FET emulation of tubes without trimmer?

[JDoyle]

Anyone here ever mess with overdriving LinCMOS chips like the TLC2262. The SansAmp XXL uses one for distortion.
Got myself a couple CD4007s to play with the other day.

Not that chip but let me toss out a much, much older one: CA3130

Here is the datasheet: http://www.intersil.com/data/fn/fn817.pdf

CMOS differential input, BJT Voltage Amplifier Stage (VAS), and a CMOS output all in an incredibly simple circuit for an opamp.

But think about what you can do with pins 1, 5 and 8.

ESPECIALLY pin 8. And no, one doesn't HAVE to use it as the datasheet says, just think of it as access to a part of the opamp you never have.

For example, if you use the 3130 in a non-inverting configuration, the signal at pin 8 is inverted relative to the input, so you could theorhetically use that point as either an inverted output (with an incredibly high output impedence or, in other words, a CURRENT output), a point to directly connect something to the input as negative feedback in addition to 'standard' feedback, a point to INJECT something into the signal path, etc.

You could also use pin 8 as a way to lower the gain of either the VAS or the CMOS output stage. A resistor from pin 8 to ground would lower the gain of the VAS stage and a resistor from the output pin to pin 8 applies local feedback around the CMOS output stage. Of course, either of these things is going to severely degrade the performace of the chip, but in ways that us guitar FX simpletons don't really care all that much about, and besides, isn't that the point???

Anyway, just throwing that out there...

Regards,

Jay Doyle

[JDoyle]

What he means is operating in Class A and with minimal negative feedback thereby producing lots of 2nd harmonic content as a result because vgs is allowed to span a significant portion of the transfer characteristic.

First: OK - but that sure ain't what he said in the phrase "class-A (curvature) like behavior"

Plus, it's not really negative feedback: http://en.wikipedia.org/wiki/Negative_feedback

"Negative feedback feeds part of a system's output, inverted, into the system's input"

I'm going to chuck this match into the gas and step back, but technically, adding a source resistor is not the same as negative feedback.

If it was, you should be able to put two back to back diodes on the source, instead of a resistor and get a feedback clipper, right?

But you don't - you get damn near to the op amp example JC mentioned, except with fixed voltages, and quite possibly a LOWER impedence - because the output doesn't feed back to the input.

Adding a source resistor is SIMILAR to negative feedback, and has some of the same effects, such as a reduction in gain and a linearization of the output signal, but it's not technically negative feedback.

Regards,

Jay Doyle

[StephenGiles]

Could it be forced into having a current controlled input?

[gez]

I'm going to chuck this match into the gas and step back, but technically, adding a source resistor is not the same as negative feedback.

BOOM! 

Well, you could argue that it is in that the actions of the source resistor curtail the gain of the amplifier...having a negative effect on gain (if you like).

I've actually seen it described as negative feedback in one or two textbooks.  Makes sense as the inclusion of a source resistor swamps the effects of the current-dependent, internal emitter-resistance, in turn improving linearity.

Anyway, I understood the 'offending' post.  As you well know, Jay, without collector-base feedback and with the source resistor decoupled/omitted, things go banana-shaped pretty quickly if the signal is above 10mV or so.

[puretube]

Must`ve read the same textboox... 
Has been discussed controversally before... 

and: yes, this thread was intended about having FETs sound pseudo-tube-similar
(esp. "classical" triodish, i.e. sounding like a single "tube" biased for class A operation)*,
so I too agree with alanlan...

*but without having to bother with the production-spread;
not, however, about "matching" or "selecting"...


[EDIT:] J-FET, that is...

[puretube]

Anyone here ever mess with overdriving LinCMOS chips like the TLC2262. The SansAmp XXL uses one for distortion.
Got myself a couple CD4007s to play with the other day.

The 4007's are nice because they're so closely matched, which helps for certain things. 

By the way, found the appnote I was talking about earlier:

http://www.colorado.edu/physics/phys3330/phys3330_fa06/pdfdocs/AN102FETbiasing.pdf

Figure 8.  I tried to re-do this procedure for a J201/9V supply, but I'm not sure if I have it correct.  Others are welcome to try, please share if it works.

see: reply #6...

[puretube]

BTW: take a look at the simple basic "Phase Splitter": here the cathode, emitter, or source IS an OUTPUT...

[StephenGiles]

Could it be forced into having a current controlled input?

The 3130, that is?

[puretube]

Anyone here ever mess with overdriving LinCMOS chips like the TLC2262. The SansAmp XXL uses one for distortion.
Got myself a couple CD4007s to play with the other day.

Not that chip but let me toss out a much, much older one: CA3130

Here is the datasheet: http://www.intersil.com/data/fn/fn817.pdf

CMOS differential input, BJT Voltage Amplifier Stage (VAS), and a CMOS output all in an incredibly simple circuit for an opamp.

But think about what you can do with pins 1, 5 and 8.

ESPECIALLY pin 8. And no, one doesn't HAVE to use it as the datasheet says, just think of it as access to a part of the opamp you never have.

For example, if you use the 3130 in a non-inverting configuration, the signal at pin 8 is inverted relative to the input, so you could theorhetically use that point as either an inverted output (with an incredibly high output impedence or, in other words, a CURRENT output), a point to directly connect something to the input as negative feedback in addition to 'standard' feedback, a point to INJECT something into the signal path, etc.

You could also use pin 8 as a way to lower the gain of either the VAS or the CMOS output stage. A resistor from pin 8 to ground would lower the gain of the VAS stage and a resistor from the output pin to pin 8 applies local feedback around the CMOS output stage. Of course, either of these things is going to severely degrade the performace of the chip, but in ways that us guitar FX simpletons don't really care all that much about, and besides, isn't that the point???

Anyway, just throwing that out there...

Regards,

Jay Doyle

While you`re at it:
CA3160
Dual: CA5260 no "pin8", however...
and similar: CA3260

and: all Non-J-FET...

[Eb7+9]

... to get class-A (curvature) like behavior from a jFET ...

Could someone, ANYONE, explain to me what in the hell this sentence fragment ('class -A (curvature) like behavior') means? 

What he means is operating in Class A and with minimal negative feedback thereby producing lots of 2nd harmonic content as a result because vgs is allowed to span a significant portion of the transfer characteristic.

that's right ...

in analogue signal processing HARMONICS are a product of a non-linear transfer function ... a non-linear transfer function is loosely analogous to a curved mirror at the fun house ... it makes you look fat, thin, distorted ... when the transfer function is linear a sine wave going in comes out a pure sine wave ... when it's non-linear more than one sine-wave comes out - the difference is called Harmonic coloration ... now, the thing to note here is that Harmonics are a direct by-product of a curved transfer function ...

although many folks think that WE ARE ALL after tube emulation what we really are after is a simulation of the non-linear transfer response of a class-A circuit running on tubes - usually triodes since they produce more class-A MOJO per pound of signal ... I'll say it again, we're after class-A curvature, and all the other fun stuff that comes with it ... in other words it's not just about Harmonics production - I think this needs to be made clear since it's really not a very obvious thing ...

to explain this a little, class-A curvature does two things for us ... (I) it provides a dynamically varying addition of class-A harmonics via transfer (provided there's enough signal swing as ALANLAN pointed out) , and (ii) the transfer curvature produces an elongation of the sinewave that results in baseline (bias) shift ... if you've ever stuck a scope on the cathode of the 6V6 in a Fender Champ you know what I'm talking about ... this opens us up to many exciting potential musical doodads to explore ... one is bias shifting of a circuit that follows this one - I'm not sure I want to spill the beans but you might already get my drift of what might happen when you stick this directly in front of the first stage of a guitar tube amp ... it's awesome in a booster setting and I'm really digging playing the circuit implementation I came up with ...

so, in effect we're not just adding harmonics through spectral means, like one can do/approximate with a mixer and clean octaver, but rather we're now doing what a single ended tube circuit does, but at a scaled down voltage ... essentially simulating, and even exaggerating, the curvature of the response ... I'm saying exaggerating because a triode circuit (and not the tube alone) can have at most power 1.5 curvature while a jFET circuit can exhibit power 2 curvature at most (that's the mode we want to be running it if we can - Danyuk shoots for 1.5 emulation in his paper) ... hence jFET circuits can be used to go a little further in terms harmonic coloration than what you can get from a class-A circuit running on a triode ...

---

back to the jFET matching thing for a second ... I did an extensive search to see if there's anything solid on the net and pretty much everybody is doing the same non-thing ... I wrote out the math to show where/how the tests are faulty and I'm pretty sure most of the established members on this forum don't want to see that stuff dragged out ...

I'll just say this because it applies to phasors ... the kind of test where a fixed current is drawn into the DUT's, a Vgs voltage is measured and devices are lumped together when close in value will produce a locus of possible points for which devices with Idss and Vp values all produce the same Id/Vgs pair response at the test current level ... it is easy to show that if you change the current test level then  devices that matched cannot match any longer unless their Vp and Idss values are identical - which to trained engineers is the only real terms of criteria for matching // the data-sheets are written for people who know and understand this ... to make proper use of the proposed test jigs a second pass needs to be performed at a reference current considerably different than in the first pass ... this can be demonstrated also with simple math ... in other words, if you were to single out seven devices in the first pass of that test and from those seven you singled out a remaining four after dropping the ref current in a second pass ONLY THEN can you claim matched devices throughout their operating ranges and therefore ALSO guarantee that the matching will apply to the VCR aspect of the jFETs ...

what this means is when people are using such a test they are likely yielding a set of devices for which the Vp/Idss values that are distributed along the curve locus (unless the devices where all matched to begin with by high chance) ... and not knowing what the exact values for Idss and Vp are this means that true resistance levels can be all over the map ...

since the equation for the incremental resistance  rds = |Vp|/2Idss(1-Vgs/Vp) is strongly dependent on Vp and Idss then the resistance be all over the map just as the Vp/Idss values will be (lying on the locus curve between Vp min and Vp max) ... now there's a happy conclusion to this story ... many of the people who go through the bother of using this false-match method are probably ending up with devices that exhibit resistance level differences that are considerably large without realizing it ... since the shift frequencies are the result of a product of capacitance and channel resistance, then if a builder uses identical cap values the shift frequencies will be staggered according to the channel resistance differences ... well, as Mark Hammer once pointed out, it would produce the same effect as if we had matched channels but staggered cap values - which we all know is how the Univibe does it's thing ... so people are VERY LIKELY (as in Gaussian likelyhood) thinking they're matching their devices when they're just randomly staggering them ... and since staggered phasors seems more pleasing or interesting to the ear it is no surprise hearing them say their phasor sounds great ... it also explains why many (probably the majority) of people who simply finish their phasors by swapping in devices by ear get there without having to go through this trouble ...

~jc

[DougH]

I think the supposed importance of 2nd harmonic content is way overblown. Typical good sounds are made up of a mixture of a multitude of harmonics and the goodness has more to do with EQ than it ever will with harmonic content.

You guys want to see how many angels you can get on the head of a pin, get the wave shapes on your scopes to come out right, and otherwise play the triode emulation game - have fun. I'll just sit back and watch...

If you want good sounds from JFETs or hell, from any other device, it's been done. Comb the archives. The rocks have been uncovered for a long time...

[gez]

I've actually seen it described as negative feedback in one or two textbooks.  Makes sense as the inclusion of a source resistor swamps the effects of the current-dependent, internal emitter-resistance, in turn improving linearity.

Internal source resistance, rather.

without collector-base feedback and with the source resistor decoupled/omitted, things go banana-shaped pretty quickly if the signal is above 10mV or so.

Likewise, should have been drain-gate feedback.  Slip of the brain...

[puretube]

Brain-internally, I always refer to those FET "gates" as: "Gitter" = german for: "grid"... 

[R.G.]

back to the jFET matching thing for a second ... I did an extensive search to see if there's anything solid on the net and pretty much everybody is doing the same non-thing ... I wrote out the math to show where/how the tests are faulty and I'm pretty sure most of the established members on this forum don't want to see that stuff dragged out ...

I'll just say this because it applies to phasors ... the kind of test where a fixed current is drawn into the DUT's, a Vgs voltage is measured and devices are lumped together when close in value will produce a locus of possible points for which devices with Idss and Vp values all produce the same Id/Vgs pair response at the test current level ... it is easy to show that if you change the current test level then  devices that matched cannot match any longer unless their Vp and Idss values are identical - which to trained engineers is the only real terms of criteria for matching // the data-sheets are written for people who know and understand this ... to make proper use of the proposed test jigs a second pass needs to be performed at a reference current considerably different than in the first pass ... this can be demonstrated also with simple math ... in other words, if you were to single out seven devices in the first pass of that test and from those seven you singled out a remaining four after dropping the ref current in a second pass ONLY THEN can you claim matched devices throughout their operating ranges and therefore ALSO guarantee that the matching will apply to the VCR aspect of the jFETs ...

JC, let me say this once again.
- The Vgs test is not bogus. Using closely matched Vgs devices is a necessary condition to getting matched JFETs for phasers. In fact, it is a sufficient condition, as experience has proven.
- It is not a test designed for perfection, as this forum hashed through years ago. As we see on the forum, about one member a week is too confused by even this level of sophistication to make the JFET matcher work. Adding more tests further selects AGAINST the number of people who can get JFET matching for phasers to work. Increase the complexity of the test needed to get a JFET phaser to work and you DECREASE the number of people who can get it to work. If people give up on so simple a test, how many more would give up on an even more complicated test?
- It does produce the most working phaser stages per device bought per test, as Idss matching alone will NOT produce working phaser stages, as Idss matching does not produce devices that are necessarily within their linear range for the LFO and biasing in the typical JFET phaser. As you point out yourself, Rds pefection is not needed for a phaser to work well. All the JFETs being in the linear region simultaneously is necessary.
- Let me just add the observation that the Vgs matcher I published is not very germane to the amplifiers discussion you brought it up in. It looks to me like you simply saw a chance to take a pot shot at me and did. 
- You have done nothing in terms of proving fact behind your statements nor a retraction. Your statement about intentionally misleading people, thinly veiled as it was, is straighforward defamation. I would like a retraction or facts behind it. I will look for other methods of redress if I do not get either a factual basis for that or a retratction.

[puretube]

Don`t sweat it, people:
I gladly solved the problem by finding my own answer to the initial quest myself,
which has led me to a rather different problem:

which of the roughly 65 different effex-designs that emanated from the experiments
should I concentrate on

[DougH]

Haha!

That sounds familiar...

[puretube]

Well, the choice already has been made:
6 out of 65 ain`t bad... 

[soulsonic]

Thank you very much, Jay! These are the kind of things I really enjoy learning about.

-M.

Not that chip but let me toss out a much, much older one: CA3130

Here is the datasheet: http://www.intersil.com/data/fn/fn817.pdf

CMOS differential input, BJT Voltage Amplifier Stage (VAS), and a CMOS output all in an incredibly simple circuit for an opamp.

But think about what you can do with pins 1, 5 and 8.

ESPECIALLY pin 8. And no, one doesn't HAVE to use it as the datasheet says, just think of it as access to a part of the opamp you never have......

[JDoyle]

While you`re at it:
CA3160

Ton, the only reason I didn't mention that one is because it has internal compensation, in series with a resistor no less.

I wanted to toss out the 'raw' version on the chip for the most flexibility.

[JDoyle]

Thank you very much, Jay! These are the kind of things I really enjoy learning about.

-M.

Of course!

The tubescreamer and it's ilk has caused so so many arguements/discussions/debates about DUAL op amps, that I honestly think there is an entire realm being overlooked - SINGLE op amps.

In duals everything is already decided for you, offset adjustment, compensation, operating current, etc. But with singles, you normally get access to points of the circuit where you can seriously mess with what is going on.

Regards,

Jay Doyle