I have an F5, single pair of IRF mosfets, 24V rails biased at 1.8A. 50C heatsinks. 5U case. Sounds great. Like everyone I want more power...which means more mosfets (bias) or more rail voltage (wattage) or both.
What is the advantage to going to multiple pairs with the same overall bias, same rail voltage as a single pair? I realize I could probably up the bias a little more due to the heat being spread out better but is there another advantage? I realize the people are paralleling mosfets to get more bias but what If I just crank the bias on a single pair?
Does a single pair of outputs sound different than multiple pair of outputs when biased the same?
Also, I have done some dissipation calculations playing w/rail voltage differences. This math may be completely wrong but bear w/me:
24V rails/1.3A bias = 31.2W needing to dissipate. This is stock.
24V rails/1.8A bias = 43.2W dissipation. This is what I am running now. I am comfortable with this heat level (50C on the sinks) even during the 90 degree F summers we have been getting around here.
Let's bump up the rail voltage and keep the dissipation the same:
32V rails/1.5A bias = 48W dissipation. So, to take the rail voltage of an F5 up to 32V you lose some bias assuming same dissipation. How much? In my case you lose .3-.4A of bias.
What sounds better:
A single pair F5, 32V rails at 1.5A-ish overall bias
or
Multiple pair F5, 24V rails, 2A-ish overall bias
What is the advantage to going to multiple pairs with the same overall bias, same rail voltage as a single pair? I realize I could probably up the bias a little more due to the heat being spread out better but is there another advantage? I realize the people are paralleling mosfets to get more bias but what If I just crank the bias on a single pair?
Does a single pair of outputs sound different than multiple pair of outputs when biased the same?
Also, I have done some dissipation calculations playing w/rail voltage differences. This math may be completely wrong but bear w/me:
24V rails/1.3A bias = 31.2W needing to dissipate. This is stock.
24V rails/1.8A bias = 43.2W dissipation. This is what I am running now. I am comfortable with this heat level (50C on the sinks) even during the 90 degree F summers we have been getting around here.
Let's bump up the rail voltage and keep the dissipation the same:
32V rails/1.5A bias = 48W dissipation. So, to take the rail voltage of an F5 up to 32V you lose some bias assuming same dissipation. How much? In my case you lose .3-.4A of bias.
What sounds better:
A single pair F5, 32V rails at 1.5A-ish overall bias
or
Multiple pair F5, 24V rails, 2A-ish overall bias
KEF R700, a 3 way manufacturer rated 89db/8 ohm/3.2 ohm min. In reality probably a 87db, 6 ohm speaker. I would like to have a little more volume but it's not a must. I don't know if I need more headroom or amps...I just want "better". Not that it is bad now.
I listen at about 85db in a large room which uses about 9 volts AC at the amp.
9V into 6 ohms in 13 watts. Or 10W into 8 ohms. Let's just say I am using about 1/2 the F5's rated power, not including peaks.
I listen at about 85db in a large room which uses about 9 volts AC at the amp.
9V into 6 ohms in 13 watts. Or 10W into 8 ohms. Let's just say I am using about 1/2 the F5's rated power, not including peaks.
Attachments
Interesting. So multiple pairs have a higher damping factor vs single pairs at the same bias? I always thought the ability to drive low-z loads better with multiple pairs was due to an increase in overall bias.
I am guessing here but...since multiple pairs increase damping factor and raise distortion (slightly...acording to the F5T article) would you say a multiple pair amp would sound less lively, more controlled and perhaps duller than a single pair amp when compared at the same bias?
That's one way to do it I guess...
I guess I will try adding A set of mosfets to the F5 which won't get me more watts but I might be able to bias it higher. If I don't like it, I'll just take 'em out.
More pairs in the output stage reduce HF distortion as well, because of margain in the output stage. Also reduce output impedance. But are harder for the input stage to correctly deal with, because of the increased drive requirements of more gate capacitances.
I felt the F5 had a more precise and focused sound, whereas the F5T had more smooth and relaxed signature. I could not measure the F5 when I had it with me about four years ago, but I recently ran some RMAA tests on the Turbo alongside one of juma's BJT versions using a similar topology and the same output devices, and you could not really tell the difference by looking at the results or graphs.
Both are great amps. Both sound a bit different from each other. I live with both, and like them equally. This winter I should have my behemoth balanced F5T-based amp running, the last version of this topology I will be building. 250W bias per channel. Been in the making for five years now.
I felt the F5 had a more precise and focused sound, whereas the F5T had more smooth and relaxed signature. I could not measure the F5 when I had it with me about four years ago, but I recently ran some RMAA tests on the Turbo alongside one of juma's BJT versions using a similar topology and the same output devices, and you could not really tell the difference by looking at the results or graphs.
Both are great amps. Both sound a bit different from each other. I live with both, and like them equally. This winter I should have my behemoth balanced F5T-based amp running, the last version of this topology I will be building. 250W bias per channel. Been in the making for five years now.
Thanks Sangram. That's my fear...that a multiple pair F5 will be more relaxed. My speakers are kind of laid back...and I like a more precise sound. Imaging and focus are more important to me than bass and drive. I will either try raising the rail voltage or another pair of mosfets and see what I like...obviously raising the rails requires a little less surgery to the amp.
I've heard the q300 with both amps, and those sound super with either. I don't think you should worry, it's just a matter of degree. The extra pair won't lull you to sleep, if that's your worry. It's plenty precise.
I find that low distortion gear tends to sound smoother, in fact. I think its the absence of harmonics.
Good luck with your experiments.
I find that low distortion gear tends to sound smoother, in fact. I think its the absence of harmonics.
Good luck with your experiments.
Last edited:
Tend to agree. Unless you are running into the maximum current rating of the mosfet (eg. class B into low Z loads), there's no benefit in running 2x the number of ops at half the standing bias each.
Doubling the ops mosfets but running them each at 0.5 the total bias gives you something like 1.5 x the original transconductance (yay!) but a heckuva lot more than 1.5x the original total input capacitance (groann!) .....
Doubling the ops mosfets but running them each at 0.5 the total bias gives you something like 1.5 x the original transconductance (yay!) but a heckuva lot more than 1.5x the original total input capacitance (groann!) .....
Most of the MOSFETs get more linear as drain current increases (until it gets burning-up high anyway). There is a region where the Id vs Vgs is extremely linear, the envy of most triodes even, but sadly it's up where the dissipation of the part is too much for it to be used there. So for that, fewer mosfets running higher current each should have less distortion. Lowered capacitances helps, too.
As kasey197 said, you do get more transconductance with the multiples. And better reliability of the parts, and better distribution of heat on heatsinks (and possibility to use more heatsinks).
As kasey197 said, you do get more transconductance with the multiples. And better reliability of the parts, and better distribution of heat on heatsinks (and possibility to use more heatsinks).
As I recall, the F5 Turbo addresses some of these issues:
www.firstwatt.com/pdf/art_f5_turbo.pdf
When you parallel Mosfets with the same overall bias, the individual
transconductance declines by the square root of the lower bias current of
each device, but then you add the individual transconductances, resulting
in more. For the example of the two devices in parallel, the improvement is
by the square root of 2.
The capacitances will start to add up, but this has not been seen to be
much of a problem for just a few devices.
www.firstwatt.com/pdf/art_f5_turbo.pdf
When you parallel Mosfets with the same overall bias, the individual
transconductance declines by the square root of the lower bias current of
each device, but then you add the individual transconductances, resulting
in more. For the example of the two devices in parallel, the improvement is
by the square root of 2.
The capacitances will start to add up, but this has not been seen to be
much of a problem for just a few devices.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.