Varistor Characteristics for Magnatone Vibrato Circuits

[trobbins]

Just keep in mind that the maggie varistor is SiC and nothing like that modern MOV device, except for the name.

[trobbins]

Did a quick comparison of varistors in a test amp. Attached are comparison plots and test notes.

I was varying speed and depth between tests, and appear to have made the plots with different speed settings. That said, there are differences between the SiC and the two zener clone variants, but I suggest they're pretty minimal and it would need a well constructed comparison listening setup to try and detect a difference.

There is definitely no 'clicking', which I can only suggest may have been in the test amp and perhaps associated with the LFO. Well worth a recheck.

There does seem to be some noise difference, although it only shows a bit with the 33/56/82V clone. But that may be my test setup. I may set up my EMU-0404 and check, but I think the hassle is the closeness of LFO and preamp circuitry to OT primary. The test amp has been a somewhat messy experimental test bench.

The maggie circuit contributes both phase modulation (shown as frequency spread in the spectrum plot), and some amplitude modulation (as shown in the scope plot).

[martin manning]

Interesting! Looks like maybe the steps in the "humps" are the Zeners switching on? Two guys here did tests in an amp, and found noticeable clicking with the Zener-resistor substitute. Perhaps with an additional zener voltage it becomes less apparent.

I rebuilt my sim from scratch and the new version does not show reverse polarity. The circuits appear to be identical, but I'm beginning to think there was something wrong with the earlier one.

[trobbins]

If there was a significant discontinuity in the varistor V-I characteristic then it would be quite plausible that as the 12AU7 PI cycles the voltage across the varistor clones, the resulting distortion component when the 56V zener stopped conducting would be an added spray of LFO harmonic frequencies, as there is feedback from anode to grid. A maggie circuit low pass filters the LFO signal prior to the vibrato modulator, so any PI stage harmonics could well couple straight in to the audio path and would likely be quite a high level given the LFO signal level across the varistors. That distortion would I reckon sound like a click.

I have an oilcan that showed such a LFO 'click', apparently from some rubber wiper debris that had stuck to the platter whilst the pedal lay dormant for decades. At the time of restoration, I was able to attenuate it somewhat with little noticeable high frequency loss. I now know that the disk can be flipped, so will try that when I have time. The oilcan and the maggie are both great 'early adoptors' of a vibrato effect, although the oilcan never had the benefit of a 'speed' control, and the maggie used the tremolo by-product as a signature 'advantage'.

[johnnyreece]

I have no input. I'm just really excited to see this come together. I've got two varistors, but would like to build a bigger version someday.

[trobbins]

By bigger, do you mean another amp, or a two sequential stage maggie such as 260 series, 450, 280A, M10-13-15, and 480 models - for that extra level of spacey vibrato?

[trobbins]

Did some more measurements to get a better indication of differences between clones and original SiC varistor.

I cleaned up the grounding a bit on the test amp. Got REW going with my $2 usb soundcard, which allowed a bit more control over the spectrum analyser setup.

The results show differences, but whether the actual sound of the 33-56-82V clone is noticeably different will require others to do some live guitar testing.

The spectrum results identify that the differences appear to be the introduction of additional LFO harmonics by the clones, which then couple to the signal frequencies and hum signals. It's easy to see the extra LFO harmonics in the 33-56-82V clone results - and this is the reason for the higher 'noise floor' level in the earlier measurements (which used a different analyser and setup).

[martin manning]

That's a real mess.

Here's another idea: Use the same basic arrangement of resistors but substitute MOV's for the Zeners. Littlefuse makes compact MOV's with approximately the same "on" voltages (I = 1mA @ 35, 61, and 90V), which might soften the transitions.

Would you mind sharing the schematic for your 5E3 with Maggie vibrato?

[johnnyreece]

By bigger, do you mean another amp, or a two sequential stage maggie such as 260 series, 450, 280A, M10-13-15, and 480 models - for that extra level of spacey vibrato?

The second one the two varistor models are cool, but I'd like to try the other. For purely academic reasons, of course.

[trobbins]

Martin, for some, a clone using MOVs may be easier to construct using the same voltage part (eg. V27ZA05P, with 1x, 2x and 3x in series).

I have some V100ZA4 in my parts bin and will try and compare their turn-on characteristic with a 90V zener to see if it is less abrupt.

That said, I'd be thinking the 'mess' may not be so audible - the LFO harmonics are more than 20dB down, and being harmonics they are going to be less/obtrusive/noticeable. The sound character may even be 'nicer' - its strange how the mind works, and there was a very interesting blind study by a uni group that played guitar music through an amp with mains ripple at 'typical' levels for such an amp, and at significantly attenuated levels, and everyone rated the sound 'better' when mains ripple was modulating the audio signal.

My test amp schematic is attached. There may be some easier means to modify the generation of LFO harmonics by the PI stage, especially related to the 2n2 anode-grid feedback cap.

Ciao, Tim

[martin manning]

Thanks for the schematic, Tim. That version of the vibrato appears to be a bit different from the one in the M213. It looks more like one unit from the M280. There are some other versions as well, such as the one in the M410. I wonder if that is the difference in the clicking being audible?

I was interested in keeping the parts count down by using varistors with different breakdown voltages. In any case this MOV substitute will be an interesting experiment. I don't have a Magnatone amp or clone to try it out, but maybe some others here can give it a try.

[Reeltarded]

As long as there is no whump on the turn I bet you are right about the added lowend junk and it being nicer.

I have never driven a Maggie with the vib engaged. That'd be awful. At the edge is good though. I want to hear this thing.

[trobbins]

Thanks for the schematic, Tim. That version of the vibrato appears to be a bit different from the one in the M213.

??

I appreciate the LED bias on the LFO is different, and no LFO footpedal switch, and the signal 12AU7 is not direct couple fed from previous stage, and the output of the signal modulator is not straight to amp PI, but I think its pretty close. The 280A certainly has differences around the varistor modulator, and the LFO PI is quite different, given its double stage modulator with common LFO driver aim.

I'll see how Stephen is progressing with his setup.

[martin manning]

I did not look too close, and you are right they are more similar than they first appear due to the schematic layout. The grid leak/bias reference and LFO signal coupling to the vibrato inverter (specifically the two 2M2 resistors) is like the M260 on your scheme vs. the M213. Did you happen to try it both ways?

[trobbins]

On the schematics I have, the LFO PI stage has the same grid leak and grid stopper values for the 213 and the 260A, although the cathode bias is different (3k9 versus 2k7).

My clone has a 2n2 coupling cap from varistor midpoint to next stage, versus 470pF in 213 model. I can't recall why I did that - maybe a thought at the time due to the clone having different loading, and I wanted the option to send out the signal.