col

That's a shame, I was looking forward to having someone with a more scientific approach to share ideas and results with. well, good luck anyway Col

Sometimes its good to clear the air, but arguing isn't pleasant - sorry folks So here's some more positive news I've been working for ages on trying to get a circuit with low parts count that doesn't compromise too much on performance. Well, today after my last agro post, I spent some time again working on this. I have got a circuit that _seems_ to be just about what I was hoping to achieve. WARNING: I've net tested it IRL yet, so there maybe some gotcha or other (although so far everything that has simulated has worked exactly as expected when soldered up) It's a basic circuit with input buffer, virtual ground buffer, 'brickwall' limiter and LM386 power stage. Parts count is about as low as I think its possible to go without compromise and/or specialist hardware. phase response is similar to other circuits I've used. 30 discrete components, 3 op-amp sections (leaving one free for adding a harmonic mode switch at some point), and the LM386. This has the option of an external sensitivity control and a drive control. The AGC reaction time is fast enough (I hope) at around 60ms, and the recovery is similar. The AGC clips during that short time which should protect the power stage from melting. The clipping will only last for that 15th of a second, so hopefully it will not be noticable. It is pretty good at squashing everything in the range we will need to exactly the same level. Minimal distortion - this could be improved slightly at the expense of reaction speed, so I'll need to wait until the real world tests before getting too excited. So, fingers crossed. If this one works well, it will be relatively compact, and easy enough to build. cheers Col

I've never claimed these Ideas were my innovations, however, I do still stand by the fact that when I started here, you were still using the LM386 set up with very high gain. There's with the setup you had that you were using a drive signal. At the same time others were having lots of trouble with squeal, fizz and noise. My suggestion of using the minimum gain settings on the IC helped a few folks at that time. It was also this assertion that preventing clipping was important that drove the AGC development forward. I'm well aware that you and others discussed AGC and attempted to use various existing compressor circuits. However again, when I joined (about 2 years into the project), this area had not been developed, I believe because it's importance was not realized, other than by LK who IIRC was only advising, not actively developing in this project? I was in discussion with LK about AGC when he tragically died. He was going to send me some part to try. In the same way, your only innovation as far as I can see has been the piggyback design - the rest all existed prior to your work, so from that point of view, we're pretty much in the same boat (and you can be certain I won't be trying to take credit for stacking a driver on top of a pickup). We and others have all brought ideas to the table, most of those ideas have not been original. However, their application to the project, and the development work involved in going from a concept to having a functional device that improves on previous devices as you know is not trivial. From that point of view, I still feel justified in taking some credit within this project for pushing both 'clean/no clipping signal' and 'AGC'. I just wish Some of the other folks who came with different ideas had persevered a bit more and taken their ideas to their conclusions. cheers Col

Can you provide any evidence that involves measurements or even references in 'the literature' that can substantiate any of this stuff about 'thin' designs being "faster"(whatever that means). And the size of the core can be useful at all without also knowing the core materials other properties such as permiability, remnance etc? (I know its not possible to obtain values for the second hand core materials we use, but it IS possible to test a variety of cores in the same coil, e.g. measuring their inductance. This will give useful empirical data rather than less useful subjective information - like you get from listening tests) OK...well, I'd like to see the "data". I and others for instance have wound a single coil without blocking it up to the 3mm or similar with the same wire, and then the exact same thing with a thinner coil with a remarkable difference. You yourself have indicated in the past "ideal" coil dimensions as I recall of depth, width and such in which the thin coil falls pretty close. I this is so, that's good, but it is not what my experimentation indicates...above a depth of 5mm there are problems. I wouldn't class a depth of 5mm as 'thin' for a coil of only 100 - 200 turns. Neither would you have a year or two ago. Aww, the straw man raises his ugly head Saying that thinness doesn't matter is NOT the same as saying or implying that any coil will work. Neither is it saying that any old wire will work. What we need to do is explain what does really matter based as much as possible on empirical evidence. We are at the stage where using the term 'thin' in a way that implies that thinness is a crucial factor for a system to work is getting in the way IMO. Nothing wrong with a driver that is thin, but its the other properties that make it work or not. I can make you a thin driver that doesn't work and a not thin one that does work (I'll only do it if you pay me but I CAN do it) Agreed, the LM386 may be 'shitty', but is has been proven to work. It would be interesting however if we could look at the performance of our different LM386s I'd bet that some are better than others. That kind of information is not really very useful in engineering (which is what we're doing). Its great to build lots of versions and try them out, and it can - as you have proven so well - get you to a system that works well. The kind of data that Hank is talking about is precise measurement. Hanks sort of data lets you take two systems that both work well, then find out whats good and whats bad about both, you can then keep the good and ditch the bad resulting in an even better design. In that way, I took your coil specs, studied some theory, worked out why they work and others don't seem to, then after a few false starts due to me not being a genius at either maths or physics, I came up with some compelling reasons why it [your driver] works. Since then I have taken that understanding and used it combined with empirical measurements and simulation (based on accurate numerical modeling) to design a driver that so far is performing better, even though it is not thin, has heavier gauge wire, and was much easier to build due to it's conformance with existing pickup bobbins and magnets. You don't need to be wise or clever to understand it - at least as well as I do. As long as you are willing to accept the word of Many people who have really studied the field, some of whom really were true geniuses. The inductance does not change with frequency. The inductive reactance does - that is the part of the impedance that is caused by the combination of inductance and frequency. The impedance is the combination of the DC resistance (your 8ohm value) and whatever extra ohms of resistance that are being caused by the inductor at whatever frequency the signal is at... You don't need to know the maths to use this knowledge. You do need to know what the inductance of your coil is, and you do need to have a URL for a calculator that does all the hard stuff for you Why not spend a few minutes using that calculator, I guess your driver will be between 1mH and 1.4mH, try those with a resistance of 8ohm and various frequencies to get a feel for it, then try taking the DC resistance down to 4ohm, notice how the final all important 'impedance' is over a wider range now for our purposes (82Hz to about 1200Hz is what we're interested in IMO). Now try with 8ohm and say 3mH... wow, look how that effects the higher frequencies in the range we use... That's it - it's as simple as that.. the impedance is what the LM386 has to drive, a higher impedance means less current which means weaker drive. So for a voltage amplifier with an 8 ohm driver, as you can verify with that calculator, your coils specs (other than thinness) are pretty close to optimal (which is a damn good result for an informed trial and error approach). With This understanding, and the Zeigler pickup calculator you can then try out imaginary coil specs. You use the calculator to get the turns vs DC resistance, then use some simple arithmetic to work out your inductance for you imaginary design. first the square of the turns count for the unknown coil divided by the square of the turn count for the known coil then multiply that by the inductance of the known coil e.g. if 150 turns around a particular core gives 1.2 mH, 300 turns will give ((300*300) / (150*150)) * 1.2 = 4.8mH This process is not super accurate, because coil linkeage will be variable, and our hand winding approach means every coil will be a little(on the best of days) different but you will get 'ballpark' figures that you can easily hone in a final development stage. The only kit you need for this is an inductance meter which can be had for as little as £10. The meter is great because it also lets you see just how much of an effect different core materials have (and it can be a big effect). So you can tweak the inductance of your driver by reducing or increasing the mass of the core after you've made the coil in order to get as close as you need to your goal inductance.... Remember though that if you use a big chunky lump for a core you really must use laminations. If the data is gathered accurately using measurement equipment from a system made from components with known specs, then it is not misleading. If your initial hopes based on that data turn out not to work, you can then use that data to understand why and to inform the development of new tests and systems. Misleading data is the stuff that you get from A/B testing two systems using your ears and hands.. late at night after hours of work and when one of those systems is your cherished favourite, and the other is an 'impostor'. Or when one system is a great new idea that you REALLY want to be an improvement. It's amazing how we can fool ourselves into really believing one is better just because we want it to be. I wish I had more than just a DMM with Inductance meter. The lack of a scope makes it much more difficult to proceed at any kind of pace because I'm reticent to waste energy building stuff based on hunches and 'possibly' ideas. Hmm, amateur science is orders of magnitude better than mumbo jumbo every time !! My suggestions turned you from using a heavily clipped signal to using a clean one. I also have presented reasons why your design works as well as it does (based on amateur science). I have built your basic system and found it to be interesting but the perfomance was unacceptable IMO, amateur science lead me to a better "solution" The people who have been unhappy with the performance of their systems and have taken up my suggestions of using AGC (you included) have been much happier with the results. These developments have come from amateur science. Col

@Hank have you looked at any of these as options? They're not cheap, but may be ideal - unity gain, compact, high quality and able to supply plenty of current. LT1206, LT1010 & LT1886 (That last one is a dual op-amp that can provide up to 200mA !! - I wonder if it can help to simplify the circuit by letting us combine AGC and output ?) cheers Col

You seem to be getting on the defensive a bit there... no need to though. I'm not using the term 'piggyback' in a negative way - If you can come up with an equally concise term that is as obvious in its meaning as piggyback, then I'd be happy to use that instead. That last bit about 'the same design neck to a pickup' made no sense whatsoever, can you try to explain that again please. As far as the 'thin coil' idea, IMO the only way in which it has any advantage over an less thin coil is if you're trying to stack it on top of a pickup, or in another extremely confined space. Other than that, it's just a coil - the turn count, core properties, resistance and permanent magnet are the significant variables - I have thought this for a long time and have more recently proved it to myself through experiment as well - thinness is irrelevant unless you need to get it into a thin space. Running the coil butted up next to the pickup would be pretty much the same as piggybacking it - agreed. However, not much has been done in that area, so no need to mention it. I was referring to specific problems that you had been having (and still are AFAIK) with a piggybacked driver, and that those problems may be related to the idea i was explaining. Maybe if you read the post from that point of view, it could give you some idea of how to take things forward with your design and maybe even solve some of the problems ? So far, the only solution you've found is to disconnect both ends of the pickup. That doesn't mean that there aren't other solutions. It is unfortunate that so few people have the same problem though. No one is suggesting that there is a significant difference. (although ironically, this would be one of the few times where a 'thin' coil would cause a difference as if it were beside rather than on top of the pickup, the coil coupling would be reduced) Well, I'm not sure. A couple of things - driver next to pickup, and the complex switching you have found to be the only solution so far to the issues it causes - are candidates. I'm not sure what else though? Not exactly lots eh? I've also been using and experimenting with ideas that others are mostly steering clear of and/or ignoring (or waiting to see the outcome of), so I know that it's not so bad - it means that if you get it working, you have a better system than the others, and if you don't get it working, then you don't have to worry about leading other people into wasting time and money on a bogus idea. What it does also meant though is that you need to develop a really good understanding of what you're working with. Hank is doing exactly the right thing - collecting data and as much information as possible. I have done the same. I don't have the same kit as him, so my energies have been slightly more on the theoretical side, but its not all guesswork and hunches, a lot of it is studying the physics of it and using tools to simulate parts of the system in order to prove or disprove theories. Where I consider it to be most important I've put the money down - e.g. bought an inductance capable DMM Maybe its time for you to start paying more attention to the textbooks (or websites). You may find possible answers hiding away in there, or at least find out why things behave the way they do. It may be that you find there is no easy fix, but in the process of that discovery, you find a better approach that doesn't have the same associated problems. cheers Col

There are likely two main reasons why you can't... 1. Pickup wire is incredibly thin...0.063mm (42 AWG) - this size of wire just won't be able to handle much in the way of current (& current is what we need here!) although, if you double the number of turns, half the current will give same power of magnet. Yes, its the inductance that's the main issue here rather than the current handling for a guitar pickup. I have wondered if it might be possible to make a small coil that sits next to a guitar pickup and acts as a transformer primary inducing a string driving current in a standard guitar pickup - a low inductance 'primary' wouldn't cause the same issues with frequency... I wonder if this might be related to whats happening with Petes piggy-back drivers causing the squeal when the pickup is not disconnected at both ends - maybe its not something to be avoided, but something to be taken advantage of ? Have to think some more about it... cheers Col

ERROR 137 "Post too long!" in module 182.12.5 of unit 5: community member 'Col'. DUMPING Memory to file. Please allow up to three days for reboot procedure. If fault persists, replace unit. Possible workaround: concise posting.

Although it might be a 'quasi-complementary' push pull output, in which case the main output transistors could be identical. cheers Col

Electronic switching is used also because it is cheaper and more compact - it doesn't make sense to have a massive wadge of expensive custom switching hardware when some cheap and compact off-the-shelf switches and a few tiny very cheap transistors will do a better job. when you tried these alternatives, where did you get the circuits from - were they existing tested designs, or your own designs (or heavy modifiacations)? alternatively, they could dip in to anywhere in the last two thirds of the thread - 99% of the pages have repetitions of extensive summaries of the project Class-D amps use push-pull output stages - e.g. there's one in the sustainiac patent circuit. cheers Col

Before you get over excited about remarkable performance, it would be wise to measure the inductance of your coil. If you don't have one, a cheapo inductance meter can be had for maybe £10 - £15. (I bought a cheap, nasty DMM form B&Q that measures inductance !). Then you can go here and quickly check what the impedance will be at different frequencies! Making a rough estimate based on the inductance of a 6 string driver with very thin core, I guess yours would be in the region of 8mH that would put the impedance at 300Hz at about 17ohm, which combined with the weaker pickup signal and the less efficient driving due to thinner strings would all add up to make it need a whole lotta extra drive gain. If the string is being actively choked, all I can think is that either it is a 'wolf tone' effect (in which case you would have to increase the gap between the string an the driver and/or pickup) or that your system - circuit+driver+pickup+pickup/driver gap is taking the phase response past 90º (or -90º) near the frequency of the B string ? ..unless the driver is somehow having a negative effect directly on the pickup...?? none of that made much sense...off to Google! hehe, I banged on about the difference between feed-forward and feed-back limiters and how it is particularly significant for a sustainer for quite a while until anyone 'got' it. More recently I've decided to try using the less well suited feed-back variety due to simplicity. My theory is that a normal feedback AGC takes the output of the AGC as the control voltage which is fine except that it can't react quite as well to the variable response of different string/fret combos With a feed-forward topology, the AGC tries to even out the signal that's coming from the pickup rather than the signal that's coming from the AGC circuit - this means that the weaker strings will be driven slightly harder than the more responsive strings - you get a more even response. Where this is even more evident is as you play further up the neck and the gap between driver and strings becomes less - a feedforward AGC will take this into account while a brickwall feedback AGC one will provide exactly the same drive level independant of the different responses of different strings and frets. That's the theory. In practice, I have found it more difficult to get a feedforward AGC to work with low distortion, fast reaction, high compression ratios and wide dynamic range (that last one is a killer BTW) So I'm hoping that a compromise will be 'good enough' - have to wait and see. cheers Col

I have been aware of the TDA7052A which is just a mono 8pin version of the 7083A. The only problem for me right now is that I'm still developing the whole mixed/current mode amp idea, and it doesn't look so easy to do that with those chips. That's one of the nice things about the LM386 - its innards are quite accessible. I've looked at lots of IC output stages in the last week or two, revisiting many I've already seen to see if any of them would be easy to set up in mixed mode, and so far drawn a blank - that's why I'm looking at a discrete amp as an option. I'll most likely get my most recent mixed mode AGC LM386 design tested before moving on to a discrete output stage. some brainstorming - you've most likely thought of this lot already - : some signal with a near 180º phase is getting to the driver (too obvious) : there is a significant DC current through the driver (but if that, why only with those strings?) : some sort of parasitic oscillation is causing a choking effect on the string when the gain is increased (you're the man with the scope, so you would have spotted that already) Are you still using a single string driver? Are you triple sure you didn't inadvertently flip the coil, the magnet or the connections over (I know, I know, but I gotta ask... )? Are you just testing the open strings, or going further up the neck? Are you using the input from a pickup or from a sig gen? From my experience, a separate driver for B & E strings is not required - it is possible to get them to sustain well using a full width 6 string driver coil. Bearing in mind the double whammy of weaker pickup signal and weaker drive force, you really need plenty of gain and a brick-wall limiter. Ideally the limiter is feed forward - that way it takes into account the weakness of the E & B at the driver as well as their weakness at the pickup. Unfortunately, it seems to be easier to setup and tweak a feedback limiter... at least to give a better dynamic range and lower distortion with faster response. cheers col

Its a shame this thread seems to be turning into a cat fight again :| I just spent an hour or two reading back through some posts from the time Juan was involved, and it was as I remembered. I normally wouldn't want to add fuel to the fire in this way, but if I don't present things from my POV when dirt starts to fly, I'll get sick of the project and give up on it again - I don't want to do that right now so: ..... Juan(zfrittz) was a cool guy - an electronics repair man, but NOT an engineer/designer (he said himself he had never designed a circuit before). He tried out our ideas, then implemented his own circuit based on those ideas. After some days and weeks of explaining how it worked better than my designs, but was not finished?! and us repeatedly asking him to post some audio clips, he seemed to disappear. Our natural conclusion was that maybe it wasn't as successful he had initially thought. (I would still honestly like to see proof to the contrary though) Throughout this time all the discussion form Juan, Pete, myself and many others was polite, friendly and productive. David on the other hand during this same time period seemed to me to try and take over. Particularly once he had taken on the role of interpreter. He started presenting pre-existing ideas as though they were Juans ideas (just because juan had also had success with them). He also began touting juan as some sort of electronics wizard and used this as a basis for criticising others who disagreed or were sceptical. There was one post where David tried to persuade Pete to join him and Juan as a 3 man team. This was presented in a way that ignored the work others had done on the project. When this request was rightly ignored, the post was repeated with sections emboldened... it was VERY aggressive attempt to try and take over what is very much a community style project, by someone who didn't have anything useful or positive to offer. It is fortunate that the nature of forums like this means we can go back and check if our memories are accurate - on this occasion, sadly, mine were :-| here are some choice examples: (remember that these are from a guy who at that point had not managed to get anything to work sustainer wise, about a guy who was unable or unwilling to provide evidence that any of his ideas worked.) [to Pete] "...I feel that if at all possible you and Juán should team up with me as the translation intermediary if you like, because he has the electronics expertise of a professional - no offense, but it was his job, so he should be good at it. He is also working on a rotary switch option, although he still considers the micro switch to be the best option..." followed a few posts later by: "...Apparently the 6v circuit is much more compact than the 9v one he sent me originally, so when it arrives I will post some picks (with his permission obviously). I feel that if at all possible you and Juán should team up with me as the translation intermediary if you like, because he has the electronics expertise of a professional (I remember you saying a while back that if we had any professional electronics techs in our team, the project would already be done and dusted, well we do, so why don't we make the most of his free contribution to the forum and his clearly altruistic attitude?!) - no offense, but it was his job, so he should be good at it. He is also working on a rotary switch option, although he still considers the micro switch to be the best option..." here's another from a little later that is a fairly typical expample of the "bigging up" that we were subjected to: "And unless my eyes are deceiving me, it also looks as though he has succeeded in combining the driver with a stacked humbucker type s/c, and if that is the case, then he has cracked it! Given that in his last contribution he presented a fully working dual coil driver of extremely small dimensions, I really think he deserves much of the merit for the experimentation now being undertaken by several others....so you might like to also consider his circuitry designs, as unlike others, who are basing their circuits on what is most commonly available, Juán has the electronics expertise to experiment with circuits that are not so commonly used, which is also why he favours the 6v circuits and is therefore also able to drastically minimize the real estate used by the pcbs. That certainly aids in the design of modular and plugin type prototypes, and probably will also bring us closer to Pete's ideal of a modular device that could actually be mounted inside a p/up cover, circuitry and all..." David continued to try to claim credit on behalf of Juan for much of the projects successes - including the work and ideas of others. I imagine that by this time, Juan must have been getting very embarrassed. a sad state of affairs indeed. Col

I know.... ;p I've already made one AGC that has an output that's at least as pure as the driven slush - its good enough for a no fizz sustainer. More recently, I've been working on various simplified ones with fewer components. Also want to improve the controlability and the reaction times somewhat. hmm, I know for a fact that you can get good (better) results with heavier wire. I do have a concern about using finer wire - even 0.2 wire. That is it will be exceeding recommendations for current load, so you will get more problems with heat losses and possibly other issues related to heating. I've just been going by what the recommendations are for transformer design, but I worked out that 0.31 wire was going to be bang on in that respect, and the 0.28 that I am using is within a reasonable margin. 0.15 may be getting too fine though ? An alternative you didn't mention is a dsPIC based system where 6 channels are conditioned (gain, AGC, filtering for harmonic modes) in the digital domain before being fed directly into 3 stereo digital-input class-d amps. If you can work with SMD, this could be very compact with a low component count. cheers Col

My only objection to using that chip and others like it is that it's obsolete, so it will get harder and harder to find. Col Agreed, it wasn't so easy to source (but I only needed a couple)...are there an other alternatives? Using such a device obviously keeps the component count (& complexity) low ..After posting, I gave it another 15 mins & did get the chip to work using a sig gen as an input (proving my theory, it's no good for direct connection to a mag pickup),....one alarming aspect is that it really did markedly as some distortion on to the sine wave as seen on the output (again it was late & I might have slipped up) I think there are other chips (SANYO LA4150), but even more difficult/impossible to source. I have been told that these all-in-one tape recorder ICs do tend to be nasty noisy distorting lumps which ties in with your findings. I've decided to eat my words and start looking again at designing a discrete output stage. After what you've said about the LM386 and the distortion it adds, I figure that if we can get some relatively simple push pull output stage to work, we can tailor the efficiency vs distortion, get the gain exactly as we need and also fine tune the mixed mode aspect. I've no experience in this area, but I've got a good book with a step through with examples including a low power one. I also have found a couple of schematics that include a transistor gain stage with feedback etc. So I guess that I should be able to cobble something together. I'll just have to over engineer the thermal protection and hope for the best from that point of view. This is going to slow development down at my end, but I have some time of work starting middle of next week, so I'll get more done then cheers Col

Aaah, I just revisited your post... there wasn't enough detail! - but in series makes complete sense now! My only thought here is that we're trying to fight EMF that was created using 200+ turns on the main driver - is a simple one or two turn reversed loop up to the job? (also, it's not exactly the most concise, aesthetically pleasing solution ... but nevertheless worth spending a short while toying with) yes, one loop is unlikely to be enough. And having one loop will also make the adjustment require more precision (maybe an impossible amount). It is just a little test to play around with - worth doing though. Maybe you will decide to run with it and develop it further. My only objection to using that chip and others like it is that it's obsolete, so it will get harder and harder to find. cheers Col

No worries about current, because its in series with the 8ohm driver coil. Think of the bell wire as the first turn of the driver coil, but detached and moved. Remember, you don't need to invert the signal and split it, just turn the loop over and the current is going the other way so the polarity of the magnetic radiation flips. (all that RWRP talk is bogus, it's either RW or RP that you need, both together would get you back to where you started!) In this case, using RW gets you what you need for your "simple loop of wire test". That idea (a brainstorming - may be nonsense one) will only work if the permanent magnets all have the same polarity - if you try to do some sort of alternating NS SN NS SN deal, it will fail. although, it seems to me that you might be as well using just one big old full width permanent magnet to magnetise the strings with your individual coils siting on top ? cheers Col

hehe, not so easy eh? I got 0.1 and 0.22 from a little local independent shop, but they are big chunky wire wound jobs. If I had a meter that could measure very low resistances reliably, I would probably make one. If you trust the manufacturer to make the winding wire with close enough tolerances, you could wind one yourself by working out what length of wire you would need and wrapping some suitable bobbin with a few turns. I even though of using the final few turns of the driver coil as the low value resistor, but decided that it really needs to be on the circuit board and mounted in a sensible position to prevent parasitics effecting the functionality. Re your thought about using a reverse polarity coil to reduce coupling between driver and pickup, there was an interesting idea proposed a while ago (can't remember who by), but I tried out some variations. what you do is take a loop of wire from just before the driver - some single core bell wire works - and bend it into some shapes putting it between driver and pickup or even in a loop around the pickup (leaving an inch or two gap all round). This is an interesting experiment, and results were occasionally surprising but for it to provide a practical solution it would have to be researched a lot more. Definitely something you should try out though because its quick and easy to play with. Re Petes suggestion that a dual core driver works in a similar way - I don't think that correct (although I will have to double check at some point). The sustainiac does work this way (I think?) but suffers from weak sustain near the centre where the two coils meet. A humbucker style config just has way too much cancellation. What my coils are doing is both pushing at the same time or both pulling at the same time - this shrinks and expands the permanent field. The reason this gives better performance with reduced fizz is because: # the permanent field is more contained and focussed # you get twice(roughly) the magnetic pull from the same level of drive signal, so you can reduce the gain more. # (possibly, but not sure) more core material means core saturation may be less of an issue. It is possible to contain the field better with a single coil driver, you need to use side (and bottom) plates - basically external core pieces - these focus the field much more tightly. However, they will also increase the inductance, so the coil would need some tweaking in order to achieve similar drive performance. I have tested this and it does improve things - my old single core driver was almost able to match my old dual core driver fizz wise (the old dual core driver was not wired in parallel, so does not benefit from the 2x magnet power thing) I wonder if a hex driver could be improved by using an open topped E core (similar to what I just described for the single core driver). The outside plates would between the coils, with a small air gap between each unit - might this help to minimise losses due to coupling between the cores ? cheers Col

but that wouldn't be the case when the battery is dying (when youve said fizz is especially prevelant) You get fizz when there's hard clipping. This occurs with too much gain. It also occurs when the battery is very low - As the battery drains, the voltage drops and the headroom is reduced so clipping occurs with lower voltage signals than when the battery is fresh. cheers Col I'd say that fizz could possibly be better defined as distortion as seen on the signal across the driver vs the input signal .....this could be either Hard clipping (driving the output signal past VCC) or soft clipping (VCC is fading away below that of output stage's normal working 'region)....OR any other other form of interference on the driver signal. In short it's distortion...but even though fizz seems a bit of a girlie description...it's very apt! Hmm, so fizz is a girlie term... I guess next you'll be calling magnetic pull gay ? Fizz is a particular type of sound that is produced by the system that is a SYMPTOM of some type of distortion of the signal. I use the word fizz because: #1 everyone can understand it - it obviously not a low frequency rumble or a musically usefull crunch, or even a soft breathy fuzz. #2 if I start talking about distortion artifacts in the range 1.5k - 4k (or whatever it really is) then I'm going to go crazy and most of the people here are going to get confused and misunderstand my posts even more. What's seen across the driver is not as helpful as you think - altough it would be a nice coup for you - you know none of us have scopes, so if we agree that we can only make worthwhile comment based on measurements taken accross the driver, we'd all have to shut up and listen to you. :-p What would be more useful to see is measurements taken across the pickup (we have to depend on your scope for this). We only care about distortion that gets to the pickup so that's where to start IMO. In that context it would be worth seeing if those match the distortions in the driver - is your scope dual trace ? fizz could be caused by distortion in the driver circuit fizz could be caused by flux being transported via magnetically saturated strings to the pickup fizz could be caused by transformer coupling between pickup and driver generating voltages in the pickup high enough to cause clipping there. fizz could be caused by high frequency distortion generated by the shitty LM386 getting into a badly designed ground configuration fizz could be caused by eddy currents in the driver core fizz could be a symptom of the phase gap between driver and pickup 'selecting' some small atonal bands of high frequency noise to promote I would guess its at least two of these things that cause the high frequency nasty sizzling artifacts, so by far the most sensible name for them is fizz - which is why its ended up as the defacto term here. I agree that a systematic approach to discovering which are the significant causes would be the best way forward - I spent some time promoting using a test rig to work on this, but I took a long break from the project at that time because I got sick of pissing into the wind Don't forget that the response of your guitar will have an impact on which notes sustain easily. My guitar significantly favours the G string over all others. Unless you can use a range of different guitar shapes to test on , your systematic approach will be weaker in this area. If you can, you will probably find that to get the full intended value (per string tailoring of drive) from your hex approach, you will have to make the cores adjustable or just accept that each new install will require a custom hex driver design. If you hold down the A string at the twelfth fret, it will be physically closer to the driver and therefor need less power to drive the sustain. This issue is worse for me because I like quite a high action, so the difference is very noticeable. So improved performance in this case may be related to the string gauge, or it may be completely down to gap from string to driver. (dead horse flogging time: the field strength drops off with the square of the distance, so a small change in the gap from string to driver can be significant, particularly if you are taking scientific measurements to base a theory on) There will be coupling between the coils. That they will affect strings other than their own is not really an issue I would guess, but the change in inductance of the coil when it is stacked with two other coils either side will be if your not ready for it. A coil that has optimum inductance for you system when stacked with other coils will not be optimal when used stand alone. My guess is that the math and physics involved in calculating this effect would be a major chore and not accurate due to not having specs for core materials etc. so the only way to get this 'just right' will be through iteration. Just as well you've built a winding machine hehe. cheers Col

but that wouldn't be the case when the battery is dying (when youve said fizz is especially prevelant) You get fizz when there's hard clipping. This occurs with too much gain. It also occurs when the battery is very low - As the battery drains, the voltage drops and the headroom is reduced so clipping occurs with lower voltage signals than when the battery is fresh. cheers Col

I outlined something similar a long time ago for my 'ideal' sustainer drive signal conditioning. along side the Limiter, there would be an expander to squelch low level signal and noise - anything below threshold would be silenced, anything above amplified to maximum drive - this way you could have a guitar that didn't squeal away all the time and try to jump out of your hands, but you would get strong immediate sustain when you did actually play a note. There were 2 channel compression chips around a few years ago that would have been perfect for this. These days its not so easy as most parts are only smd which is really fiddly or just impossible to solder by hand. This sort of thing might work well if we can get it in a big enough package. You are right about the LM386 for sure - I've heard the same things said in other forums related to various applications. However () Its cheap, its small and its very easy to get hold of. I have managed to get very clean results using it, you just have to stop it from clipping and keep a resonable distance between driver and pickup. I think that its the gain it forces on us that is the bigger problem, but anyhow, lets find a better alternative that is readily available (and has access to the feedback resistor)... I agree, however, I don't think its ever going to happen here - what ever language you use, if it's technically specific, it WILL be misinterpreted, then drowned in a sea of techno babble That sounds like a reasonable theory, but how do you account for the fact that I can run an LM386 based system at 9.3V with good sustain and NO fizz whatsoever - none ? What I would be interested in finding out is that if you have a power stage that doesn't have such nasty distortion characteristics, can you then use a square wave or similar as drive signal with less fizz ? ... btw, have you tried other drive frequencies yet? if so, how did you get on? how do they compare to the 330Hz? cheers Col

70mA and 565mv => 39mW. That seems very low. Possibly a combination of no adverse phase cancellation and using a higher inductance coil (try that at 1200Hz) Yes, the LM386 is not ideal - we could do without the gain - just an on chip power stage would be really cool. The first thing I did when I first tried this project was minimize the gain. It's really annoying, and must cause significant noise issues having to siphon off so much juice before the LM386 input only to apply all that gain to the noise added by the resistors... aaarrrrgggghhh The thing is though that building a power stage from discretes has a whole bunch of extra pitfalls - its nice not to have to think about those. Anyone know where there's a chip with just a push-pull class AB power stage with nice pinout access to the internals ? cheers Col

That's good progress. One thing I would suggest is don't use 330Hz as a test frequency - that will potentially give you over optimistic results. A coil with as many turns as yours will likely have a MUCH higher impedance at higher frequencies. e.g an 8ohm 6mH coil (yours may be more than that) will have an impedance of around 14.5 ohms at 330Hz. At 660Hz it would be 26ohms and at 1320 (24th fret?) the impedance would be a whopping 50ohms. A high inductance driver like this will push the strings very hard with relatively low current at the lower frequencies, but struggle badly at the higher frets. To avoid potential dissappointment, it would make sense to test at a higher frequency - say 1000Hz or 1200Hz ? That way you know you will have at least as good a response for the full frequency range of the guitar. Stick a capo of some sort on so you don't have to hold the string down to run the tests. cheers Col

I didn't understand that, can you try and explain again ? I've been using SC strat bobbins. Not the cheapo type with metal poles and a bar magnet (though I have a few of those as well), but rather 100% magnetic pole pieces. My construction has been: unwind stock PU, bock up all but upper 3mm, wind coil in upper 3mm. This leaves the lower remainder of bobbin empty except for the lower portions of the pole pieces (total height minus the 3mm coil). I've not measured it, but I'd guess it's around 7mm to 10mm. What is the effect of having these poles left in their entirety as opposed to having polepieces only within the 3mm bounds of the coil? I know this is a theoretical best guess type of question at this point, but intuitively, I would think there is some waste with the extra length or an increase in fizz or some other EMI-related philoso-babble. I add this as question number 379 under "sustainer questions that shall never be (other than empirically) answered". Intuitively, I would think the system would like it better to have an equally-sized permanant magnetic field both aboe and below the coil. I would also think that as is, there would be more phase lag or shift on one half of the AC wave form than the other as there must be some sort of effect on destroying and rebuilding the field in the opposite polarity. The magnetic core will give you a weaker electromagnet than a good soft magnetic material - like silicon steel. I think its something to do with magnetic moments and reluctance... anyhow, the permanent magnet will resist the change in the coils field whereas a steel core would enhance it. If you reduce the magnetic poles to 3mm its unlikely that there will be enough permanent field to magnetize the strings - the pickup probably has just the right amount of permanent field. Why would the longer poles cause EMI ? I don't get it. Although you're right to call a lot of the talk of EMI on here philoso-babble. Magnetic flux is a necessary part of the process - we just need to have as much as possible of the varying flux from the coil over the strings, and as little as possible everywhere else. I don't know why you would want equal sized permanent magnet above and below the coil ? The field is not destroyed and rebuilt, it is reduced and enlarged. There will probably be some effect associated with the hysteresis of the core (is that not just an efficiency thing though?) To limit it you would want to avoid permanent magnets in the core and stuff like pure iron - best with more esoteric alloys. More significant possibly is the fact that the string is physically closer for one half of the wave than the other. The pull of the magnet is reduced with the square of the distance, so will be a lot stronger when the string is near the driver compared with when it is far. Another related effect is that any flux that gets passed to the pickup via the strings will be much more evident while the strings are closer to the driver. (this may be why the fizz is so nasal - more is getting through on that 'close' half of the waveform?) I have wondered in the past if it might be worth contriving some circuitry that increases the power when the string is further from the driver and reduces it when it is closer - an asymetrical drive signal - using some sort of 'squaring' circuit. This might let us take the driver closer to the pickup as it would reduce the problem of the strings as flux conduit. However it might also cause other problems like DC offset and audible distortion due to normal field coupling, so I've not got around to actually trying to design something to test the theory. Maybe that something that would be more practical using a PCM class-d type of drive circuit... Hmm, I guess that takes us back to an old question - why not only pull?... this is like the converse of that - why not only fight the permanent magnet. That would probably increase the need to start playing with rare earth as they will be less susceptible to losing their strength over time with the coil fighting them constantly... so many potential traps though... I'm rambling now... back to your drivers - why don't you try using one of the cheaper pickups with a bar magnet and steel poles ? you might find it works better or at least as well. cheers Col

I didn't understand that, can you try and explain again ? cutting any permanent magnet will have an adverse effect on the magnetic field. Afaik in industry, the raw stuff is sold un-magnetised so it can be cut to size first and then magnetized. My designs are based on a different philosophy from Pete's, mix and match may not be worthwhile. Without some measurements and understanding, a dual core driver could turn out better or worse than a single core one - It has the potential to be better, but you gotta get it right for that to happen. I believe that there is very little point in trying to make super thin coils unless you are focused on piggy-backing them on a pickup as a priority. Personally I'm more interested in trying to use the driver as a pickup than piggyback a driver. cheers Col