flatten the small hump? Gate resistor to 22R does not seem to do much.
> What hump is that?
>> The second plot when the amp is loaded with a dummy speaker load (reactive, about 4 ohm).
Thanks,
Patrick
http://www.diyaudio.com/forums/showthread.php?postid=1567732#post1567732
>> What would you suggest to flatten the small hump? Gate resistor to 22R does not seem to do much.
> What hump is that?
>> The second plot when the amp is loaded with a dummy speaker load (reactive, about 4 ohm).
Thanks,
Patrick
!
> Somebody please remind me again, for use in an F5 topology, what is the reason to choose the IRF or Fairchild MOSFETs over 2SK1530s and 2SJ201s?
>> Transconductance.
According to my measurements, at 1A, IRFP240 has 3.5S, 2SK1530 has 3.2S, so within 10% and IMHO no issue. If you are really fuzzy, then lower the source resistor will compensate for that.
Toshibas are more costly (at least 3x at retail price), and more difficult to get, and not from an American manufacturer.
Patrick
>> Transconductance.
According to my measurements, at 1A, IRFP240 has 3.5S, 2SK1530 has 3.2S, so within 10% and IMHO no issue. If you are really fuzzy, then lower the source resistor will compensate for that.
Toshibas are more costly (at least 3x at retail price), and more difficult to get, and not from an American manufacturer.
Patrick
Well at this low THD-level pcb-artwork, wiring, grounding, trafo shielding are already more important and can dominate THD.
But, it's still tempting to use the Toshibas
Nevertheless, the F5's frequency response is very sensitive to parts choice, so if you want to achieve the same results as Nelson shows off in the article one has to stay with the Fairchilds. Otherwise trimming is necessary which not everbody has the equipment for.
Have fun, Hannes
> Transconductance numbers of the 9240 and SJ201 at 1A-25Vds are more interesting
Don't trust the datasheets.
In any case, 2SJ201 has more than FQA12P20, but that is not the entire story.
> Nevertheless, the F5's frequency response is very sensitive to parts choice, so if you want to achieve the same results as Nelson shows off in the article one has to stay with the Fairchilds.
I have both versions built, and they have very similar freq response. You just need to read what I already published.
In any case, don't do armchair DIY and start building ...
Patrick
Don't trust the datasheets.
In any case, 2SJ201 has more than FQA12P20, but that is not the entire story.
> Nevertheless, the F5's frequency response is very sensitive to parts choice, so if you want to achieve the same results as Nelson shows off in the article one has to stay with the Fairchilds.
I have both versions built, and they have very similar freq response. You just need to read what I already published.
In any case, don't do armchair DIY and start building ...
Patrick