Variable Harmonizer

[ervinl]

Hi,

Interesting discussion.

[Wavebourn]:
- I'm still confused about Rk adjustment, it does adjust the gain/volume, but in one of schematics, there is a bypass C, so gain/volume does not matter here, so I'm just curious if it related to amount of odd - even harmonics generated.

- About FET, is it only MOSFET, or including JFETS? And in what topology 2nd harmonics will be more than 3rd and higher orders? Common Source vs Follower? Or SE or PP/Complimentary?

[keantoken]:
Which complimentary topology will generate most 2nd than even and higher order harmonics? Common Source or Follower, or both the same?

And is there matter about VAS, as there is a VAS with SE driving PP, and there is also symmetrical VAS (dual pure SEs are combined in PP)?

Thanks,

Ervin L

 

[keantoken]

[keantoken]:
Which complimentary topology will generate most 2nd than even and higher order harmonics? Common Source or Follower, or both the same?

And is there matter about VAS, as there is a VAS with SE driving PP, and there is also symmetrical VAS (dual pure SEs are combined in PP)?

Thanks,

Ervin L

The amount of high order harmonics depends on the loading of the transistor. After that, peripheral components like biasing resistors may emphasize this distortion but the difference between these mechanisms is that the transistor's distortion will have a steeply slanting line of harmonics at low levels and a non-sloped trailing line of harmonics at high levels; the peripheral devices assuming they are passive, will emphasize all harmonics equally. This goes for any transconductance device I believe, even tubes; the difference is tubes have more space between the two extremes.

So to compare between topologies we must set the baseline to be that AC current of the transistor must be the same for each test; this way we can measure the effect of peripheral circuits on the transistor's unchanging distortion.

Using this test we can see the distortion of a VAS depends on the B-E resistor; the smaller this resistor the greater the distortion induced by the Vbe transfer. So it seems in most applications the distortion of the VAS will be larger, relative to the input signal strength, simply because the resistor emphasizes the distortions already present. However remember that the resistor won't discriminate between benign and non-benign distortions, the type of distortion benign or not will be the same, just more of it. For low Hfe VAS's, Ib linearity adds to this distortion. As long as you operate the transistor in a region where the Hfe/Ic line is either straight or straight and slanted upwards (not in the hump or "linear region"!) on a logarithmic graph, the distortion will either be small or it will be the same type and polarity as the Vbe distortion, so in keeping low levels it should not be an issue. Another consideration is Early affect. I don't have concrete knowledge on this but in my experience, Early affect is usually benign and only injects low order harmonics (though I can't be sure about this). However I still recommend cascoding because Cob has the potential to inject switching trash from later EF stages in power amps.

In the follower position, if there is no B-E resistor or resistors whose current directly or indirectly depends on the Vbe of it in the audio spectrum, it will usually distort the least, less than the VAS. With this distortion mechanism out of the way, the second consideration is Ib. Many japanese BJT's have a straight Ib/Ic curve, these will have lowest distortion. Most other have an Hfe curve that rises logarithmically with Ic, these will inject monotonic (which means a straight line of decreasing harmonics) distortion as well, but of the same polarity as the Vbe distortion so they won't cancel and leave rogue odd harmonics; still benign at low levels.

All complimentary SS combinations I know of produce dominant odd harmonics, except the one I have spoken about. It is good to know that a good VAS will ignore the Vbe errors of later follower stages; if you can avoid ruining this by an over-sized Cdom, and cascoding will greatly help.

The trouble with complimentary stages is that they have the potential to completely cancel the even harmonics and usually the solution which does this offers the best "engineering" bang for buck. This may not be best for sonics, although it always depends on the application. Luckily, this canceling can be easily interrupted by changing the operating points of the devices so their curves don't match and cannot cancel completely. IE in symmetrical arrangements, give one transistor say twice the Ic of the other. Most times it is only the "null point" where all evens are canceled which we want to avoid, and this is like a notch filter, a small bump in either direction may be enough. However since the 2nd harmonics in these stages are always opposite phases, you cannot avoid canceling some even harmonics and thus emphasizing odds.

Wavebourn:

Gotcha! It does work and I think it will work for all transconductance devices with +- symmetry.

- keantoken

 

[Wavebourn]

Take some JFET and experiment with it, you may find that it is a good voltage controlled resistor.

So this thingy will generate almost 2'nd order harmonics:

 

[keantoken]

My circuit uses 4 BJT's (2 for simple version) and one pot for "3rd null".

Does your circuit generate a profile where the 4th harmonic is higher than the 3rd? Mine takes two monotonic profiles and cancels the odd harmonics...

I worry about the pot in the feedback loop, if this disconnects output will stick to a rail.

- keantoken

 

[keantoken]

Okay, time to spill the beans...

Look at this post to see the specs for the device as incorporated into my headamp:

http://www.diyaudio.com/forums/head...phone-amp-jlh-output-stage-7.html#post2289775

Here is the standalone circuit:

http://www.diyaudio.com/forums/head...phone-amp-jlh-output-stage-4.html#post2220416

Up a few posts I posted several variations of the circuit, and a shot of it working on the scope (the peaks are uneven, but this is because of a distorted input signal).

- keantoken

 

[Wavebourn]

I worry about the pot in the feedback loop, if this disconnects output will stick to a rail.

Don't worry, it is a basic principle to show how did I generate almost 2'nd order harmonic for electric guitar.

 

[Dobson]

About that guitar harmoniser, I see no plausible way to get that tube in my country... And what FET it is?

 

[Wavebourn]

About that guitar harmoniser, I see no plausible way to get that tube in my country... And what FET it is?

Can you find 6BJ6 remote cut-off pentode and 6N1P triode?
You may use them, paralleling both sections of 6N1P

MOSFET is used a source follower in a voltage stabilizer. Any MOSFET with breakdown voltage 250V or higher may be used, like IRF730

 

[otto88]

Wavebourn (or anyone else!)

What circuit do you think would, with taper dialled in allow controllable 2 or 3 HD,
yet be most transparent when you want it to be – for well recorded music?

(Putting aside the choice of power amp – fyi I’m thinking of doing an OTL power amp for a very transparent sound, fine detail, etc; but using a Variable Harmonizer version ?, to allow me to listen to & enjoy my many 60 - 70s rock, 50s blues etc discs without having to lower the volume).

 

[otto88]

Hi Wavebourn . .

"Flipping the circuits’ use around" to know its real flexibility, what are your comments on its ability to be transparent?

 

[Bigun]

If you have an amp with negative feedback, it's relatively easy to create even order harmonics (in theory):

http://www.diyaudio.com/forums/solid-state/168503-tgm-amp-goes-tubey.html