To get 2nd order distortion, take almost any valve or transistor and bias it badly (on a curved part of the characteristic - many people do this accidentally and call it "tone"). If you want no gain then use an anode load equal to 1/gm. For fairly pure 2nd you could try an ECC82/12AU7, as this does a good approximation to a square-law curve.
For higher orders too, use a misbiased overloaded cathode follower or a variable-mu valve. Variable-mu gives an output which is related to the exponential of the input, just like a BJT. You could make a valve guitar amp which sounds like a solid-state one!
For higher orders too, use a misbiased overloaded cathode follower or a variable-mu valve. Variable-mu gives an output which is related to the exponential of the input, just like a BJT. You could make a valve guitar amp which sounds like a solid-state one!
Hello DF96,
Correct me where I go wrong if you will.
Harmonic distortion is produced by the rate of change and slope of the transductance curve of the amplifier over the range of the input voltage. Triodes typically follow the 3/2’s power law and have a downhill stair step sonic signature of harmonic output; 2nd , 3rd, 4th . To the extent a device follows the square power law it produces only 2nd harmonic distortion (think JFET’s). Now take a look at graph on page 3 of this data sheet:
http://www.mif.pg.gda.pl/homepages/frank/sheets/030/p/PCC189.pdf
You can select a bias point to tailor your desired sonic signature. Yes even primarily 2nd harmonic.
All this assumes that we are building an “effects” amplifier.
DT
All just for fun!
Maybe, maybe not. The graph shows anode current and gm changing roughly with the exponential of grid voltage - note that the vertical axis is logarithmic. The plots are not straight lines, so the characteristic is not pure exponential but it is quite close. You may be able to find a bias point which reduces higher order terms, typically towards the higher current end, or add a small cathode resistor to bend the curve downwards.
Note that this exponential characteristic for variable-mu curves the opposite way from the normal 3/2 power law for a straight valve. This typically means that 3rd-order has the opposite sign. Exponential contains all orders (up to infinity!), just like 3/2 power.
Provided they are not driven too hard, all valves (and BJT) have declining amplitude of higher orders. This is simply a feature of any reasonably smooth curve described by a convergent infinite series.
There are a couple of VTs with some gawdawful plate characteristics you might want to try.
12AV7 -- medium gain dual triode
6SF5 -- high gain, Octal singleton. This one designed for use as a front end pre for AM BCB receivers. Since this is a Lo-Fi application, this type wasn't designed for linearity, despite nominal characteristics that look a lot like half a 12AX7. It's not. The 6F5 is the same type, but with a funky top cap connection to the control grid.
Just about any loadline will give H2 > 6%. For music reproduction, they're utterly nasty. For music production, may be what you're looking for?
12AV7 -- medium gain dual triode
6SF5 -- high gain, Octal singleton. This one designed for use as a front end pre for AM BCB receivers. Since this is a Lo-Fi application, this type wasn't designed for linearity, despite nominal characteristics that look a lot like half a 12AX7. It's not. The 6F5 is the same type, but with a funky top cap connection to the control grid.
Just about any loadline will give H2 > 6%. For music reproduction, they're utterly nasty. For music production, may be what you're looking for?
Cathode Follower
The Follower can do nice job of lowering the impedance of the driver section and ,depending on the output section , provide a better impedance match.
I recently built a 6SN7 preamp section with triode 1 CCS loaded direct coupled to triode 2 which is configured as a follower. The follower opened up the sound stage when compared to a regular plate coupled driver section. I attribute this to the match between the input and output sections.
Stuben🙂
The Follower can do nice job of lowering the impedance of the driver section and ,depending on the output section , provide a better impedance match.
I recently built a 6SN7 preamp section with triode 1 CCS loaded direct coupled to triode 2 which is configured as a follower. The follower opened up the sound stage when compared to a regular plate coupled driver section. I attribute this to the match between the input and output sections.
Stuben🙂
Using a suitable varistor as a cathode bias resistor should also give a simple exponential characteristic. Old telephone click suppressors had nominal resistances of a few kohm in vicinity of 0.1 to 1mA.
- Status
- Not open for further replies.