F5 with 2SK2013/2SJ313

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
M

masag1

Member
Joined 2008
S

Samuel Jayaraj

Member
Joined 2001
I am working on a PCB layout for this amp. Today I went out and got all the parts required. The NTCs I got are 5k instead of 4k7, but I guess it is not a problem. I found that 47,000uF caps were big and expensive. So I bought 50 Elna caps rated at 4,700uF 50V and plan on parallel them to achieve the desired capacitance.

4 questions I have are:

1) I plan on using the caps in this configuration - CCC R CCC R CCCC - hence, 10 caps will give me 47,000uF. Should I also include an R component in the GND path? I have seen this employed a few times.
2) I have 8 each of the Toshiba devices. Is there any benefit in using 4 per bank or adhere to Juma's scheme of 3 per bank? Conflicting thoughts here as I would like to bias as high as possible (1.6 Amps per channel, 4 devices per bank) but am tempted to go like Juma with 1.2 Amps, save on the 2 pairs of Toshiba Mosfets and try out Juma's 25watt ClassA with Hitachi/Renesas laterals.
3) I also managed to get 70 odd Nos. of 10uF 100V plastic film caps. I am very tempted to use one or two near each output device thus totaling about 80uF on each rail per PCB. However, I have not yet conclusively gathered as to why there are no caps on the amp boards.
4) Is it a good idea to have the Cap multiplier on the amp board itself and have the same heatsink for the regulator mosfets and Output stage? Maintaining short distances is one good thing; but I have also seen that in some amps, one can tune the taughtness of the bass by the twist of the power supply cables.


Thanks everyone for your inputs and support.
 
juma

juma

Member
Joined 2005
Samuel Jayaraj said:
....
1) I plan on using the caps in this configuration - CCC R CCC R CCCC - hence, 10 caps will give me 47,000uF. Should I also include an R component in the GND path? I have seen this employed a few times.
Click to expand...
Avoid resistors in rails (sound is better without them). That's one of the reasons I used a cap multiplier. You can expect some improvement if you put 10000uF after the each cap multiplier.

Samuel Jayaraj said:
....
2) I have 8 each of the Toshiba devices. Is there any benefit in using 4 per bank or adhere to Juma's scheme of 3 per bank? Conflicting thoughts here as I would like to bias as high as possible (1.6 Amps per channel, 4 devices per bank) but am tempted to go like Juma with 1.2 Amps, save on the 2 pairs of Toshiba Mosfets and try out Juma's 25watt ClassA with Hitachi/Renesas laterals.
Click to expand...
It's OK to use 4 pairs per channel, especially if you go for 1.6A bias. 3 pairs are OK if your speakers don't go much lower than 4 ohms.

Samuel Jayaraj said:
....
3) I also managed to get 70 odd Nos. of 10uF 100V plastic film caps. I am very tempted to use one or two near each output device thus totaling about 80uF on each rail per PCB. However, I have not yet conclusively gathered as to why there are no caps on the amp boards.
Click to expand...
I tried some Philips 1uF MKC without audiable effect. Original F5 doesn't use bypass caps too. It's easy enough to try and see for yourself.

Samuel Jayaraj said:
....
4) Is it a good idea to have the Cap multiplier on the amp board itself and have the same heatsink for the regulator mosfets and Output stage? Maintaining short distances is one good thing; but I have also seen that in some amps, one can tune the taughtness of the bass by the twist of the power supply cables.
Click to expand...
MOSFETs in the cap multipliers dissipate less power than output MOSFETs and I mounted them on the aluminum bottom of the amp to keep them cooler and to avoid adding dissipation to the main heatsink. If you have that big heatsinks you can put them all together.
 
S

Samuel Jayaraj

Member
Joined 2001
juma said:
Avoid resistors in rails (sound is better without them). That's one of the reasons I used a cap multiplier. You can expect some improvement if you put 10000uF after the each cap multiplier.


I tried some Philips 1uF MKC without audiable effect. Original F5 doesn't use bypass caps too. It's easy enough to try and see for yourself.
Click to expand...

Thanks for the quick reply.

I am surprised that a CRC filter does not improve the sonics of the already excellent F5 with Toshiba Mosfets. Elektor had reported atleast twice that 0.1E-10,000uF-0.1E(10 Watt)-10,000uF(RCRC) type of filter cleans up the signal below 500Hz and helps a little in preventing current inrush. I routinely employ this in all my amps now and have not checked a supply without Rs to verify its sonic impact. Thanks for the reply. But do you have any idea why it was the case in your test?

What about Resistors breaking GND into 2 or 3 stages? IIRC Dean Hughes had shown something like that and it is supposed to result in a quiter supply. I always do this isolation when the Front End of an amp has a separate/higher voltage compared to the output stage. But when high currents are involved, is this a good idea?

As for trying out plastic film caps on the amp board, I was only worried whether this would lead to oscillation. Years ago, I read that this used to happen on Leach amp builds.
 
Zen Mod

Zen Mod

Official Court Jester
Joined 2003
zenmod.in.rs/
Paid Member
Elektor reported that for which amp ?

in any case - it's hard to beat ( without resorting to some active solution , of which cap multi is almost best possible ) brute force approach - hefty C bank , without any resistive or inductive series element , along with rectifiers and xformer(s) capable to feed that C bank ....... and last , but not least - hefty bypassing that C bank with solid caps ; motor run ones , in range of 50-70uF , are my fave .

strictly speaking of A class amp as load/governor of TES delivery to your precious speaker :clown:
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom