F5 power amplifier

[npapp]

3A I'll need more aluminum

 

[Melon Head]

Even with more aluminium I am not sure I would push these mosfets that hard.
I think 2A is fine and then if you still want more I would start paralleling ouptut devices.

 

[bobodioulasso]

I think it sounds better with higher bias

Is that super duper class A?
What are tree huggers?

 

[bobodioulasso]

What about the currents going in and out of the capacitor ?

I meant: can you see, looking at the waveform, crossover distortion on a scope with bias upper to 100 mAmps? I do not think so.

 

[AndrewT]

100W of output + 60W of dissipation = 160W of input.

And compared to a class ab amp?

this is ClassAB.

The ClassA limit is ~2.6A (= 2 * 1.3A)

 

[tinitus]

Are you saying that it leaves classA when loaded into low impedance

F5 manual dont mention this, so to me not any issue
Its rated as 100% 1-2ohm stable
No mention of not being classA anymore
Why do you say such

Anyway, not all amps doubles output whenever load impedance is half

Its known that some of the bigger PASS designs doesnt do that
(sure hope Im right on that point, but thats what I have been told)

Are you certain that F5 doubles output in half impedance load?

Anyway, if it does, why suggest that the classA bias will be inadequate

 

[AndrewT]

My speakers are 4 ohms. My understanding is with a 4 Ohm load I am only in Class A for 12.5W. I am hoping to extend the Class A rating by increasing the bias for the max that my heatsinks can dissipate. If I move to 1.5A bias doesn't this give me up 3A output in Class A increasing the rating a few watts before crossover to AB?

If I did this right 1.5A bias would give me 18W class A into 4 Ohms & 1.7A bias yields 23W. Will the increase bias change the character of the amp?

your logic is right.

But, you normally set power and sensitivity and normal listening volume so that the peak SPL at the listener's location to about 20dB greater than the average music level.

Based on that 20dB headroom for transients, your maximum average power input to your speakers will be ~ 0.25W.
This is well within ClassA into 4ohm speakers.

If the output goes to 2.5W into your 4ohm speakers (average +10dB) then the peak voltage is still only 4.5Vpk. The maximum current to meet this transient peak is likely to be <=3.2Apk for severe reactance speakers. This is just outside the ClassA limit for a standard F5.
If you increase the bias to 1.65A or 1.7A, then that +10dB transient will be within the ClassA limit.
Only transients that are above the +10dB level will drive the speakers harder. and only if that transient is either very fast starting or very fast stopping will it demand the highest of current to be sent to your reactive speakers.
There's a good chance that you will only go out of ClassA once or twice a recording (if it's very dynamic) when your average listening level is <=0.25W

 

[Melon Head]

Hi tinitus
This is a push pull design not a single ended design.
Therefore it stays in class A up to twice the bias current then goes into class B.

If you had a look at my FFT results, you would see that the distortion doesn't look to pretty once you start approaching Class B.

 

[Melon Head]

Hi Andrew
I didn't see what the efficiency of his speakers were.
Are you making an assumption here?
If so what assumption are you making?

 

[pro]

Is it a good tweek, to put a cap beetwen the mosfet gates? Did anybody try it?
Another question.
I have 6 couple of IRFP140/IRFP9140N. Are they good for this amplifier?
Thankyou

 

[AndrewT]

you normally set power and sensitivity and normal listening volume so that the peak SPL at the listener's location to about 20dB greater than the average music level.

I didn't see what the efficiency of his speakers were.
Are you making an assumption here?
If so what assumption are you making?

do want yet more numbers?

I have already posted them.
Gedlee has also posted numbers.
Many others have posted numbers.

 

[tinitus]

Hi tinitus
This is a push pull design not a single ended design.
Therefore it stays in class A up to twice the bias current then goes into class B.

Obviously I dont understand as I expected bias current to be the same for all loads

So its 25watt classA in all loads, is that what you are saying

Anyway, I dont think anything else have ever been promissed

Maybe thats why 300VA trafos are still ok for both channels

25watt classA in 16ohm, 25watt classA in 8ohm, 25wattt classA in 4ohm, 25watt classA in 2ohm, 25watt classA in 1ohm

And the rest is just rubbish
Makes sense

 

[jackinnj]

Is it a good tweek, to put a cap beetwen the mosfet gates? Did anybody try it?

No need -- there is no instability in this amplifier until you run it up into the Megahertz.

The F5 has that wonderful combination of speed, stability and very low distortion.

If you have to tweak, some electrolytics like the Elna Silk 220uF on the rails almost immediately next to the MOSFETs will help. I tweaked up mine to adjust the gate input resistors to the JFETs which has no effect on sound, but makes the slew perfectly symettrical.

 

[pro]

It's not for stability, but it, from simulation, looks to lower alot the even harmonics.
Regards

 

[tinitus]

If you have to tweak, some electrolytics like the Elna Silk 220uF on the rails almost immediately next to the MOSFETs will help.

Or even better, place your last main supply cap close to amp board

I usually prefer a single pair of caps
But this time I may have multiple caps placed along the entire supply line from bridge to amp board conection🙂

Anyway, I have never understood the use of those small signal supply caps in poweramps
They are not fit for high ripple currents
How that works is beyond me

 

[Melon Head]

Sorry I am a bit too lazy to give a long explanation.
But it is like this if the output is 25W Class A into 8 Ohms, then it will be 12.5W Class A into 4Ohms, and 6W Class A into 2 Ohms.

 

[Melon Head]

Sorry Andrew
I was asking a genuine question.
If the speakers he has are 100db/w/m then your calculations would be a lot different for someone elses speakers which might only be 85db/w/m.
Wouldn't they?
Please explain?

 

[jackinnj]

It's not for stability, but it, from simulation, looks to lower alot the even harmonics.
Regards

That's because you bandwidth limit the amplifier when you add more capacitance to the output device -- usually you do something like this to cure some instability arising from poor lead dressing etc. When you bw limit the amplifier you slow the slew rate as well.

 

[Melon Head]

Hi Pro
Just to be sure I hope you are not talking about my suggested tweak.
I never suggested putting caps at the gate.

 

[R-K Rønningstad]

your logic is right.

But, you normally set power and sensitivity and normal listening volume so that the peak SPL at the listener's location to about 20dB greater than the average music level.

Based on that 20dB headroom for transients, your maximum average power input to your speakers will be ~ 0.25W.
This is well within ClassA into 4ohm speakers.

If the output goes to 2.5W into your 4ohm speakers (average +10dB) then the peak voltage is still only 4.5Vpk. The maximum current to meet this transient peak is likely to be <=3.2Apk for severe reactance speakers. This is just outside the ClassA limit for a standard F5.
If you increase the bias to 1.65A or 1.7A, then that +10dB transient will be within the ClassA limit.
Only transients that are above the +10dB level will drive the speakers harder. and only if that transient is either very fast starting or very fast stopping will it demand the highest of current to be sent to your reactive speakers.
There's a good chance that you will only go out of ClassA once or twice a recording (if it's very dynamic) when your average listening level is <=0.25W

Seems the First Watt or so are Class A ..... 😎