.
Good to have extra pair of eyes. Some time it's difficult see errors even when it's in-front of you.
KSC3503/KSA1381 100mA device MJE340/350 500mA.
I think 2 pair vertical mosfet require a buffer (emitter follower). Gate capacitance of vertical mosfet much higher than lateral mosfet. If VAS directly drive 2 pair vertical mosfet, the slew rate will be reduce.
You can consider use vertical mosfet output design in this thread. I think Still4given built it.
You can consider use vertical mosfet output design in this thread. I think Still4given built it.
That's why the unwritten rule that after you finish the layout , shut shop and look at it again only the next day !
Apart from making small corrections it looks very nice. What software are you using to make your layout ?
I also noticed Bimos warning. Both the power HEXFETS are driven from a collector. Just a short while ago this problem was covered on the forum. Not sure if it was this thread or somewhere else. In that design the output pair had emitter followers to drive them. I can't remember all the posts but a few that had emitter followers to drive the output stage also apparently had very good sound . Everything is relative of course so nothing like trying them out .
General Patton started a stalled tank by giving it a kick . That method was not mentioned in the guide book ! So you can try everything and see which one actually works acceptably !
Cheers.
Last edited:
Would it be as simple as swapping the positions of T1 and T2 to make them emitter followers?
The strange thing is that it looks pretty good in the sims.
Remember that BJT has only around 0.7 volt between base and emitter. So when you invert the emitter and collector (without adjusting everything else) the operating condition is simply not right (for proper voltage gain).
Vzaichenko wrote elsewhere:
"With regards to "no drivers" approach - HexFETs have got rather high and non-linear input capacitance (depends on Vgs), noticeably influencing VAS performance. Live measurements just confirm the theory - spectrums contain higher number of odd harmonics with higher levels. That's why I always add the follower - as soon as you drive the HexFETs with low impedance drivers, capacitance issue does not hurt any more."
Last edited:
This mod I'm doing is for me and I'll still build. Final schematic doesn't exist yet, I used part numbers that we found on FX8 and FH9's existent schematic.
I hope you enjoy
Guys you are great!
Many thanks to all who provided data to me!
I'm now going to bundle all data (schematic, layout, gerber, BoM) on the recent level of FH9(HV). Of course everbody can still use what he has in his drawer. I only want to give a flash of the current status of discussion what seems to be optimal. I will not count up the version number, but make a difference between the 'standard' version rails 35V to xxV? and the 'high voltage' version xxV? to 52V called as FH9 and FH9HV.
In case you have better ideas, please let it know me! I also expect some suggestions for 'xxV'.
Don't get me wrong, I like to do it primary for myself to get a good overview and though it might be useful to share it with community.
After you provided me with all data, I can do it rather easily. There is only one thing, I don't have: a schematic which can be edited. xrk971, you must have a schematic from u simulations. Ofc I don't know in which program u did it. Maybe u can tell me the program and provide the working file? In case, the schematic is in eagle, it would be great!
The schematic is not emitter follower, it is Complementary Feedback Pair (CFP).
Example of driving HEXFETs with low impedance:
http://www.diyaudio.com/forums/soli...e-old-ideas-1970s-ips-ops-31.html#post4540474
http://www.diyaudio.com/forums/soli...e-old-ideas-1970s-ips-ops-31.html#post4540474
Has this any relevance for FH9? Or is there only a need if u use more than 1 hexfet?
If u use more than 1 pair of HEXFET it is necessary because the VAS current simply cannot charge and discharge the parallel HEXFETS' input capacitance properly. This is why vzaichenko suggest to applying an unity gain buffer between the VAS and HEXFETs. If u using single pair of FETS it can be omitted. Higher the VAS current lower the HF distortion
Last edited:
Prior to the development of the FH9, I was given similar warnings and admonition of how one simply cannot just change the use of lateral FETs for vertical FETs in an amp design and that hexFETs just wouldn't work on an FX8. Well, it just took one more transistor to make a Vbe multiplier. The model showed it would work and my prototype also worked to confirm the model. Now we are talking about adding another set of gates to drive. Again, the model says it should work. I would rather put my bet with a model vs putting my trust in general good practice guidelines. Guidelines, are not always correct and we may be a special case once again.
I share several hexfet amplifiers A9, FX14, HV23... with or without drivers, F100 is upgrade with BJT or mosfet drivers version, HV23 is without drivers but with 30mA bias in VAS... FH9 with 20mA bias in VAS can drive 2 output pairs without drivers.
There is F250 (mosfet version of SR250).
There is F250 (mosfet version of SR250).
Attachments
It WOULD work. No doubt.
We are talking about very small differences here. Something that might be not audible in an ABX...
Sorry Mr. Apex, I didn't know some of these amp. That's why we need an extended directory of Your Amplifiers.