JFET overdrive, how to preserve/enhance dynamics?

[Pojo]

Yes.....another JFET topic. 

So this is more of a general thing instead of anything circuit specific. That said, my latest build was a modified plexi-drive which almost has it all....the tone/eq, distortion quality, etc are all perfect. The only thing I don't like is that when playing in a band context, the sound gets lost a bit because it lacks dynamic variation compared to (for example) my Carvin Legacy's lead channel. It's like the tone lacks punch and isn't as 'alive'. I know any kind of overdrive adds compression to the signal, but somehow my favorite tones seem to preserve some dynamics regardless of being super saturated.

Basically I'm looking for ideas on how (if its even possible) to get that out of a cascaded JFET circuit. Maybe I should try more voltage? More gain stages with each stage lower in gain? Aged unicorn droppings?

[Kipper4]

I dont know about jfet cascaded but mosfet circuits seem to do this rather better IMO.
what about a mosfet booster do you think this might help out?

[Pojo]

As in Box of Rock style cascaded Mosfets or Mosfet clippers?

May have to breadboard a mosfet plexi-drive and see 

[Mark Hammer]

Overdrive sound happens because dynamic range is deliberately limited.  So the goal of acquiring noticeably more harmonic content, while retaiing as much dynamics as possible, tends to be best achieved by "collecting" a bit of harmonic content from each stage, over multiple stages.

One of these days, I'd like to see a stage-by-stage THD analysis of some of the more popular or classic tube amps, such that one could see how much harmonic content was added after stage 1, 2, 3, phase-splitter, power-tubes, and output transformer.  It'd be useful to know how much each stage contributes to the eventual tone.

[Jopn]

One of these days, I'd like to see a stage-by-stage THD analysis of some of the more popular or classic tube amps, such that one could see how much harmonic content was added after stage 1, 2, 3, phase-splitter, power-tubes, and output transformer.  It'd be useful to know how much each stage contributes to the eventual tone.

Well Mark, you DO have a long weekend coming up...

[Kipper4]

I was thinking put a mosfet booster beforethe plexi drive to see if it would improve your dynamics. At least its a quick low parts count build and wont cost much to experiment. Unless anyone things i'm way too wide of the mark of course.
Maybe cut back on the distortion on the plexi (less compression).
It might work?

[Pojo]

Good one, Mark...more lower gain stages does seem promising.

Kipper, I've been running it (on and off) with a JFET mini-booster in front of it. May build up a SHO and see what happens. Lowering the gain brings back some dynamics naturally but I'm looking for ways to retain the dynamics and keep the saturation.

Any thoughts on upping the voltage for more headroom?

I have a feeling maybe I'm asking too much from solid state devices, as it's something tubes seem to do very well....and it's indeed a tube circuit which has set the bar for what I'm going for. Hmm :/

[Mark Hammer]

One of these days, I'd like to see a stage-by-stage THD analysis of some of the more popular or classic tube amps, such that one could see how much harmonic content was added after stage 1, 2, 3, phase-splitter, power-tubes, and output transformer.  It'd be useful to know how much each stage contributes to the eventual tone.

Well Mark, you DO have a long weekend coming up...

Yes, but I don't have the instruments or the know-how.  All I can do is ask the questions and cross my fingers that the right folks get curious. 

[tca]

I have a feeling maybe I'm asking too much from solid state devices, as it's something tubes seem to do very well...

I think you are asking to much from a single bjt stage, compare with a single tube pre-amp. How much compression/dynamics can you take out of it? My suggestion would be to add a least 3 gain stages, including the power amp section.

I find the Plexi-Drive sound very dull... = a buffer, one amp stage and a second to be overdriven, stiff as a rock!

I've done some kind of similar exercise... the number of gain stages (3 of them) are not enough, you should have some and must have some kind of "uncontrollable" transients, it is not only the final wave form that counts.

[Morocotopo]

I think the question is, what is dynamics? Or better yet, how we percieve dynamics? It´s not only volume. Actually the waveform or timbre of the sound directly influences the perceived intensity, so also the perceived dynamics. Try the following: listen to a sine wave and a square wave of the same amplitude and frequency and tell me wich one sounds louder to you...
Speaking about distortion, what I think the tube amp does in relation to a pedal is, it changes more notoriously the output waveform in response to the variations in the input signal, so you perceive a more dynamic outcome. So maybe that´s the direction you should look into.

[teemuk]

The overdrive thresholds in those battery-powered FET thingies are usually kinda "fixed", which produces the monotonic sound.

For dynamics you may want to enhance schemes that make overdrive interact with signal envelope, not only its amplitude.

Try, for example, enhancing circuit characteristics that:

- with sustained high amplitude signal content produce DC offset shifts that result to harmonic variation due to shift from symmetric to asymmetric clipping or vice versa

- with sustained high amplitude signal content variate the clipping threshold (e.g. power supply sag, clipping references tied to envelope followers tracking the signal amplitude)

- with sustained high amplitude signal content introduce additional distortion mechanisms (e.g. crossover distortion)

- with sustained high amplitude signal content modify the frequency response (e.g. bandwidth reduced at higher signal levels)

- with sustained high amplitude signal content introduce modulation of the signal (e.g. ghost noting from 120Hz ripple)

Note the word sustained. You want these effects to build in gradually, not just come on when certain voltage threshold is exceeded. Inroduce also a time threshold.

All this will variate the harmonic content in relation to signal's overall dynamics and envelope instead of just clipping the signal at a fixed threshold, which produces very little variation in harmonic pattern of the distortion.



The main fault of most of those JFET -based "emulator" pedals is that they entirely ignore the power amp portion of the amp they "emulate". And the power amp portion happens to introduce many of those characteristics.

I find the Plexi-Drive sound very dull... = a buffer, one amp stage and a second to be overdriven, stiff as a rock!

Case in point.

[chptunes]

good stuff.  i've read that the OCD pedals yield a very dynamic response, because their clipping diodes (MOSFETS) are tied to Vr instead of Ground.

any thoughts?

-Corey

[teemuk]

^ Well... "Vr" or "Vbias" of those single-rail circuits is usually filtered with a capacitor to ensure a steady DC potential in the said node. Whenever you sink DC current to that node (such as through a clipping diode, which is essentially a half-wave rectifier) that current gradually charges the filtering capacitor, causing a -slight- DC bias drift, which leads to -slight- modulation of the clipping thresholds and most likely to -slight- asymmetry in clipping. The Vbias node also sets the crossover point of the waveform so any DC shift in the Vbias node will also reflect to all stages using the Vbias as reference. These stages will also experience  a -slight- DC bias drift, which may further -slightly- enhance asymmetry of the clipping.

I use the term -slight- because usually the effect is quite small due to these typical reasons:

- The filtering capacitor can perfrom it's duty - filtering - because it essentially decouples all AC signals to a low impedance node (common of the power supply). Though the current through the clipping diode is essentially DC it also has a large AC component, which means that the charging process of the filtering capacitor due to "rectified" current is actually so slow that any possible effects are typically completely unaudible or undetectable - except during very, very long sustained clipping, which will likely never take place with typical input signals from guitars. The RC time constant could be shortened with lower filtering capacitance but that degrades the operation of the filter itself. (This likely would not make big difference with battery-powered circuits but could be an issue with circuits deriving DC power supply from rectified AC voltage).

- The clipping scheme may be (and often is) symmetric, thus charge from positive side of the waveform clipping is quickly drained by clipping of the negative side - and vice versa. This even further slows down the DC offset drift at Vbias node.

In other words, in such scheme the process is typically way too slow and its effects way too small to cause significant difference in operation in comparison to schemes where clipping diodes simply shunt directly to ground. Therefore, IMO, there are several ways to achieve similar effects in much more pronounced manner than connecting clipping diodes to Vbias node.

Yes, I'm aware of schemes using individual Vbias references (preferably ones with poor filtering) for each clipping stage. They work better but it still isn't even near as efficient as deliberately generating a (variable) DC reference voltage for the diodes with a circuit specifically designed for the purpose. Such circuit can have low impedance to effectively sink the diode current and it will not need capacitive filtering, which would slow down the modulation of the DC reference considerably.

Another point is that charging the capacitor actually -increases- clipping threshold, so if you want to emulate effects like power supply sag where clipping threshold -decreases- during sustained periods of high amplitude signal content the scheme is actually working just the opposite to that. Under heavy clipping the capacitors would eventually charge up to potential where clipping would almost completely seize.

But yes, there is some truth in that claim - and yes, in certain designs the effects very likely are more pronounced than in some others. There's also another scheme based on very similar principle of operation: Connect the other end of the clipping diode to a circuit consisting of parallel resistor and capacitor. The other end of those is connected to common as usual. Both clipping diodes will need their own RC circuit due to aforementioned reasons. Now, each time a diode clips the DC current charges its corresponding decoupling cap and slightly alters the DC potential at the junction between the diode and the RC circuit. This is a crude "crowbar" approach.

It just can be done way more effectively if needed. ...Not to mention a specifically designed reference circuit allows way more control and possibilities: The time constants of "attack", "release" and "decay" characteristics are more tweakable, the operation doesn't interfere with circuit portions it should not interfere with, you can decrease the clipping threshold, and you can hook the reference to signal sources like envelope followers, oscillators, etc. ...you name it.

[Gus]

You should read the jfet specs before you raise voltages. 

[DDD]

Maybe 2x9 Volts instead of 1x9 Volts battery?

[Pojo]

Gus: Yep, J201's are good for 40V max and 2N5457 (used for Q2) are good for 25V...at least how this novice mind perceives the datasheets. I would push to no more than 18V.

teemuk: Thank you for your very detailed thoughts and insight. I'll need to re-read a couple times to digest but I'm interested the concepts you introduced by tracking the signal envelope and such. I'm thinking an OTA like the LM3700 would be a potential candidate for this? Kinda seems like it would be combining a compressor/expander and overdrive with the 2 interacting in certain ways....hmmm. I haven't a clue as to how to implement a time threshold though. Can that be done in the analog realm? Also I think you hit the nail on the head with jfets having a very fixed clipping threshold. Another idea that just came to mind would be maybe using a crossover filter to split the signal into a couple frequency ranges and individually clipping them at different thresholds....is there anything out there that does this already?

tca: I get what you mean, but the plexi-drive is actually just 3 cascaded jfet common source amplifiers. No buffer. Regarding the power amp section: For me, there's still an actual KT77 x 4 power stage which is pushing pretty hard in gig scenarios. But the dynamics are lost before the signal even gets there with my pedal engaged. I'm not really going for a complete tube amp emulator, just something to give me a different flavor of overdrive that can be just as usable as my amp's natural overdrive...if that makes sense.

As a side thought, I have a 5 stage jfet overdrive on vero (discussed when I was trying to come up with a layout in my first topic on this forum) sitting in my scrap bin. If the plexi circuit has taught me anything its that these types of circuits really need to throw away a ton of low frequencies before the clipping starts or else they sound more like a bad fuzz pedal. The coupling cap between Q1 and Q2 is only 470pf! Maybe just for shits and giggles I'll pull that out and see how it sounds with that cap in the same position.

[tca]

^ Sorry for that, you are right the first stage is not a buffer! It is worst than I thought . The biggest problem of the plexi-drive is that with Vth=-1.5V and a guitar signal about 100mV the first stage will put the signal almost at 1.5V near |Vth| and after that its all hard clipping (give or take).

P.S. (edit) Take a look:



There is no "dynamics" there.

P.P.S.

You could try and run it at lower voltages 4V and get something like this, this could seem a paradox, but it is not :



You probably should also correct the freq response...

[Keppy]

One thing that hasn't been explicitly stated in this thread: If only one side of the waveform is clipping, then the unclipped side retains its dynamics. Asymmetry provides some benefits here, but once both sides are pushed into harder clipping the dynamic benefits are lost again. In a normal JFET overdrive with no clipping diodes, you could experiment with biasing the drains away from 1/2 supply to achieve asymmetrical clipping. Combine this with low gain and you may hear some more dynamics than you're used to. It's fun to play with, even if it doesn't get you to the specific sound you're after.

As Mark said, once significant clipping occurs, the actual dynamics are pretty much gone. There may still be perceived dynamics due to the effects teemuk and others have mentioned.

[Steben]

Old thread, but some thoughts.

1 You cannot make a Jfet circuit sound like a power amp.
It is impossible. It would have been done over and over already by commercial builders already.

2 Dynamics and "headroom" are very vague things. Many don't see the same meaning. Running the plexidrive on 4V may seem "soft" on electronic emulation pc software, but those rarely express what we feel and hear in reality. Sometimes assymetrical and soft distortion gets labeled "mushy" and "dull" as well. Analyse a fuzz face and you will see soft assymetric distortion as well

[tca]

It would have been done over and over already by commercial builders already.

In a strict logical sense that sentence is false, i.e., that's not the all truth.

I think that what is lacking in all those JFET distortion makers is IM distortion:  The elephant in the dance floor that nobody sees!

Cheers.