MPP

If i take the circuit as it is...Then

Gain is app 65 dB at 1 KHz.
The Riaa curve Fits perfectly will app 0.2 dB low-end lift.
Distortion is not super low at app -50dB for the second order, but only second and third order exists. 3.order app a hundred times lower...
Juggling a little with the offset....It's low (app 2 mV/out) when the circuit is floating.. but at -1.5V with fixed rails..hmm
 
This effect with second harmonic is interesting. Should sound like tubes ?
Distortion could be brought down a bit by lowering the gain to 60dB maybe. I am happy that my RIAA values worked out so well. I simply used Baxadall´s formula and interpolated the output impedance of the mirror. How did the buffer perform with the floating supply ? Was the output current limited in any way as Calvin predicted ?
The DC offset could be trimmed away in the fixed rail version by using the trim components like in the output buffer maybe. That is a little bit more noisy with the 5 Ohm resistor in series. I could life with that. Thank you Michael that you put serious work into this.
 
The supply for the buffer also needs a tie in the center..but it could be that in real life it can be floating...think often the simulator goes basurka (Viking like) when something is floating,, where in real-life it'll find a balance..

Gain can be lowered 5-8 dB by scaling the Riaa

If the trimmer in the buffer is 15 ohm.. its possible to trim the offset...so it in the uV range..but thermally I don't know how it'll drift. Don't fancy a servo...and to be honest would A cap in the output do any damage....?? apart from the fact that DC-coupling would be nice..

Think your realistic experiments will reveal this if thermal drift is a problem..

I think the distortion specter is super nice....only second order...and the rest is super super low...even if distortion was 10 times higher is would still be nice...I believe distribution matters much much more than level.

Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 2.306e-01 1.000e+00 179.98° 0.00°
2 2.000e+03 1.082e-04 4.695e-04 -90.05° -270.03°
3 3.000e+03 2.606e-09 1.130e-08 -170.18° -350.16°
4 4.000e+03 2.327e-11 1.009e-10 69.02° -110.96°
5 5.000e+03 2.104e-11 9.123e-11 176.24° -3.74°
6 6.000e+03 1.155e-11 5.008e-11 -169.19° -349.17°
7 7.000e+03 6.884e-12 2.985e-11 -79.82° -259.80°
8 8.000e+03 1.891e-11 8.202e-11 -15.56° -195.54°
9 9.000e+03 1.334e-11 5.786e-11 2.59° -177.39°
Total Harmonic Distortion: 0.046946%
 
There's a problem with adding the DC-blocking cap before the buffer stage... While it's nice to keep it small..I also raises gain in the low frequencies at the front current stage... this to a degree of app + 40 dB..from at Riaa gain at 10 Hz..at app 80 dB to more than 120 dB.

Don't see much change wit the different mirrors...somewhere i saw an analysis.. that the the difference in current precision between the different types mirrors is only 0.something percent.. So frankly I would settle on at simple 4 transistor mirror.. and maybe have a bleeder to aid in setting the impedance...

Distortion and profile remains the same and so does the amplification...
With 2 x improved Wilson mirrors the resistor in each branch must be lower (app 200 ohm) to get the right working conditions..
 
Regardless of the mirror-type...I think is a bloody clever circuit...You get enough amplification with super low noise for a MC stage...and only using N-channel fets...add to that a distortion that is decreasing with frequency and as such has lower distortion at 20KHz tnan at 1 KHz..and also that the distortion profile is limited to only 2. order....I bet you I'll sound fantastic.. It also a fast circuit that extends into the MHz area....If nicely implemented...and tuned...I believe It'll be hard to beat...One contender could be My One shot Riaa that bears some of the same strengths...
 
This here seems to work very well.. if the DC cap is placed in front of the buffer... we need the high impedance of the input jfet.. If we add something like a M-ohm we start loosing base...The reason for my trouble was that i already added 100Kohm to gnd.. and thus screwed up the riaa by a few dB..even with 1Mohm you loose app 0.5 db in the base...

One worry is the need for matched resistors in the current mirrors, if just 1 ohm unbalanced we could run into problems...Here again ...practical implementation may show something different.
 

Attachments

  • Joachim Riaa.jpg
    Joachim Riaa.jpg
    145.9 KB · Views: 388
  • Joachim Riaa-deviation.GIF
    Joachim Riaa-deviation.GIF
    79.2 KB · Views: 374