Nachbauer Dogzilla amp

[Mark Hammer]

I was sifting through the offerings of a components place in Canada that I had never used before, and passed an offering of fairly inexpensive 5818 Schottky diodes, in relatively modest quantity.  That got me thinking about whether I might have use for them.  "Hmmm, what things do I know of that need a lot of diodes, or that I might end up making for myself in quantity?".

For whatever reasons, I was reminded of the late Fred Nachbauer's "Dogzilla" amp project, and the soft limiter circuit he included, which is shown below and more extensively described here: http://www.dogstar.dantimax.dk/tubestuf/dzart-6.htm



So I started wondering, could Schottky diodes be used in similar fashion for a soft limiter that was running off lower supply-voltages?  Schottkys have a noticeably lower forward voltage than silicon types, so one could presumably use a diode ladder like he shows, without requiring gains so high that the amplifying device has already run out of headroom before that complete ladder can be made to work.

Whaddya think?  Worth pondering?  And if a 9v supply won't cut it, let's assume 12v or perhaps a charge-pump-assisted 17v supply.

[R.G.]

I did something similar on a skunk works amp that never got marketed. I had the diode ladder and resistors pretty much that way, but used a pair of MOSFETs diode-connected for the final and biggest limiter.

It worked fine. Well, at least that part of it worked fine. 

Schottky diodes would be good for something like this, allowing you to get the transfer function into a lower voltage. But I had a 25V power supply to work with.

One thought is that this might be a good place to use a signal transformer to step the signal up from +/-3V to +/-10V or so for the limiter, then resistor-padding it back down for output. That lets you get around the power supply limitation and also advertise it as "sounding just like the output transformer on a tube amp" if you're advertising-inclined ( or is that 'advertising-infected'?).

The other thought is to just use diode-connected MOSFETs, which are a similar transfer function to what this probably is trying to do anyway. That will get you down to +/-3 to +/-5V depending on the MOSFETs you select.

I liked the way the amp sounded. It was not marketed for other reasons than sound.

[PRR]

The Schottky drop is about half a Silicon drop. So you could scale the whole thing to half voltage.

DogZilla works with a 420V supply, make that 210V.

The three NE2 lamps suggest the signal level can touch 200V. That can be 100V.

> if a 9v supply won't cut it, let's assume 12v or perhaps a charge-pump-assisted 17v supply.

I think you are way under the mark.

Even a LM386 driving a 70V 1W speaker-transformer barely gets into this zone. (The pocket-radio transformers are way out of their league.)

There *are* low-volt "diode" curve-fitters based around op-amps. I think there's an idea in the Simulation section.

[kaludjerko]

The Schottky drop is about half a Silicon drop. So you could scale the whole thing to half voltage.

DogZilla works with a 420V supply, make that 210V.

The three NE2 lamps suggest the signal level can touch 200V. That can be 100V.

> if a 9v supply won't cut it, let's assume 12v or perhaps a charge-pump-assisted 17v supply.

I think you are way under the mark.

Even a LM386 driving a 70V 1W speaker-transformer barely gets into this zone. (The pocket-radio transformers are way out of their league.)

There *are* low-volt "diode" curve-fitters based around op-amps. I think there's an idea in the Simulation section.

On that DogZilla page you can see that he feeds 10V peak-to-peak signal into diode limiter. Than limiter limits it to around 3V peak-to-peak. For 10V signal we would not need 200V power supply, would we? Also, 5 x 0.6V drop that Si diodes produce results in around 3V peak-to-peak limited signal, so that fits. To my opinion, there is no need for high voltage in power supply.

You are right that power supply is 420V at DogZilla, however diode limiter section is decoupled from tube gain stage by capacitor. I got the impression that NE's are used to set operating point of tube, not to define level of amplified signal, so I'm not sure that reasoning is correct.

[brett]

Hi
plotting V and I for Schottky and Ge diodes show them to be similar (ie the 'knee' and 'height' are almost the same, in both relative and absolute amounts), but there might be some mileage in looking for small differences.
I assume that Ge is more sensitive to heat than Si, and heat lowers Vf, so I wonder whether you could warm a Ge diode for even lower Vf?
cheers

[kaludjerko]

Hi brett,

another option to get it to lower voltages would be to decrease number of back-to-back diode stages.
Dogzilla uses 5, but I have impression that even 2 stages connected in this fashion would be large improvement over single diode clipper (read, would not produce hard clipping).

[R.G.]

And then there is using one pair of diode-connected MOSFETs for the clipper array. The voltage and limiting characteristics are similar.

[analogguru]

Do you mean in a configuration like this:



or this:



or more like this from Elektor Februar 1985 (portable guitar amplifier).



analogguru

[kaludjerko]

I played with CircuitLab today. Let me share what I found out.

If this is regular diode clipper:



than it's response at 3V peak-to-peak is:



On the other hand, two stage diode ladder, like one from Dogzzila, has response:



schematic for a reference:



While amplitude is very similar, shape is not, it is much less saturated in case of ladder ...

When I crank V1 to 30V, diode clipper, stays at 0.6V peak voltage, while ladder goes up to 5V.

[tca]

A DFG (diode function generator)? Look at old analog computer manuals http://www.analogmuseum.org/english/library.html

Cheers.

[kaludjerko]

And then there is using one pair of diode-connected MOSFETs for the clipper array. The voltage and limiting characteristics are similar.

Hi R.G.,
I enjoyed your site since it was one of the only few on net on this topic, in 90's ... :-)
Anyway, what I was curious that you mention that MOSFETs would have similar response. By playing with simulation software, I got the impression that these ladders have peculiar behavior. If input voltage increases well above forward diode voltage (Vin = 30V or so), it behaves more like restive divider than clipper and distortion actually decreases. I will post few plots shortly.

cheers, kaludjerko

[R.G.]

A lot depends on how the ladder/limiter is designed. The fundamental way of designing these things is to make the diodes (or transistors, in another refinement) be switches, and switch in a different resistance to the overall divider as signal amplitude changes. The biggest steps in the "ladder" can be designed as true clamps to keep the output signal from further increasing, or to let the signal continue increasing at some unchanging rate corresponding to the last resistance/divider value.

So yes, there are ladders which become effectively a resistor network once the changes near zero are over with. Seen this way, the shaping becomes a kind of "crossover distortion" to much larger signals.

[elseif]

Because I play a Hammond organ clonewheel (Nord C2D), which is a rather clean-sounding board whose built-in overdrive is horrible, I have been looking for a tube amp simulation pedal that is subtle.  I don't want John Lord heavy overdrive, which pretty much rules out pedals made for guitar.  Rather, I want an overdrive that provides a complex IM and harmonic distortion but no buzziness.  Some subtle grind.  EQ will be needed - haven't tackled that yet.

I took Fred's design and wrapped it into the feedback loop of an op-amp.  So far, I am okay with it.  At the drive level I run it at, it doesn't really need much subsequent low-pass filtering.

This schematic is for line-level inputs.

I have LTspice simulation files, if there is an interest.

[willienillie]

I have been looking for a tube amp simulation pedal

lol