Class A sounds good but I don't like a 700W electric heater in my room, nor the electricity bill.
The solution is a quiescent current, dependant upon the output level of the amp.
I'm thinking of rectifying the amp output, low pass filtering and than clamping.
The output of the clamping circuit can drive the Vbe multiplier.
The Vbe transistor can be a photo transistor or an opto coupler.
I.e. there will be a loop that regulates the quiescent current in the output stages.
Are there any people who tried this?
The solution is a quiescent current, dependant upon the output level of the amp.
I'm thinking of rectifying the amp output, low pass filtering and than clamping.
The output of the clamping circuit can drive the Vbe multiplier.
The Vbe transistor can be a photo transistor or an opto coupler.
I.e. there will be a loop that regulates the quiescent current in the output stages.
Are there any people who tried this?
Lots of people, including Krell.
The original Aleph 1 had an automatic idling mode to reduce
dissipation when not in use (after 10 or 20 minutes) but
we immediately took it out because everybody complained.
The original Aleph 1 had an automatic idling mode to reduce
dissipation when not in use (after 10 or 20 minutes) but
we immediately took it out because everybody complained.
Nelson,
I'm just wandering if there are any more detailed circuit idea's?
I just don't want to invent the wheel again.
Making a quick drawing is never a problem. It's always the detail design that takes 80% of the time.
I cannot imagine that there are no DIY'ers who did this before.
I'm just wandering if there are any more detailed circuit idea's?
I just don't want to invent the wheel again.
Making a quick drawing is never a problem. It's always the detail design that takes 80% of the time.
I cannot imagine that there are no DIY'ers who did this before.
Another variation is a dual voltage power-suppy. The amplifier switches between supply voltages depending on signal level e.g Marantz PM-64 and similar "Quarter A" amplifiers.
another option is PWM modulating power supply. Modulation by the music.
In this case the supply voltage on the ouput stage transistors is just a little higher than the signal level.
Very efficient but complicated.
In this case the supply voltage on the ouput stage transistors is just a little higher than the signal level.
Very efficient but complicated.
There appear to be two fundamental approaches; A system that tracks the signal level and makes a continuous adjustment or one that switches between two states dependent on some threshold.
But which one to choose...?
😕
But which one to choose...?
😕
A PWM modulating power supply is too complicated. However it may be possible to build an amp that allmost has the efficiency of class D and the sound quality of class A.
My idea was to make a simple loop that regulates the quiescent current.
My idea was to make a simple loop that regulates the quiescent current.
sliding class A
Kees,
You may want to check out
Electronics World (formerly Wireless World) June 2001
issue where the Circuit Ideas section has a
class A circuit that is biased dynamically by the
input signal. Class A in this case means that the
amplifying transistor never switches off, but idles
at low current.
Basically, it uses a small resistor in the collector of the
output emitter follower and a feedback loop using an
opamp tracks the voltage drop across the resistor. The
emitter follower is biased by a transistor current source.
The output of the opamp then biases the current
source transistor dynamically by amplifying the differential
signal of the input to the follower and the resistor.
Unfortunately I do not have a scanner, or I could send
a scan.
Hope this helps.
Kees,
You may want to check out
Electronics World (formerly Wireless World) June 2001
issue where the Circuit Ideas section has a
class A circuit that is biased dynamically by the
input signal. Class A in this case means that the
amplifying transistor never switches off, but idles
at low current.
Basically, it uses a small resistor in the collector of the
output emitter follower and a feedback loop using an
opamp tracks the voltage drop across the resistor. The
emitter follower is biased by a transistor current source.
The output of the opamp then biases the current
source transistor dynamically by amplifying the differential
signal of the input to the follower and the resistor.
Unfortunately I do not have a scanner, or I could send
a scan.
Hope this helps.
Designing a smart class-a amplifier with all the benefits and few of the disadvantages is certainly desirable and there have been many commercial designs claiming to have achieved this, but how close do any of these designs come? Why aren't we all listening to 'new class a' type amplifiers?
I think in many cases the performance of this type of amplifier is dominated by the extra control circuitry rather than the class-a aspect, and without care you could end up with class-a/b in disguise.
The idea is certainly worth pursuing but analysing such a design could prove difficult and it is unlikely to ever out-perform a true full-time fixed bias/PSU class-a design since the extra complexity will not do anything to improve sound quality.
Having said that it is tempting to want to have a go.
😀
I think in many cases the performance of this type of amplifier is dominated by the extra control circuitry rather than the class-a aspect, and without care you could end up with class-a/b in disguise.
The idea is certainly worth pursuing but analysing such a design could prove difficult and it is unlikely to ever out-perform a true full-time fixed bias/PSU class-a design since the extra complexity will not do anything to improve sound quality.
Having said that it is tempting to want to have a go.
😀
>The original Aleph 1 had an automatic idling mode to reduce
dissipation when not in use (after 10 or 20 minutes) but
we immediately took it out because everybody complained.
How 'bout your original Threshold "swinging class A" ? Wasn't that the idea with it ? ............ mike
dissipation when not in use (after 10 or 20 minutes) but
we immediately took it out because everybody complained.
How 'bout your original Threshold "swinging class A" ? Wasn't that the idea with it ? ............ mike
I always thought that the main benefit of highly biased class A was low level resolution. In Pass amps using mosfets, the higher you bias, the better the sound quality. So, it seems like you would be shooting yourself in the foot by reducing bias for low level signals. I could see running a good amount of bias, and increasing it for more output power, but I think you would hurt the sound by reducing it too far.
No so Dutch idea
In my opinion (having succesfully built my second JLH-amp last summer) Kees' idea might be worth exploring a bit more. However my first note would be that the design should not go for maximum reduction of power consumption, but for some siginificant reduction.
Besides Bias-current reduction, Rail-voltage reduction should be considered (this is however not a reduction when using one fixed mains transformer)
One idea might be just to relate bias to the position of the volume control. (crude, but not totally insensible)
With volume between 7 o'clock and 9 bias is halved. Over 9 bias is full.
In my opinion (having succesfully built my second JLH-amp last summer) Kees' idea might be worth exploring a bit more. However my first note would be that the design should not go for maximum reduction of power consumption, but for some siginificant reduction.
Besides Bias-current reduction, Rail-voltage reduction should be considered (this is however not a reduction when using one fixed mains transformer)
One idea might be just to relate bias to the position of the volume control. (crude, but not totally insensible)
With volume between 7 o'clock and 9 bias is halved. Over 9 bias is full.
Re: No so Dutch idea
But the output level of Source equipment varies and is not standardised yet. Hence, the position of the volume control is not necessarily an indication of amplifier drive.
One idea might be just to relate bias to the position of the volume control. (crude, but not totally insensible)
With volume between 7 o'clock and 9 bias is halved. Over 9 bias is full.
But the output level of Source equipment varies and is not standardised yet. Hence, the position of the volume control is not necessarily an indication of amplifier drive.
Re: Re: No so Dutch idea
Correct Samuel,
However there are some standards (which are not adhered to very well) eg
a normal line output should have something like 0,7V @ 0dB (a CD player makes 2V@0dB).
And of course you have CD's which don't come anywhere near -12 dB and there are ones which do contain 0dB signal.
So my idea should work, has limitations, but it has some simplicity too.
Jos 🙂
Samuel Jayaraj said:
Correct Samuel,
However there are some standards (which are not adhered to very well) eg
a normal line output should have something like 0,7V @ 0dB (a CD player makes 2V@0dB).
And of course you have CD's which don't come anywhere near -12 dB and there are ones which do contain 0dB signal.
So my idea should work, has limitations, but it has some simplicity too.
Jos 🙂
nobody special said:
I agree, and this is why I prefer and use a simple mute/standby switch to limit power consumption between listening sessions. The front end and all power supplies stay "live" so warmup/break-in time is minimal. Very basic, and works like a champ.
sliding bias control chip
The Linear Technology chip LT1166 is a sliding bias control chip for mosfets, it maintains the product of the upper and lower currents as a constant. Also provides dual slope SOA feature.
The Linear Technology chip LT1166 is a sliding bias control chip for mosfets, it maintains the product of the upper and lower currents as a constant. Also provides dual slope SOA feature.
I think I can lay claim to the original patent on this subject,
and you'll notice that I tossed the idea after 2 years. If
you want to adjust the bias, I recommend a knob and a
meter on the front panel.
and you'll notice that I tossed the idea after 2 years. If
you want to adjust the bias, I recommend a knob and a
meter on the front panel.
I think the core issue of the idea of Kees (hi Kees) is that music is highly dynamic. Adjusting the bias to the average signal level means that at high level peaks it is underbiased, unless you have a very fast acting bias control. I would be very sceptical on the quality aspects of a bias control that adjusts itself along the instanteneous signal level, however.
But I have never tried it, maybe somebody finds the holy grail of bias control.
Jan Didden
But I have never tried it, maybe somebody finds the holy grail of bias control.
Jan Didden
- Status
- Not open for further replies.