I would put it another way.
ClassA & ClassAB of the push-pull topology clip on voltage overload and only current clip if current limiting has been implemented and triggered. Poorly designed current limiting may trigger far too early and result in valid audio transient suffering limiting when the signal should be passing through.
ClassA single ended can both current clip and voltage clip. Both will sound horrible.
ClassA & ClassAB of the push-pull topology clip on voltage overload and only current clip if current limiting has been implemented and triggered. Poorly designed current limiting may trigger far too early and result in valid audio transient suffering limiting when the signal should be passing through.
ClassA single ended can both current clip and voltage clip. Both will sound horrible.
Tinitus,
i think you are confused.
A bridged/ balanced amp doesn't see half of the load it has.
It is composed of two amps seeing eachone half of that load.
So, if each amp is designed for minimum 4 ohms, each amp will see that load when the bridged amp is loaded by 8 ohms.
If you bridge two amps accepting 4 ohms min. that bridged amp will not accept less than 8 ohms.
i think you are confused.
A bridged/ balanced amp doesn't see half of the load it has.
It is composed of two amps seeing eachone half of that load.
So, if each amp is designed for minimum 4 ohms, each amp will see that load when the bridged amp is loaded by 8 ohms.
If you bridge two amps accepting 4 ohms min. that bridged amp will not accept less than 8 ohms.
Last edited:
confused ? a bit, yes
but nice explaing, its more clear now
problem may be that very few speaker out there are 8ohm
other than maybe fullrange drivers
and those are exactly what we are not aiming at, and the main reason for more more
8ohm nominal multiway speakers are 6ohm at the best, but often more like 4-5ohm
ok, each half of the balanced amp are optimised for a 4ohm load, making the 8ohm speaker ideal
but as explained, there are no 8ohm speaker, or only very few
so, in reality, we may have to optimise each half of the balanced amp for 2ohm, to optimally drive the "common" 4ohm speaker
well, I admit I left the idea of calculating for a 2ohm load with F5
seriously, if I really needed my amp to drive a 2ohm speaker I would definately not choose classA
No, I was not talking about balanced amps.
I was talking about biasing of push-pull (eg F5) vs single ended (eg F3).
It looks like you understand the difference between Balanced and Regular amp designs now. So no need for me to explain that one.
Last edited:
ahh, ok
still looks good for the F5 then well, single ended is something with tubes, eh
Don't panic yet.
Just because you might need 50W or 100W that does not mean it has to be all in class A (assuming you are using a push-pull design). You could quite happily design for 10W to 20W class A biasing and the rest in class B
A lot of guys who biamp put a 10W to 20W class A amp on the tweeter and then a 100W class B amp on the woofer, knowing that the tweeter uses very little energy and the woofers tend to suck more energy.
The same idea can be applied to a single amp (without biamping) just design a 100W amp but bias it so you get 20W class A and the rest in class B.
Obviously you can't use this approach with a single ended design.
However you could pinch Neslon's idea of putting a current source in parrallel with the p-channel mosfet biased up for a few watts. I think this is probably a better idea than the Aleph current source for most reactive speakers around 83dB to 87dB.
Last edited:
I think this is what you want:
http://www.passlabs.com/pdfs/articles/leaving_class_a.pdf
In general, a Class A (8 ohm) push-pull amp will do better
into into 2 ohms than a Class AB because it has a higher bias.
http://www.passlabs.com/pdfs/articles/leaving_class_a.pdf
In general, a Class A (8 ohm) push-pull amp will do better
into into 2 ohms than a Class AB because it has a higher bias.
Ideal setting for two different F5s?
If one has nominal 8 ohm speakers (OB ala Martin King - Alpha 15 and FF85K), and one was choosing a rail voltage for a Casoded F5 and a Balanced F5, where would you recommend the rail voltage for each amp?
In the Cascoded case, let's assume two outputs per leg of the PP circuit (eight Mosfets total). For the Balanced, let's assuming one output for each leg of the PP (again eight Mosfets). Assume one is seeking the greatest Class A performance with the Mosfets above, at what bias and rail voltage would you recommend?
For the Cascoded F5, I'm not sure what the optimal outcome is.
Patrick has chosen to model his Balanced F5 around a 6 ohm load. He's biasing to 4A with 15V rails. I don't have 6 ohm speakers, so perhaps another voltage is in order? I intend to etch a Balanced channel tonight and put together the Balanced Amp tomorrow to go head to head with the Cascoded.
If one has nominal 8 ohm speakers (OB ala Martin King - Alpha 15 and FF85K), and one was choosing a rail voltage for a Casoded F5 and a Balanced F5, where would you recommend the rail voltage for each amp?
In the Cascoded case, let's assume two outputs per leg of the PP circuit (eight Mosfets total). For the Balanced, let's assuming one output for each leg of the PP (again eight Mosfets). Assume one is seeking the greatest Class A performance with the Mosfets above, at what bias and rail voltage would you recommend?
For the Cascoded F5, I'm not sure what the optimal outcome is.
Patrick has chosen to model his Balanced F5 around a 6 ohm load. He's biasing to 4A with 15V rails. I don't have 6 ohm speakers, so perhaps another voltage is in order? I intend to etch a Balanced channel tonight and put together the Balanced Amp tomorrow to go head to head with the Cascoded.
If you know the load and the desired wattage, then it's
pretty easy.
Take the case of 25 watts into 8 ohms:
The peak wattage will be 50 watts, Wpk = 2 * Wrms
Then we calculate the peak current:
I^2 * R = 50, I^2 = 6.25, Ipk = 2.5
Ibias = Ipk / 2, so the bias is 1.25A
To calculate supply, take the peak wattage
V^2 /8 = 50, V^2 = 400, Vpk = 20
Now you add a couple volts for OS loss and you
get rails of 20+2 = 22V
For the balanced case, the currents are the same,
but the voltage swing is 1/2 the unbalanced case
so it is 10 volts. You still add a couple volts for
OS losses for 12V rails
I think that's right....
This means a bias of 3.125A.
pretty easy.
Take the case of 25 watts into 8 ohms:
The peak wattage will be 50 watts, Wpk = 2 * Wrms
Then we calculate the peak current:
I^2 * R = 50, I^2 = 6.25, Ipk = 2.5
Ibias = Ipk / 2, so the bias is 1.25A
To calculate supply, take the peak wattage
V^2 /8 = 50, V^2 = 400, Vpk = 20
Now you add a couple volts for OS loss and you
get rails of 20+2 = 22V
For the balanced case, the currents are the same,
but the voltage swing is 1/2 the unbalanced case
so it is 10 volts. You still add a couple volts for
OS losses for 12V rails
I think that's right....
This means a bias of 3.125A.
Another way of looking at the problem is, lets assume you don't want to dissipate more than 130W per channel using your heatsinks.
So your options are (for F5 cascode):
40V rails with 1.625A
30V rails with 2.17A
20V rails with 3.25A
or anywhere in between.
If you are interested I could possibly do the simulations of 4V pk into 8Ohms and 4Ohms for each scenario.
aha
read everything about X and XA amps
I would like to but didn't find precise infos. I always thought it was a secret.
Nelson, as always, this is very helpful, thanks.
Now a question a bit more complex.
For the Balanced:
I'd been led to believe the IRFP devices were not linear enough in the 10-20V range to make a suitable amp. The Toshiba's are more linear in this region. Is linearity of the outputs a major issue at 10-20V in terms of outcome? Would you favor Toshiba over IRFPs in this range?
Cascoded:
On another note, is cascoding useful at 20-22V?
Now a question a bit more complex.
For the Balanced:
I'd been led to believe the IRFP devices were not linear enough in the 10-20V range to make a suitable amp. The Toshiba's are more linear in this region. Is linearity of the outputs a major issue at 10-20V in terms of outcome? Would you favor Toshiba over IRFPs in this range?
Cascoded:
On another note, is cascoding useful at 20-22V?
If you know the load and the desired wattage, then it's
pretty easy.
Take the case of 25 watts into 8 ohms:
The peak wattage will be 50 watts, Wpk = 2 * Wrms
Then we calculate the peak current:
I^2 * R = 50, I^2 = 6.25, Ipk = 2.5
Ibias = Ipk / 2, so the bias is 1.25A
To calculate supply, take the peak wattage
V^2 /8 = 50, V^2 = 400, Vpk = 20
Now you add a couple volts for OS loss and you
get rails of 20+2 = 22V
For the balanced case, the currents are the same,
but the voltage swing is 1/2 the unbalanced case
so it is 10 volts. You still add a couple volts for
OS losses for 12V rails
I think that's right....
This means a bias of 3.125A.
I'm quite sure Nelson was not endorsing 10V Balanced F5 and 20V Cacode F5 as the optimum voltages.
It was merely to show you how to do the calculations once you determine what power you want to run at.
It was merely to show you how to do the calculations once you determine what power you want to run at.
Last edited:
I prefer higher voltages due to the increased capacitance
of the devices at low voltages. Any circuit output
approaching the rails will experience this in any case.
The Toshibas are very nice parts. Since they have been
discontinued, I only have enough for some prototype
work out of interest.
Cascoding the inputs is always useful. Cascoding the
outputs chews up voltage, so you will always need higher
rails to do that.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.