Varistor Characteristics for Magnatone Vibrato Circuits

[Smokebreak]

Oh that's right, my bad, you're doing the 210. I'm looking forward to seeing how it goes!

[andyfromdenver]

Oh that's right, my bad, you're doing the 210. I'm looking forward to seeing how it goes!

Thanks Smokebreak!
I appreciate your kindness
I know now I'm like, duh, why didn't I photo the amp guts while it was at my friends shop.

[trobbins]

I updated my cloned varistor and measurements a few months ago. The DC measured results using 36x 3V6 zeners with some matching resistors was excellent. I noticed that my original measurements had a subtle inaccuracy, which is shown in the new results in the linked file.

http://dalmura.com.au/projects/Varistor ... 0zener.pdf

I got side-tracked before I had time to do some 'in amp' signal measurements, and it may take a while to do that.

Martin, your 56V/82V zener clone measured curve doesn't have a logarithmic scale for current - were you able to get a close comparison of clone V-I when current is less than 0.1mA? If so then I'll cross-check the measurements using my test jig - I've got some 56V zeners, and can make up some 82V types.

Ciao, Tim

[martin manning]

Hi Tim,
I'm afraid the substitute using two Zeners and resistors to approximate a varistor I-V curve is a dead end due to the unacceptable level of switching noise. This was shown in Jeremy's test using his amp, and I don't see any way to get around it.

[trobbins]

Martin, you seem to have designed it for bipolar function - was that just to make the clone 'fool proof'. The clone I prepared is polarity dependant - but I didn't think that was a major problem if it emulated a maggie varistor correctly

[martin manning]

Current flows both ways through the varistors since the audio signal has positive and negative swings.

[trobbins]

Yes AC current will flow, but the net current is of one polarity due to the split load PI configuration.

The PI forces the voltage to cycle from idle condition of about 55-60V across each varistor, to the signal pk-pk limits of about 20-30V to about 90-100V (ie. up to a 70-80V pp swing). The varistor current never flows through zero, but sweeps between near 0.01mA to near 1mA.

[trobbins]

Martin,

I measured your clone circuit - file 1 result below. A bit wonky below 0.1mA.

I removed the back-to-back zeners to make the clone easier to make, and added a 33V zener to iron-out the lower voltage response. A few resistor value tweaks and voila, the revised clone is pretty accurate (file 1a), and pretty simple to make up. The zener polarities have to be observed when fitted to a maggie circuit.

I'd be thinking that two of these clones would have a much better chance of being matched, compared to the original varistors, so thumping should be almost non-existent.

Many thanks to Dinko, who most kindly posted me the two original varistors gratis that are plotted for reference - much appreciated.

Ciao, Tim

[martin manning]

Tim, I need some time to look into this. I built a sim of the Mag vibrato, and as I recall the varistor (faux varistor, really) current did go through zero. I'll get that out and look again.

I don't know what that s-curve is in your I-V plot. I traced the original assembled clone parts using a curve tracer, and the result looked like the theoretical straight-line approximation of the curve.

The real issue is the Zeners are very noisy. It worked in a real circuit, but the clicking noise is unacceptable.

[trobbins]

I don't recall any noticeable additional noise when testing the clones previously, and certainly no clicking. I'll try and get my maggie 213 clone amp out over the weekend and do some measurements, as I've got my soundcard cro working again.

My reverse engineering notes for the nominal maggie vibrato circuit are as follows:

The 12AU7 LFO PI for the varistors is mid-biased at about 1.9mA and 7.5V Vg, with about 0.1mA bypassing the 12AU7 via the varistors. Full swing level across the series varistors is up to ~90Vpk with an 8Vpk input. Voltage swing across each varistor therefore can span from about 20V to about 105V. At 105V across a varistor, the current requirement is about 1mA, and the effective resistance is about 100k each. This non-linear parallel loading across the 12AU7 is equivalent to progressively changing the slope of the loadline at higher grid voltage levels, which progressively lowers the gm and introduces even order harmonics into the LFO signal variation of varistor R. Maximum design level heater-cathode voltage is 200V, so this stage is within limits, and cathode peaks to about 135V. The 12AU7 grid is ac coupled from the LFO stage, and the grid bias is set by the 3k9Ω.

At min varistor resistance of about 100kΩ each, the RC time constant of 50kΩ and 820pF is about 41ΩuF, which achieves about 150deg output phase lag at 1kHz from the two arms. At max varistor resistance of about 2.5MΩ each (assumed), the RC time constant of 1.2MΩ and 820pF is about 1000ΩuF, which achieves ~10deg output phase lag at 1kHz from the two arms. [Reference to Moses, 1953]

[martin manning]

Is the Moses reference available on-line somewhere, or can you send me a copy?

I dug up my sim and investigated the varistor current. With enough drive (a not unreasonable 2V peak grid signal applied to triode driving the split-load inverter) the voltage across the varistors will reverse. But, as long as the reverse voltage is not larger than the lowest Zener voltage, the behavior of the Zener Varistor (Zenistor?) will approximate the Varistor characteristic as well as ever.

Adding another Zener voltage and resistor is not too painful if only one of each value Zener is needed; the total is the same seven parts per unit as the bipolar version. To better-define the very low current part of the curve I would suggest values as follows (Vz, R):

None, 1M8 to 3M3
33, 680k
56, 120k
82, 62k

[trobbins]

Stephen Keller has the reference on his informative website
http://thermionic.info/

Stephen is about to try your variant, as we were both quite interested

I'm not sure how a split load cathodyne PI can drive the plate voltage below the cathode, and hence change the polarity across the string of the two varistors ??

For those measurements I just made, the zeners were '1.5W' type, and the 82V zener was a string of 33+33+15. The type of zener may subtly influence the characteristic at such low current levels. Some more subtle tweaking of resistor values may align the clone somewhat better, although the crossover regions between sections can sometimes require further tweaks. The variation between the two original varistors Dinko sent me, and the likely variation within a varistor population, indicates to me that clone repeatability should be pretty good, and any subtle non-alignment using just those two varistor examples as the reference would be pretty imperceptible.

[martin manning]

Thanks. I'm very interested to hear how the experiments come out.

[Reeltarded]

Wow. Neato.

Cool


Ok I will be t3h quiet now and watch.

[angelodp]

Thought this might be of interest and informative.

https://www.youtube.com/watch?v=CiBXN82vzNQ