Does the JFET Doubler really work?

[daeg]

In the half dozen or so related threads, I didn't come across any positive build reports or sound clips of the JFET Doubler, Mu Doubler, MosDoubler.

So, are they:
a.) Functional (albiet finnicky)
b.) Unicorns (a few claim to have gotten them working, but success is not repeatable)
c.) More of an Conceptual / Experimental / R&D thing

[antonis]

Have you built any of them without acceptable result..?? 

[Marcos - Munky]

What I know on them is:

1) they require matched fets. So that's a very probable source of the "non funcional" builds. And that's one of the reasons I didn't tried them.
2) they're published by no other than RG, with an article saying they work. That's enough for me to trust them as working and functional.

[daeg]

Have you built any of them without acceptable result..?? 

No, I haven't tried yet.

RG's cancellation method of octaving has me very interested, but I'd like to set realistic expectations before going down another rabbit hole.

[Ben N]

I thought RG pulled those from his site because they were too finicky to get right?

[Marcos - Munky]

Two of them are still on RG's site:
http://geofex.com/Article_Folders/jfetdoub/jfetdoub.htm
http://geofex.com/Article_Folders/jfetdoub/mudoubler.htm
And both says matched fets should be used. That's why they're finicky. The MOS Doubler, on the other hand, was removed for a reason I'm not sure.

Maybe RG shows up to say a bit on these ones.

[antonis]

I'd like to set realistic expectations before going down another rabbit hole.

So, you're looking for rabbits already dug their own holes for yourself..

[Axldeziak]

The Mos Doubler is still linked on his site but according to the wayback machine has returned a 404 since 2002.
Here's a working 2001 link.
https://web.archive.org/web/20010816085417/http://geofex.com:80/FX_images/mosdoub.gif

[amz-fx]

Some discussion about the jfet doubler from 18+ years ago: 

https://www.diystompboxes.com/smfforum/index.php?topic=3879.msg22297

Enjoy!

regards, Jack

[daeg]

Is there a single sound clip of this thing out there?

[anotherjim]

I have to admit that this idea is one I just haven't managed to figure out.
It seems to take signal, split into 2 opposite phases, then cancel them out exactly -  leaving...?
I must be missing something!

[11-90-an]

I have to admit that this idea is one I just haven't managed to figure out.
It seems to take signal, split into 2 opposite phases, then cancel them out exactly -  leaving...?
I must be missing something!

from my limited understanding...
signal split to 2 opposite phases, 1 gets sized down a bit (lack of better words... think volume control) and the other is untouched. stuff gets cancelled, amplifying the harmonics and other stuff... gah I give up...
here's what RG says in here:

P1 is called an octave balance control. When it's full up, the mirror imaged signals from Q1's source and collector drive the gates of Q2 and Q3 out of phase, so the the signal that appears on the joined drains is cancelled. If Q2 and Q3 were distortionless, there would be NO signal at their joined drains. However, Mother Nature being the lady she is, there are no distortionless devices, and so Q2 and Q3 both distort in their own way. It's a peculiarity of distortion that even order distortion is single directional; this means that Q2 and Q3 generate second order distortion that zigs the SAME way instead of opposite, so the second order distortion products reinforce at the drains instead of cancelling.

The signal is rejected by the amount that the devices are matched, the distortion is reinforced and amplified, so what appears at the drains is largely second harmonic distortion and a smattering of other junk.

But that only happens for perfect balance on the signal we feed in. If we deliberately unbalance the signal in, the cancellation of the base signal is not perfect. That's what P1 does. It lowers the signal from the source of the phase inverter by as much as half - P1 is a 50K pot, R6 is a 51K resistor. This allows us to unbalance the cancellation and get mostly signal with a little second harmonic. Since the signal is about 20db (10 times) as large as the distortion to start with, even a little imbalance puts a lot of signal on the drains. That's why you need to match Q2 and Q3 - if they're too far off, you can't get a good, solid octave.

from what I get... basically harmonic noise is amplified... and now my crazy thinking...
what would happen if we were to feed this "harmonic noise" and the clean signal into another differential amplifier of sorts? would it cancel out all the "noise"?

I don't thing so... am I correct?

[antonis]

Can't get what you're missing, Jim..
(in the mean of canceling them out..)

Q1 is a traditional phase splitter where Q2/Q3 are an octave up configuration  (full wave rectifier)
For a signal possitive waveform, Q2 is ON and Q3 is OFF and vice versa..



Or do I miss something regarding your query..?? 

[11-90-an]

Q2/Q3 are an octave up configuration  (full wave rectifier)

Wait.. Full wave rectifier? 
I thought differential amplifier... or even long tailed pair...

looks like i've got more reading to do... 

[antonis]

looks like i've got more reading to do... 

Plz feel free to do that.. 

http://moosapotamus.net/files/stompboxology-going-discrete.pdf
pages 6 & 21..

[anotherjim]

I was trying to get my head around talk I saw of the pair (as in Q2, Q3 here) as the "canceller". I thought it was something especially clever for using FET for making more second harmonic, cancelling the clean, then amping up the remnant harmonics!

[duck_arse]

the two fets form a short-tail pair.

[11-90-an]

the two fets form a short-tail pair.

Is this part of a plot, Stephen...?  or should I say, duck?

[GFR]

It can be a FWR or two amplifiers with the outputs shorted (a "summing" amplifier of sorts), it depends on how its's biased (class B for FWR or class A for diff amp).

The Univox Superfuzz has a similar topology, wiyh BJTs. Also, the Mutantes "octave doubler" (CCBD) too.

I remember testing two versions of the MOS doubler, with a signal generator and scope (I ended up not testing with a guitar and amp for lack of time). The older one showed a clear octave on the scope, but I couldn't get the newer one to work.

Some old discussions from AMPAGE, maybe you find something useful.

https://archive.ampage.org/threads/0/fxdiy/002093/MOS_Doubler_Lives-1.html
https://archive.ampage.org/threads/0/fxdiy/004690/Re_MOS_Doubler_Problems_etc-1.html

[R.G.]

That's an oldie! Yes, it works. Yes, it's finicky. It requires matched devices in the diffamp and probably tuning on each unit. Even some matched monolithic pairs don't work well, largely because for this you want devices that are not very linear so there's more distortion to bring up to audible. And it requires good attention to low noise techniques because at best you're generating a distortion product that's 1-2% of the original signal and then amplifying that up to as big or bigger than the original, so you're looking at a 20db reduction in signal to noise just because of that. But in a world of "I need MORE gain!" fuzzes, that's not insurmountable.

The descriptions are pretty accurate in the page.It's not a full wave rectifier, in spite of the visual similarity to the rectifying section of something like a Superfuzz.

FETs of all kinds generate a preponderance of second harmonic distortion, plus other stuff if you overdrive them. Fourier's theorem says that any repeating waveform can be deconstructed into a series of sine waves. In the case of FETs, this is mostly fundamental and second harmonic. with other stuff generated the harder you drive it.

So to bring up the second harmonic/octave good stuff, you construct two equal signals from a FET and subtract them. To the extent that this "cancels" the fundamental, all you're left with is the harmonic distortion products. And in doing this, you are immediately faced with the problem of trying to isolate small differences between two nearly equal things.

This was considered for a commercial product by at least one company, but the tuning on each unit made it impractical for mass production. Too much skilled labor required.

To set expectations, it will require a lot of tinkering and tuning, and selecting parts.