col

Nobody is denying that it is feedback. However, the sound is nasty (but not in a good way), IMO. You have lots of feedback - its not all what I would call 'musical' feedback though. If that's just the sound of your guitar, then the problem doesn't lie with the sustainer, but you should get the rest of your kit tested. If you can't hear the grit and atonal ugly fizz in there, then you really should go to the doctor and arrange to get your ears tested (that's not a joke BTW, hearing damage is common amongst guitarists, and its the higher frequencies that go first). To me it sounds like the natural sound of the guitar is being choked. Most of the discussion and hard work on this project has been getting it from something like what you have - strong sustain with a LOT of artifacts, background noise and 'wolf tones' - to a clear strong even sustain that doesn't change the natural tone of the instrument. If you don't want that and are happy to have all the nasty background junk killing your tone, then that great, you won't have to go through the rest of the process. Its definitely a good start though, well done. Col

Would it be possible to make a short clip of just the guitar and sustainer through a clean amp with no effects (even better through a DI box to avoid room reverb and amp warmth). That way we can get a better idea of what's making the noises. You definitely have something happening, but it sounds like it's going to take some work to get it under control. There's a lot of fizz (our term) and grunge(sustainiacs term) in there for sure. There are lots of things that can cause this. The main problem is that there are different ways in which the drive signal can get mangled and get to the pickup. In order to keep everything sounding nice, there are two strategies we can use (there may be others, but these two both work). #1 keep the drive signal the same as the pickup signal - if we can prevent distortion getting into the drive signal, then if it bleeds through to the pickup, its not possible to tell it from the pickup signal. It is effectively masked -- Turn the gain down as much as possible; use a higher supply voltage for more headroom; use AGC/limiting/compression; filter out higher frequencies as much as possible within the circuit; avoid cheapo unregulated power supplies... #2 reduce the channels for drive signal to reach the pickup - careful design of driver; dual core drivers; humbucking pickups; tightly focussed fields; shielding and fancy wire; keep as much distance as possible between driver and pickup; thoughtful positioning of circuitry; high quality constuction throughout; fancy **** like isolation transformers; avoiding magnetic pickups... #1 is by far the most important in my experience, but as many things from #2 as possible will also help a great deal. if Pete is right and you are using long runs of wire, particularly if they are from circuit to driver, it would be worth making up some inter-8 weave wire - this limits the EMI from the wire. Its actually pretty easy to make this stuff I've posted about it before. just search the thread (botttom left of page) for inter-8 weave and check a few of the posts, there a pic of my very own wire, and some diagrams of how its woven. keep working on it - you're getting there. cheers Col

The only part of my explanation I'm not fully convinced about is the chain reaction part (which is a remnant of someone else's explanation). Its more likely that the fundamental and all harmonics up to the point where the Phase gets closer to 360º (or to 0º) are quickly cancelled together leaving the harmonic at that tipping point to ring out. Yep that's more like it - otherwise we'd hear a weird sound as the note rises through the harmonics - which doesn't happen. You seem to have an aversion to the whole idea of phase response though ? There's nothing really complicated or special about it. Its just part of the whole system, and in the case of a sustainer, a very important part. Ignoring it doesn't make it go away (although not having a scope or software simulation must make it hard to get a handle on it. I'd like a scope to get a more accurate idea of what the physical gap is doing, but for now, at least I can look at simulated bode diagrams of the circuits phase response) ------------------- As for pics, I'm not interested in the weedy ones that work with the beginner kits or cheap programmers. I would want to use a dsPIC with decent specs to do some real DSP processing on, the kit for that is more expensive. I guess it would be possible to build a programmer, but I'm not interested in that and I have limited time, so that would 'cost' even more. It would also be nice to add some AI in there and you need some real grunt for that... I did spend some time researching whats available, so I do know what the options are, and I can't afford the stuff I personally would want to use. cheers Col

I'm pretty sure about this and have explained it in the past (it's guesswork, but nobody's come up with a better explanation). The phase response of the system is frequency dependent - both electrically and physically (pickup driver gap). At a low frequency, an inverted signal cancels the fundamental promoting the octave harmonic and the chain reaction you describe starts. At some point though, the sum of the 180º invertion, the phase gap between pickup and driver* and the phase response of the circuit as frequency rises adds up to some figure closer to 360º than 180º... as you know 360º is equivalent to 0º in this application, so at the point when the phase sum is closer to 360º than 180º the chain reaction ceases and note settles at a harmonic. *remember that as the frequency rises, the gap between pickup and driver causes a bigger phase offset If you could have a system in which driver an pickup were in the same position and the circuit had no phase distortion, the signal inversion based harmonic mode wouldn't work ! cheers Col

As far as PIC goes, it's not the chips themselves that I can't afford, its the programmer - the cheapest I can find that works with the chips I'd want to use is over £100. =============================== Current mode amp. I found a nice explanation of current mode a loooong time ago here I thought it was interesting and would probably have some application to the sustainer, but I was busy with other things and the article is theoretical - it doesn't give actual circuit diagrams, just simplified topology diagrams. Much later I was studying the Sustainiac patent, and realised that what they call a "current source" amplifier is basically the same thing as the current amp described in that sound.westhost page (patent speak always tries to make simple things sound dramatic and complicated). I decided to try and get a simple current amp working for the sustainer based on our existing circuitry. After trying lots of approaches and getting to the stage where I was giving up on a simple LM386 solution in favour of having to design a power stage from scratch (with all the extra hassle that that would entail - thermal issues, layout etc.), I had a brainwave! I checked the internal diagram of the LM386 in the datasheet - damn if this worked it would be a very elegant solution.... and it did The LM386 exposes some of its internals to allow gain control and bypass filtering. One of the gain control pins (pin1) allows you to connect a resistor or cap in parallel with the 1.35k gain resistor. What it also does as a side effect is give access to the 15k voltage feedback resistor. So I thought, what if we stick a tiny 'current sense' resistor after our driver, and feed the current back to pin1, that current feedback combined with the existing internal voltage feedback should give us a 'mixed mode' amp. well, I tried it and yes it does work - good ol' LM386 Here's the circuit. you can see the tiny (though physically big) 0.1 ohm resistor after the driver. the current is fed back to pin1 via a 10ohm resistor and a coupling cap (to keep out unwanted DC voltages). Its set up with lots of current mode and not much voltage mode to keep the current steady through the coil. It may be that a custom design could get this balance better and be more efficient, but you can't argue with the simplicity of the LM386 version. The 1.5k resistor and 133n cap form a low pass filter that rolls off frequencies above the clipping threshold... this has to be tweaked depending on the inductance and resistance of the driver. This isn't a stand alone circuit, it needs a pre-amp section, I have two in development, a simple one and an AGC one. Note also that this hasn't been road tested yet, its just been simulated and had a basic 'does it work at all?' test. It's possible that the LM386 could die quickly or deteriorate due to being used in this way, I hope not, but don't blame me if it does. I was worried that it might have some sort of protection not shown in the simplified diagram that would kick in and spoil things, but thats not the case. Oh, and don't mess around with the component values unless you know whats what or you'll probably toast your LM386. So folks, you saw it here first - LM386 in mixed mode! All rights reserved etc. ------------------- This might be good in things other than a sustainer e.g. a modified version could work well in a 'ruby' style small guitar amp - the westhost article explains how the technique is useful to give a more valve type sound to a guitar amp ! cheers Col

Here's the deal for the dual core parallel driver: # having two coils in this configuration helps to keep the field focused - so you don't have as much trouble with EMI getting to the pickups. # more efficient driver - more pull for the same power input! This second one needs some more explanation so here goes: For a simple electromagnet, the pulling increases with the current in the wire and the number of turns. Increase either and you get a stronger magnet. For a driver there are two problems with this: #1 the amount of current we can provide is limited by the power amp and power supply. #2 adding more turns to our coil increases the impedance as frequency rises Its easier to explain the next bit in relation to the current mode amp, so I need to describe that The current mode amp tries to provide the same current through the load even if the load changes - so as the frequency rises, it will push harder to keep the current at the same level - this means we get the same drive at different frequencies. ...so back to it... There is a point as the frequency rises where the amp can't push hard enough and starts to clip. This is the point where #2 meets #1. If you add more turns (or a better core), the frequency where clipping starts goes down – which is BAD. However, as you add more turns (or a better core), the magnetic pull gets stronger which is GOOD. What we need is the most powerful coil that still provides clipping free drive within our frequency range and power requirements. My idea was that if we decide what the highest frequency we want to drive fully. And choose a suitable maximum dissipation figure (de-rating to about 80%), and possible more... we can then optimise the driver by setting it up so that its Inductance is as high as possible without the clipping point dropping inside our desired response vs power range... Heres where dual core parallel comes in: Two parallel coils let us use two coils with double the inductance - wiring them in parallel halves the combined inductance, but also the current. So the circuit output sees 1 0.8mH 2Ohm inductor while the strings are being driven by 2 1.6mH inductors. Now, as we have wired in parallel, we can double our inductance, BUT we lose by halving the current through each coil... Overall we gain though because we have half the current going 4 times as many turns because we have two coils! In reality, We don't get exactly double the power, there will be other effects due to coupling between the fields. Its also important to get enough separation between the two coils otherwise the field doesn't reach over the strings enough – not enough projection. It may be that the best performance is with the power input much less than the maximum – this would be great to save battery, it would also mean changing the design of the driver somewhat – lowering the power rating means that the inductance can be increased while still keeping the clipping point at the same frequency. So thats basically the plan (working so far). Use a current mode amp to even out the current through the coil. Then tailor a dual core driver to maximise the magnetic drive within a set of desired parameters. Cheers Col

The main thing holding me back from even considering a hex system is all the drive circuitry. I came to the conclusion a long time ago that for any real quality of sustainer you need some sort of AGC. This is not trivial circuitry, and a hex system requires a 6 channel drive circuit - thats 6 buffers, 6 channels of AGC and 6 power stages. I would need a semi-acoustic to find enough room for all that circuitry unless is was all SMD. Thats the spirit . I'm jealous of your PIC experiments. I've looked at a few, but the ones I liked the look of are out of my price range. A decent DSP PIC should have enough power to process all your 6 channels, and condition them - including AGC, phase correction etc. so 'all' you would need would be the power stages. Are there any with that many ins and outs? I bet there are (I've done some DSP tinkering and lots of programming - assembly, C, C++ etc.). ============================================================ I have also been sustainer tinkering ! I had a first trial run to test out some of my theories. LM386 in current mode driving a 2.5ohm load WORKS! woohoo Parallel Dual core driver with 140 turns per coil of 0.28 wire WORKS ! woohoo I will have to do a lot more work to refine these, but for now I am feeling very positive. (after a bad start! I wired up the driver wrong and was worried the whole premise might be erroneous) The coils are wound on the guts of an old humbucker. There is no 'padding' so the coils are full depth, not 'thin'. The core is made from laminated steel. I used the steel from PC power supply casing - this seems to have very good magnetic properties, is readily available, nice and thin and you can cut it with tin snips. I have still to insulate and glue the laminates - figured I would wait until I'd run a basic test. The humbucker had rails, so the bobbins have slots in them, which is lucky for me, but a shame for the 'project' because these seem to be unavailable as parts, so it's a case of searching for an old pickup with rails (although a normal humbucker with pole pieces should work well with a few more turns of wire). With the laminations finished, I hope to get slightly better efficiency and less noise - right now the laminations are vibrating against each other lol. The thing DRIVES the strings! even with a knackered old battery 7v Its nice that it works without having to fabricate bobbins or mounting hardware - just winding some chunky wire onto big easy to handle bobbins, then use the magnet, wooden spacers and fittings that the pickup came with - nice and simple, and of course, it looks like a pickup!. I'm sure it would work with the cores the pickup had, but not as well - the inductance was too low, and eddie currents would also have been an issue. Testing this really brings it home how limited a non-AGC circuit is though. It takes a while to get a good drive started, then it tries to shake the guitar to pieces! AGC gives that hard drive at the start, then eases off when the strings are ringing.... anyhow - it's getting there. When I have finished the driver and installed it into the guitar. I'll be better able to appraise the 'simple' non-AGC version of the circuit. If I think this is good enough for starters, I'll post it. I have also been working on a more compact AGC, so I need to get that breadboarded and see how it compares with my old circuit. This new AGC is a feedback variant which is a compromise. Partly to minimise the part count and partly because its easier (for me) to build a feedback variant with a wider dynamic range based on some other circuits I've found. There are a some other things I'm trying out in this which might be nice. One idea is to roll-off everything above about 2kHz. The idea is that as I explained a while back, due to the space between driver and pickup, the sustainer actively damps the higher harmonics. The idea is that if we filter them out, the driver wont drive them so they won't get damped. When its all up and running, I'll play around to try and get the best frequency... Just now its just a simple RC filter in the power stage feedback. I haven't had much luck design wise with 2-pole active filters - they mess up the phase too much. ...enough for now.... When I get some more work done, and more tests, assuming its still working as I'd hoped, I'll explain the theories behind the amp and the design approach that lead to the driver design. cheers Col

Here's a quick attempt at a possible explanation of what Pete is talking about. First, go here and read about why star grounds are used in valve amps. Now consider Petes 'piggy-back' driver design. A driver coil with relatively few turns in very close proximity to a pickup coil with thousands of turns. The two coils act as a transformer, the current in the driver coil inducing a current in the pickup coil (the pickup coil will develop a relatively high voltage). With the pickup coil connected only at the ground, any current induced will flow through the ground connection. Surely it's possible that, depending on the driver circuit, ground layout, quality of components and construction etc. this could add noise and potentially cause parasitic feedback? Think about the river analogy on that web page... cheers Col Those are my bolds - it seems we're trying to come up with theories why a simple switching solution won't work! (& I thought we all sought simplicity?) Yes we are - the simple solutions that have been tried don't work, so it helps to think about why that might be. well, that depends what you mean by test conditions - I don't think Pete was using a laboratory standard rig with scopes and meters, but he built it and it didn't work. I trust that he built it 'correctly' ie. all wiring was checked multiple times for errors, bad workmanship etc. ok - so ground potential is ground potential, a perfect 0 in a textbook. But in reality, thats not always the case, wires are wires and all have resistance capacitance and inductance. Ground potentials can drift, different parts of a circuit can have slightly different localized ground potentials, unwanted parasitic currents can flow. Personally, I've not tested the circuits Pete was using, but hey, if you think more work needs to be done developing that side of things, get to it - then tell us we were all wrong (or right) one way or the other, that's how things get done. FWIW, Pete spent loads of time (and posts) explaining the weird effects he was getting with his switching setup. The only thing that cured the feedback and oscillations was disconnecting both of the pickup connections. I'm suggesting a possible reason for the symptoms he described - I haven't had those problems because I'm not trying to piggy-back a pickup. The thread search facility works pretty well, so you should be able to find some relevant posts from way back when, with a little effort.

Here's a quick attempt at a possible explanation of what Pete is talking about. First, go here and read about why star grounds are used in valve amps. Now consider Petes 'piggy-back' driver design. A driver coil with relatively few turns in very close proximity to a pickup coil with thousands of turns. The two coils act as a transformer, the current in the driver coil inducing a current in the pickup coil (the pickup coil will develop a relatively high voltage). With the pickup coil connected only at the ground, any current induced will flow through the ground connection. Surely it's possible that, depending on the driver circuit, ground layout, quality of components and construction etc. this could add noise and potentially cause parasitic feedback? Think about the river analogy on that web page... cheers Col

That's a great source of magnets. I wish there was somewhere in the UK with such a good range and good prices. There are a few ideas I want to try out that need a very compact magnet, however after adding shipping and tax, I'm not sure if I can justify the expense just now.

That's good, always good to hear about a success, although its hardly surprising! The low frequencies are no trouble for a sustainer, its the higher frequencies that introduce design limitations. In fact, it should be possible to create a more powerful and more efficient system for bass guitar as you can use a more powerful electromagnet withought having to worry as much about the effect this has on the higher frequencies. FWIW, my latest research suggests that the optimum wire gauges for a normal 6 string guitar driver are in the 0.25 to 0.31 range. Unfortunately, this assumes a current mode amplifier set up for the purpose, so don't expect the heavier guages to work with a basic datasheet voltage mode LM386 poweramp. I will be testing this all out soon, and if it does turn out to be correct, I'll post a lot more details. I'm worried I might pop my LM386 in which case, I'll have to design a custom power stage using discrete transistors - this would take a lot longer... so don't hold your breath Although I would say that anyone who is about to build a driver or circuit should hold off for a week or two! Also, hold off buying wire until I've done a few more tests - buying wire is by far the most money I've spent on this, and I wouldn't want others to go through the same process and end up with loads of reels of winding wire that they don't have a use for ! My current thoughts are that the ideal guage is 0.25 or 0.275... I've just bought some 0.275 to test this with and will do that asap. BTW, I just had the pleasure of stripping 330 turns of epoxy potted 0.23 wire from a pickup bobbin - not a pleasant task, but interesting that it's even possible cheers Colin

yep, the LM324 section in the top left is for virtual ground - the op amp buffers the half supply voltage created by the divider (2 x 100k resistors). The op amp can prevent multiple references from loading the divider and pulling it away from half supply voltage - as long as the load doesn't exceed the ammount of current it can supply (I think?) In that circuit I have LOTS of things using the virtual ground - no extra dividers are needed, and it all works as it should, although the filter cap should be MUCH bigger to prevent low frequencies from the signal getting into the ground - should be the same as the output cap I think. There are lots of great docs on the web that explain various techniques for providing virtual ground, but this works, is simple and uses readily available components. Anywhere you had a half supply divider, scrap it and connect to the virtual ground Theoretically sure, but what when you pull two batteries out of their wrappers and one reads 9.8V, while the other is 9.4V ? Suddenly your reference isn't half supply any more. As the batteries are used, this could get worse limiting headroom. I've seen circuits attaching the external ground to the virtual ground, effectively making that 0v and the + and - on the battery become -4.5 and +4.5 rather than 0 and 9v, but I havn't looked into this any further as it doesn't simplify things a whole lot anyway, and may cause issues with components/circuits designed specifically for a single supply - e.g. LM386?. Coupling caps are pretty useful just for keeping DC offsets and drift out from stage to stage. I have enough trouble understanding whats going on without worrying about stuff like that The centre point of two batteries is the centre point of two batteries. if the -ve of the 'lower' of two 9v batteries is connected to earth, then it is at 0v, it's +ve terminal (and the -ve of the upper battery are at 9v and the +ve terminal of the upper battery is at 18v. If the centre point is connected to earth, then it is at 0v and the lower and upper terminals are at -9v and +9v respectively. cheers Col

I've considered them, but I want to use a battery for this project. The extra op-amp and two resistors that are needed to setup a virtual ground, and the very few extra components elsewhere are a bargain to pay for being able to use a single 9v battery IMO.

Progress Report: I've been playing with circuits again. I've found a simple and effective way to use the LM386 as a current mode amplifier - it can be set up to push a constant current through the driver that doesn't change with frequency (up to the point where it starts to clip). It also has a very good phase response (I think - the phase is fairly level at around 0º after the driver!). This really works well, however because it pushes harder the higher the frequencies get, upper harmonics and any harmonics associated with clipping or noise are causing lots of fizz (what a surprise). It should also cope much better with differences between different drivers - it keeps the current constant over a range of driver inductance values and can handle various output impedances (as low as 2ohm). So to counteract this, the next step is to test a low pass filter on the input - a simple RC network doesn't cut it, so I'm going to try a second order low pass. This means an extra op-amp 2 caps and 2 resistors . The filter has been designed and simulated, and works so far - just need to breadboard it up and test it for real. So far the circuit uses 2 op-amp sections (1 for input buffer and one for filter) the LM386 and a bunch of capacitors and resistors. Testing this has reminded me just how fundamentally important it is to have ACG in a sustainer system. Without it, here is just a huge discrepancy in response between the notes that are easily driven and those that struggle. Still, I will post a version sans AGC for those who don't want the complexity. cheers Colin

The problem with your design is that fets are not that consistent, specs can vary a lot. That's why I haven't posted my design, based on the Ross compressor. Fet needs to be biased properly, not all ca3080s work exactly the same(bias, transconductance), transistors have different hfes. What we need is a design where all the calculations can be made based upon component values like with the THATS. Could be as simple as this one Looked it up in the dictionary, still think I ment exotic Cheers Fizz My circuit worked as I expected first time with the first FET (J201) I tried. If you're having trouble, buy a handful (they're pretty useful for all sorts of guitar effects circuits anyway - so worth having a bag full), and go here for a very simple method for measuring their gate-source cutoff voltage (that's what we're interested in here). If someone isn't willing to put in that amount of extra effort, then this project probably isn't for them, what with all the driver coil construction, magnets to try, selecting and testing core materials etc. There are loads of tasty JFET based AGC circuits all over the net to experiment with. (Thanks to this latest discussion, I found yet another that looks like a really good bet - might give limiting over a wider range of signal voltages than the others I've seen.) You ment exotic. I however, meant esoteric . 1. Intended for or understood by only a particular group 2. Of or relating to that which is known by a restricted number of people. 3. Confined to a small group any electronic component that fits the above definitions isn't going to be easy to get and isn't going to be around for long - the THAT chip in question fits nicely

AGC is just a general term for something that varies the output volume depending on the input volume. This can be a compressor, a limiter an expander or some combination thereof. [this isn't a reply directed at Fresh Fizz, just at this discussion ] Yes it is important that we don't use parts that may be difficult to obtain. FWIW, the THAT2159 is a voltage controlled amp it's not an AGC chip. You still need to supply it with a control voltage, and thats where most of the work in an AGC is anyway. e.g. in the AGC approach I've been using, the voltage controlled amp is one op-amp, a FET and a few resistors and the ood capacitor - no big deal really. The complexity is in the circuitry to supply the control voltage - this uses 2 op-amps and a whole bunch of caps resistors and diodes - maybe two thirds of the AGC is to create the control voltage. There have been DIP package compressor limiter AGC chips in the past, but they went out of production when tape recorders bit the dust. There was a cracking chip of this type that had two channels of AGC - we could have used one channel as a compressor/limiter to level out the input we wanted to use, and the other channel as an expander (squelch) to kill and input below a threshold to prevent un-tamable feedback from low level signals. Unfortunately this is another obsolete out of production item - it would be crazy to base a new design on something like that. There were also others that had input buffer, AGC and a power-amp all on the same IC - think 'instant sustainer driver circuit'. Again these were intended for radio cassette recorders and are unavailable these days. We discussed this whole topic a long time ago in great depth. Afterwards, a number of AGC approaches and circuits were proposed. The one I ended up with works very well for the purpose. It is maybe a little over complicated, although to put things in perspective, my whole circuit including input, AGC, harmonic modes and output amp is less complex than that cs3 circuit. Its possible to create a pretty good AGC for our project using 3 op-amp sections, 2 diodes, 9 resistors, 1 capacitor and a FET (give or take a few resistors) Thats pretty good value and no esoteric parts required. As for cloning a sustainiac - GO FOR IT. When you get it up and running, it would be great to hear how you got on and see some documentation on how to do it. cheers Col

Probably not. And it would have been no fun. And it would have been more expensive than buying one and installing it, so why bother ? If you really want to build a clone, just download the Sustainiac patents - I'm pretty sure its legal to build one for your own private use? Good luck though I've heard its not an easy job building a class-d amp from discrete components and getting it tuned and stable - your layout design will be critical... Oh and how are you going to find out exactly what alloy is used in their driver core? and when you do, where will you source it from? etc. etc. etc.

I would say up to now that this is the way to go driver wise. two parallel 16ohm coils will - due to the relationship between, current, coil turns, impedance and magnetic pull - give roughly twice the magnetic force while keeping everything else about the same - so that's twice as efficient ! Unfortunately, my driver was wired in series because I'd not worked this out until a little later. Since then I've don't a bunch of other pencil pushing that suggests it may be possible to use a wide variety of drivers, and that the success will depend on the circuit used to drive them ! we'll have to wait and see, but all the stories about pickups being used as drivers may be perfectly true, and the idea of using a pickup as a driver just needs some refinement on the circuit side! Here are some thoughts I've had on this - this is not an area I've spent a lot of time mulling over, so there may be other problems e.g. losso due to temperature etc. A pickup has a very high resistance, so there will be a very low current through it - maybe 100 micro ohms - when driven by a 9v battery circuit. However, a pickup has a LOT of windings around it, and therefor a very high inductance - in the region of 2.7Henries the magnetic pull of the driver is directly proportional to the square of the current (so we lose out there) and to the square of the number of turns in the coil (where we lose on the current, we gain here). The inductance is also directly proportional to the square of the number of coil turns, so we can look at the ration of inductances and currents and draw some (rough) conclusions about the magnetic pull we might get from that pickup when driven by a small circuit. my working single coil driver is 7.5ohm and 1.2mH(milli-henries) my half a humbucker rail is 5.69kohm and 2.9H.... 5690/7.5 = 758.666... 758.666.. x 0.0012 = 0.9104... interesting more interesting is that when I substitute my 8ohm 1.2mH inductor in my simulation with a 5.69k 2.9H inductor, the current through it comes out at about 183microAmps. 2.9 / 0.0012 = 2416.66... 2416.66 x 0.000183 = 0.442 e.g. it gives us about the same magnetic drive as our 8ohm 1.2mH driver does with 442mA going through it (which is lots) What I'm suggesting is that the loss we get due to the high impedance of the pickup as driver is more than made up for by the increased inductance of that driver. Downsides: Major one is that with a big inductor, there is a big discrepancy in impedance over the frequency range of the guitar. (This is the reason (IMO) that Pete ended up with his 'thin' driver config. any more turns of wire or core mass would make it's effectiveness much more frequency dependent.) Another is that I'm not sure ho well an LM386 is going to work when driving such a high impedance. If it doesn't handle it well, we'll have to find another solution... It seems that it might be pretty simple to go some way towards solving the first issue with some clever circuitry. This is what Sustainiac call 'current source amp' (I think?) I have seen it better described on the web as 'current mode feedback' the idea is that with a tiny resistor after the driver, we can use feedback to 'tell' the amp to work harder (or not) depending on the current at the other end of the driver - so it will balance up the frequency dependency of the system. Even if we can't use a pickup as a driver, we should be able to use drivers with much higher inductance, therefor get much better efficiency. In simulations, this works pretty well. Still to try it out for real cheers Col

the little red X's are to show that that op-amp is already getting power - it is part of a dual or quad amp and one of the others on that chip is wired to the power rails

You probably know a lot more than me then ! I use a lot of trial and error combined with bits of other circuits pasted together in simulations. I've learned enough to understand most of whats happening, but I often overlook problems due to ignorance that I later understand. For example, in this circuit, I'm concerned that the multi harmonic mode switching has different input impedances depending on which mode is selected. Unfortunately, I don't have the chops to work out if this is true and if so, is it significant... Here is a post I made a while back that has links to a schematic and a layout diagram. It also has some very important notes and instructions. Since this was designed, I have simulated much simpler versions that work just as well - more similar to a schem curtisa posted eons ago, however this is the last circuit that I built and installed, and its the one I used to make most of my clips with. You should also note that I used a dual coil 'rail' driver. If you do decide to build this, I'd like to hear how you get on. Otherwise, no worries. I might get around to building and testing a simplified and improved version some time. EDIT: oh, and (obviously really) the layout diagram doesn't include the power-amp section because I just used an existing mini LM386 power amp section. cheers Col

Humour me here ....who/what are ROG?!! ROG = runoffgroove ... www.runoffgroove.com

Well...I guess I have the same problem too...my solution is the 100uF cap, which produces a harmonic bloom on the lower strings (that I quite like) and better high string control. I've suggested that perhaps a switchable output cap might be another approach...maybe a frequency dependent switchable output cap if we wanted to get tricky with it! the 100u cap is a poor solution IMO. The problem is that the guitar and basic sustainer system are more responsive on the lower strings for a variety of reasons. What the cap does is kill off some of the low end response so that the gain can be turned up without those lower strings getting 'too lively'. Unfortunately it is a clunky approach, it is not easily tweakable. It still doesn't offer a very good even response, it removes the option of fundamental sustain on the lower strings, and most significantly its success is dependant on the particular guitar, driver, pickups that are being used - so each time a system is build, experimentation will need to be done. What is needed is an approach that will just work. AGC is one approach that is proven to work, it works very well. Unfortunately, by your description of your system (that the notes keep getting gradually louder), your AGC is not particularly effective - I would be interested in seeing the mechanism. This is not the only way however. One of the things the Sustainiac patent shows is a 'current mode' amp (they didn't invent this idea btw). thats one of the things I'm exploring and hope to test out really soon. My initial system worked maybe too well, the drive was extremely strong, but because it was so good over the whole range of the guitar, the unbalanced response of the different strings of the guitar, and of fretting position combined with a medium/high action, it was impossible to get a compromise that I was happy with. changing from 0.23 wire to 0.2 wire wouldn't have 'fixed' this neither would making a small change to the thichness of either driver or core - it's a fundamental problem of the systems design. Sure, it's possible to get it a little better with lots of tweaking of all the parameters, but that would be needed with each new installation! The only way to improve things radically is through designing a better circuit. (not necessarily a fancy one though) The synthetic quality of the raw drive from the basic system wasn't appealing to me either, so I spent time trying to control it. I achieved this very well, unfortunately as I've explained, there are other issues that this has highlighted - not caused by the AGC system, but more obvious when using it due to it being a more refined approach. Why would they give a different response? I have explained how the gap between driver and pickup radically damps higher frequency content. With this in mind, the tiny gap between two rails of a humbucker style driver has a comparatively insignificant impact, this checks out in listening tests. The main difference, apart from reduced EMI, is that its harder to design and construct a dual core driver, and that you need a bigger hole in the guitar. It should also be possible to achieve significantly better drive efficiency using a carefully designed dual core driver (I've worked this out on paper and explained it a long time ago, although I've not tested it out, someone else did and was very happy with the results - Truth Davids buddy as I recall) The F/R is never going to give anything like optimal results with any driver. Maybe you should build one and test it before selling it any further? It wasn't ever 'designed' and I for one really dislike the way ROG have been shown in a bad light here because someone else cobbled together parts of two of their circuits and presented them publicly as a solution for problem they were never intended to solve. If ROG had developed a sustainer driver circuit, you can bet it would have been **** loads better than the F/R (i've built some of their projects and they are GOOD) whos suspicious ? not me! I was highlighting the fact that your clips SEEMED to be there as evidence of what could be achieved with the DIY system as 'presented' here. I know that you are using 'secret' circuitry. I also know you used some tricky 'advances' techniques to compress your coils (making them more efficient). I also know that you've never pretended otherwise. Unfortunately I also know that this thread is so filled with enormous bloated blog like posts that finding the little useful bits of information like that can be very difficult. soooo I was suggesting that you should have made it clearer that no-one should expect the results you got in your clip by building the basic driver and F/R circuit that seems to have been recommended by you and others as a good starting point many times here. I doubt it! A basic easy standard circuit is never going to give the same quality of results as a more advanced one. There are just too many things that need to be achieved to make the best magnetic drive sustainer possible. You need AGC to even out the massive difference in response over the guitar. You NEED a highly efficient system, for which you need things like class-D amplification and a fancy pants driver (with twin coils, laminated core, high quality construction etc.). You need nice harmonic control modes to offset the fact that the sustainer kills the natural harmonic response of the guitar. Each of these things adds a great deal to the complexity of the project. Anyone who has success with a basic design without these features is going to be left feeling like the cat that didn't get the cream. The sustainiac patent is a pretty good example of a design that is about as simple as it could possibly be and not have any major compromises, and I don't know about you, but I'm not going to attempt that any time soon Still, its not going to stop me. We can still get a much better simple circuit that the F/R. And I believe we can also get a more efficient simple (not super compression techniques) driver than the single coil 'thin' driver. Unfortunately it'll only be a bit better - it's not going to light up the sky. So... no Nobel prize for sustainer development for me then Col

Just because there is not intentional 'phase control' doesn't mean that the phase response of your amp will be the same as or anything like that of the F/R. Each stage of the circuit - input buffer/pre-amp, any filters, the AGC and the output cap could alter the phase response. Your amp could be 50º, 60º or more different over import parts of the guitar range when compared to the F/R or other circuits - there's no way of knowing without measuring or simulating it. Just because your unaware of whats happening doesn't mean it's not happening or that its not good If it is doing something nice at some sort of 'sweet spot' it would be good to analyse it and find out WHY then use that knowledge to create of improve some standard circuit. Or just demonstrate that its bog standard and it really is all about the driver ! A bigger cap (220u or 470u) should not effect the higher frequencies, it should just give a more transparent low end. If a bigger cap really does have a negative impact on the high end then there must be some other reason. In other news, I've been working on a simple circuit (fewer than 20 components) that might be a nice beginner option, but at this stage, i don't have all the parts to go for a test run (super low value resistors aren't easy to find locally). When (if) I do, and if it works, I'll post it for others to try. cheers Col

I think that if you're going to say things like that, you have to make it VERY clear that you are using circuitry that you have kept secret and that these clips were not achieved using a Fetzer/ruby based system. If I were new to this project I would be extremely suspicious that someone was implying that circuitry is not crucial, but at the same time refusing to post a schematic for their circuit. If its a legal issue, then wire up your driver to an existing public domain circuit (e.g. Fetzer/Ruby) and then post clips made with that, at least then, you're not giving the Fetzer/Ruby users false hope

Feedback from a loud amp in a room damps some overtone harmonics and boosts others, as you move you can control which. Feedback from the sustainer (neck driver bridge pickup) actively damps all of them, and you don't have control over this. So obviously the two are not the same. playing with loud amp feedback, pinch harmonics come alive, as does the harmonic complexity of the guitars sound. playing with sustainer feedback the opposite is true - pinch harmonics and artificial harmonics are killed and the sound becomes sterile very quickly. we can go some way toward a fix by forcing the system to favour a specific harmonic overtone, but this then becomes the sterile note with no overtones or complexity. You just don't get the shimmer and excitement of a real loud guitar tone from the sustainer as it is currently - with or without AGC, single or dual driver etc. - it just isn't a raw rock'n'roll vibe. sure it can fool you for a while, and can work very well in a mix, but for hours of playing it gets old pretty fast once the novelty has worn off. When you think carefully and in detail about how real loud amp feedback works, you realise that there are many differences. The main one is that the pressure waves in a room work on the whole string and on the body and neck. It's easy and natural to 'catch a wave' by standing in the right place and positioning the guitar, then moving around as you play to maximize the energy of the air in the room to your advantage - just not an option with the sustainer as is. I would like to think that there will be some way to get around this issue, I just don't have a working solution yet. One possibility might be to introduce some sort of variable complex delay with reflections that provides a similar reverbation type effect as a room. I briefly looked into this a few months ago, but the PIC chips I could find that were affordable didn't have the processing power or bit depth to make it an attractive proposition. Might be worth experimenting using pc software to check if it would actually work, although having said that, the latency would have to be really low, and there's still the issue of driving only one point on the string but with a big mess of inputs. All the ideas I've had about using esoteric types of string sensing are problematic due to extra hardware getting in the way of playing, or because they require precision of construction that would be beyond any DIY project. anyway, still thinking away here Col