Lm386 Based Sustainer Note

[col]

Just a quick 'heads up' for folks messing around with LM386 based sustainer projects (eg. Fetzer Ruby).

The Zobel network in the LM386 datasheet uses a 47n cap and a 10ohm resistor. These values are necessary because the circuit it is intended for full range audio - songs etc. The downside of this is that it doesn't do enough to offset the problem of inductive reactance rising at higher frequencies in a sustainer system.

Explanation:

Assuming a driver with inductance of around 1.2mH (the basic design as developed by Pete), even with the datasheet zobel network, the impedance at 2.5kHz is about 16ohms, at 5k it's more like 40 ohms.

Look at the graphs in the datasheet, you can see that the maximum power available before the distortion (THD) gets high gets much lower as the impedance rises... with an 8ohm load, you hit 10% THD at about 0.75W, whereas with a 16Ohm load, 10% THD comes at about 0.45W.

This becomes a problem as soon as there is clipping in the signal chain - clipping generates high frequencies in the signal. With a standard zobel network, these will cause nasty distortion a plenty from the LM386.

Solution:

Change the values of the zobel components.

We don't need to drive anything like the full audio range, we only need from 82Hz to say 1.5Khz. So if we change the values of the resistor and cap in the zobel to 8ohm and 10u, we don't lose anything we need, and as the signal rises to the point where the inductive reactance of the driver causes the impedance to rise to distortion levels, the zobel provides an alternative route for the current So that the LM386 just sees a nice friendly 8ohm impedance.

The result is that if you drive the circuit hard, instead of a really nasty fizzy distortion, you get a more musical rumbly crunch type of thing (at least I do), so if yo are having trouble with fizz, this might help a bit.

remember, Zobel values: cap 10u, resistor 8ohm.

cheers

Col

[psw]

Good Tip Col...the zobel does seem to be important, some have tried to leave it off

I checked what is in mine and it is a 10 ohm resistor and a 0.1uF cap...plus the 100uF cap that also bridges from pin 5 to the output...I wonder if the value of the outpuct cap has a difference as well...

I admit I just used the values that were working from the old CHAmp kit days...fairly standard data sheet stuff...so no science in the choice other than it works.

10 ohm does seem very close to 8 ohms though but that is a bigger cap you are suggesting.

I may attempt this next time I build this circuit to see what the change of sound is...might be a valuable tweak. Another reason to like playing with LM386's...sop tweakable.

good work!

[col]

...plus the 100uF cap that also bridges from pin 5 to the output...I wonder if the value of the outpuct cap has a difference as well...

Of course the output cap makes a difference.

For a level impedance, and a flat frequency response, you need at least a 220u output cap, but a 470u will give flatter amplitude a phase response at the low end of the guitars range.

The reason you have found the 100u to be successful is related to the fact that your AGC is maybe not doing an aweful lot.

Assuming an 8ohm driver at about 1.2mH, a 100u cap causes the output to drop off below about 330Hz, which is the high E string frequency. The response stays flat up to about 600Hz, and drops off above that.

So what it is doing for you is providing more drive at the 'difficult' frequencies, and less where for various reasons the sustainer works more easily. This sounds great, but there is one problem - although I know you think of it as a 'feature' - frequencies lower than 300Hz or so will not be driven as fundamentals, they will bloom to harmonics. This is because the 100u cap favours the harmonics because it acts like a bandpass filter pushing 300Hz to 600Hz.

It's not a big problem really, the sound of the lower notes blooming to harmonics is natural and similar to what a loud overdriven guitar does anyway. The only other downside I can think of is again the issue of impedance. The drive drops off below 300Hz because the capacitors reactance causes the impedance to rise. This has the potential to again cause problems with the LM386.

I guess that the strange 'modified zobel' setup you have talked about with some sort of feedback might be in some way counteracting this... dunno as I've not seen a schematic. Anyhow, it might be possible to create a very basic sustainer with fairly even response by using a 100u can and having a zobel to counteract the rising impedance at the top end, combined with a chunky inductor to perform the same role for the low end.

AGC will give more control, as you can drive the fundamental of all strings, but for those who want a simple circuit to build, this might be something worth looking into. (the G string will be VERY lively though, as it can be driven easily, and it is not attenuated much by this approach.

cheers

Col

[psw]

Thanks col

Yes, as I say, I mucked around with output caps to get a good drive on all strings...100uF has been consistent, but with the symptoms of low end bloom described. I realize that this biases the circuit towards the treble response and so has trouble sustaining low end fundamentals in any mode (the harmonic mode produces another harmonic in the low end).

I meant, does the output cap have an effect on the zobel calculations you present?

So what it is doing for you is providing more drive at the 'difficult' frequencies, and less where for various reasons the sustainer works more easily. This sounds great, but there is one problem - although I know you think of it as a 'feature' - frequencies lower than 300Hz or so will not be driven as fundamentals, they will bloom to harmonics. This is because the 100u cap favours the harmonics because it acts like a bandpass filter pushing 300Hz to 600Hz.

Yes, my stuff does tend to be a compromise, hopefully towards a 'decent' outcome. I do see it as a 'feature' but am not hyping it up, just explaining the limitations of what it does and how such simple tweaking can even out the response across strings say...less active predominant low strings, powerful drive in the high strings that more matches things. The g and b strings can get lively, but not too excessively, you just need to learn to "play" the things differently...don't hit them so hard, use the power to good effect, dampen as necessary.

But the real trade off was to work towards a decent out come that filled the other criteria I set for "success"...approachable DIY technology, adapatbility of design with low mods to the host instrument, smaller size...given the different guitars and applications I ahve applied it to and that the driver and circuit are both 1/4 of the sustainiac...I do see it as something of a success. Having to put up with this "feature" to have achieved it is fine by me.

With much bigger caps, the high end can struggle (the source pickup can have a difference though to the high end response of course).

...

"My Circuit" is not that "magical"...it really just takes the established blocks of the things I had been doing, so a basic buffered LM386 but when the output across the zobel network exceeds a certain level, it triggers a diodes threshold and turns off the input. Some capacitors slow down the rate at which the source signal is turned on. All of this then is attached to the zobel network effectively I suppose, but only when the thing automatically turns off for a bit. If the level of the strings still vibrating through momentum is still too high, the thing will remain off by recharging the caps for another cycle. If the "drive" control is set to a very high level, it would appear that this cycling is effectively defeated or too fast to have any appreciable effect, at lower levels more so. I had considered that perhaps a similar mechanism could be used to switch through different output caps or be frequency sensitive and therefore "fix the feature" of low string harmonic bloom in the fundamental mode...but it really doesn't bother me and may make such a circuit overly large for my criteria. But I am sure that we have discussed a lot of these detail at the time I was designing them, and certainly some of the circuits that influenced the design. It is nothing like as elegant a solution as the circuit you presented col, and in truth acts much like the more basic circuits most of the time.

Oh...and I know I don't have the knowledge and equipment or the nature to model these things in my head, to 'tweak' the parameters towards a workable solution, I ran the things through a speaker so I could hear the cutting off of the AGC as a kind of tremolo like effect...I just took the values that seemed to give a good result when hooked to my driver design and made them fixed. The only onboard control is a trimmer to adjust overall gain to below squealing...on the tele interestingly enough, it can take the full gain.

...

It's not a big problem really, the sound of the lower notes blooming to harmonics is natural and similar to what a loud overdriven guitar does anyway. The only other downside I can think of is again the issue of impedance. The drive drops off below 300Hz because the capacitors reactance causes the impedance to rise. This has the potential to again cause problems with the LM386.

Yes, my guitars exhibit a very natural "loud guitar" kind of response...that also was the kind of organic sound I was looking for. Often, particularly the low strings, if they are humming along and everything is controlled, the device can sound a little 'sterile' for me...but a fortunate coincidence I suppose.

I am not sure what "problems" you are referring to...distortion?

I am sure that people who know about circuits, obviously far more than me, can derive the principles I came to on my latter circuit, maybe do a little better...the AGC thingy does seem to give a slightly better battery life and I suspect perhaps a little less distortion (at lower levels) for a similar effect as if it weren't there...but it is a very simple thing. Again, it is a credit to the LM386 how easy these things are to work on that you can do stuff like this, output caps, zobel networks and AGC's and such can be constructed to suit pretty easily even with 'doofus' methods like mine.

Of course it would be easy to add a switch to alter the output caps and thereby create different responses perhaps...I have been avoiding these kinds of overly complicated controls for my own use. I suppose this is something of the nature of the Floyd Rose patents switchable caps phase correction circuitry that could be looked into for guidance on getting an even response.

...

I am not sure how important it even is for people to drive the fundamentals either you know...clearly I am not overly fussed and people seem to be attracted to the harmonic and blooming harmonics feature or sounds. I can see how it might be important to some though, in a different musical context than the things I use it for (to simulate a very loud guitars response in most cases and create predictable harmonics) it could be important to me also. For others they may be seeking completely even fundamental polyphonic sustain of a chord, for me 'that might be nice' but it is also cool to hold a chord, play the higher strings, clean and natural without sustain and have the lower notes gradually morph into harmonics above the chord.

I am not sure that you can have 'everything' that this technology can do in one system, or that it is necessarily worth the efforts for most peoples uses to aim for some of these more difficult responses. If a simple mod of the output cap provides a better string response and a 'feature' of multiple blooming harmonics in the lower strings, I am kind of content with that myself as a 'decent' solution.

[col]

I meant, does the output cap have an effect on the zobel calculations you present?

No. The response of the Zobel will be the same, however, changing the output cap will change the overall response, so depending on the value, one might want to tweak the Zobel response to compensate.

With much bigger caps, the high end can struggle (the source pickup can have a difference though to the high end response of course).

No, you have this upside down. A bigger cap will have no effect on the high end. It will however help the low end, so if you don't have AGC, you have to turn the gain down because the low end is too strong, this compromises the high end. If you have a functioning AGC, this is not an issue!

so a basic buffered LM386 but when the output across the zobel network exceeds a certain level, it triggers a diodes threshold and turns off the input. Some capacitors slow down the rate at which the source signal is turned on. All of this then is attached to the zobel network effectively I suppose, but only when the thing automatically turns off for a bit. If the level of the strings still vibrating through momentum is still too high, the thing will remain off by recharging the caps for another cycle. If the "drive" control is set to a very high level, it would appear that this cycling is effectively defeated or too fast to have any appreciable effect, at lower levels more so. I had considered that perhaps a similar mechanism could be used to switch through different output caps or be frequency sensitive and therefore "fix the feature" of low string harmonic bloom in the fundamental mode...

Sorry, Pete, but none of this makes any sense. You seem to be trying to describe some sort of AGC, however, if you had a working AGC, you wouldn't need a 100u cap to even out the response.

Also, a Zobel is an impedance adjustment network, if you start attacking feedback lines from it, its not a Zobel any more, so to call it one is misleading.

You aren't going to be able to 'fix the feature' of low string harmonic bloom until you switch to a bigger output cap. A 100u cap favours the harmonics of the lower strings by something like 10db over the fundamental. To 'fix' this you need a level frequency response, and AGC to even out the amplitude response.

...

Again, it is a credit to the LM386 how easy these things are to work on that you can do stuff like this, output caps, zobel networks and AGC's and such can be constructed to suit pretty easily even with 'doofus' methods like mine.

It's no credit to the LM386. A Zobel can be added to any amps output. Unfortunately, the response of the LM386 is so sensitive (bad) that a zobel is essential. AGC can also be applied to the input of any amp.

I'm not sure about your circuit because I haven't seen it, but from the descriptions, and things you've said about the way it sounds, there 'aint any AGC happening in there - you wouldn't be getting the best response with a 100u output cap if you had functioning AGC. EDIT, I'm not saying that it doesn't work, or that your feedback circuitry doesn't help, just that it isn't AGC (at least in the usual sense of the term).

I am not sure how important it even is for people to drive the fundamentals either you know......but it is also cool to hold a chord, play the higher strings, clean and natural without sustain and have the lower notes gradually morph into harmonics above the chord.

Yes I said as much in my post. However, it is also cool to strum on a low chord and for the low notes to be driven as fundamentals, but the high end to bloom to harmonics, so you end up with a deep fundamental bottom end simultaneously with high pitched harmonics. My system can do this, but it's not possible without a 470u output cap and AGC.

I also get reasonable sustain of clean chords, but I can switch to blooming harmonics if I want.

cheers

Col

Edited May 14, 2010 by col

[Fresh Fizz]

Just a quick 'heads up' for folks messing around with LM386 based sustainer projects (eg. Fetzer Ruby).

The Zobel network in the LM386 datasheet uses a 47n cap and a 10ohm resistor. These values are necessary because the circuit it is intended for full range audio - songs etc. The downside of this is that it doesn't do enough to offset the problem of inductive reactance rising at higher frequencies in a sustainer system.

Explanation:

Assuming a driver with inductance of around 1.2mH (the basic design as developed by Pete), even with the datasheet zobel network, the impedance at 2.5kHz is about 16ohms, at 5k it's more like 40 ohms.

Look at the graphs in the datasheet, you can see that the maximum power available before the distortion (THD) gets high gets much lower as the impedance rises... with an 8ohm load, you hit 10% THD at about 0.75W, whereas with a 16Ohm load, 10% THD comes at about 0.45W.

Yes, but the same input signal doesn't distort more with a 16 ohm load. There is only less efficiency with a 16 ohms load. It works the other way around, the bigger the load impedance the more headroom. More voltage but less current.

This becomes a problem as soon as there is clipping in the signal chain - clipping generates high frequencies in the signal. With a standard zobel network, these will cause nasty distortion a plenty from the LM386.

Solution:

Change the values of the zobel components.

We don't need to drive anything like the full audio range, we only need from 82Hz to say 1.5Khz. So if we change the values of the resistor and cap in the zobel to 8ohm and 10u, we don't lose anything we need, and as the signal rises to the point where the inductive reactance of the driver causes the impedance to rise to distortion levels, the zobel provides an alternative route for the current So that the LM386 just sees a nice friendly 8ohm impedance.

The result is that if you drive the circuit hard, instead of a really nasty fizzy distortion, you get a more musical rumbly crunch type of thing (at least I do), so if yo are having trouble with fizz, this might help a bit.

remember, Zobel values: cap 10u, resistor 8ohm.

I've been playing with all the parameters in a spreadsheet. A 10 uF cap for Zobel keeps indeed the overall load impedance between 8 and say 11 ohms. Good work! It should surely help when the high frequencies take the zobel route.

You got me thinking. Have you ever thought about an ultralinear like design? Create a negative feedback loop with a tap on the driver coil.

cheers

FF

[col]

Yes, but the same input signal doesn't distort more with a 16 ohm load. There is only less efficiency with a 16 ohms load. It works the other way around, the bigger the load impedance the more headroom. More voltage but less current.

That makes sense, but then why does a zobel network help ?

What is it about the amp that requires it to feed a resistive load?

cheers

Col

[col]

You got me thinking. Have you ever thought about an ultralinear like design? Create a negative feedback loop with a tap on the driver coil.

Do you mean a 'current mode' amp? with a 'current sense' resistor between the driver and ground to provide a feedback current?

I have attempted this, but I tried to do too much in one go (I messed around with driver specs at the same time)... So it's still to be tested properly. I attempted it feeding a very low current back through pin1 of the LM386 via a decoupling cap.

cheers

Col

[psw]

Oh well...out stripped my technical terms and my ability to concentrate with the flu I've got. It does indeed "compress" and "limit" the signal, and was based on known designs for that purpose, hearing it with a speaker does show this effect. Not sure if you consider this AGC.

Anyway, the thing is an improvement, but still nowhere near perfect. The last I heard of you system col it was an effect that seemed extremely "mild" and controlled, but that is possibly the effect you were after. I am sure that a lot more could be done and you are right of course, unless I up the output cap value I am not going to get a true bass response and so drive. I'm not that concerned about "fixing it" at this stage.

I have not heard, nor quite understand, except by biasing the frequency response of the circuit through filters (much the same effect as my dirty output cap solution) where you have low end fundamentals with high end harmonic bloom...sounds cool though.

Not sure about it being wrong that it improves high end response...that is the reason to change to lower values...the higher output caps did not produce adequate drive and quickness of response at any gain levels, it was not to even out response of over active low strings, though this is a positive side effect to a degree. Lower gains will make the fundamental sustain longer on the low strings before blooming.

But yes, I have always described my circuits "AGC" as a very simple thing that is defeated anyway by high gains and until switched, it leaves the zobel as per a standard typical amp. This is of course why I am confident that this project can "work" with simple circuits and the LM386 circuits remarkably well. The LM386 is not the "greatest" thing to use, but with SMD's taking over and all kinds of quirks, it has been the best and most versatile that I tested...I am sure there is better 'fidelity' to be had if one has a mind to search them out. Quite a few have been doing this including me and still there seems not to be a match to replace it in this application or general small amp designs...witness things like the "noisy cricket" and the "gem", "Ruby" and any other number of 'designs' floating around that use it. You don't find an equivalent with any other chip in terms of usage.

Anyway...looking forward to trying this tweak one day...my designs have always sought to be as small as possible...at around an inch square, I could make it a little bigger, but this compact size has always been important to my criteria, just like compact drivers and I personally am unlikely to go or have the ability to go to a vastly more complex or larger design.

Will be interested to know what you guys come up with though. I do wonder though the influences of the drivers as well, both FF and cols are dual coil systems and I expect quite different from the simpler single coil drivers I have generally used. I noticed quite a bit of difference with things like both my bi-lateral and compact rail drivers...seemed to need to be extremely close to the strings in comparison and not to drive things as hard though the circuit and guitar remained identical.

[col]

I do wonder though the influences of the drivers as well, both FF and cols are dual coil systems and I expect quite different from the simpler single coil drivers I have generally used.

you expect quite different what?

I noticed quite a bit of difference with things like both my bi-lateral and compact rail drivers...seemed to need to be extremely close to the strings in comparison and not to drive things as hard though the circuit and guitar remained identical.

Did you design the bi-lateral and rail drivers carefully so that their permanent fields were of similar strength in the area around the strings, and so that they had the same overall inductance as the single coil you were comparing them with?

If you didn't, then your comparison cannot tell you ANYTHING about the relative merits of single coil, bi-lateral, or rail drivers.

I could easily knock together a single coil drive that work much less well than my rail driver - equally, I could make a rail driver that works less well than my single coil.

If you were trying things out without taking measurements (DC resistance doesn't cut it I'm afraid), then your results cannot be used to generalize beyond those particular drivers.

cheers

Col

[psw]

No...I was not advocating one over the other or any direct comparison, just that the various types seemed to be quite different animals. It is not about the numbers and measurements that I have already conceded to you, but the action of the driver with multiple coils and magnetic polarities...ideas about "throw"...opposite magnetic rails tend to be attracted to each other for instance...

But, I am not anti- multi coil devices...remember I'm the guy who spent a year pursuing six coils and 12 magnets and such. I advocate the present single coil design as a good entry level and working device, it does the job. I have not heard that a multicoil design that was significantly better running from a simple circuit, I have not heard of side by side tests with the same circuit and guitar...things that I have at least tried to do.

Much of the original ad vocation of multicoil systems were to address EMI issues, however, it is possible to build single coil drivers that do not suffer badly from such effects. My experiments in such designs were in an attempt to use such qualities to push the boundaries further than what the single coils designs can't do...like a mid-position coil. To justify the size (and all that typically entails like guitar mods and or losing the neck pickup) of a multicoil driver, it would have to exhibit necessary qualities that the single coil lacks (not related to circuits, AGC and such) in order to meet the criteria of overall success that I demand for my work.

The neck pickup is vital to much of the things I use my guitar for, often combined with the bridge pickup, and I just can not justify a dedicated guitar to produce this "effect"...the reality seems to be that this is less important to others active in the project at an advanced level. So, I cant see how I could commit myself to a dual coil system...add to this that they are at least twice the work to construct...and I would have to see and hear some significantly better performance over a single coil compact design under the same conditions.

I do think though that they need to be developed further and a 'formula' presented as to how to build the things before I personally advocate them to the kinds of posters typical here. I understand you new directions, but I don't know exactly what you would be advocating for people to replicate your results, particularly with the skill set and expertise that seems evident. The idea of dual coil systems have been around for quite some time now after all, but I have yet to be convinced that it is "in fact" a better way to go at this point or that the costs are worth it to someone like me.

But, looking forward to be proven wrong...

I appreciate your efforts with your new found inductance meter, and that some valuable improvements like the subject of this thread will improve things for any style of driver, but I simply don't have nor am likely to engage in a number crunching war about it. That you have given measurements for the driver typical of what I and others make, I assume you have actually build one and measured it? But the real crunch comes with testing such devices side by side and objectively annotating the pros and cons...it might "work better" (not sure what the criteria for this is) and it could be skewed that way, but I have to consider a lot of other factors. I hear someone proposing huge digital circuits and six amps and plugging the guitar into a wall...and I just shudder...my criteria is quite the opposite, compact, low impact, no loss of pickups, cheap, replication, desirable, works...

As for my circuits, I am sure that you can and have made far more sophisticated compressor circuits, but I have used elaborate limiter/compressor preamps for years and years as I was reminded by a recent old photo that predates me even having a digital camera (back then that technology was rare and there was no reliable picture hosting which accounts for a poor record of what was actually bing done. The pictorial for instance was the first thing I used the digital camera for). I know it is "compressing", I can hear it and I used established principles when it was designed. It is clear that presenting it will do nothing for anyone and only attract more criticism of my feeble attempts with a lot of hyperbole, or indeed, just modifying it and calling it their own. I find it remarkable that except for a few notable exceptions, for all the words and such, we are still talking about the LM386 and that no one really has come up with a better basic amp for people...obviously I am no0t the one to do that, especially as the competitive spirit is such...but also, that many people have made a huge deal of their expertise but in the end come up with virtually the exact same thing, also with faults, that has been around all along!

So, I think after all these years, and present company partly excepted, it is time for some others to step up and prove their devices and advocate for their features in line with the pros and cons and costs involved...as a total project, not just a single pickup dedicated guitar with any amount of space for circuits and such...these are the things I have always had to design around.

[Fresh Fizz]

Yes, but the same input signal doesn't distort more with a 16 ohm load. There is only less efficiency with a 16 ohms load. It works the other way around, the bigger the load impedance the more headroom. More voltage but less current.

That makes sense, but then why does a zobel network help ?

What is it about the amp that requires it to feed a resistive load?

cheers

Col

Some stability issue, keeping the load impedance low and real: Boucherot cell

I've never bothered to try but perhaps the zobel could be omitted. With a tiny LM386 there isn't that much at stake. The self resonance of the driver must be in the hundreds of kiloherzes. I can remember Hank McSpank wasn't able to trace the resonance frequency when performing tests for the piggyback driver.

And the self inductance of the driver coil isn't that dangerously high. It's nothing compared to that of the primary coil of an output transformer of a valve amp. If a valve amp isn't loaded with speakers or a dummy load then the induction voltages can destroy the output transformer or power valves.

Do you mean a 'current mode' amp? with a 'current sense' resistor between the driver and ground to provide a feedback current?

More like this:

cheers

FF

[Fresh Fizz]

@col and psw

what if one of you guys started a new general sustainer driver discussions topic?

And keep this thread for lm386 related issues only.

FF

[psw]

FF...where does the arrow from between the two coils go...perhaps you could explain what is being achieved here?

I have found as others have reported to, that things like the zobel network are important. Tweaking the values though may well produce improved results.

Originally, I had even used an LM386 with an output cap alone and got things working (the original single coil drivers and ansils sustainer mod are this basic) but I did find that if using the things at high gain, and in line with the data sheet for such applications, a cap between pins 1 and 8 and from 7 to ground does help...in the CHAmp thing I originally used, these were all 10uF and have had no reason to change from those values.

Point taken FF, my only real query about the drivers is that is there a difference in the values of things such as the zobel with different output caps or indeed radically different loads...ie drivers. So, what may be expected to work with one type such as a dual coil system like yours or cols...may not be the same for single coil drivers such as the ones I typically use and is most commonly made.

As is often the case, or my failure to communicate this, it was taken as an advocation of the single coil design over dual coil designs or personal bias. It seems to be a common theme that if I use a circuit I believe it is "best" or that the LM386 is the bees knees, or to ignore that some people are working with different criteria for their projects.

In this regard, for years my criteria have been publically listed...these criteria are at least as important an influence on things for me. The LM386 has been at the forefront of circuits because it is so simple and cheap and available everywhere...and somewhat adaptable. This is not the case for other options often spending weeks obtaining a similar chip for $15 than say $2 at a local store for an LM386 here, parts for my more recent circuit ahve been increasingly difficult to obtain as well, and very tricky where I now live...though I dare say I'd still find an LM386 about. As the public project has advocated the use of such simple circuits that are obtainable to the average DIY'er the LM386 has persisted.

Anyway...will be interesting if people were to try out this zobel mod on an existing simple sustainer and see if it makes for an improvement...

[col]

Point taken FF, my only real query about the drivers is that is there a difference in the values of things such as the zobel with different output caps or indeed radically different loads...ie drivers. So, what may be expected to work with one type such as a dual coil system like yours or cols...may not be the same for single coil drivers such as the ones I typically use and is most commonly made.

A has been explained MANY times before, the 'load' from the driver depends on the combination of its inductance, its DC resistance and the signal frequency. whether it has 1, 2 or 74 coils is irrelevant. A zoblel that works for an 8ohm 1mH single coil driver will work just the same for an 8ohm 1mH dual coil driver or an 8ohm 1mH 17 coil driver made from unobtanium by aliens from uranus.

In this regard, for years my criteria have been publically listed...

Again as has been mentioned many times, this is NOT TRUE.

You have never posted a diagram of your driver circuit. This is significant, because its amplitude and phase response will have an impact on how different output caps effect the overall result.

e.g. It may be that a 100u cap works for you precisely because of an unusual phase or amplitude response from your driver circuit.

IMO, you should be clearly stating this as a caveat when you recommend various selected component from your system. Unfortunately, in this case, you seem to be doing the opposite.

You are very quick to criticize the abilities of "the kinds of posters typical here".

"I do think though that they need to be developed further and a 'formula' presented as to how to build the things before I personally advocate them to the kinds of posters typical here. I understand you new directions, but I don't know exactly what you would be advocating for people to replicate your results, particularly with the skill set and expertise that seems evident"

Maybe the real problems is that they are doing a great job building adequate drivers, but having no ultimate success because, crucially, the driver and output cap are only one part of the system, and the rest is just as important and in your case is *secret*.

I for one have tried a well built 'to Pete's spec' driver with a fetzer/ruby, and I can tell you the results are not acceptable to me.

Your normal response here is to state that you don't and never have recommended the Fetzer/ruby... but this time before you do, maybe you should consider where that leaves folks who want to follow your philosophy that a sustainer needn't be complex and that fancy AGC circuits are not necessary... that's right, it leaves them with no alternative to the Fetzer/Ruby!

cheers

Col

Edited May 16, 2010 by col

[psw]

In this regard, for years my criteria have been publicly listed...

Again as has been mentioned many times, this is NOT TRUE.

I'm afraid the criteria has and been repeated for many years...

The so called "secret circuit" only appeared two years ago, not long before everything was closed down and before that I have used the CHAmp and PreCHAmp circuits and a whole raft of alternatives many photographed, some using chips that have since been recommended...

A bit of "poisoning of the well" logic amongst others, I didn't recommend the F/R that has continuously been attributed to me...that would be outright plagarism to pass off these mash up circuits as ones own...yes, just like Hank did with his chip and tillman and most other things proposed...but then it is not that much different than many circuit I used and got results with for years...a buffered LM386 circuit...I stated that is what I was using, G-mike came up with the F/R solution, no one else bothered to come up with their own.

I for one have tried a well built 'to Pete's spec' driver with a fetzer/ruby, and I can tell you the results are not acceptable to me.

Yes, but as you have stated since joining this forum, you have sought a completely different response than I have and your posted sound clips sound completely different in many ways...so each to their own...

Many people clearly have built these things to spec with success on their terms and I have posted my results and what the thing sounds like.

My criteria for "success" is not achieved by drivers so big that they prohibit a neck pickup, require a lot of room for circuitry, have to plug into an outboard power supply and a range of other ambitions discussed years before you guys turned up in the Hex era...so, in fact THIS IS TRUE.

It is worth noting that the continued lurker HmcS had the same tone and ridiculous suggestions to promote elaborate hex systems and the like, only to settle for something remarkable the same as what I have been doing with the simpler things all along, and then trying to sell the things! When challenged about the piggy back driver thing...he builds one remarkably similar to the kinds of things I dhave for years, and originated in concept...oh, and that works too!!!

I have suggested that any number of small amplifiers can fit the bill, that as far as cheap and easily obtainable the LM386 fits the bill and has shown to work over and again...but I am not going to endorse the F/R in it's present form or even promote my own circuit as "the solution" when there are clearly so much animosity and it is against my beliefs that there is even one "best circuit" or my circuits alone are the kind of things that work...

So...now...out of nowhere, you say you have made an F/R with a 100uF output cap and preferably the mods I described for it, and a driver like mine that I am only too aware that someone like you could have done in 10 minutes with materials you are known to have on hand, able to test it with your own new inductance meter and get direct comparisons with your own work...yet you want to pressure me into some crazy dance about inductance figures?

Are you suggesting that you were unable to get a response on all strings, that the thing does not work?

No, it does not fit your understated goals and criteria. You have never considered it important that the guitar have any more than a bridge pickup, I've seen no work that even shows how to implement such systems.

So...you don't like the response of these things, it's not to your taste? We'll each to their own really, there is no "correct" response to things...is there a perfect distortion pedal, if I promoted such a thing would anyone not eviscerate me for making such claims?

There is obviously a misunderstanding on my question about the action and qualities of the load...not the load (simply in terms of inductance)...of single coil and dual coil and even hex coils...but your argument screws that to simply throw a few mocking 'unobtanium and uranus' shots...helping no one and again threatening any threads on this subject...clearly a good tutor there, well done col!

I would expect that a two coil design with say two 16ohms of wires, many, many more turns...to have quite different resonant frequencies and interactions to a single coil driver...I spoke of the "load" as the driver the circuit is designed to drive, not just reduced to the things measurable on you inductance meter. So, it would appear that this new found measuring devices main purpose is to imply that because I don't have one or don't work in the McS approved manner (that ended up with a remarkably similar solution regardless) is largely to show up my inadequacies on technical terms and "hard data" and not to forward the project at all or actually produce something convincing in the time spent mocking me and the work so far on those grounds. Well, hope that this is the acceptable result of such expense and time on the project, for me it was to actually create these things, but this impetus seems to have long ago become a side line to this "main event"...

I know, and common sense will show you, that magnets in close proximity will interact...they will attract and detract...so with electromagnetic fields...so the interaction of multiple drivers will generally result in a different thing...the question is whether such things as you put forward could be expected to have a difference...

You are very quick to criticize the abilities of "the kinds of posters typical here".

"I do think though that they need to be developed further and a 'formula' presented as to how to build the things before I personally advocate them to the kinds of posters typical here. I understand you new directions, but I don't know exactly what you would be advocating for people to replicate your results, particularly with the skill set and expertise that seems evident"

I've not added the McSeemian shouts...but the reality is that the typical posters have tended to not stick to the formulas and basic procedures for making any driver, they have attempted to pot coils in playdoh and run 20 watts into them to compensate...a huge amount of the original threads were trying to convince people that these things are not the way....spawning a host of latter day detractors who claimed that people generally had problems with the actual project...oh, i couldn't get the buffer transistor so I left it out, I had pickup wire so I used that...does it matter if I just tie the broken pieces together...come on col...these are the typical posters of late, are you seriously saying that suggesting that someone goes straight in to winding a dual coil driver without any explanation about wire gauge or coil values and how to wire them to the appropriate load is some how better than my specific instructions given for a working basic system?

So...yes, typical posters do give the impression are lacking in some of the skill set and expertise to make this successful in it's basic form, let alone some glib...I'd make a dual coil driver but I won't tell you how, response for which I made that remark! I still stand by it also.

It may be "quick" but I do spend a lot of time trying to encourage people to stick to something that has shown to "work" and for which they intended to embark. It may not meet your standard...but then, what is that standard and how do you know what you seek is what others want. Al S started a thread recently that seemed to seek a lot of what my systems typically aim for in response...almost the opposite of the pure fundamental sought by you. Neither are wrong...but as far as criteria goes, the reasons for dual coil systems in conventional applications was originally and primarily to address EMI problems...if a good compact single coil driver can adequately cope with it and meet the range of other criteria I ahve marked for my own success...then that is enough for me. I don't build to your criteria, I can't even assume what they are, but I have made mine clear and consistently worked to those aims. The single coil systems so far...and before that the ultra compact hex systems that were smaller...were the best ways I could come up with to achieve that.

Also, I have made dozens and dozens of these, I have had them independently verified and even endorsed by an american custom maker on a spare no expense guitar having never met me...and it met the criteria sought there too. I sent a hex driver and explanations to LK who confirmed that the things worked and the principle that was suggest was breaking the "laws of physics" was in fact sound...but then I got this same language back then because the 5 LED's in some of them were assumed to be the things driving the damn strings.

...

Ok...so you have some work in the works? Some amazing new driver and circuit are going to be presented that is going to blow me and everyone else away? I'm so looking forward to it...but I have not seen anything like this in a very long time. Those who suggested I was being unfair to Hank should have a good hard look at the end results for all teh hyperbole and abuse...no hex drivers, no innovations for all the scopes and measurements and a years of arguing and abuse of my work...he comes up with a single coil driver and a basic amp and a plagiarized tillman front end...couldn't even design an amp circuit from scratch, let alone purpose built for the thing...and yet, as one might expect from something based on a known design (copied even down to the 5 LED's from 2005!) it worked. And, when challenged about speaking about the piggy back driver that "couldn't work" he makes one in a day and it does work...however of course, the same argument put to me now, right here, and in much the same way...oh...but not to my unspecified criteria.

Oh well...when I see something come of all this, when a full system is proposed by anyone else and fully disclosed...well then, I'll be a happy man. If it can not accommodate my important criteria (leaving the functioning of the host instrument in tact) I will probably still be desirable of a compromised system that gets the job done.

Of course "fancy AGC's" and the like that I appreciate you have a huge investment in, are necessary to get the kind of response you apparently desire...but they are not necessary for infinite sustain on all strings, blooming harmonics, harmonics on all strings or any number of the things the systems without these can and do achieve and been independently verified again and again. Though I greatly admire your own work col in this area and always cite and deffer to you on such matters, or indeed the patents before you, your desires and biases for 'how these things should be', your apparent subscription to the old "perfect sustainer myth", your non consideration of peoples desire not to have extensive mods or loss of pickups or to desire something other than what you consider to be "acceptable" does not mean that those who have made a successful simple driver, yes even with the F/R, did not find what they created was not acceptable...TO THEM.

I'm not sure where this new round of vitriol towards me is coming from, but I suspect more would be gained if you simply put these words into practice and blew us all away with your version of how things should be and acceptable to you, explain why, show how it is so good that it justifies the space required for the circuit or the loss of something like a neck pickup...and then spend countless hours coaxing the typical poster on the matter that playdoh is not acceptable, don't wire circuits together with winding wire, and defend yourself when they fail and blame the project of the guitar that these naive attempts are attached to...then come up with the big reveal...oh, so I rand 20 watts into it and it still didn't work. These are the "typical posters" for which I speak, and I ma not overly quick about it...I have spent years discussing the same things over and over again...yet exactly the same things are repeated adinfinitum. And then the myth that 'it doesn't work' is perpetuated in mocking tones without any qualification (are you suggesting that the thing did not sustain, or just not to you acceptable unstated criteria?) by people who should know better and frankly do know better than that col.

So, time to put these theories into action, state clearly what you are aiming for, and present how you achieved the "ultimate sustainer" and watch as people line up and suggest they wanted or expected something different!!!

[col]

Some stability issue, keeping the load impedance low and real: Boucherot cell

Cool, I'll check that out (again) EDIT: that makes sense,

More like this:

cheers

FF

That looks(?) like a current amp to me (assuming the in signal goes to the LM386 +ve input and the feedback line goes to the -ve).

this is a very good article on the subject.

I messed about with this idea in simulations taking the feedback to the LM386 input with poor results.

Then after a look at the internal diagram of the LM386 in the datasheet, saw that I could take the feedback directly to the input of the push-pull stage of the LM386 via pin 1. This worked very well in simulation, and the test circuit I built seemed to work fairly well, but I still don't have a scope, so haven't been able to test it thoroughly enough to be confident about committing to it.

Here's the basic idea:

You would have to either low pass filter the input signal, or stick a cap and resistor from pin1 to pin 5 to low pass the amp. otherwise frequencies above 2k or so will cause clipping as the peak to peak voltage required to maintain the current is too much for the LM386 with a 9v supply.

EDIT: or of course, use a 'boucherot cell'.

You could use a coil tap instead of the 0.1ohm resistor, but I think you would get into trouble with precision issues, and you would gain so little in efficiency that it's not worth the hassle.

I can't remember if I ended up using an 0.1 resistor or slightly bigger... it was very low resistance anyway.

A variation of this might work quite nicely as a Ruby style amp going by that esp article - he talks about using the approach to get closer to the response of a valve amp!

I'm afraid the criteria has and been repeated for many years...

The so called "secret circuit" ... 8< snip

It's simple Pete, I'm annoyed because you claim to have made your specs public. This frustrates me because it is misleading, and may have an impact on anyone new who comes here and tries to build a sustainer - I explained why in my previous post.

You don't seem to have a straightforward answer to this, so I'll leave it there for now in order that we can try to keep things on topic.

Cheers

Col

Edited May 16, 2010 by col

[Fresh Fizz]

That looks(?) like a current amp to me (assuming the in signal goes to the LM386 +ve input and the feedback line goes to the -ve).

'...'

I messed about with this idea in simulations taking the feedback to the LM386 input with poor results.

Yes, a current amp and connected like you assumed, or I could stick an inverting opamp in front of the LM386 to increase the open loop gain (inverting opamp as mixer of input and feedback line). Perhaps the open loop gain has to be a bit more than 200?

I don't know if I'm right, but this is what I was thinking:

Let's say we have a tap at 10%. (upper coil 90%, bottom coil 10%) Total self inductance is 1.2 mH, resistance is 8 ohms.

First component of self inductance based on ratio of turns compared to total amount of turns

Upper coil is (0.9)² x 1.2 mH = 0.81 x 1.2 mH

Bottom coil is (0.1)² x 1.2 mH = 0.01 x 1.2 mH

But the total self inductance is 1.2 mH, not 0.82 x 1.2 mH

Mutual inductance is 1 - (0.81 + 0.01) x 1.2 mH = 0.18 x 1.2 mH

Mutual inductance equally divided across both of the coils is added

Upper coil : (0.9 x 0.18 + 0.81) x 1.2 mH = 0.972 x 1.2 mH

Bottom coil: (0.1 x 0.18 + 0.01) x 1.2 mH = 0.028 x 1.2 mH

For low frequencies (if open loop gain = ∞ ) gain = 11

For high frequencies gain = 0.972 / 0.028 + 1 = 35.7

So there will be some compensation at high frequencies but never more than 35.7/11 times (some 10 dB)

My next driver will have a tap so I can check this out.

No man overboard if this idea proves out to be totally incorrect.

Take it easy, guys.

Cheers

FF