Blitz said:
You can set the input impedance of an F4 pretty arbitrarily, so it
makes it easy to construct a single-pole filter, or if you want 2 poles
(the 12 db/oct kind) you can use the high input impedance coupled
with the low output impedance to make a Sallen-Key type filter with
fairly high Q. This is the kind of thing you can pick up in a book on
filters, like Lancaster's "Active Filter Cookbook" where the active
element is a buffer without voltage gain.
I will eventually be covering some of this in the Buffer Extravaganza
Pt 2 article whenever it gets out.
so, how much will each device dissipate given the bias listed in the service manual? At some point I calculated 14.47W. I'm estimating the die as 16mm x 20mm, does that seem reasonable? Also, I noticed that Rjc is different for the p-channel and n-channel devices. I'm doing a (elementary) study on which board orientation will be best. I'll post results in a day or two for critique.
papa please forgive me for i have sinned--i found the source of my hummmm in my F4!!!!!! i kept wondering why the dc offset acted so wierd and different between channels---- then i had an oh sh@# moment"do you think i really did that!" yes i did-when i laid out my parts for the boards i put both p channel jfets on the left channel and both n channel jfets on the right. i clipped the top of the wrong fet off of each channel and tacked on the correct part and viola!!! quiet as the proverbial church mouse and no trouble dialing in the bias and offset.and i fixed it myself!!!---bet it even sounds better!!thanks for your help-bubba
ps-ordered the plitron trannie and parts for the f5 this week
ps-ordered the plitron trannie and parts for the f5 this week
Variac said:Don't worry bubba, that's the secret break-in method. They'll sing like angels now.....
My secret break-in method is to reverse connect the input and ground OR output and ground connections...then hook up to an analyzer and watch the circuit scream. It would be better if I put BNC panel jacks on everything.
jacco vermeulen said:
So, here I am again; because of this problem (I can set a maximum bias of about .4A, with the amp still getting a bit too hot for my hands), I decided to black-anodize the whole thing. I've got friends who do that for free to me.
Hopefully, with the thermal efficiency increased by the black finish, I'll be able to reach the correct bias point and live happy.
The question is: being the anodized finish self-insulating, do you still recommend me to put anything between the FETs and the sinks?
Thank you,
Giacomo
Zen Mod said:
Uhm, I still have the teflon tape pieces which I used until now... Are they bad?
(I mean these: http://img254.imageshack.us/img254/8369/img1959rkm6.th.jpg )
Giacomo
giacomo_pagani said:
Giacomo,
excuse me, i forgot your Q.
Yeah, it does make a lot of difference.
The relationship between heatsink height and C/W value is not a linear one, and beyond a heatsink specific height value there's little to be gained.
Your heatsinks look like a Fischer SK56 model, similar to the KL146 by Seifert.
Their efficiency is pretty good up to the 125/150mm mark, beyond that height the spacing of the fins is too narrow to get enough airflow and the heatsink chokes.
Plus, the C/W value is measured at a higher heatsink temperature than you use it at, means a 0.25C/W heatsink is more like 0.30C/W at 50C.
Check page 70 of the Seifert catalog for the temperature graph of the KL146 : here
Nelson Pass said:
Here is a 3rd order Butterworth LP filter (458Hz) with a Sallen-key topology I'm working on (any comments are welcome...). I used a JFET type buffer (thanks Juma/Nelson) instead of opamps.
So Nelson, you're saying the last stage could be the F4? In my example, the last stage is the one pole, but it could be I guess the 2 poles one. Is there any advantage for that?
Thanks,
An externally hosted image should be here but it was not working when we last tested it.
