Author Topic: Fender brownface vibrato in a stomp...does it exist? What about with JFETS?  (Read 39910 times)

Steben



Well, it looks tubey enough. Bringing "Ugs" (gate bias) down gives smaller amplitude. In a Fetzer II, the curve is more linear, which is in fact more triode-like, but reduces the amplitude change, which means you have to bring it down even more.
Rules apply only for those who are not allowed to break them

R.G.

I did a quickie markup of an old Phase 90 schematic to show one way to do a solid state version. It's up at GEO now.

I think this will get the right feel, if not be a perfect repro. The JFETs will probably introduce some soft distortion. That's NOT going to be the same as tube distortion, but it should be nice.
« Last Edit: March 09, 2007, 03:27:39 PM by R.G. »
R.G.

Quick IQ Test: If anyone in a governmental position suspected that YOU had top-secret information on YOUR computer, how many minutes would you remain outside a jail cell?

Sir H C

Having a brown amp, the trem is amazing.  That is why those amps are so cool. 

Now they are too expensive for me to ever think about buying again...

Eb7+9

discussed here ...

http://www.diystompboxes.com/smfforum/index.php?topic=49413.0

class-A tremulant circuits won't sound the same exported to op-amp "loss-less" technology because of the low-end cancellation that takes place in the class-A case ...
DISCOVERY happens to prepared minds

R.G.

JC, did you even look at the schematic?

The circuit drives exactly the same network as in the Fender amp, then the result is buffered by the high impedance input of an opamp and the output of that is modulated by open-loop JFETs into a mixer. It's not a lossless opamp technology.

Next time, understand the circuit first, OK?

And that wasn't a discussion, it was a discharge into the ether. Discussion implies much more two-sided give and take than advice to look at patents for "electronic tremulant" and some amp schematics. I didn't see any discussion at all.
« Last Edit: March 09, 2007, 07:50:42 PM by R.G. »
R.G.

Quick IQ Test: If anyone in a governmental position suspected that YOU had top-secret information on YOUR computer, how many minutes would you remain outside a jail cell?

Dragonfly

I did a quickie markup of an old Phase 90 schematic to show one way to do a solid state version. It's up at GEO now.

I think this will get the right feel, if not be a perfect repro. The JFETs will probably introduce some soft distortion. That's NOT going to be the same as tube distortion, but it should be nice.


Hi RG...
 Interesting...i might "give this a go" on Sunday. I'm pretty curious as to the results, and it looks like an easy enough build. If I get a chance to build it, I'll be sure to report back with the results.
  Thanks,
    AC

R.G.

Cool.

The other thing I thought about was using a variant of the EA tremolo to do the amplitude variation before the mixing, but I didn't see a simple way to do the inverted LFO. At least not as simple as the design based on the Phase 90.
R.G.

Quick IQ Test: If anyone in a governmental position suspected that YOU had top-secret information on YOUR computer, how many minutes would you remain outside a jail cell?

Eb7+9

JC, did you even look at the schematic?

The circuit drives exactly the same network as in the Fender amp, then the result is buffered by the high impedance input of an opamp ...

right ... to be fair, at first I didn't look at the schematic because I kinda know by now how you "stand" so to speak on basic loading/drive issues ... :icon_wink:

so, I had a look and I was right, it's a lossless mixer ...

since you're a big fan of understanding, take a look at the redraw of the 6G7 Vibrato circuit

http://www.lynx.net/~jc/6G7vibrato.gif

at first glance it just seems like there's a split in the signal path, a low-bandpass leg and a high bandpass leg, the LFO causes an asymmetric change in gain to the diff pair these filtered legs are fed to and then into a resistive mixer to "add" the difference ...

at second glance there's something more going on ...

the oscillator puts out a sine wave that's around 6 v pk-pk ... by the time that voltage cross es the voltage divider and hits the grid there's about 3 to 4 volts pk-pk hitting the grid - that's with Intensity full up ... since the LFO has an inverter on one leg leading to the other side of the differential circuit that's a total of 6 to 8 volts of input voltage across the differential input circuit  ...

that means the diff pair is swinging almost (at least somewhat near) across it's fulldynamic range ... at the top end of these ranges the grid leak current rises to a considerable level relative to idling levels ... this translates in a (severely) reduced input resistance, well below 100k at the top input voltage extreme - vive-versa they go to higher than nominal resistance during the low part of the cycle ...

since this input resistance plays a dominant part in setting the high-low band-pass freq's then what happens is the bi-amp frequency moves up and down with the cycling ...  it's very much a part of the effect

in your circuit that frequency be static ...

so, what we have is varying bandwidth's being mixed at varying levels and drive ratios ... does this remind you of any other circuit ??

right ... the Univibe

btw, the very same one-stage variable-bandwidth variable-mixing idea is available as output in the the Jax Vibra-Chorus and Shin-Ei Resly Tone - but not the Univibe  ... though, again, the varying bandwidth component takes places through another mechanism in those circuits ... really interesting how this variable-bandwidth variable-mixing stuff pops up in different ways ...

then, there's a second amplitude thing going on at the plates of the differential triode circuit since the driving point impedances also varies considerably around it's nominal values as the bias on the tubes goes up and down ... this is another component that will sets the character of the signal envelope - the partial Trem component ...

the main thing is the variable frequencies occuring "before" the mixing - what I was referring to in  the post mentioned above ...  it's a product of wide-range loading variations // the opposite of losslessness ...

now, I'm not saying your circuit won't work or won't be appealing ... it just won't have that varying bandwidth component in it that's all - what I was saying ...

just to add to this, if we were to actually do a jFET emulation of the class-A mixer (with really well-matched FET devices) as it appears in the Fender schem then this component would be missing as well since the input resistance on a jFET hardly changes with bias ... combine this with the two-way device matching problems and this would explain why you don't see that version being explored much ... yeah, you could always rig up photocells and make the whole thing look even more ellegant ... then you've got that lower shut-off limit that tends to hit harder that the more gradual cutoff of tubes ...

very specific non-linear manifestos are built into this Fender circuit and it is what it is for those reasons ... otherwise, there's not much else going on ... it's one of those tube circuits tranny or op-amp emulators probably can't nail ...

let's see what the clips sound like ...
DISCOVERY happens to prepared minds

George Giblet

  • Guest
JC, thanks for the write up.  When you start looking at the finer point there's a lot going on.  It's the kind of circuit that easy to fob-off with a first pass glance only to realize it requires more detailed study.

> this translates in a (severely) reduced input resistance,

I'm not so sure that will happen to a large extent in the steady-state.  What I expect will happen is the coupling caps, 5n to the depth pot and 50n to the X-point will act as peak detectors with the grid diodes and eventually (after power up) a new set of DC conditions will be established such that the low resistance region is minimal.  I haven't actually analysed the circuit but the couple-cap grid diode DC shift is a common thing.

The other thing is the there's a lot of phase shifts due the low-pass filtering on the inverted LFO path - that's go to do something.

It's too late for me to think about this anymore today - I'm pooped.

Thanks for peaking some interesting into this ckt.


d95err

Quote
RG, I looked in more detail at the 6G5, 6G14, 6G15 shcems available at the FIS, and don't see any phase inversion between the lowpass and highpass filters.  Apparently the extra inverter you mention wouldn't be needed, or if installed would produce a different effect from what we have.
That's correct, there isn't any signal inverter in the stock circuit. I put that in there because I believe that it would produce a more intense effect. In fact, a switch to feed the filters from either just the buffer or buffer plus inverse would be a good idea.

The tremolo used in the old Vox AC30 and AC15 uses a similar design which, if I remember correctly, also inverts the phase somewhere in there. I think it's actually even more complex...

R.G.

Quote
right ... to be fair, at first I didn't look at the schematic because I kinda know by now how you "stand" so to speak on basic loading/drive issues ...
OK. Your reply said as much.
Quote
so, I had a look and I was right, it's a lossless mixer ...
But neither your first reply nor the "discussed here" thread said anything about a lossless mixer. I guess if one looks hard enough they can find something to grouse about.

I don't suppose you read it where I said "I suspect that this gets 90% of the sound of the original, maybe even close enough not to need the other 10%." or "I think this will get the right feel, if not be a perfect repro. The JFETs will probably introduce some soft distortion. That's NOT going to be the same as tube distortion, but it should be nice."

But having not said that it would be perfect and then being told it's not perfect for thus-and-sundry reasons, lets look at it a bit, since I really am a big fan of understanding.
Quote from:  JC
at second glance there's something more going on ...

the oscillator puts out a sine wave that's around 6 v pk-pk ... by the time that voltage cross es the voltage divider and hits the grid there's about 3 to 4 volts pk-pk hitting the grid - that's with Intensity full up ... since the LFO has an inverter on one leg leading to the other side of the differential circuit that's a total of 6 to 8 volts of input voltage across the differential input circuit  ...
That's a great theory. But it's not what happens. I didn't happen to have a circuit board set up to try this, nor a 6G5 to measure, so I put the circuit into a simulator.

Before you read me the manifesto on simulators not being perfect and not trusting them, I'm well aware of the shortcomings of simulators as I used them for decades professionally. The simulator I'm using has certain issues, but biasing points with the tube models is not one of them. It's been very accurate, in that its results are inside the variations I see in real tubes when I breadboard the same circuit. So in lieu of a real amp to measure, I simm'ed the whole LFO/modulator circuit.

It turns out that the simulator seems to think that the voltage excursions are not that big. It shows a peak to peak on each of the modulator tube grids of about 1.5V. And the nonlinearities on the plates were at cutoff, not saturation. So forward grid conduction wasn't happening. Allowing for normal tube variation, I'm sure that some tubes would go into grid conduction somewhere, somehow in this circuit, but it's not the central effect. The plates hit cutoff only with intensity over 80%. Under that, they didn't even cut off.

So the theoretical basis for variable loading with LFO signal doesn't seem to apply. The LFO signals at the grids isn't that big, it doesn't seem to push them into positive grid bias range.

Yes, it is possible that my simulator is wrong. But it's been right many, many times about things like this. I trust it more than just eyeballing a schematic or my sometimes-faulty calculations.
Quote
this translates in a (severely) reduced input resistance, well below 100k at the top input voltage extreme - vive-versa they go to higher than nominal resistance during the low part of the cycle ...
There is an abrupt change from near-infinite impedance to low impedance at a grid when the input voltage gets just a hair over the source voltage. And I'm using the word correctly - abrupt. It's not a lazy, slew-the-response kind of thing. Tubes will give razor-sharp distortion when driven into positive grid voltages from a high impedance, as the circuit here is doing. One can soften that transition with a low-impedance drive to a grid, but that's not what's happening here. The grid goes from repelling electrons to sucking them in.  The forward biased grid resistance for a 12AX7 tube is something between 5K and 10K depending on the degree of forward bias, by the way.

The grid impedance may go from near-infinity to nearer infinity as you make it more negative, but given the rest of the circuit component values, this produces no change. Normal tube grid impedance is negligible loading on a 1M grid resistor. Making it many times negligible, even if that were to happen, has no effect.

Since I wanted to be sure of my facts, I dug out the Radiotron Designer's Handbook, fourth edition. They say:
Quote
When a valve is used at audio frequencies, it is sometimes assumed that the grid input impedance is infinite. In most cases, however, this assumption leads to serious error, and careful attention is desirable to both its static and dynamic impedances.
EEK! I could be wrong. But then, I always consider that to be a possibility. Only the naive believe they're always right.

There follows a long string of equations describing how to calculate the input admittance (inverse of impedance). The meat of the discussion boils down to the equation showing that for frequencies up to 10MHz, the input admittance is
G = 1/rg +jw(Cgk +Cgp)

There follows the two comments:
- "If the grid is negatively biased to prevent the flow of positive grid current, the grid resistance becomes very high and 1/rg may be negligible in the expression for G. [G being the name for the overall grid admittance.]
- "the short circuit input admittance is affected by the transit time of electrons, the inductances of the electrodes (particularly the cathode) and the capacitance between grid and cathode. The capacitance term is practically constant." (my italics)

So out of this - there are a bunch of special cases to consider in the general case for critical applications. But the special cases don't seem to matter to this discussion. The 1/rg term is neglegible under these conditions and the capacitive term is constant. No varying bandwidth based on grid bias.

The rest of the discussion digs into the effects of interelectrode capacitance, gas leakage, grid emission of electrons, and transit time of electrons across the tube. All reasonably esoteric, but something the professional tube designer needs to take into account for industrial designs.

Quote from: JC
the main thing is the variable frequencies occuring "before" the mixing - what I was referring to in  the post mentioned above ...  it's a product of wide-range loading variations // the opposite of losslessness ...
But I thought the only losslessness you found was in the mixer. That's what you just said. Or wait - what was it you said? Can you explain how the losslessness was the mixer, but then wasn't, it was before the mixer?

And as we've seen, the variable loading doesn't happen, so the losslessness doesn't matter, eh?

Quote from: JC
really interesting how this variable-bandwidth variable-mixing stuff pops up in different ways ...
Sure is. Especially if you look really, really hard for it.

Quote from: JC
now, I'm not saying your circuit won't work or won't be appealing ... it just won't have that varying bandwidth component in it that's all - what I was saying ...
Where I said "I suspect that this gets 90% of the sound of the original, maybe even close enough not to need the other 10%." or "I think this will get the right feel, if not be a perfect repro. The JFETs will probably introduce some soft distortion. That's NOT going to be the same as tube distortion, but it should be nice." it wasn't enough?

By the way, Merriam-Webster says that "manifesto" means:
Quote
  a written statement declaring publicly the intentions, motives, or views of its issuer
I didn't know that circuits had manifestos. I didn't know they could write. But is is a fancy sounding word, all right.
R.G.

Quick IQ Test: If anyone in a governmental position suspected that YOU had top-secret information on YOUR computer, how many minutes would you remain outside a jail cell?

Ghandi

anybody tried it jet?

Ben N

  • Awesome!
  • ****
  • Posts: 4326
  • Total likes: 448
  • Ben N. - really! Now Israel, once Baltimore
Easy, now fellas!

(Ducks, runs...)

Processaurus

RG, I was looking over the schematic for your fender style tremolo, and was wondering if it was an error that the two 1M resistors from the complementary hipass and lowpass filter's opamp's non inverting inputs meet, and then go through another 1M to Vref, rather than straight to Vref as a typical opamp design would?  As is the channels might bleed a bit into each other.

Very cool circuit, by the way, I'm thinking of stealing the filters to add onto the end of my stereo EA based trem mutant, since the panning is happening already.  I think I can do it in stereo too (one channel getting darker as the other gets brighter), with a six pack of opamps, 4 for the filters and 2 for mixing/recovering gain with each channel's respective filters.  I'm excited to hear how it will sound combined with the eccentricities of the phase shift LFO.

nordine

forgot to say it, but i breadboarded that emu-circuit RG made

have to say its a very beautiful sound.... really difficult to explain.. its bouncy, but solid through all the swirl

one thing that i had while testing it, is that the oscillator worked only at fast speed, and did cease to function if i went past certain (lower) speed.. had me wondering what could it be

R.G.

Quote
the oscillator worked only at fast speed, and did cease to function if i went past certain (lower) speed.. had me wondering what could it be
Thanks for the build report.

If the 10uf cap is leaky it might cause that. Can you try a different 10uF cap in the LFO?
R.G.

Quick IQ Test: If anyone in a governmental position suspected that YOU had top-secret information on YOUR computer, how many minutes would you remain outside a jail cell?

rockgardenlove

Did you by any chance make a PCB?



nordine

Quote
the oscillator worked only at fast speed, and did cease to function if i went past certain (lower) speed.. had me wondering what could it be
Thanks for the build report.

If the 10uf cap is leaky it might cause that. Can you try a different 10uF cap in the LFO?


i already took the circuit apart, but will try if it was the cap, then i'll report

ps: no man, no pcb (i've never etched a circuit), sorry

QSQCaito

  • Awesome!
  • ****
  • Posts: 862
  • Total likes: 1
  • Diego Andrés. Buenos Aires, Argentina.
Wow.. I was thinking of making a tremmolo,  for what said, this would be a good solution unique sounding.

Is there anything that can be done by me to help?

A PCB design I could make if the circuit is correct, I'd have no problem.

Bye bye

DAC
D.A.C

joelap

I'd also be interested in keeping up to date on this circuit.  Actually, I've had the watchful eyes (aka my friends) keeping a lookout for me for any '62 or '63 blonde bandmasters that might pop up, which also has this same type of oddball vibrato/tremolo circuit.  If I can snag one, I'd gladly do a side by side comparison if a PCB layout is made of this circuit.  That's a big if, since they are more difficult to find, but should I get a hold of one, I'd gladly record clips of the real deal versus R.G.'s work.
- witty sig -