You can still get 2SK2013/2SJ313.
Next choice is FQP3N30/3P20, also NOS.
Someone has cleaned them off Rochester.
But you can still find them.
Patrick
without bias spreader(vbe multiplier) every HEXFET/lateral will blow. this circuit will not work proberly .
Rs resistor you can quit if you have one Laterals with more MOSFETS you need 0,1 - 0,22R. each.
at F5 you have a thermal contact with a extra componet to the MOSFET to be aware of blow up the POWER MOSFET!
TH1 + TH2
kr
chris
Have you measured them ?
Are they still good ?
BTW they are Hitachi laterals.
Essentially the same as 2SK1058/2SJ162.
Or similar to Exicon single die.
Patrick
My F5X has been working for years with 60°C heatsink temperature.
It's all down to thermal management.
All explained in detail scientifically in my article at Linear Audio.
You can run Exicons and Hitachi laterals without Rsource if biased above 100mA single die.
They will then have negative tempco and hence self stabilise thermally.
Patrick
Yes; And even paired them. I even got a few that had N and P with same threshold. But it is just dumbo check on I think it was a CCS with 100mA. And no heatsink, maybe that is moot with the neg tempco.
I did a quick simulation of that GrannyMOS (using parts/figures I inserted in the sim to stabilise, differing from original component values. . . Does not look bad:
Try FQP3N30/3P20 for Q26,27.
You will need to adjust R72/74 for proper biasing of the second stage.
For 4 ohm load, I would use at least 4 output pairs and 2A bias.
Will only give you 32Wrms class A.
Patrick
.
You will need to adjust R72/74 for proper biasing of the second stage.
For 4 ohm load, I would use at least 4 output pairs and 2A bias.
Will only give you 32Wrms class A.
Patrick
.
Last edited:
Patrick, will the (still venerated?) 2SK2013 2SJ313 be usefull? I have some used laying around; I can't see the FQP's readily in the shops (Farnell/Mouser) - else I'll get the quality the Red States like: imho fake.
As I have the potmeter on the board, I'll set for say 100mA and reduce the bias threshold resistors R82 R83 as needed in vivo.
As I have the potmeter on the board, I'll set for say 100mA and reduce the bias threshold resistors R82 R83 as needed in vivo.
I built the original GrandMos amplifier with Hitachi laterals, I did not like that much the result. Maybe some tweak would help to bring some life (dynamics) into that amplifier. After I took out the power mosfets into the ProFet (4PC bridged) amplifier. All I can say Pass F5 with Toshiba mosfets outperform both of those amplifiers! At least in my system, I prefer the Pass F5. Juma has something similar but he used VAS Toshiba mosfet drivers.... I have the layout for 3 pairs lateral or maybe two pairs of dual die mosfets.... I want to follow the thread...............
Gabor. . . . Ouch. Dynamics. Backstage sound. Violins. The female voice.
Part of the hard stuff. -->
A nice thing on the F5 is the amplification (common source); that increases the internal resistance of the output. Adding a small source resistor (see post 1, .5Ω) will dramatically influence the output resistance.
A 'GrannyMos type' source follower with FB over the whole loop might make it all dull...??
- An initial idea I had was to 'stay true to F4' & make the mosfets into followers without feedback over the whole loop. Just over the driver stage, and please do not forget to check the DC drift. The output on 4Ω is just 800mdB down, bandwidth is not impaired.
Note I differentiated the source resistors... in this small FB Loop=F4 output configuration.
Part of the hard stuff. -->
A nice thing on the F5 is the amplification (common source); that increases the internal resistance of the output. Adding a small source resistor (see post 1, .5Ω) will dramatically influence the output resistance.
A 'GrannyMos type' source follower with FB over the whole loop might make it all dull...??
- An initial idea I had was to 'stay true to F4' & make the mosfets into followers without feedback over the whole loop. Just over the driver stage, and please do not forget to check the DC drift. The output on 4Ω is just 800mdB down, bandwidth is not impaired.
Note I differentiated the source resistors... in this small FB Loop=F4 output configuration.
- On LeClasseA - One can by just a small intentional deviation of the emitter resistance of the output pair (that, being the emitter of the driver) change the sound completely, on paper at least. The Hfe of the driver amplifies the smallest difference and then - - varies the contribution, such that the lower half -even in full class A- will only contribute 20% of the AC. I'm working that out now, almost ready with soldering. Intention: A LeClasse-as-driver for 2SK182 ...
We have done :
No feedback from the output stage
Loop feedback from the output stage
Mixed feedback from both frontend and output stage.
www.diyaudio.com/community/threads/the-f5pi-f5p-voltage-gain-stage-m2-output-stage-perfectly-integrated.406024/
www.diyaudio.com/community/threads/the-m2-output-stage-in-class-a-b-and-maybe-a-power-whammy.390636/
😊
Patrick
It is probably an unfair comparison between a Class A amp and low-bias, low-NFB Class AB amps.
It doesn't mean the topology is inferior by any means.
Patrick
Yes, first choice.
FQP only second choice.
I only suggested DWP because you were using laterals in your spice file.
Patrick