
Home  Forums  Rules  Articles  The diyAudio Store  Gallery  Blogs  Register  Donations  FAQ  Calendar  Search  Today's Posts  Mark Forums Read  Search 
Solid State Talk all about solid state amplification. 

Please consider donating to help us continue to serve you.
Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving 

Thread Tools  Search this Thread 
27th December 2012, 04:34 PM  #111 
diyAudio Member
Join Date: Nov 2003
Location: Amsterdam

/ Semantics on
And the gm variation is caused by bias voltage modulation. So the bias voltage modulation is the root of all evil and gm variation, or more precisely, effective gm variation, is just a consequence. Gm varies in a way as depicted, simply because I have included the bias voltage in the definition of gm. If we define gm as the sum of gm of the top respectively bottom tranny itself, thus the bias voltage not counted, we get a completely different picture. But that's not a realistic approach, because in a real amp, we can't ignore or exclude the effect of the bias voltage. / Semantics off Cheers, E.
__________________
Een volk dat voor tirannen zwicht, zal meer dan lijf en goed verliezen dan dooft het licht…(H.M. van Randwijk) 
27th December 2012, 05:01 PM  #112 
diyAudio Member
Join Date: Jan 2011
Location: London

I disagree with the "bias modulation" approach. The most logical way to discuss this distortion type is to consider the OPS gain vs. Vin dependency (or "gain modulation" if you prefer). Plot the OPS gain (always < 1) vs. the OPS input voltage Vin and you'll get the whole picture. Then think Gain=Rload*gm, valid at both small and large signal, therefore you have an equivalent "transconductance modulation" effect. An equivalent "bias voltage modulation" doesn't help as much understanding the causality beyond this distortion type.
The "gm doubling" denomination is confusing, it should be "gm variation". There is a significat difference between bipolars and mosfet OPSs. If the mosfets are (auto)biased far away from the subthreshold conduction region (where Id(Vgs) is as much exponential as Ic(Vbe)), then the "gm variation" distortions (excluding the gm drop at high currents) are much smaller. When discussing dependencies on the OPS Iq, are you sure you are considering bipolars, rather than mosfets? As a side note, "switching distortions" are only important at HF, where storage effects in bipolars can no longer be ignored. See the Leach paper I quoted above. 
27th December 2012, 07:49 PM  #113 
diyAudio Member
Join Date: Sep 2006

Let us try to summarize the problem in terms acceptable to everyone (I think all the participants to this discussion agree in fact on the core nature of the problem, but not on its semantic description).
When only (ideal) semiconductor junctions are present, no serious problem exists: the resulting transconductance is the sum of all the transconductances of active elements present, whatever their polarity, and when the current is transferred from one side to the other, the transconductance remains identical. That is not true anymore when class B currents become large, but practically this effect is relatively minor, given the typical bias current and output load magnitude of "normal" amplifiers. A more serious problem occurs with "normal" values of emitter resistors: if we simplify the problem and decide they dominate the output resistance, the output resistance in class A becomes ~Re/2, and in class B ~Re. That's where the (so called) gm doubling effect comes from. Ideally, both effects should be taken into account, but there is anyway a mitigating effect: the OPS is normally not voltagedriven (there are exceptions). Such differences should therefore be mostly accessory, except for the frequency compensation which renders the drive voltagemode at higher frequencies. Are the differences that important?
__________________
. .Circlophone your life !!!! . .
♫♪ My little cheap Circlophone© ♫♪ 
27th December 2012, 09:47 PM  #114  
diyAudio Member
Join Date: Nov 2009
Location: algeria/france

Quote:
output current dependant voltage drop through RE and Re wich will add/substraxt to the requested Vbe once the output current start varying , then we can interpret it as the varying gm being the cause of Vbe modulation. Quote:
seems right to me. When looking at the output stage , Gain = output/input = RL/(RL + 1/gm) , we can immediatly see that if gm is constant there will be no distorsion but if gm is varying with output current then the gain will vary as well , hence there will be distorsion. 

27th December 2012, 09:55 PM  #115  
diyAudio Member
Join Date: Jan 2011
Location: London

Quote:
Good night. 

27th December 2012, 10:07 PM  #116 
diyAudio Member
Join Date: Nov 2009
Location: algeria/france

There is no Re that is taken into account , actualy that is the
formulae for Re = 0. RL is the load impedance and the formulae just show that the gain is less than 1 unless gm is infinite. 
27th December 2012, 11:07 PM  #117 
diyAudio Member
Join Date: Jan 2011
Location: London


28th December 2012, 10:25 AM  #118  
diyAudio Member
Join Date: Nov 2003
Location: Amsterdam

Quote:
Gain = output/input = RL / ( RL + 1/gm + RE' + RE) , where gm ~= Ie / 26mV, RE' is the internal and RE is the external emitter resistance. Now let's define R_{top} = 1/gm_{top} + RE'_{top} + RE for the 'top' tranny and R_{bot} = 1/gm_{bot} + RE'_{bot} + RE for the 'bottom' tranny, then gain = RL / ( RL + R_{top}R_{bot} ) and the gm of the whole circuit is 1 / ( R_{top}R_{bot} ), okay? Now, let's plot this stuff and look at the black curve, which depicts the combined gm (of top and bottom tranny). Contrary to my previous post of gm, the green curve, this one doesn't show the 'doubling' at large currents. Instead, only a small increase in the crossover zone. This is because the circuit is a bit over biased (which is another story). Despite the fact we have taken into account the effect of Vbe, RE' and RE, this kind gm does not explain the excess of distortion caused by a sliding bias. Please tell me what I did wrong. Cheers, E.
__________________
Een volk dat voor tirannen zwicht, zal meer dan lijf en goed verliezen dan dooft het licht…(H.M. van Randwijk) 

28th December 2012, 01:58 PM  #119 
diyAudio Member
Join Date: Jan 2011
Location: London

The usual confusion between a small signal, linearized, and a large signal analysis. gm=Ic/26mV holds for small signal only. For large signal, Gm(0)=gm/2 where "0" stands for Vo=0. In fact, Gm depends strongly on the output level Vo (otherwise, for example, a bipolar oscillator would not work).
There is no simple way to calculate the large signal transconductance. The large signal analysis of an emitter follower with Re does not have any known analytical closed solution. 
28th December 2012, 03:35 PM  #120 
diyAudio Member
Join Date: Nov 2003
Location: Amsterdam

I wrote: gm ~=Ic/26mV and notice the tilde.
Whether this equation is correct doesn't matter, as I didn't use it for any calculation at all. My simulator did the calculations, which also takes care of large signal and other effects. And please don't start whining again about the accuracy of simulators. For the purpose at hand they are accurate enough. The point is that the composite gm of the bipolar output trannies plus emitter resistors did not reveal the gm doubling at large currents. That means you can't use it for explaining the distortion. Perhaps you forgot it, but we are talking about the real cause of distortion: is it bias voltage modulation or is it gm modulation. According to my last plot (black curve) it is not gm modulation. edit: the astute reader would notice that the 'green' gm curve does show gm doubling. So, wtf are we talking about? Well, the 'green' gm has taken into account the effect of bias voltage modulation, while the 'black' gm does not.
__________________
Een volk dat voor tirannen zwicht, zal meer dan lijf en goed verliezen dan dooft het licht…(H.M. van Randwijk) Last edited by Edmond Stuart; 28th December 2012 at 03:45 PM. 
Thread Tools  Search this Thread 


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
ClassAB meets ClassD: Yamaha's EEEngine Topology  where are this Diy Projects?  tiefbassuebertr  Solid State  62  30th May 2015 07:53 AM 
Collection of Class B topologies <100mA Idle and Sound closest by Class A  tiefbassuebertr  Solid State  37  27th July 2012 08:04 AM 
Can a Class AB PP amp be said to be operating in Class A at low signal levels?  ray_moth  Tubes / Valves  19  23rd January 2009 07:52 PM 
How about a roundup of Class A kit power amps, or collectable vintage class A?  Brisso57  Solid State  4  14th February 2007 10:30 AM 
New To Site?  Need Help? 