Germanium investigations

Ok did some more testing with a simple Vbe multiplier to try taming the wild offset. I tried placing Q5 on various points on the heatsink and even directory on top of one of the output transistors. It doesn't seem to match the coefficient of the output devices but that's not surprising. What is surprising is how it effects the THD.

680 ohm resistor 0.2%,
Vbe with MP26 (PNP) 0.55%
Vbe with MP10 (NPN) 2%

Why does this simple ccs add so much distortion? Also interesting that the distortion is mostly second harmonic.
 

Attachments

  • 1T806 Thermal Comp.png
    1T806 Thermal Comp.png
    39 KB · Views: 290
Why does this simple ccs add so much distortion?
I can't say anything about this. Everything needs to be well researched. Still, it would not hurt to replace the MP10 with a silicon transistor. The deviation of the "zero" voltage at the output should be more stable. The author got such a THD.
1. 1 kHz, 1 V, 8 ohms;
2. 1 kHz, 6 V, 8 ohms.
 

Attachments

  • 1.png
    1.png
    38.8 KB · Views: 94
  • 2.png
    2.png
    40.8 KB · Views: 89
Interesting patent.

Did some reading on the JLH class A update site https://sound-au.com/tcaas/jlhupdate.htm and decided to implement their two transistor current source and it works great! THD is just as low (if not lower need more testing) than a resistor.

I don't understand why replacing the first transistor with silicon would help with offset drift. It draws 2ma and only heats up from the surroundings. The middle transistor is what really makes the difference. A small heatsink does help, but the designs from post #588 seem to agree.
 
Thought someone could make use of these files. Those large transistors require a clamp to hold them to the heatsink. Mounting multiple transistors on the same heatsink can be a problem because of the metal clamp. Here are step and dxf files for anyone wishing to 3d print or laser cut some clamps from plastic. Now you can put a thermal pad under the transistor and clamp them down without anything shorting. Is there a proper name for this particular package?
 

Attachments

  • Soviet_Transistor_Clamp.zip
    16.9 KB · Views: 85
  • Soviet_Transistor_Clamp.png
    Soviet_Transistor_Clamp.png
    12.8 KB · Views: 76
  • Screenshot 2022-06-06 at 10-12-15 4.56US $ 2 PCs transistor 1t806v 1t806 v Transistors - AliEx...png
    Screenshot 2022-06-06 at 10-12-15 4.56US $ 2 PCs transistor 1t806v 1t806 v Transistors - AliEx...png
    62.9 KB · Views: 96
  • Like
Reactions: 1 users
Here are step and dxf files for anyone wishing to 3d print or laser cut some clamps from plastic. Now you can put a thermal pad under the transistor and clamp them down without anything shorting. Is there a proper name for this particular package?
You can do it, but the plastic must be durable and 3 ... 5 mm thick. These are clamping washers (flanges). We ordered the same ones in China, but for installing diodes as a temperature sensor.
 

Attachments

  • 4084790.jpg
    4084790.jpg
    21.7 KB · Views: 85
  • DSC03043.JPG
    DSC03043.JPG
    172.6 KB · Views: 85
Ok did some more testing with a simple Vbe multiplier to try taming the wild offset. I tried placing Q5 on various points on the heatsink and even directory on top of one of the output transistors. It doesn't seem to match the coefficient of the output devices but that's not surprising. What is surprising is how it effects the THD.

680 ohm resistor 0.2%,
Vbe with MP26 (PNP) 0.55%
Vbe with MP10 (NPN) 2%

Why does this simple ccs add so much distortion? Also interesting that the distortion is mostly second harmonic.
Because it is not a current source but a voltage source. It is a vbe multiplier! 😳
 
Account Closed
Joined 2010
RE the "germanium triode" experiment mentioned sometime ago: Firstly, the linked article is not clear on how exactly they connect the DUT (for the first, "not nice" set of curves, they say "a PNP transistor gets simply used as NPN transistor", while for the triode-like curves they say "take a ... PNP Transistor, connect it simple reverse, so as NPN Transistor" :-/). I tried the various combinations, and the one where "simply" the collector and emitter are switched does indeed result in triode-like curves... for any transistor, even silicon ones, and you can easily simulate that in Spice:

View attachment 1024674
(P210 is a germanium PNP, [2S]A1186 is silicon; disregard the MOSFET-related stuff - that was just a hack to make the Peak tester do voltage drive instead of current drive.)

View attachment 1024675
I think early guitar fuzz effects used germanium trz like that.

I may be wrong (I often am :-D), but aren't these just the Vbe vs. Ic curves, offset by the base voltage source step? Meaning there's no practical triode-like behaviour as such...?
 
Incredible good sound from this little design I created based on old schematics.
In fact, so good and sweet to listen!

PS.: I need to see on PCB the values missing in schematic. :smash: I calculated and tested when I stuffed the PCB :spin:
 

Attachments

  • 20220628_135745.jpg
    20220628_135745.jpg
    181.5 KB · Views: 160
  • Like
Reactions: 1 user