Hi Gaborbela
Regarding 2sk246, 2sj103, even though I have got them working quite nicely, I am not convinced they are better than using a low noise bipolar as cascode.
If you have them 2sc2240 and 2sa970 are quite nice as cascode devices as well as the already mentioned BC550/560.
No need to appologise, I think a lot of people here admire your stubborn enthusiasm.
Regarding 2sk246, 2sj103, even though I have got them working quite nicely, I am not convinced they are better than using a low noise bipolar as cascode.
If you have them 2sc2240 and 2sa970 are quite nice as cascode devices as well as the already mentioned BC550/560.
No need to appologise, I think a lot of people here admire your stubborn enthusiasm.
stop right now ;
if you're able to wrote that - you must be sentenced to ( pleasure!!) re-read F5 article at least 20 times !!!!!
both input jfet's drains and output mosfet's gates are referenced to rails , so - including any variable there ( as two independent supplies ) is big no-no ;
any fluctuation between two PSUs will find direct way to output gates and result in ( if nothing worse ) in horrendous sound
Thanks for the warning
I suppose that means best to dump the Renesas as well
Which I accidentally bought in my confusion, believing Toshibas to laterals as well
Obviously those laterals will continue to push me in the wrong direction
Pretty daft move
No more mess
Back to basics
I suppose that means best to dump the Renesas as well
Which I accidentally bought in my confusion, believing Toshibas to laterals as well
Obviously those laterals will continue to push me in the wrong direction
Pretty daft move
No more mess
Back to basics
Last edited:
I agree 100% with that. Independant power supplies for front end, and output, is gonna cause a lot of grief.
Tinitus if you haven't built the original version yet start with that first and then move on trying different variations.
It seems you are trying to make an excellent design more complicated for very little benefit.
If you are using 20V to 30V rails you won't need cascoding at the input.
If you want to put them in then feel free to do so but you won't get any significant benefit at low voltages.
Last edited:
I think you should try the lateral devices. They will work
a little differently, but you might be pleasantly surprised.
Yes I am definitely considering the laterals as my next variation.
They seem to perform quite nicely at lower bias currents compared to the IR devices.
I also like the look of the Toshiba devices 2sk1530/2SJ201 however I have no experience with them sound wise.
While with the laterals, I can say they do have a pleasant sound to them. In fact they are used at the output of my favourite bass amp of all time ie Trace Elliott AH 500-12
I think you should try the lateral devices. They will work
a little differently, but you might be pleasantly surprised.
I've been dying to remove the front end of my Borbely DC100,
which has three matched pairs of J56/K176 Hitachi TO3s per
channel. The ongoing back & forth about laterals had given me
pause. The DC100 is a good amp, more powerful but darker with a loss of detail compared to the F5.
I'm building up a second set of CVillers boards, to be used for a balanced F5. But, thought I'd try them, as built for the F5, on the Hitachi's first.
I hope to be 'pleasantly surprised'
and not up in smoke 
Sometimes it is hard to predict how measurement will correlate with what we hear.
It would be nice to get other peoples opinions on sound quality of these devices in the F5.
From what I understand (which is very little in fact) the laterals have lower transconductance so the amount of feedback will be less, so the measured performance does not look as good in terms of absolute numbers, however the higher order harmonics fall away very quickly with the lateral devices.
It would be nice to get other peoples opinions on sound quality of these devices in the F5.
From what I understand (which is very little in fact) the laterals have lower transconductance so the amount of feedback will be less, so the measured performance does not look as good in terms of absolute numbers, however the higher order harmonics fall away very quickly with the lateral devices.
One of my concerns is the higher input capacitance they have and three pairs will only triple that. If I recall correctly Erno B. had to go with heftier drivers, mosfets in case of the DC100, inorder to drive the outputs.
Tomorrow, I may put the 'pedal to the metal' and hobble the F5/DC100 Frankenstien together. Using the F5s 24v supply for the input board and the 56volt ps for the outputs. Grounds tied together.
I was under the impression that the laterals had lower input capacitance compared to the IR devices.
It looks like what you are proposing is exactly what Zen Mod warned against. ie independant power supplies for front end and output. If the voltage ripple is different between front end and output, then the PSRR is reduced dramatically.
It looks like what you are proposing is exactly what Zen Mod warned against. ie independant power supplies for front end and output. If the voltage ripple is different between front end and output, then the PSRR is reduced dramatically.
I have to check that, and lower capacitance would be great. I'll have to dig out the old Audio Amateurs and see what the drive requirements are.
Independent supplies, I guess I skimmed over Zen Mods post. I vaguely recall someone posting that they were dropping the idea of separate supplies.
I will go with 24 volts one supply. I recall a class A amp article in AA using the 35/150 Hitachis. Multiple pairs and if I'm correct 24 volt supply.
Which allowed them to be run safely at high bias.
[QUOTE
If you are using 20V to 30V rails you won't need cascoding at the input.
If you want to put them in then feel free to do so but you won't get any significant benefit at low voltages.[/QUOTE]
Hello Thanh
I was looking for these sentence almost two weeks !
It is great to know you tested in the real life and share your opinion .
Now I know were to start and which way to go .
Thank you so much!
Greets
If you are using 20V to 30V rails you won't need cascoding at the input.
If you want to put them in then feel free to do so but you won't get any significant benefit at low voltages.[/QUOTE]
Hello Thanh
I was looking for these sentence almost two weeks !
It is great to know you tested in the real life and share your opinion .
Now I know were to start and which way to go .
Thank you so much!
Greets
I think you should try the lateral devices. They will work
a little differently, but you might be pleasantly surprised.
Tinitus
I built amplifier like the F5 with similar laterals you have and it sound great ,one of the best amp I built in the last 20 years
.That gave the idea to try out the F5 X or F5 in bridged mode .
Wich one will work out to me .
Greets
The other benefits from cascode are improved PSRR and possibly lower input capacitance.
These benefits while not insignificant, probably won't be audible, espescially if your power supply is already up to the task.
The main reason for their use was to be able to make a high powered version of the F5 (eg using 40V rails) in which case it is necessary to share some of power dissipation with the cascode device
These benefits while not insignificant, probably won't be audible, espescially if your power supply is already up to the task.
The main reason for their use was to be able to make a high powered version of the F5 (eg using 40V rails) in which case it is necessary to share some of power dissipation with the cascode device
Last edited:
Vital numbers for a quasi-complementary output stage, but irrelevant for a source follower setup.
The capacitance of the F5 output stage devices that has to be charged and discharged is Cgd, Cgd = Crss.
Datasheet numbers for Crss are 10pF-25pF for the Exicon devices and 160pF for the IRFP240.
And in reality these numbers can differ a great deal, it has been posted numerous times not to have blind faith in datasheets.
Straight from The Mastah : www.diyaudio.com/forums/pass-labs/129040-mosfet-output-stage-capacitance.html#post1599135