clipping diodes as voltage or current limiters?

[idy]

On another forum, discussing The Wampler Ecstasy, it was compared to the "Timmy", and paulc wrote:

"The Timmy has 3 choices of current limiting. The Ecstasy looks like it has one of the same settings along with no limiting and voltage limiting. These are very different."


I have not heard of feedback loop diodes and ground shunt diodes called respectively current limiting and voltage limiting. Anyone else? I have heard the terms "soft" and "hard" used. Can anyone explain why feedback loop clippers are current and not voltage limiters?

[thehallofshields]

That's a great question.

I'm going to guess its a mistake, because with guitar amplification were almost solely concerned with Voltage Gain and not Current Gain. Id be very interested to know if Feedback Loop Diodes are used to limit current in Power applications.

[anotherjim]

Seems like an unnecessary distinction. Maybe stemming from the the fact that an opamp is mostly all about current (the current mirror in the inputs). The resistors/caps/diodes or whatever, into and around it , are converting voltage into input currents. In the end (I don't have complete understanding of op-amp action at internal level), it might well be limiting current BUT as we observe the output, it's limiting signal voltage!
Clipping diodes across the signal path and ground or vref are for sure going to be voltage limiting - actually increasing output current (from the source) once they conduct.

[merlinb]

"The Timmy has 3 choices of current limiting. The Ecstasy looks like it has one of the same settings along with no limiting and voltage limiting. These are very different."

I have not heard of feedback loop diodes and ground shunt diodes called respectively current limiting and voltage limiting. Can anyone explain why feedback loop clippers are current and not voltage limiters?

Simple: the person who made the quote doesn't actually know what he's talking about.

[teemuk]

I have not heard of feedback loop diodes and ground shunt diodes called respectively current limiting and voltage limiting. Anyone else? I have heard the terms "soft" and "hard" used. Can anyone explain why feedback loop clippers are current and not voltage limiters?

Those are not probably the best terms.

...But if you evaluate current flowing through the diodes (which also imparts its effects on overall clipping characteristics) you find that diode forward currents in feedback loop vs. shunt clipping are usually quite different. Shunt clipping diodes usually receive full, non-clipped, signal swing from the preceding amplifier stage, therefore higher voltage swing will drive diodes with more current when clipping occurs. In feedback loop scheme the diodes already limit the signal amplitude locally so they also limit the forward current at the same time (within that gain stage). That's why you generally get hard clipping from the other and soft clipping from the other, unless you specifically compensate for different diode currents.

[Johan]

If you want to. You can think of the opamp inputs as +input=voltage input and -input= current input.  Perhaps that's what he's getting at

[idy]

teemuk wrote
"you find that diode forward currents in feedback loop vs. shunt clipping are usually quite different...."

I think you have explained it pretty well. Thank you all for your ideas.

[PRR]

> an opamp is mostly all about current (the current mirror in the inputs).

There may be a current mirror near an opamp's inputs (none in 709, for example), but not AT the inputs.

The classic Ideal opamp has NO input current.

While real opamps have real current, we usually design so that's not a factor.

OTOH there is/was the "Norton Amp", AKA Current Differencing Opamp (CDA), LM3900 and LM359. The inputs were virtual ground, current sensitive. Clever idea, very cheap implementation, worse for some chore and better for others. Mostly didn't catch-on, but does lurk in a few corners.
http://www.experimentalistsanonymous.com/ve3wwg/doku.php?id=cda_opamp

[anotherjim]

Well, there you have it!
But I have often read of op-amps called "current operated" devices, and aren't the inputs of a 741 BJT bases? But then again, I wondered how CMOS op-amps got on.

But Paul is right, the inputs should be high impedance and although current does figure (be it picoamps), it's not an ideal part of the internal operation.

I've got some LM3900 - bought yonks ago. Haven't found a use for them yet, although there is a simple looking VCO design in the National app notes I might use some day.

Cheers

[merlinb]

But I have often read of op-amps called "current operated" devices,

You've been reading some funny book then!  Virtually all amplifying devices are voltage controlled, and that includes all transistors, from which opamps are made.

[anotherjim]

Pretty sure that transistor action requires flow of charge carriers,  be it -ve electrons or, in the opposite sense, +ve "holes". That's current in other words. Potential differences make it happen  (makes most electrical phenomena happen apart from magnetic and thermal  - that's a given), but the action depends on carrier flow, even in a FET.

Now I ask, what are Q1 and Q2 in the inputs of the 741 above going to do without current flow?

[induction]

I guess it depends on what you think 'current operated' means. I always thought it meant that the signal at the output is controlled by the current of the input signal, not that current flow happens somewhere in the device.

[merlinb]

Potential differences make it happen 

So it's voltage controlled, then! No voltage, nothing happens.

[PRR]

> what are Q1 and Q2 in the inputs of the 741 above going to do without current flow?

Yes, BJTs always have input current.

Q1 Q2 are flowing 10uA and probably have 0.2uA base current. This is a DC term, we allow for that in biasing, and forget it for audio.

For Q1 Q2 to switch FULL current, to swing the output, we would need another 0.2uA of signal current to the base. But they never switch ALL 10uA. Counting on thumbs, we only need 1uA of swing at Q15 base to throw pin 6 to its limits. So only 0.02uA change of base current at Q1 Q2.

So the input current change is "small" for most any application where a BJT-input amplifier makes sense.

There's current. If the amplifier designer has done well, we can often ignore it.

Further reading

Note that input current always becomes "not small" at some high frequency where input Capacitance starts sucking. The many-Megs of a tube or JFET can fall to a few hundred K Ohms at the top of the audio band. This is significant current. And in that regard, a BJT is sometimes the easier design chore, because its relatively large input leakage swamps capacitive current up to quite high frequency. It sucks, but uniformly over a wider frequency range than we audio guys care about.
__________________________________________

> transistor action requires flow of charge carriers

Cherry-Hooper's iconic book (Amplifying Devices and Low-Pass Amplifier Design) treats vacuum tubes, FETs, and BJTs "all the same", except that BJTs have non-negligible "input leakage".

But everything leaks. A sheet of dry glass leaks a teensy current. This is apparent when you get too crazy with MOSFETs. Their input current (leakage through a "glass" gate insulator) is many orders of magnitude down, but hardly zero.

Vacuum tube grids leak significantly, as you'll know if you ever forgot the grid resistor. Sometimes the grid will drift down and shut-off the tube. But when it really hurts: floating power tube grids seem to always leak up, turn-on the tube to infinity or smoke.

Vacuum tubes and JFETs have input leakage maybe 1,000 times smaller than their output currents. Negligible in many situations. However the trends of these leakages are complex with inflections at odd points.

Oddity: Widlar noticed that at high temperatures, a carefully crafted BJT would have less input current than a JFET, hence the LM308 super-beta opamp. JFET leakage rises sharply with temperature. BJT leakage drops slightly when fairly hot (because beta rises with temp and other temp-related leakages are small).

Cherry-Hooper treat all devices as "voltage controlled", classic transconductance. This is perfectly valid for BJTs.

Because BJT input current is a fairly consistent fraction (1/beta) of output current, we *may* instead treat them as "current controlled". In many cases the currents are obvious while the voltages are tiny and hard to estimate. However the output current is always (if also) a function of that input voltage.
_________________________________________

I don't understand what paulc is saying in "another forum", and wonder if that side of the discussion ought to be left over there where paulc can explain his remarks.