2nd Harmonic, even harmonics

[joaquins]

Simple BJT biased and driven for high 2nd harmonic:

Was going to post this too but PRR beat me to it 

I'll just point out that

• below about 10% THD, the scaling PRR mentions is almost perfect.  1mV at input gives 1% THD etc
• V2 has to be a true, low Z voltage source.
• If not, THD will be reduced in proportion to the source Z compared to the input Z of the BJT

thanks ricardo,
what exactly do you mean by the  "true, low Z voltage source"

Driving it directly with an opamp will do.

JS

 

[kambo]

ohhhh got it
thanks again

 

[joaquins]

I don't know your exact application or needs, be aware the level is really low so to get that low output from a operational you have to be careful, noise could be a problem but burning the transistor as well. You need some controlled DC to bias the transistor for it to work properly, probably something between 0.6V and 0.7V will do but I'd try to get an adjustment to it just to be sure.

Once you have the proper DC in place applying too much signal from a low impedance source will make it saturate, no problem for the collector since it has a resistor to hold it in place, but the base is not so tough than can hold the output current from any opamp without trouble (depending on the transistor used) 2N2222 shows a 50mA current on the base for the max saturation condition, I'd try to say below that. Using a wimpy opamp is protection enough I guess. A TL07X is in that range of maximum output currents. You should use an input attenuator to get the signal to few mVpp as desired.

I wonder if a higher power (lower hFE) would work as well, it would stand the abuse much better and have less gain which may help to reduce the noise.

JS

 

[kambo]

sounds like i better order half a dozen 2N2222 

 

[PRR]

> You need some controlled DC to bias the transistor

That sketch will ONLY work in the phantom world of SPICE.

In real life the base DC bias will be different for every transistor and for any change of temperature.

You need NFB. At least a DC loop around the transistor. Since the point is to hold it "in distortion", but the waveshape takes a severe offset when it gets big, it may need to be a gated servo.

Base blowout should not be an issue with any modern transistor.

Bigger transistors should not have lower gain, Shockley's Law applies.

 

[Bo Deadly]

A transistor is a very poor substitute for a tube when it comes to making musical harmonics. Something that should be considered is that the harmonics increase over some range of input that is 6-10 dB with a tube but much less with a transistor. It's IV curve is much sharper. A JFET with a high voltage supply, a bootstrapped load to get the gain up and some emitter resistance to reduce the IV curve exponent will be close. But that doesn't get into grid current limiting which I think is a very important part of the tube sound.

I've messed around with trying to make a tube-like circuit from conventional components for years. I never came up with anything that I thought sounded even vaguely like a tube. There is a long history of failure in this area. If you want something that sounds like a tube, use a tube. If you want your rig to sound like Led Zeppelin, you'll have to have Jimmy Page come over and play it for you.

 

[kambo]

If you want your rig to sound like Led Zeppelin, you'll have to have Jimmy Page come over and play it for you.

100000000000000000000000 +
i totally agree with you

 

[ricardo]

• V2 has to be a true, low Z voltage source.
• If not, THD will be reduced in proportion to the source Z compared to the input Z of the BJT

what exactly do you mean by the  "true, low Z voltage source"

In PRR's #11 circuit, Q1 is running 3V/R1 = 0.6mA.  The emitter resistance is 26R/0.6 = 43R3.  If we assume hfe is 100, the input resistance of Q1 is 100 x 43R3 = 4k33

The source resistance from V2 must be MUCH less than that.

In PRR's example, he quotes 3.3mV for 2.9% THD .. which is what you get if V2 is truly low impedance.

If the source resistance from V2 was 4k33 (ie same as Q1 input), 2 things happen.

First the input to Q1 is halved which also halves the THD.

But because Q1 is now fed from less of a voltage source, the THD is reduced again ... halved again cos the source resistance is the same as Q1's  input resistance.  So I expect about 2.9% / 4 = 0.73% THD.

I used to do THD estimates for new PA designs like this.

Du..uh!  it was about 40 yrs ago before we had stuff like SPICE.    I now expect PRR and da other SPICE gurus to tear me to bits for half remembered BS  8)

 

[merlin]

is there a general rule for increasing even harmonics in tube pre design.

Yes, use a variable-mu triode.

Alternatively, a degenrated pentode with the suppressor connected to ground (not to cathode) can generate so much second harmonic that you get frequency doubling! http://ax84.com/bbs/dm.php?thread=379279

 

[kambo]

thank you all for all the input and valuable information...

i should be able to make one Melin's Sitar fx this week, i have couple of EF86 in my stash

 

[PRR]

> half remembered BS

Like you, I prefer to count on my thumbs, using SPICE only to get 8-place answers more impressive than matchbook scribbles.

The transistor is *sweeping* almost zero mA to 2mA.

Using your "26" number, re goes down to 13 Ohms, up to (oh, say) 260 Ohms.

Ass-uming hFE=100, base impedance swings from 1.3K to 26K. Far on each side of your 4.3K simplification.

Source resistance must be "much" less than 1.3K. If source is 1.3K, that spike on the downside is only half as severe, THD significantly less (as Ricardo says, perhaps half). (It does however get much easier to pass through later stages; headroom management gets tough with very lop-sided waveforms.)

And signal level is very low, putting us at risk of hiss. Coming out of "normal" sources, like opamps, it is probably reasonable to use a resistor divider. Assuming we can make 3V signal, we want say 100:1 reduction. Taking "much less than 1.3K" as 100 Ohms for the lower divider resistor, the upper is 10K Ohms, an easy load.

If the source impedance is "much" higher than the high-side 26K, then the Vbe/Ie curve drops-out. Now we look at the flatness of the hFE curve in the small-mA to 1mA area. For most modern transistors this will be super flat, "no" distortion. (Next best thing to common-Base drive, where output current IS input current, just at bigger voltage.)

 

[kambo]

EF 86 merlin's Sitar FX...

i dont know if i made the whole thing working the way is suppose the work, but
it sounding good to me

i didnt use the 10nf input cap  :

i have 600:600 on input, and 10k:600 on output. 250VDC for B+

THD %36.4
2nd %36.2
3rd %0.69


edit  : frequency response is pretty flat too...

 

[kambo]

i didnt use cathode resistor, do i have to ?

i am getting very nice even harmonics at unity gain...

 

[leadbreath]

Interesting... might try this circuit out...

 

[Bo Deadly]

i didnt use cathode resistor, do i have to ?

i am getting very nice even harmonics at unity gain...

The cathode resistor is important for biasing. You are basically just biasing the circuit low (meaning anode voltage is low). Without the cathode resistor that could contribute to asymmetric gain because of greater swing in current. However, at some point if the bias is low enough the signal will simply start to clip sooner. If you had an adjustable negative supply like the bias in a power amp and then you could use that to pull the grid negative a few volts, look at how the signal clips in the scope and then adjust the bias so that clipping is even on positive and negative swings. That will maximize the 2nd harmonic from asymmetric gain but give you odd harmonics from clipping which might be more interesting. Just running with the bias very low is strange.

You could also put an LED in place of the cathode resistor to adjust the bias but still get the increase in current swing. You probably want the anode voltage to be more like 150V not 85V. Try different color LEDs and see what anode voltage you get.

 

[kambo]

thanks for the technical explanation. much appreciated
i was biasing with ear and watching the harmonics from the room EQ.
85V might be wrong... schematic was from the old drawing, so i will re_check, and try your suggestions.
i was really lazy to hook up my scope... i will do that too....
cheers