Understanding a Zves SHO? (Someone smarter than me)

[sevenisthenumber]

I would love to have a breakdown of what each part does! Anyone board enough??? 

I would love to add a tone knob or toggle switch... Maybe something else useful...

[frequencycentral]

http://circuitworkshop.com/forum/index.php?topic=460.0

[earthtonesaudio]

Harder than it looks... good luck!

[sevenisthenumber]

Yea..

Any suggestions on a tone control or tone switch?

[earthtonesaudio]

How about you tell us what you would like to do.  Do you want a treble cut?  Put a small cap across D and G.  Treble boost?  Split the "upper" 10M resistor into two 5M resistors, at the middle of the two, put in a cap to ground.  Put a resistor in series with the cap in either case for a variable tone control, or a switch for a preset tone switch.


These are just two of the many, many possibilities.

[sevenisthenumber]

Great ideas! I would love a 3 way tone toggle i think. Middle would be stock, up brighter than stock  and down would be darker than stock.

[PRR]

It's a basic one-stage amplifier.

earthtonesaudio and GeoFex give excellent overviews of MOSFET amplifiers like this.

If you really want to understand "what each part does!", it would be good to have some understanding of basic amplifiers.

And don't think of "what each part does". That's like asking what each note in a song does. Think of chords and melodies. Think about circuit loops. Battery, R3, Q1, VR is one loop. Mainly R2 and Q1 are in a tug-of-war to split the battery voltage.

> Yea.. Any suggestions on a tone control or tone switch?

I sense that you are not going to explore amplifier theory today, just want a recipe.

"Tone" in the sense of "brighter darker" is different gains at different frequencies.

The gain of this amp is roughly R3/VR.

Using pure resistors, gain is "flat" at all frequencies.

If you shunt a resistor with a capacitor, "resistance" (impedance is a better word) goes down with frequency.

Therefore a 0.01uFd across R3 will give normal gain at low frequency and less gain at higher frequencies. "Darker".

OTOH, a 0.01uFd across VR will give normal gain at low frequency and more gain at higher frequencies. "Brighter".


Alex-

The maximum gain (VR=zero) won't be what you calculated because "If we assume the BS170 here has the same transconductance as the one in the GEO article (0.320S)" is false. For tutorial purpose R.G. took the datasheet Gm, but this is specified at 200mA!. This plan works nearer 1mA. Gm is less at lower current (obviously Gm falls to zero at zero current). How much less? Unfortunately the Fairchild sheet does not say.

Also from observation of many amplifiers: gain of 16V/0.01V= 1,600 in a single stage is unlikely, and never this easy.

I'm going to use some FET behavior and predict that Gm falls as square-root of current. We are working at 1/200th of datasheet current, we expect Gm to be like 1/14th of 0.320S. Say 0.02S.

There's variation from one FET to another, and the accuracy of my square-root interpolation is dubious, especially for two decades down. So we won't try for "exact" calculations, and I'll round R3||R4||Rload to 5K for E-Z figuring.

Now when we have a big cathode resistor impacting voltage gain, it may be convenient to invert Gm to get an equivalent cathode resistance. 1/0.02S is 50 ohms.

When VR is 5K the total resistance under the FET is 5,050 ohms. And when VR is zero it is 50 ohms.

With R3 as 5K the gain runs from 1 to 100. Well, the load may be more like 4.3K, so gain is more like 0.8 to 80 at best. Even 80 is a large gain for a single low-volt stage, so I would not bet on it. But "unity to high" gain, which is interesting.

That all assumes the source impedance is Zero, so that NFB from 1Meg Source-Gate is absorbed. Since R3/(VR+Rs) gain is not even 100, a 1K source is "zero-enough", a 10K source is "hmmmm..." And in guitar work, we often have sources over 10K.

Input impedance: You know Mr Miller. R2 directly loads the input. R1 is Miller-ed by gain. At unity amplifier gain R1 acts like half its value. At gain of 100 it is about 1/100th its value. So with 1Meg resistors the input runs from 330K to nearly 10K. You see why this is often built with 10Meg resistors: then it only droops toward 100K, a more tolerable load for guitar sources.

This is why I didn't suggest "treble cut?  Put a small cap across D and G". This is equivalent to a cap across the source. Its action depends on the source. In a particular case, wound-pickups, the source is inductive and we get a peak-and-drop resonance. That may be good; but has to be set-by-test. OTOH a cap across R3 (or Drain to ground) works mostly against R3 and gives a smooth predictable action.

[earthtonesaudio]

Excellent points, Paul.  Thanks for the corrections.  Would you mind if I quote you (with credit) in the write-up I did at the other forum?

[Paul Marossy]

What I always thought was odd about the SHO is those diodes on the input.

[earthtonesaudio]

The reason for the diodes is to try to hold of the almost-inevitable death by electrostatic discharge.  The concept is relatively simple: if a static charge comes in that makes a Gate-Source potential greater than about 20V (even with less than microamps of current), the insulation between gate and channel is punctured and the MOSFET dies.  The reverse-biased diodes at the gate are supposed to clamp any voltage greater than the supply voltage or less than ground. 
The real-world scenario is slightly more complex.  The problem with ESD damage is it takes place over an extremely short amount of time.  The time required to kill the FET may be less than the time it takes for the diode(s) to go into forward conduction and protect it.  This is because the diodes have parasitic capacitances which make them appear as tiny (just a few pF) caps to extremely fast signals.  If the gate capacitance is greater than the diode capacitance, it will take longer to charge and chances are the spike will be diverted through the diodes: happy happy!
However, this is not guaranteed to work.  If the diode capacitance is greater than the gate capacitance, a fast ESD spike will kill the FET. 

This is probably why Zvex switched to the Gate-Source zener diode in the later models.  This is much better:  one diode has less capacitance than two diodes, and zeners often have very low capacitances anyway.  In addition, the Source follows the Gate, which makes the quiescent capacitance appear much smaller than it really is, which keeps the zener from having much effect on the tone of the circuit.

The main problem with both of these protection schemes is the fact that the input is capacitively coupled.  There's nothing in there to slow down an incoming ESD pulse.  Simply adding a series resistor somewhere between the input jack and the diode(s) would slow down any incoming pulse and give the diodes a better shot at clamping the sucker.
If you check out old CMOS datasheets which show the input ESD protection, there is always a limiting resistor in series with the diodes.

[PRR]

> Would you mind if I quote you (with credit) in the write-up I did at the other forum?

There's no secrets there, and credit won't raise my pay-grade.

Absorb it, understand it, and write it over yourself (possibly clearer than my mumbles).

And..... if you have a BS170 powered-up and handy, and enough tooling to measure gain, you could determine the *actual* Gm of BS170 at typical amplifier currents (not at the whopping 200mA point). That would be a sweet thing for the community to know.

[Paul Marossy]

The reason for the diodes is to try to hold of the almost-inevitable death by electrostatic discharge.  The concept is relatively simple: if a static charge comes in that makes a Gate-Source potential greater than about 20V (even with less than microamps of current), the insulation between gate and channel is punctured and the MOSFET dies.  The reverse-biased diodes at the gate are supposed to clamp any voltage greater than the supply voltage or less than ground. 
The real-world scenario is slightly more complex.  The problem with ESD damage is it takes place over an extremely short amount of time.  The time required to kill the FET may be less than the time it takes for the diode(s) to go into forward conduction and protect it.  This is because the diodes have parasitic capacitances which make them appear as tiny (just a few pF) caps to extremely fast signals.  If the gate capacitance is greater than the diode capacitance, it will take longer to charge and chances are the spike will be diverted through the diodes: happy happy!
However, this is not guaranteed to work.  If the diode capacitance is greater than the gate capacitance, a fast ESD spike will kill the FET. 

This is probably why Zvex switched to the Gate-Source zener diode in the later models.  This is much better:  one diode has less capacitance than two diodes, and zeners often have very low capacitances anyway.  In addition, the Source follows the Gate, which makes the quiescent capacitance appear much smaller than it really is, which keeps the zener from having much effect on the tone of the circuit.

The main problem with both of these protection schemes is the fact that the input is capacitively coupled.  There's nothing in there to slow down an incoming ESD pulse.  Simply adding a series resistor somewhere between the input jack and the diode(s) would slow down any incoming pulse and give the diodes a better shot at clamping the sucker.
If you check out old CMOS datasheets which show the input ESD protection, there is always a limiting resistor in series with the diodes.

Ah, I didn't think of it as an ESD protective measure. I was thinking it was like diode clamping on the input. The dual arrangement is what looked really weird to me.

[PRR]

> an ESD protective measure

Yes. Relative to Source pin, the desired signal is roughly +1V to +3V. The FET itself will "clip" anything outside of that, no diodes needed. On a good day you do not NEED these diodes for any reason.

The "BAD!" zone is +20V to -20V. Anything outside of that (or maybe 30V) will ZAP! the FET Gate, and then it won't work at all. On a bad (dry) day, without the diodes, you may kill the MOSFET without knowing it.

Since Source is within supply voltage, and supply voltage is 9V or maybe 18V, we can simply constrain the gate within the supply rails, clamping anything more into the robust battery (or power cap). 2 diodes does that.

-OR- we can use a Zener diode. One way it clamps at 0.6V (ordinary forward diode). The other way it clamps at Zener voltage, which we can buy from 3V to 100V, and 5V to 20V will give fine protection.

Difference? Zener costs more each, but counting mounting the 1-part solution may be a cent cheaper than the 2-part solution. In mass production that matters; in DIY it doesn't. You may be guided by what's in your drawers: a few-Volt Zener or a couple of plain diodes.

Personally I doubt capacitance matters.... we are coming off 3 to 10 feet of guitar cable so 100pFd 300pFd already. And there's 40pFd-500pFd of input capacitance from the BS170. And the guitar pickup design evolved for the additional ~~100pFd of a 12AX7 grid. Another diode or two hardly matters.

Yes, I do think there should be a 10K-50K series resistor to control peak ZAP current. Guitar amps generally have 34K in series with the tube grid. This does not reduce treble and hardly affects hiss noise.

Sharp eyes on the BS170 datasheet will spot a diode. But this is Drain-Source. It does not protect the Gate. It is really an "accident": the way they bake Silicon, the bottom layer tends to diode to the layer above, so we normally bias this parasitic diode "off".