Small signal myths and real measurements

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
More and more posts in the forum deal with pseudo theories and unsupported claims about mysterious small signal distortions in circuits with negative feedback. I would like to post some of my measurements that prove that opposite is true in case of 'normal' design. I would like to encourage the others as well to post their results.

I am starting with a link-level preamplifier based on OPA627 and DRV134 and measurements related to 0dB = 146mV. I will appreciate your comments on 'small signal distortions'.
 

Attachments

  • ccif.PNG
    ccif.PNG
    73.1 KB · Views: 158
  • fft_1kHz.PNG
    fft_1kHz.PNG
    72 KB · Views: 161
  • fft_10kHz.PNG
    fft_10kHz.PNG
    69.4 KB · Views: 155
Pavel, would it be possible to do some IMD measurements on that component, where the level of the 2nd signal is down -20, -40, -60, etc, compared to the 1st - to satisfy my curiosity ...

Thanks,

Frank, I appreciate your curiosity. Reducing amplitude of the 2nd signal only makes things easier for the amplifier.

1) 19+20kHz. All intermodualation products disappear and the only product you can see is the 2nd harmonic of 19kHz, i.e. 38kHz, about 96dB below 19kHz line level. Further reducing of 2nd signal amplitude brings no change so please allow me not to post images.

2) 1+20kHz. No intermodulation products available.

The conclusion is that smaller the signal components less work for the amplifier, in this circuit under test.
 

Attachments

  • ccif3.PNG
    ccif3.PNG
    68.4 KB · Views: 129
  • ccif4.PNG
    ccif4.PNG
    68.7 KB · Views: 119
https://web.archive.org/web/20101203210751/http://www.analogzone.com/tmt_1002.pdf

I find the intersting part starting ~ p15 - the ever popular external Class A bias of an op amp output - but with actual measurements

Thanks for the link, this is a good and interesting publication.

We may also replace rail resistor by self-biased JFET.
My question - in case of 1mA output stage idle current and 10k load the output pp stage would remain in class A up to 20Vp output voltage (never used in a link stage). Does it make any sense to use output stage supply resistor then? I understand it may make sense for 600 ohm load, but then I would go for a BUF634 or discrete buffer.
 
Also interesting is no big influence of heavy load in case of small signals. Output impedance is 50 ohm, please compare effect of 10k and 50 ohm load. We can see the plot shifted down by 6 dB (50R : 50R divider). 0dB = 150mVrms approx., as shown in 1st post of the thread.
 

Attachments

  • load 10k.PNG
    load 10k.PNG
    70.9 KB · Views: 40
  • load 50R.PNG
    load 50R.PNG
    69.7 KB · Views: 40
You know guys I like the recommendations. But I do not like unsupported claims and pseudo-theories that are often offered by 'gurus' that in fact try to support their design philosophy and commercial success at first place. I am in both camps, objective and subjective. IMO, good parameters are necessary to get good sound, i.e. I cannot tolerate 'mask and hide' approach. And that's what you really often get, no names. If you read in other threads, you will see that my key target is the fight with EMI interference and its influence to audio signal. This is the second must, together with perfect audio band parameters.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.