Discrete TubeLess...

[gez]

The standing current might be quite high because theres 4.5v bias on the gate. A lower supply voltage or bigger source resistors would reduce it.

Good point!  Removing the source resistors and running things at 5V might improve the current consumption (yet retain the softness resulting from the exclusion of these resistors).

I wonder how a power stage would fair?  I tried this last year, but (assuming that the power inverter wouldn't bias at half supply) I incorporated a servo circuit to do the job.  End result was oscillation that I just couldn't get rid of.  Had I known they balance up nicely, I wouldn't have bothered with the servo! 

[puretube]

only done this with those smallsignal babies - dunno how the powerdevices would behave ()



So as not to show off the usual schoolbook curves/waves of the common smooth clipping,

instead I got a nice experiment with 2 of them discrete inverters + 1 cap & 1 resistor:

creeping frequency "dubbler"
(looks like someone writes the initials "m m" onto the screen... 

blending frequency "multi plyer"

(both envelope-dependant: no knobs moved during recording).

The clips don`t sound as seary as it comes out of the speakers, due to the microphone pointing towards the damped ceiling...

[Elektrojänis]

This morning I did some random mixing of several BS170 (from 3 different manufacturers (dunno, which)),
with several BS250 (from one batch):
they all stayed within 150mV of Ub/2!

Wow, that's really interesting.  I wonder why (seems counter-intuitive)?

Hmmm... I don't really know anything about mosfets, but could the nice bias be related to the same effects that cause the soft clipping?

This is what got me thinking about it (quote from the other thread):

Transconductance of each device drops off when pushed into saturation, hence the compression. 

So if the transconductance is a smooth curve where transconductance goes down when SD-voltage goes down, this kind of symmetrical circuit probably will seek an operating point whre they balance or something...

I'm not too sure I understood this stuff at all... Just quessing.

[The Tone God]

I like!

Andrew

[puretube]

Petri: I had posted some nice basic FET links in an old thread,
(see also reply#24, there)
which I had deleted a while later after having been annoyed/molested by the usual gang, for whom I thought
it wasn`t worth pointing to at that time...

I dug them out again for explanations in some other recent threads.

they`re interesting, but don`t reveal all questions:
FET principles & circuits pt.1,
pt.2,
pt.3,
pt.4, by the great Ray Marston.

[puretube]

I like!

Andrew

I wonder, whether the MO or the PM-Club is the better place to disclose the schematic for the Waveply,
or if I just ought to wait to submit it for the monthly contest (being a neg-FB-circuit)...

[Elektrojänis]

they`re interesting, but don`t reveal all questions:

Cool stuff! Thank you! I did not read it all yet but sooner or later I will. Even if they don't answer everything, they seem to be just the kind of basic info that is relatively easy to use when experimenting.

Got to try this mosfet circuit too when I get some other things out of the way... Your sounclips sounded very cool.

[The Tone God]

I wonder, whether the MO or the PM-Club is the better place to disclose the schematic for the Waveply,
or if I just ought to wait to submit it for the monthly contest (being a neg-FB-circuit)...

Its your design so the choice is yours.

Andrew

[gez]

The standing current might be quite high because theres 4.5v bias on the gate.

I'm beginning to think that this is the reason why everything biases at half supply.  As there is so much voltage on the gate, both channels are conducting nicely.  If manufacturers produce devices that have consistent Rds when in such a state, it would make sense that a discrete inverter would bias around half supply.

Edit: bias is one thing, but how a pair behave as an amplifier is another matter: I should think that 'unmatched' devices would produce a little asymmetry.  Ton, out with the scope again! 

[puretube]

no need to... 

the different sounding ("different"/"harsher") fets of various manufacturers that I mentioned in reply#16,
were indeed accompanied with deviations from an otherwise ("cool"=warm) symmetrical waveform.

The "good" ones:,

which stayed closer @ Ub/2 than the mentioned ~150mV deviating ones
happened to be from the same distributor, but obviously not from the same manufacturer...


concerning: a day out with the scope, again 
I tried to capture the frequency tripling  on the "WAVEPLY" vid,
but the cam wouldn`t wanna focus sharply... (but with a little imagination you can see 2cuts descending from the top of the wave,
cutting it into 3).

another setting gives a larger multiple of peaks (3-7 or8),
again not to be seen very well,
but maybe better heard here,
where for the first 10 seconds the guitar volume was rolled down, then turned on to full volume...


Petri: more food... from R.G., this time.

[puretube]

Without the (admittedly) dirty "add 1 resistor + 1 capacitor"-trick,
that made a "WAVEPLY"
out of a dual (cascaded) "SANSVALVE"

the following pic shows, that a "cleaner" ordinary frequency-tripleing
(as is well known from other circuits...)
can also being achieved, by using a less mean trick,
to bring forth the: "TRIFOLD":



(now the sun doesn`t shine onto the tube anymore, the image became clearer...).

It`s obvious to be seen, how during (the imposed) one cycle of the input sine
which shows one "peak" up and  one "peak" down,
the derived wave shows 3 peaks up and 3 peaks down.

[gaussmarkov]

referring back to the sansvalve, you can get some asymmetry by putting a resistor "inside the inverter".  for example, between the drain of the PMOS and the "out" junction.  sort of a joe davisson kind of idea.

[oops.  editted "resistor" back into the post.]

[gez]

referring back to the sansvalve, you can get some asymmetry by putting a resistor "inside the inverter".  for example, between the drain of the PMOS and the "out" junction.  sort of a joe davisson kind of idea.

[oops.  editted "resistor" back into the post.]

I posted a circuit that has that very implementation a few years ago.  It's hosted by Marcos.  I think I called it the Nut Cruncher.

[The Tone God]

the following pic shows, that a "cleaner" ordinary frequency-tripleing
(as is well known from other circuits...)

Hey, keep out of my waveforms!

Andrew

[gaussmarkov]

referring back to the sansvalve, you can get some asymmetry by putting a resistor "inside the inverter".  for example, between the drain of the PMOS and the "out" junction.  sort of a joe davisson kind of idea.

[oops.  editted "resistor" back into the post.]

I posted a circuit that has that very implementation a few years ago.  It's hosted by Marcos.  I think I called it the Nut Cruncher.

  and exactly (not "sort of") a gez kind of idea!  thanks gez.

for those who are search-challenged:  gez's nut cruncher at Munky DIY

[puretube]

the following pic shows, that a "cleaner" ordinary frequency-tripleing
(as is well known from other circuits...)

Hey, keep out of my waveforms!

Andrew

I`m doing an initials-collection:
in reply#21 it was the "MM"
a little while ago I had a very nice "Z"-sign 
and recently, a quite realistic looking (fastood-)"M" logo... 

[gaussmarkov]

does the sansvalve need matched MOSFETs?  i see that gez used a CD4007 for his nut cruncher and that gives nicely matched pairs.  but given that asymmetry introduced by a resistor is interesting, it seems possible that "odd couples" might be a good thing too. 

ton, you just grabbed a pair out of the pile for your experiments, right?

[puretube]

The standing current might be quite high because theres 4.5v bias on the gate. A lower supply voltage or bigger source resistors would reduce it.

Good point!  Removing the source resistors and running things at 5V might improve the current consumption (yet retain the softness resulting from the exclusion of these resistors).

I wonder how a power stage would fair?  I tried this last year, but (assuming that the power inverter wouldn't bias at half supply) I incorporated a servo circuit to do the job.  End result was oscillation that I just couldn't get rid of.  Had I known they balance up nicely, I wouldn't have bothered with the servo! 

Wait!:
4.5V bias?
isn`t the bias: V<sub>GS</sub>?
the "R<sub>S</sub>" & "R<sub>D</sub>" inserted between the sources and the rails reduce the voltage between the rails
and the sources (in the sansvalve case by ~2 - 2.5V, with the 1k "inline"-resistors),
so there`s only 2 - 2.5V left for "bias"...

[puretube]

does the sansvalve need matched MOSFETs?  i see that gez used a CD4007 for his nut cruncher and that gives nicely matched pairs.  but given that asymmetry introduced by a resistor is interesting, it seems possible that "odd couples" might be a good thing too. 

ton, you just grabbed a pair out of the pile for your experiments, right?

yep, the "Nutcruncher" doesn`t count here, because it`s not discrete  ( );

upon jaytee`s request in reply#10, I tried a lot of combinations of all (~30) mosfets of the named types (170/250) I had at hand,
as noted in reply #16.
The (small) variations in the established bias was audible and visible,
but close enough to be acceptable for the usual circuits, and especially for a "trimmerless-field-effect-stage"
(that very first breadboard test was a completely "blind" test, but perfect).

Gotta keep in mind, though, that it`ll always be a "classA/B" configuration, with a potential for crossoverdistortion ( ).

Back to your question:
it obviously doesn`t need matched components (for the types used), and no adjustments.
(which is the fact I loved about the hex-inverters, and always thought is immanent only to them...).

It seems that the manufacturers are able to produce tight V_t tolerances in these devices nowadays;
and the use of those "inline"-resistors, that increase ("linearize") the resistance of the voltage-divider (Rds_P/Rds_N)
which in fact establishes the bias fromout its center-node, seems to add to the symmetry (me thinks)...

[gez]

Wait!:
4.5V bias?
isn`t the bias: V<sub>GS</sub>?
the "R<sub>S</sub>" & "R<sub>D</sub>" inserted between the sources and the rails reduce the voltage between the rails
and the sources (in the sansvalve case by ~2 - 2.5V, with the 1k "inline"-resistors),
so there`s only 2 - 2.5V left for "bias"...

Yeah, I had accounted for that.  Even with the reduced voltage (can't remember off-hand, but isn't typical threshold of BS170 about 1.8V??) both channels should be conducting nicely (though not fully on).