F5 power amplifier

Simplified Balanced F5

There is current through the first stage, it flows through the feedback resistors.

This does work I have created a design similar but using all Mosfets.

I really didn't have time to simulate this, so i think most of the resistors are the wrong value, but as i said "a simplified design"
 
........it seems like it was only yesterday!

Mr.Pass,

I hope you recover soon..........the guys at lab miss you. I have to congratulate you on another fine project well worthy as a replacement for the A-40.............Line stage next please or maybe a relay based attenuator to go with it.

Regards.

Jam
 
Quad feedback, quadrant balanced, hyperbalanced :)

I like these parallel feedback designs they have unique advantages.
If a normal amp is driven into current clipping where 1/2 of the totem pole is fully turned off, the single feedback path has effectively lost control. With parallel feedback, with a feedback path controlling each quadrant, there will still be a feedback path that is in control while while the other half of the amp is clipping
 
Patrick,

To answer your question, that is if you have a balanced preamp I would use two of these amps per channel ie. one per phase to obtain a bridged amplifier. I came to this conclusion while building an amp with a similar topology. On the other hand if your preamp is single ended a balanced bridged version of the amp would be better.

Just an opinion mind you.

Jam
 
Quite a bit offtopic, but...

After following this thread with great interest i'm very interested in a audioxpress subscription. I'm only not happy with the ordering page.. fill in billing details and credit card number on a (with all do respect) obscure ordering page and see what happens.. does anybody have any experience with this ordering page and willing to give me a quick word on it?

my apologies for the off topic question

cheers,

c.
 
Member
Joined 2002
Paid Member
culture said:
Quite a bit offtopic, but...

After following this thread with great interest i'm very interested in a audioxpress subscription. I'm only not happy with the ordering page.. fill in billing details and credit card number on a (with all do respect) obscure ordering page and see what happens.. does anybody have any experience with this ordering page and willing to give me a quick word on it?

my apologies for the off topic question

cheers,

c.

I wasn't too thrilled with the ordering page either.
However, the lure of the F5 was too great, so I
went ahead.

I had ordered late at night and I received a
confirmation the following day via email.

Cheers,
Dennis
 
culture said:
Quite a bit offtopic, but...

After following this thread with great interest i'm very interested in a audioxpress subscription. I'm only not happy with the ordering page.. fill in billing details and credit card number on a (with all do respect) obscure ordering page and see what happens.. does anybody have any experience with this ordering page and willing to give me a quick word on it?

my apologies for the off topic question

cheers,

c.
I had the same reservations about it not even being a secure http page. I ended up calling them myself. Another option is to send two separate emails with the numbers to your credit card spelled out phonetically, split across the two messages. Hugh Dean of AKSA uses this approach and it is a bit safer if someone is picking off random clear text messages looking to parse numeric strings that look like credit card numbers. Tom
 
The one and only
Joined 2001
Paid Member
EUVL said:
I personally do not like thermal compensation with thermistors. For one, it changes the open loop gain, and you don't really know by how much. Worst, you don't know whether the 2 halves drift and compensate symmetrically.

R11,12 are source degeneration for the MOSFETs. John Curl mentioned on the blowtorch thread that source degeneration messes up the distortion spectrum. But I guess you need something for bias stability (a sort of local current feedback). And it provides a convenient means for current sensing (for protection). I find 0R47 a bit high, and would have gone for 0R22 or even 0R1. But that also depends on a lot of things like which MOSFET and how much heat sinking.

R13, R14 are gate stoppers for the MOSFETs (to prevent oscillation). The values are quite low, but then it is supposed to be a fast amp (see the 200kHz square wave), and the drain resistors are quite high (compared to e.g. Aleph-X). Something to experiment with.

The only remaining question for me is :

If I wish to go balanced, should I just use two of these or should I use complementary diff pairs for the first stage and Xed feedback ?

Pretty much correct, except:

1) You don't have to use the thermal compensation, particularly
if you have generous heat sinking, but apart from having extra
parts, there is no downside to it. The compensation is fairly subtle
and the open loop variation is insignificant, probably less than
without compensation. Remember that you adjust the amp after
you reach operating temperature anyway, and you want it to
reside there in a stable manner.

2) You can operate with lower value Source resistance, but I am
using it to get better leveling of characteristic between P and N
channel devices to better cancel the 2nd harmonic. I have a tweak
that I'll write up about this after I make another 22 channels.

3) The optimum value of the Gate stoppers will depend greatly
on the parts. I actually managed to get one stable into a resistive
load with 0 ohms, but I wouldn't plug it into my HiFi. 47 ohms is
probably the minimum, and as in (2) I am still cataloging the
variations.

4) The question of using dual diff pairs is moot, as you will need
2 pair of JFETs in either case. The effort required to run the
feedback back to the Gates versus the Sources is not much, so
you can explore this without much effort.

And welcome back, Darth.