Geddes on distortion measurements

Status
Not open for further replies.
Jump to Latest
#101
I ahve to appologize for being unable to follow through on this post.

I found on inspection that what I was doing had some mistakes in it and I had intended to redo the measurements with the corrected calculations.

I am back to Thailand in two days and I won't have the time to follow up until probably November. I will post again once I have ironed out the situation to my satisfaction.

In the meantime I encourage everyone to do measurements at very low signal levels in amps, etc. and see if some reasonable approaches can be developed.

Later.
 
#102
scott wurcer said:


Never meant to say you weren't serious. I guess I missed the point, is it enhancing the resolution of sound card measurements or measuring the true THD at reduced output powers?

BTW one of those amplifiers appears to also have some serious power supply problems (120Hz sidebands on the harmonics).
Click to expand...
I seem to be losing track on this, but which measurement are you referring too?
 
#103
gedlee said:


In the meantime I encourage everyone to do measurements at very low signal levels in amps, etc. and see if some reasonable approaches can be developed.

Later.
Click to expand...

I will make these measurements on my class A design again and this time I will store them. I will make the same set on my class AB design with error correction, for comparison. The reasonable level of measurements would start at 100mVrms, i.e. 1.25mWrms into 8ohm load. I hope you will make some conclusion from my measurements 😀
 
#104
PMA said:
............. for comparison. The reasonable level of measurements would start at 100mVrms, i.e. 1.25mWrms into 8ohm load. ........
Click to expand...
Hi,
100mVrms is a peak current into the 8ohm load of 12.5mApk.
If the real speaker load were drawing a transient current of two to three times this value then the output stage could be delivering upto 36mApk into an 8ohm speaker from just 100mVrms.

I can see why Gedlee is asking us to examine low voltage outputs, since most ClassAB amplifiers are moving from ClassA to ClassAB at just these levels of current.

Let's see your ClassA vs ClassAB(+err cor) results. How about doing a ClassAB where there is no error correction, just plain feedback but with the optimum ClassAB bias that is below the test voltage/current?
 
#105
I believe that the image I am attaching is saying similar information. These are transfer functions of several output stages that I have built and compared. These curves perfectly fit to listening test results of the amplifiers with those output stages and show importance of smooth transfer of smaller signals.
 

Attachments

  • outstages.gif
    outstages.gif
    31.3 KB · Views: 754
#106
PMA said:
I believe that the image I am attaching is saying similar information. These are transfer functions of several output stages that I have built and compared. These curves perfectly fit to listening test results of the amplifiers with those output stages and show importance of smooth transfer of smaller signals.
Click to expand...
Is this a spice simulation?
 
#107
Hi,
based on my reading of Self's results, your's seem to fit with that model.
I am perturbed at how bad you have made Symasym (optimum bias ClassAB without error correction) look in comparison to ClassA.
Can you confirm Re and Vre for your test set up of the Symasym?
 
#108
soongsc said:

Is this a spice simulation?
Click to expand...

Yes, it is. All of them are built and working, though.


AndrewT said:
Hi,
based on my reading of Self's results, your's seem to fit with that model.
I am perturbed at how bad you have made Symasym (optimum bias ClassAB without error correction) look in comparison to ClassA.
Can you confirm Re and Vre for your test set up of the Symasym?
Click to expand...

Andrew, these are only ouput stages, without any global NFB. The symasym iddle current is about 40mA in this simulation. But, I operated it first at 150mA and then at some 500mA, after listening tests.

Class A is a CFP ("my" diamond buffer topology), much more linear, than 2EF.
 
#112
Well, what I am trying to say is, the proof is in the pudding - I am more interested in measurements on complete amps, regardless of how the nonlinearities are being taken care of. Of course the more linear to begin with, sans NFB, the better etc. But again, the proof is in the pudding of the final product.
 
#113
IMHO to uncover kind of non-linearities before NFB application is very, very important. The most important reason for me is that the listening results correspond to these characteristics. I have done too many "standard" measurements on complete amplifiers that discovered nothing, so I do not rely on them.
 
#116
That 150mA Ib gets me much closer, but I was originally struggling to get 50mA Ib + 45mA quiescent to fit the space available.
150mA will probably be outwith the dissipation capacity of the standard heatsink and I'm not prepared to remove all the external clues to the Tannoy name on the speaker.
 
#120
Hi


PMA said:
CFP is complementary feedback pair (or compound feedback pair)
2EF is double emitter follower (like Darlington)
Click to expand...


PMA thanks I got it.
http://www.dself.dsl.pipex.com/ampins/discrete/cfp.htm

I guess its basically the same topology you find in NAIM amps with the additional benefit to have the same power trannies rather than the need for a complimentary NPN / PNP pair.



PMA said:

But we are a bit OT in Earl's thread with this explanation.
Click to expand...


Yeah – more than just slightly OT I'm afraid -


Greetings
Michael
 
Status
Not open for further replies.
Top Bottom