Is a mini-Aleph using BF862 possible?

[njepitt]

Originally Posted by Nelson Pass

Last call is at the end of the year, so they can still be ordered,
although some distributors stop dealing with parts the
moment the notice is issued. We already had a bunch of
them on order and had to insist on the contracted delivery.

I think maybe I'd better pay up and order some from abroad. (As much as I hate paying 70% import taxes... and that's payable on the postage & packing too!!! ) I'd still like to know what other options are out there, though... Nelson, somewhere you remarked on a substitute by Harris, but without specifying a part number; are there other manufacturers out there with equivalent parts? ST or ON, maybe?

Cheers

Nigel

[juma]

Originally Posted by njepitt

I've searched here, and can't find any specific reference to "Mini Aleph-J"

You probably won't find it as such There is a known project called Mini Aleph, and there is another project named Aleph-J. Since they share the same basic concept it's easy to cross them and appropriately name the mixture

Originally Posted by njepitt

I have a nice heatsink which is 30cm x 20 cm with 38mm fins; ....

That should be enough for both channels - if it gets too hot (more than 55C on heatsink) you should lower the bias by using the 0R66 resistor instead of aforementioned 0R5. That will lower the amp's power, but it will still be enough for your needs

Originally Posted by njepitt

I think I read somewhere here that all the IR P-channel mosfets have the same problem (???). Does anyone have a diferent suggestion?

Yes, the problem is known and Mr. Pass wrote about it several times. It can be measured, but it's not for certain if it can be heard - Stein (member on this forum) used P-channel IRFP MOSFET's in his F5 amps with success and reported no problems (and he has some very respectable speakers to test those amps)

Originally Posted by njepitt

... circuitry around Q1 and Q4 (which is just a voltage regulator circuit, right?). ....

Not really, MJE243 (or BD139 - it's not crucial) is a CCS (16-18mA) for input LTP and Q4 delivers constant current (5-6mA) for LED1 which serves as constant voltage source for Q1 CCS. This is enough data to calculate dissipation for every device in input circuit

Have fun !

[njepitt]

Originally Posted by juma

You probably won't find it as such There is a known project called Mini Aleph, and there is another project named Aleph-J. Since they share the same basic concept it's easy to cross them and appropriately name the mixture

Ah! This means you did the work here, easy or not, so many thanks!

Not really, MJE243 (or BD139 - it's not crucial) is a CCS (16-18mA) for input LTP and Q4 delivers constant current (5-6mA) for LED1 which serves as constant voltage source for Q1 CCS. This is enough data to calculate dissipation for every device in input circuit

If the current is that low then I can almost certainly put a SMD bjt in place of the MJE243 and put all of that part on the smaller PCB too. 18mA through (at most) 56R means 0.0031V across the resistor and about 0.00005W to dissipate... No worries at all with the little 1/8W SMD resistors....

Any practical thoughts about doing the first stage SMD? For instance, in the Aleph threadsthere is a lot of discussion about thermally coupling the jfets in the LTP. In this case the BF862s certainly can't be held together in the way described for 2SJ74s, but how about gluing a little copper plate on top ofthe two of them or something... How have other people done this?

Thanks again for all the help

Cheers

Nigel

[juma]

Originally Posted by njepitt

If the current is that low then I can almost certainly put a SMD bjt in place of the MJE243 and put all of that part on the smaller PCB too. 18mA through (at most) 56R means 0.0031V across the resistor and about 0.00005W to dissipate... No worries at all with the little 1/8W SMD resistors....

Sorry, but no again
V=R*I => 56R * 0.018A=1V
P=V*I => 1V * 0.018A=18mW
Still nothing to worry about, but please be gentle with Ohm's law
MJE243 will dissipate about 17V * 18mA=306mW which might be a bit too much even for TO-92 package, let alone SMD (the rule of the thumb is that you shouldn't use the device on more than one third of declared Pd)

Originally Posted by njepitt

... how about gluing a little copper plate on top of the two of them...

It will work great, just make that copper plate small and thin (to avoid thermal inertia). I don't couple them thermally since they are easy to match - 90% of 500 pieces batch I bought are with Idss of 13 to 13.5 mA

[njepitt]

Originally Posted by juma

Sorry, but no again
V=R*I => 56R * 0.018A=1V
P=V*I => 1V * 0.018A=18mW
Still nothing to worry about, but please be gentle with Ohm's law

D**N!! That'll teach me to post before drinking coffee in the morning...

MJE243 will dissipate about 17V * 18mA=306mW which might be a bit too much even for TO-92 package, let alone SMD

Well, maybe stick with the MJE243 or BD139 (which is easy to get here).

It will work great, just make that copper plate small and thin (to avoid thermal inertia). I don't couple them thermally since they are easy to match - 90% of 500 pieces batch I bought are with Idss of 13 to 13.5 mA

I only have 20 of them, but let's see how matching goes - I presume here that I only need two matched pairs for the LTPs, and not a matched four? That is, channel-to-channel matching is not necessary, right? (And Q4 doesn't need matching at all... Does it even need to be the same model jfet? I guess not...)

Many thanks for your patience with beginner questions.

Cheers

Nigel

[juma]

Originally Posted by njepitt

... I presume here that I only need two matched pairs for the LTPs

Exactly!

Originally Posted by njepitt

....channel-to-channel matching is not necessary, right?

Right!

Originally Posted by njepitt

And Q4 doesn't need matching at all...
Does it even need to be the same model jfet? I guess not...

Both right!
Just take care that Q4 has Idss of 5-6mA or more and adjust R13 accordingly

[njepitt]

Hi again. I've just spent a very frustrating couple of hours trying to buy the p-channel mosfets online. This can sometimes be surprisingly difficult if you live in Brazil; here's a little rant:

Many companies won't ship to Brazil at all, either because of bad experiences with the import barriers, or because they have "partners" in Sao Paulo or somewhere. So, for example, Farnell's UJK and "export" websites both list the Fairchild part, but won't ship to Brazil, presumably because Farnell-Newark in Sao Paulo has "first rights" on my business, but *their* site shows the part as unavailable, even internationally... So, even if I am willing to pay the shockingly high import taxes, I just can't get the part at all.... The other company linked to above has a $100 minimum order for international purchases, so that's out too, I'm afraid, unless I wait to collect together a big enough list.

OK, that's the rant over, at least until next time, but with that in mind can I repeat a question I asked above? I know where I can get IRFP9140, and I have two IRFP9240 in my parts box. Can I make these work? If so, is it better to use IRFP9240 for Q6 or Q3? Alternatively, is there another IR part which would be better than the 9140?

[juma]

Originally Posted by njepitt

..... I know where I can get IRFP9140, and I have two IRFP9240 in my parts box. Can I make these work? If so, is it better to use IRFP9240 for Q6 or Q3? ...

Option a):
You have a GB flag in your profile which suggests that you have some connections there (friends, relatives). Can someone buy Fairchild parts in GB and send them to you?

Option b):
Be brave and use IRFP9240 (better choice than IRFP9140). Others did and were very satisfied. All the fuss about IR MOSFETs is actually about small frequency-dependent non-linearity in mids region which is hardly audiable at all. Far to insignificant to justify "Can I make this work?" kind of question. Off course it will work

[njepitt]

Originally Posted by juma

Option a):
You have a GB flag in your profile which suggests that you have some connections there (friends, relatives). Can someone buy Fairchild parts in GB and send them to you?

Have relatives, yes. I've also sent an e-mail to Farnell complaining, so maybe they will come up with a suggestion. (After all, it was quite possible to buy things through mail-order even before the internet, so it must be possible to get them *somehow*...) I hope one of these works out, although will probably be a bit slower.

Option b):
Be brave and use IRFP9240 (better choice than IRFP9140). Others did and were very satisfied. All the fuss about IR MOSFETs is actually about small frequency-dependent non-linearity in mids region which is hardly audiable at all. Far to insignificant to justify "Can I make this work?" kind of question. Off course it will work

I'm quite willing to be brave, but I only have *two* IRFP9240, and the shop has no more. They do have IRFP9140, and plenty of other mosfets, but mostly lower-power stuff, in TO-220 cases. Even assuming that option (a) works out, I'd like to understand option (b). In fact, I may do option (b) while I wait for the Fairchild parts - can always switch them in later. But what specifically makes the IRFP9240 a better choice than IRFP9140? Is there no other substitute?

Cheers

Nigel

[Zen Mod]

just use 9140 ;

final differences between mentioned mosfets are lesser than all other factors involved .... quality of your source , speakers , room

finally - that's puzzle ; besides fact that is fun learning all little blocks , and ways to sum them , you can always change any part later for better , or "better" one .

for Mini - even IRF9540 class is more than enough