[sparko311]

Ive been playing around with a class A output stage that controls some current controlled current sources (CCCS) which assist the class A portion with current drive - and keeps the class A portion in class A mode.  Its similar idea to the Quad405 "current dumping" amplifier, but with different implementation.  A schematic of the output stage is attached.  The diff amp is for bias control, its not an input stage.  This schematic is the output stage only.  The open collector NPN is temp sense / tracking.  The class B bias in the CCCS is controlled by servoing the class A bias in the class A portion.  This thing can swing 200W RMS into 4 ohms and never leave class A mode.  Idle dissipation is ~ 65W.  Nearly 1/7th less than it would be to swing that power in "traditional" class A mode.  Im trying to decide if this is something novel.  Thoughts?  Here is a web link that has photos of the finished amplifier and some more info. 

http://sparkoslabs.com/projects/class-audio-amplifier/

[G Georgopoulos]

Ive been playing around with a class A output stage that controls some current controlled current sources (CCCS) which assist the class A portion with current drive - and keeps the class A portion in class A mode.  Its similar idea to the Quad405 "current dumping" amplifier, but with different implementation.  A schematic of the output stage is attached.  The diff amp is for bias control, its not an input stage.  This schematic is the output stage only.  The open collector NPN is temp sense / tracking.  The class B bias in the CCCS is controlled by servoing the class A bias in the class A portion.  This thing can swing 200W RMS into 4 ohms and never leave class A mode.  Idle dissipation is ~ 65W.  Nearly 1/7th less than it would be to swing that power in "traditional" class A mode.  Im trying to decide if this is something novel.  Thoughts?  Here is a web link that has photos of the finished amplifier and some more info. 

I think your amplifier is class ab,pure class a amplifiers output is about 25% of idle power
,your amplifier looks familiar somehow but it uses new topologies i havent seen
before,have you measured distortion?,nelson pass makes amps class a but some of
his amps are actually class ab,idle dissipation of 65w is still alot compared to class b
,class a amplifiers may have low distortion but they have other problems...

[sparko311]

I think you are talking about single ended class A when you say they can only put out 25% power (or rather, are 25% efficient) Is that correct ?   

Push pull class A  is ~ 50% efficient. (Pass tends to use single ended class A, I think)   However this design, while push pull, is more efficient still.   I dont consider it to be an AB design, as the class A portion of the output stage (the transistors on nets 46, 12, ) never leave class A mode, and they direct drive the output.  The "dumpers" or CCCS drive from their collectors, and operate class B (~50mA per device)

Its kind of like a class A stage in parallel with a class B stage.  Only the B stage operates as a current controlled current source to assist the A stage in driving the load.  To call this an AB stage, to me, is sort of a misnomer as class AB stages are simply over biased class B, (or under biased class A) 

I called this thing an "Extended Class A" - as the dumping transistors "assist" the class A stage with current drive and "extend" their Class A operating range.

THD according to spice is really good, but, spice has a tendency to lie.  Spice will lie right to ones face.  (sometimes)  ((Depends on the models, I know)) 

[mgalusha]

THD according to spice is really good, but, spice has a tendency to lie.  Spice will lie right to ones face.  (sometimes)  ((Depends on the models, I know)) 

Have you measured the finished amp to see how it compares to the spice models?

[barrows]

Pass Labs current amplifier line works like this:  XA models are pure class A up to their rated output power (above their rated power they actually go on further in A/B), they run in single ended class A for the first handful of watts, then go into push/pull.  X models run in pure class A for only the first handful of watts, then transition into A/B, they also run single ended for the first few watts.
How does your amp sound in comparison to known references?

[sparko311]

Have you measured the finished amp to see how it compares to the spice models?

Mike - Youre local, right?  Im in Westminster CO. Arent you in Aurora or somewhere down south?   I dont have any THD testing gear.  I was looking on craigs list and considering buying an 800 dollar HP spectrum analyzer, and banging it all out the hard way.  Seems like someone around here ought to have an audio precision 1 (or some other such standard) that would be willing to swap some gear time for some beers....Or some free discrete op amps....  Or something    

[mgalusha]

Mike - Youre local, right?  Im in Westminster CO. Arent you in Aurora or somewhere down south?   I dont have any THD testing gear.  I was looking on craigs list and considering buying an 800 dollar HP spectrum analyzer, and banging it all out the hard way.  Seems like someone around here ought to have an audio precision 1 (or some other such standard) that would be willing to swap some gear time for some beers....Or some free discrete op amps....  Or something    

Hi Andrew, yes I'm local, we are south east of Denver, actually a little south east of Parker. I have a Prism Sound dScope III and an old HP 8903A, which is kind of a boat anchor but still very useful. We can certainly arrange something, ring me some time and we'll figure it out. I'll send you my number. The dScope isn't quite at the level of the AP but it's very close. I haven't fired it up lately, I'll have to find something to examine to re-familiarize myself.

mike

[G Georgopoulos]

It would be interesting to see distortion vs frequency of this class a amplifier...

[*Scotty*]

Here are some links to information on amplifier design which should prove interesting.
http://www.amazon.com/Douglas-Self/e/B001IQX7UU/ref=sr_ntt_srch_lnk_2?qid=1400454473&sr=1-2
http://www.douglas-self.com/
http://www.douglas-self.com/ampins/ampins.htm

Scotty

[G Georgopoulos]

THD according to spice is really good, but, spice has a tendency to lie.  Spice will lie right to ones face.  (sometimes)  ((Depends on the models, I know)) 

Thanks scotty for the links...

Andrew even douglass self uses spice and it's all truthful...

[sparko311]

Yes, I am totally a student of Dougless Self. 

Spice is certainly useful, but there are various things Ive found that spice just doesnt take into account.  One of which is the variance in stray capacitance of transistors (miller capacitance, etc) that varies with applied voltage.  For example, when a device is close to saturation, and the voltage across it gets smaller and smaller, the devices capacitance begins to get larger.  Its the same principle as a varactor diode, but spice doesnt seem to take this into account.  I have modified all of the models to the devices that I am using, which helps, but its still not entirely accurate.  The other thing about spice is that all of your connections are perfect wires which have no inductance, resistance, etc.  Layout paracitics can be manually put in to a simulation, but their exact value and distributed nature (rather than lumped) makes them approximations at best.

One thing that I did have to do with this design is clamp the VAS stage so as to not drive the output stage into saturation.  When the output stage gets near saturation, device capacitance goes up, and stability begins to be compromised. 

Hi Andrew, yes I'm local, we are south east of Denver, actually a little south east of Parker. I have a Prism Sound dScope III and an old HP 8903A, which is kind of a boat anchor but still very useful. We can certainly arrange something, ring me some time and we'll figure it out. I'll send you my number. The dScope isn't quite at the level of the AP but it's very close. I haven't fired it up lately, I'll have to find something to examine to re-familiarize myself.

mike

Thanks Mike.  I will be in contact with you sometime!   

[G Georgopoulos]

Yes, I am totally a student of Dougless Self. 

Spice is certainly useful, but there are various things Ive found that spice just doesnt take into account.  One of which is the variance in stray capacitance of transistors (miller capacitance, etc) that varies with applied voltage.  For example, when a device is close to saturation, and the voltage across it gets smaller and smaller, the devices capacitance begins to get larger.  Its the same principle as a varactor diode, but spice doesnt seem to take this into account.  I have modified all of the models to the devices that I am using, which helps, but its still not entirely accurate.  The other thing about spice is that all of your connections are perfect wires which have no inductance, resistance, etc.  Layout paracitics can be manually put in to a simulation, but their exact value and distributed nature (rather than lumped) makes them approximations at best.

One thing that I did have to do with this design is clamp the VAS stage so as to not drive the output stage into saturation.  When the output stage gets near saturation, device capacitance goes up, and stability begins to be compromised. 

Thanks Mike.  I will be in contact with you sometime!

Andrew,it all comes down to being objectivist or subjectivist,you cant be both..
I too think i have to do something myself..
You accept computer numbers or not..(software or hardware)
If you do then stick with it..
It doesnt matter if you use ap or other test equipment..
But if you choose ap you get a better idea where you are compared to other audio amplifiers..

cheers