Diomedian said:
The following paragraphs are a question, rather than a statement.
When using the Fairchild parts, the transconductance of the the P-channel parts is higher - so in the upper half of the circuit, you've got a lower S jfet and a higher S mosfet, and the lower half of the circuit you get a higher S jfet feeding a lower S mosfet, and these two combinations (partially?) cancel each other out.
Now when you're using the Toshiba mosfets, based on the fact that they're so well matched for transconductance, and that the jfets still aren't, you need to reduce the current output of the P-channel jfet (2sj74) and this is why you've increased source and feedback resistors.
Is this in the right ball park?
Thanks for the tips Patrick - I may end up going Fairchild unless I manage a bit better understanding here.
-j
you must look at amp as whole ....
besides xconductance , N and P type mozfets have different Ugs ;
that means that - chasing needed current and offset in/on output of amp - two pots will have different settings ............ so input Jfets will have different gains .
solution is simple .... ask Papa which THD gadget he's using ...... and what's good value of this horny resistor , for start
edit : writing about different xconductance - I have strictly IRFs on mind
Toshiba's are different animals
> Now when you're using the Toshiba mosfets, based on the fact that they're so well matched for transconductance, and that the jfets still aren't, .....
you need to reduce the effective transconductance of the P JFET by using a slightly larger source resistor. Once you have done that, of course you need to increase the feedback resistor as well to maintain the same closed loop gain.
At least that is my understanding.
> how can we redesign with different transistors ? nelson says simulations are not good enough..
I think I'll leave this question to the Master himself, if I may .....
> besides xconductance , N and P type mozfets have different Ugs ;
Not the case for Toshiba's. They have essentially the same Vgs for a particular current. That is the beauty, especially if you want to reduce the source resistor of the MOSFETs.
Patrick
you need to reduce the effective transconductance of the P JFET by using a slightly larger source resistor. Once you have done that, of course you need to increase the feedback resistor as well to maintain the same closed loop gain.
At least that is my understanding.
> how can we redesign with different transistors ? nelson says simulations are not good enough..
I think I'll leave this question to the Master himself, if I may .....
> besides xconductance , N and P type mozfets have different Ugs ;
Not the case for Toshiba's. They have essentially the same Vgs for a particular current. That is the beauty, especially if you want to reduce the source resistor of the MOSFETs.
Patrick
> how can we redesign with different transistors ? nelson says simulations are not good enough..
I think I'll leave this question to the Master himself, if I may .....
> sorry - off course that I meant strictly on IRFs , but I forgot to write that .
Even worst for Fairchilds. And the transconductances differ by miles. You just need to roll back umteen pages for the measured data I published here. This is why the source resistor has to be such high value.
Of course one can compensate for all these things by clever design. I am just too lazy and rather use better matched (complementary) devices.
If only the JFETs would be just as good .....
Patrick
I think I'll leave this question to the Master himself, if I may .....
> sorry - off course that I meant strictly on IRFs , but I forgot to write that .
Even worst for Fairchilds. And the transconductances differ by miles. You just need to roll back umteen pages for the measured data I published here. This is why the source resistor has to be such high value.
Of course one can compensate for all these things by clever design. I am just too lazy and rather use better matched (complementary) devices.
If only the JFETs would be just as good .....
Patrick
Bridged F5's
Jacco, you are no doubt correct - but I wasn't talking about altering the bias at all. The amp will easily put 2.5A into a 4 ohm load, unbridged, and 2.5A into an 8 ohm load bridged, but probably not in Class A and not steady state with music signals. Putting two F5s in parallel on each side of the bridge doubles the bias and will increase the Class A output point, but also doubles the part count, heat dissipation, yada-yada-yada. It would also probably give you an amp comparable to a DIY Pass X160. Maybe I need to order another four boards from Peter Daniel......
Jacco, you are no doubt correct - but I wasn't talking about altering the bias at all. The amp will easily put 2.5A into a 4 ohm load, unbridged, and 2.5A into an 8 ohm load bridged, but probably not in Class A and not steady state with music signals. Putting two F5s in parallel on each side of the bridge doubles the bias and will increase the Class A output point, but also doubles the part count, heat dissipation, yada-yada-yada. It would also probably give you an amp comparable to a DIY Pass X160. Maybe I need to order another four boards from Peter Daniel......
Bridged F5's
Jacco - cool. I think the last thing one wants to mess with is Papa's choice of the bias point on his selected MOSFETs.....my guess is Pass does a LOT of testing and listening to determine this, and I don't presume to think I can do any better.
I do think a "double-bridged" F5 would be a pretty cool project and would make one hell of an amplifier. In fact, given the simplicity of the circuit, ease of building these things and the high input impedance of the JFET's Pass has chosen, I suspect "scale-up" was an implicit design objective.
Jacco - cool. I think the last thing one wants to mess with is Papa's choice of the bias point on his selected MOSFETs.....my guess is Pass does a LOT of testing and listening to determine this, and I don't presume to think I can do any better.
I do think a "double-bridged" F5 would be a pretty cool project and would make one hell of an amplifier. In fact, given the simplicity of the circuit, ease of building these things and the high input impedance of the JFET's Pass has chosen, I suspect "scale-up" was an implicit design objective.
Zen Mod said:
if nothing else - go to Papa's site ( Pass Diy ) and download few pdfs ...
I mean that in "discrete OP for DIY " ( or something like that ) file you can find how to test jfets and mosfets .
use your imagination , and learn something ......
Thanks Zen Mod.
I'll build the jfet testing circuit and see what comes of it. I just thought someone might know right away if putting Q2 in backwards would fry it or not. Anyway.... I've ordered some more Q1 and Q2.
If Q1/2 check out ok on the test circuit I suppose the next step is to look at the current limiting parts. Test TH1/2. I'm a little concerned that I've actually used NTE128/129 instead of the ZTX450/550 (which is what mouser threw up for ZTX450/550 part). So that might also be the problem.
Thanks for the help.. will hope to do some testing tomorrow and will report back here with the results.
Fran
I was tapping and drilling on my F5 chassis tonight after a long day and decided it was time to just kick back and relax... lest I make a(nother) mistake 
One more heatsink to go and hopefully fire it up with no smoke. The power supply with Christian's board and the AnTek toroid works just fine so far at +/-25 unloaded.
Just remember stuff happens to all of us. Hang in there. It should be easy to fix and it will be worth it!
One more heatsink to go and hopefully fire it up with no smoke. The power supply with Christian's board and the AnTek toroid works just fine so far at +/-25 unloaded. Just remember stuff happens to all of us. Hang in there. It should be easy to fix and it will be worth it!
woodturner-fran said:
If you check that link, you might think that it shouldn't: http://www.diyaudio.com/forums/showthread.php?postid=1202052#post1202052
OK, so I done some more testing. First up was to check out Q1/Q2. I made up the circuit that someone posted a little back up here (Idss) - and I got 10mA for SJ74 and 9mA for SK170. So it would seem that they were alive still.
Next I decided to look at the current limiting circuit. I removed all the components from the circuit as detailed earlier in the thread. Now I found I could get the bias set ok across R11 and 12. Its set as 0.6V on both but my offset at the speaker terminals is high. The mosfets are heating up nicely, but they aren't that hot. I've left it on for a little while and will go back and check it again. The offset I'm seeing is "floating" ie moving around a good bit while the bias values seem pretty stable. I get 4.4-4.6V G-S on Q3 and 4 and I meaure 16V D to S on Q1 and 2.
Does this sound OK?
Fran
Next I decided to look at the current limiting circuit. I removed all the components from the circuit as detailed earlier in the thread. Now I found I could get the bias set ok across R11 and 12. Its set as 0.6V on both but my offset at the speaker terminals is high. The mosfets are heating up nicely, but they aren't that hot. I've left it on for a little while and will go back and check it again. The offset I'm seeing is "floating" ie moving around a good bit while the bias values seem pretty stable. I get 4.4-4.6V G-S on Q3 and 4 and I meaure 16V D to S on Q1 and 2.
Does this sound OK?
Fran