This is a fun debate, but borders on an argument over semantics and definitions. Bob Cordell's 2nd edition of his power amp book has a nice discussion of the LT1166. He explains that both devices dynamically and smoothly vary their conduction so the product of their currents is a constant. Both Bob and Linear Tech wisely avoid any discussion of bias class. 😉
Frequently, class A is advocated as a way to avoid crossover distortion. In this thread about class A heatsink sizing, I think my intent was to suggest there are multiple ways of ensuring low crossover distortion without requiring heat.
Steve
Frequently, class A is advocated as a way to avoid crossover distortion. In this thread about class A heatsink sizing, I think my intent was to suggest there are multiple ways of ensuring low crossover distortion without requiring heat.
Steve
Last edited:
Agreed.
Maybe they came up with something clever. Either way I'm curious what it does to distortion.
And we definitely agree there. A Class B (or AB) output will avoid crossover distortion but have distortion from the gain wobble near the zero crossing. Some may call that crossover distortion. I wouldn't. But that gets into semantics and definitions too.
Tom
Agreed on all points. I'll even concede the Sumo Nine bears a strong class B behavior due to its push-pull bridge drive, even though both sides of the bridge drive have continuous class A conduction. I can believe James Bongiorno might want to hype class A characteristics in his technical pitch.
Cordell has a very brief, intriguing passage in his preamble discussion of the LT1166: "A BJT output stage controlled by a geometric mean bias spreader can theoretically have constant transconductance, independent of output current, and thus have flat wingspread and no static crossover distortion." That's the entire verbiage. No footnotes.
Bob used to post frequently, but I've not seen anything in quite awhile. Hope he's well.
Cordell has a very brief, intriguing passage in his preamble discussion of the LT1166: "A BJT output stage controlled by a geometric mean bias spreader can theoretically have constant transconductance, independent of output current, and thus have flat wingspread and no static crossover distortion." That's the entire verbiage. No footnotes.
Bob used to post frequently, but I've not seen anything in quite awhile. Hope he's well.
The summary is: any amplifier that gets over 50% efficiency is not conventional linear class A.
Non-linearity is necessary for efficiency. The differences are all in the shape of the transfer curve.
Ed
Non-linearity is necessary for efficiency. The differences are all in the shape of the transfer curve.
Ed
Has anyone here built the LT1166 amp? It looks quite novel to me. I'd love to see some measurements from it.
I'm thinking about it. I was planning to have a pluggable bias spreader (1166 vs Vbe multiplier) and also add diff amps (single-ended measurement with a scope) across the output emitter resistors to check the top/bottom current real-time. Specifically does a Vbe multiplier cause the transistor to go into cutoff, whereas the 1166 does not? I would also like to build a fixture to study gm doubling of various output stages.
You are correct--- the LT1166 prevents the transistors from ever completely entering cutoff. Not so for conventional bias spreaders.
Your 2nd paragraph shows your heatsink has a 0.3 degC/W thermal resistance and your 3rd paragraph shows the Class AB amp has about 10W of idle power (Class A). You'll have to experiment with different loads on your AB amp and how far your PSU will take it beyond 50W.
Which one is good pure quality for class a? Bjt or mosfet class A amp
And single power or duall power ?
For small pa speaker
And single power or duall power ?
For small pa speaker
If I’m not mistaken, the three power devices in lower right of post 50 need to have their emitter and collector leads swapped. Otherwise, I believe it will explode on the launch pad.
so finally as per this link
https://neurochrome.com/pages/thermal-design
if we calculate we get the following results.
In the first set we have considered the crest factor of 6db and duty cycle of 0.5 hence for a given load of 8 ohm without any reactive parameters the dissipation is relatively less.
In the second thumb rule method we are taking 0.4x of the total power rating of the channel.
Please do let me know if anything needs to be corrected in the above excel sheet in both cases which one to be better considered as correct method.
https://neurochrome.com/pages/thermal-design
if we calculate we get the following results.
In the first set we have considered the crest factor of 6db and duty cycle of 0.5 hence for a given load of 8 ohm without any reactive parameters the dissipation is relatively less.
In the second thumb rule method we are taking 0.4x of the total power rating of the channel.
Please do let me know if anything needs to be corrected in the above excel sheet in both cases which one to be better considered as correct method.