Thanks for putting in the numbers Andrew.
Megastat16,
Don't get confused with multiple output pairs etc. A single pair at 2 amps gives the same Class A output as a twin pair at 1 amp per pair. The combined heat is the same, and the reduction in heat output the same at at a given % reduction in current. Providing they share the current equally during the 360 degrees of output then they never drop out of Class A.
Edit... actually it doesn't matter whether they share equally or not, provided the current drawn by the load is below the output quiescent current in simple terms. All that would happen is one pair would get hotter than the other due to unequal sharing.
Megastat16,
Don't get confused with multiple output pairs etc. A single pair at 2 amps gives the same Class A output as a twin pair at 1 amp per pair. The combined heat is the same, and the reduction in heat output the same at at a given % reduction in current. Providing they share the current equally during the 360 degrees of output then they never drop out of Class A.
Edit... actually it doesn't matter whether they share equally or not, provided the current drawn by the load is below the output quiescent current in simple terms. All that would happen is one pair would get hotter than the other due to unequal sharing.
Last edited:
I really like your idea Mooly.
It is definitely worth pursuing. The only problem I see is that no one has mentioned the fact that speakers don't have flat impedance vs frequency response.
A lot of 8Ohm speakers have impedance dips of 4Ohms.
So just need to take that into account when designing this
It is definitely worth pursuing. The only problem I see is that no one has mentioned the fact that speakers don't have flat impedance vs frequency response.
A lot of 8Ohm speakers have impedance dips of 4Ohms.
So just need to take that into account when designing this
Thanks thanh1973, that's very true, and I know my own B&W's have a manufacturer quoted 3 ohm minima depite being called "8 ohms" The upside to that for a large part of the frequency range the impedance is way above 8 ohms.
It's a very valid point though and applies to any amplifier when considering what you may "hang" on the end of it.
It's a very valid point though and applies to any amplifier when considering what you may "hang" on the end of it.
go back and read post 22.
It allows for three times the nominal speaker impedance current in the lower bias settings to ensure the underbiased amplifier never leaves ClassA during the lower SPL volume settings.
I have not read the thread, sorry
But I expect the combined temperature with output pairs will be less
Im thinking at less temperature difference between device case and heatsink
Thus fast dynamic heat rise wont push heat up as much as with hotter running single devices
More effective cooling and result in less combined heat, as I see it
If it really happens I dont know
Have you read Broskie latest articles on AC amps
It is about low dissipation classA, as I understand it
Tho the details is beyond me
I have had a quick glance through the Class AC articles... if you are refering to the same parts of it as I saw, then it seemed to be the same old ideas, having "extra" output devices to take over at higher outputs... which doesn't meet the accepted definition of Class A.
It's a bit like calling the Quad 405 a class A amp because that supplies the load at low levels exclusively from a low powered Class A amp and has "extra" outputs to take over (the dumpers) at higher levels.
The idea I had at the start of this thread was that the amp genuinely meets the accepted definition.
The big big question is whether a single design of amp can retain it's sonics over a wide range of different fixed bias settings appropriate to the expected output demands.
I think I understand your point on temperatures, that it's easier to keep cool multiple pairs than a single pair.
It has...
IIUC This is sliding plateau bias, present in krells ever since...
The bias and rails are switched in response to the incoming signal, not as a result of the user doing anything.
I have not followed the thread thoroughly, please forgive me if I'm restating what has already been observed.
Stuart
IIUC This is sliding plateau bias, present in krells ever since...
The bias and rails are switched in response to the incoming signal, not as a result of the user doing anything.
I have not followed the thread thoroughly, please forgive me if I'm restating what has already been observed.
Stuart
Please Lumba, an information please, to avoid me to read all thread
Have someone posted a schematic tested about this idea?
Have someone posted a picture of this idea applied into a real amplifier?
thank you...you answer will avoid me to read all thread..seems to me philosophical discussion..but maybe someone have made something real too.
regards,
Carlos
Have someone posted a schematic tested about this idea?
Have someone posted a picture of this idea applied into a real amplifier?
thank you...you answer will avoid me to read all thread..seems to me philosophical discussion..but maybe someone have made something real too.
regards,
Carlos
Hi Andrew
I did see your post which was well thought out.
Do you think it will be easy to lower voltage while still maitaining reasonable efficiency?
The only way I can think of lowering voltage is with a regulated power supply. So if the amp is designed for 50V rails and you drop down to 20V. that is a lot of heat given off by the power supply.
How would you implement the idea?
Adjusting bias should be relatively easy I am just not sure about how to efficiently (and easily) adjust voltage.
I did see your post which was well thought out.
Do you think it will be easy to lower voltage while still maitaining reasonable efficiency?
The only way I can think of lowering voltage is with a regulated power supply. So if the amp is designed for 50V rails and you drop down to 20V. that is a lot of heat given off by the power supply.
How would you implement the idea?
Adjusting bias should be relatively easy I am just not sure about how to efficiently (and easily) adjust voltage.
we are not discussing sliding or any form of auto adjusting bias.
We are looking at fixed ClassA bias to suit the maximum possible current demand at any particular fixed volume setting.
Last edited:
So you have various bias settings for the output stage depending on the volume setting, yes i can see the logic in this with a fixed maximum RMS input voltage to the pre amp. However it amounts to the same thing as the sliding bias that Krell use if you think about it. The only difference is they have the amplifier monitoring the input waveform & a bias circuit that is faster than the output stage apparently. Whether it's done at pre amp level or by the power amp itself (imo) it amounts to the same thing. Obviously you could introduce as many stages as you liked though, but then costs might become prohibitive.
One thing though, i bet the Krell can't determine the output impedance it's driving
I should imagine they'd bias things for about 6 ohms allowing for only moderate impedance variations in the load. Connect a 2 ohm load & the amp is sure to be AB.BTW, the variation of voltage to the output stage would certainly be novel.
Last edited:
Not really sliding
IIUC The krell scheme has several fixed bias points, specifically chosen to fulfill the same criteria you mention, appropriate bias current for the voltage to fulfill proper Class A operation at any of the chosen operating points. The switching is realized with hysteresis, so once a particular plateau is chosen it is held until the signal profile would indicate it has not been used for some seconds and is therefore no longer required.
I've no clue which is better sonically, sadly a 'fixed' volume level is anything but; decent music has a wide dynamic range (+20db?) from average to peak. The automatic version has the benefit of realizing the full potential of the amp (presumably with cycles of clipping before the switch occurs), while a user "volume control" will allow considerable clipping in the presence of high enough peaks, the poor guy has to get out of his chair to adjust... Of course one could simply choose a bias and voltage level that allows +20db over average, but that will probably not save as much power as the automatic version, needing 100w bias level to fulfill a 1w average.
I think it's a great idea, but it is by definition a sliding bias scheme. The rail voltage changing at the same time is a very cool optimization of the principal. The only drawback I see is the control scheme, which only responds in the presence of long term objectionable distortion.
Stuart
IIUC The krell scheme has several fixed bias points, specifically chosen to fulfill the same criteria you mention, appropriate bias current for the voltage to fulfill proper Class A operation at any of the chosen operating points. The switching is realized with hysteresis, so once a particular plateau is chosen it is held until the signal profile would indicate it has not been used for some seconds and is therefore no longer required.
I've no clue which is better sonically, sadly a 'fixed' volume level is anything but; decent music has a wide dynamic range (+20db?) from average to peak. The automatic version has the benefit of realizing the full potential of the amp (presumably with cycles of clipping before the switch occurs), while a user "volume control" will allow considerable clipping in the presence of high enough peaks, the poor guy has to get out of his chair to adjust... Of course one could simply choose a bias and voltage level that allows +20db over average, but that will probably not save as much power as the automatic version, needing 100w bias level to fulfill a 1w average.
I think it's a great idea, but it is by definition a sliding bias scheme. The rail voltage changing at the same time is a very cool optimization of the principal. The only drawback I see is the control scheme, which only responds in the presence of long term objectionable distortion.
Stuart
Member
Joined 2009
Paid Member
This idea (volume control also controlling bias of output stage) sounds like a good idea (I think it's appeared briefly in this forum somewhere once before). It doesn't rely on fancy control circuitry and the user knows what the amp is doing at any one time.
Even simpler if you allow for a separate control, a swtich with 3 positions 'Hi' 'med' and 'Low'. At least I'm now considering such an option myself
Even simpler if you allow for a separate control, a swtich with 3 positions 'Hi' 'med' and 'Low'. At least I'm now considering such an option myself
Hi Stuart,
It's not a sliding bias scheme, the input signal and it's history has no bearing on the bias. As long as the volume control remains set at one position so to does the bias. My original idea was to have a continuously variable bias but the suggestion of having a few discrete steps is (probably) more practical.