Load line plotter for unbypassed gain stage

Discussion in 'Shock Brother's DIY Amps' started by Tom Kamphuys, Jun 24, 2020.

  1. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    754
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    In this thread (https://www.tdpri.com/threads/preamp-signal-through-my-rr800v.1032352/) it became clear that I've only seen fully bypassed gain stage load lines.

    I'm trying to update my scripts for unbypassed gain stages. I'm currently using https://worldradiohistory.com/Archive-Rider/BOOKS/Inside-the-Vacuum-Tube-Rider-1945.pdf (p327 and futher).

    The first baby steps have been taken. I'll try to give updates. As this is the first time I'm reading/doing this, any feedback from you guys that actually know this stuff is very welcome.

    Sneak preview:

    unbypassed.png

    Lower left lines are an error. I need to find a reasonable starting point for my fminsearch.
     
  2. Ten Over

    Ten Over Tele-Holic

    Posts:
    802
    Joined:
    May 13, 2015
    Location:
    Central California
    I like those little Martian subatomic particles. The electron has wings, a minus sign for a mouth, and lightning bolts coming out of its cephalothorax.

    Electron Stick Figure.png
     
  3. Ten Over

    Ten Over Tele-Holic

    Posts:
    802
    Joined:
    May 13, 2015
    Location:
    Central California
    I hope you didn't give up on this because I wanted to see what you would come up with.

    Plate Curves Unbypassed 1.5K Reduced.jpeg
     
  4. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    754
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    Too many little projects going on.

    Not given up, yet.
     
    Ten Over likes this.
  5. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    754
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    Not complete wrong, not completely right:

    comp.png

    It seems that the 'standard' grid curves are already different.

    @Ten Over Where did you get this picture? Somewhere online? Book? Or is there some software/webpage where you can do this?
     
  6. Ten Over

    Ten Over Tele-Holic

    Posts:
    802
    Joined:
    May 13, 2015
    Location:
    Central California
    I did it like they did it originally. I took some tubes and compiled a mountain of data and then I took a French Curve to it.

    My graphs are slightly different than the published fully bypassed plate characteristics in that the anode voltage is from the plate to ground and the grid voltage is from grid to ground as opposed to anode to cathode and grid to cathode. I included the Vgk = 0 curve so that it would be known where grid current becomes a problem under typical conditions.

    Plate Curves Unbypassed 820 Reduced.jpeg

    Plate Curves Unbypassed 10K Reduced.jpeg
     
  7. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    754
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    I thought you were high on something, but I found out today that the image is from the 1945 book :D

    comp2.png

    Promising!

    I don't get why you Vgk curve is different between your images.
     
  8. Ten Over

    Ten Over Tele-Holic

    Posts:
    802
    Joined:
    May 13, 2015
    Location:
    Central California
    The plate voltages on my graphs are from plate to ground. The plate voltages on the published graphs are from plate to cathode. The Vgk curve hits the +33 curve on the 10K graph at about 200V plate-to-ground. The cathode is at +33V, so the plate-to-cathode voltage is more like 167V. If you scoot the Vgk line over to 167V, it gets pretty close to my Vgk line for a 1.5K cathode resistor.
     
  9. elpico

    elpico Tele-Afflicted

    Posts:
    1,190
    Joined:
    Sep 14, 2011
    Location:
    Vancouver BC
    Just making sure you guys know spice exists. If you want to make a script to calculate electronic circuits for the challenge and learning opportunity of doing it I totally understand that and by all means carry on. Just in case though, thought I'd mention those tools do exist. They're even free.

    This is a quick spice plot of 12AX7 plate characteristics with a 1.5K cathode resistor. I haven't labelled the grid voltage of each line but the range is +5 to -5

    [​IMG]



    I'm not understanding what you mean about bypassed vs non-bypassed cathode resistors though, could you explain that to me? I'm not seeing what the difference would be here. These are just steady state idle currents, there's no ability for this kind of plot to represent frequency domain things like bypass caps is there?
     
  10. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    754
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    It is a learning experience. I can recreate the lines, so I'm now ready to use existing tools.
    I know spice exists. I have it installed and have used it. I don't know how to create the plot you show though. Could you somehow provide the file?

    The bypassed and unbypassed gain stages have a different gain. See e.g. section 1.18 of http://www.valvewizard.co.uk/Common_Gain_Stage.pdf . In another thread (Preamp signal through my RR800V) we could not match measurements with simulations. @Ten Over suggested we should use the unbypassed plate characteristics graph. I don't understand it good enough to give you a clear answer, but for an AC signal, the bias will not change in the bypassed version and will change in the unbiased version. The grid to cathode voltage will change: cathode current feedback (see 1.18). I thought this effect was captured in the images in the previous posts. Maybe @Ten Over can chime in...
     
  11. Ten Over

    Ten Over Tele-Holic

    Posts:
    802
    Joined:
    May 13, 2015
    Location:
    Central California
    The way that we use the published plate characteristics assumes that we have a fully-bypassed cathode resistor. We determine the various plate voltages with an AC input signal centered on the operating point. The cathode voltage must remain constant in order for this to work. This means a fully-bypassed cathode resistor for the gain stages that we typically use.

    Having the un-bypassed cathode voltage change with the AC input voltage makes the published plate characteristics a lot more challenging to use. In fact it is so challenging that they are virtually useless by themselves for determining the behavior of an un-bypassed gain stage with an AC input signal.

    Since we are interested in what the plate-to-ground signal is with a grid-to-ground signal, I made graphs with those parameters. These are static characteristics, also, but they aren't the same static characteristics as the published plate characteristics.
     
  12. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    754
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    I was hoping I could use the new plate curves in the unbypassed case, just as we do the normal plate curves in the bypassed case.
     
  13. elpico

    elpico Tele-Afflicted

    Posts:
    1,190
    Joined:
    Sep 14, 2011
    Location:
    Vancouver BC
    Okay I think see what you guys mean. For those two edge cases, fully bypassed and fully un-bypassed, you can ignore frequency so a static plot is all you need? Makes sense.

    Tom,

    I can send you the spice file for that when I get home but there's not much to look at. The magic is all in the 12ax7 model itself. This used a particularly good one made by a fellow named Adrian Immler. I believe he used a selection of modern tungsol reissues to generate it but I'll have to double check that.
     
  14. elpico

    elpico Tele-Afflicted

    Posts:
    1,190
    Joined:
    Sep 14, 2011
    Location:
    Vancouver BC
    This is the spice file

    Hit run, then when it's finished current probe the 12ax7 anode to see the curves. The bottom scale of the plot is marked "s" for seconds but it's actually voltage. Or actually, it's setup so they're equivalent. (the charting function in ltspice can be a bit inflexible and I was too lazy to take the data over to excel to make one with more sensible labels)

    Turns out Adrian went through several tung sol 12ax7 looking for one that best matched their published curves. He then created the model by making detailed measurements of that specific "golden tube" as he puts it. It's the most accurate 12ax7 model I've ever seen, I'm really impressed by his work. You can find him on Diyaudio if you have any questions about it.
     
    Last edited: Jun 30, 2020
  15. NSB_Chris

    NSB_Chris Tele-Meister

    Age:
    56
    Posts:
    490
    Joined:
    Sep 2, 2017
    Location:
    New Smyrna Beach Florida
    This is very cool!
    After the RR800V test thread, I purchased several books, so I have a much better feel for the basics. This discussion is great!
     
  16. andrewRneumann

    andrewRneumann Tele-Holic

    Posts:
    612
    Joined:
    Mar 22, 2020
    Location:
    Cincinnati, OH, USA
    Ok I am going to play devil’s advocate. I can clearly see that developing these curves has taken a lot of effort and I totally respect that. So please don’t take this as throwing darts.

    Once you have the curves, what will it help you accomplish? The design books/websites I have read don’t seem to bother with this process.
     
  17. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    754
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    Not a strange thought at all.

    I think it would help you find center bias.

    But I also have not seen 'm elsewhere. I'm wondering what to do with them. And what about all those lines with positive grid voltage? @Ten Over has measured them, so they seem possible. But what about grid current? I need to investigate them further, but currently I'm busy with metal work for my next amp...
     
  18. andrewRneumann

    andrewRneumann Tele-Holic

    Posts:
    612
    Joined:
    Mar 22, 2020
    Location:
    Cincinnati, OH, USA
    I thought you changed the grid voltage to reference ground instead of cathode, and that was why you have positive and negative lines—to reflect the actual signal appearing on the grid centered around 0V.

    I can say cathode degeneration is playing a major role in voicing my amp. I pulled a lot of cathode bypass caps to reduce the gain in the distortion stages, but this made it really hard to bias cold. Every time I would bias cold, instead of increasing distortion due to cutoff, I would just reduce the gain and increase headroom. This left me scratching my head until I realized what was going on—cold bias means more cathode degeneration. So I’m going to have to put at least one bypass cap back in for the cold bias stage. That’s where I’m at right now.

    Good talk good talk. :):)
     
  19. Ten Over

    Ten Over Tele-Holic

    Posts:
    802
    Joined:
    May 13, 2015
    Location:
    Central California
    They are positive with respect to ground, but not with respect to the cathode.

    You could relabel the curves on the fully bypassed graph so that the curve that crosses the load line at the bias point is zero volts. The lines to the left of it would then be positive and the lines to the right of it would be negative. The curve with an absolute value the same as the absolute value of the cathode voltage would then be the line that used to be labeled zero.
     
    andrewRneumann likes this.
IMPORTANT: Treat everyone here with respect, no matter how difficult!
No sex, drug, political, religion or hate discussion permitted here.