col

Here is a good description of how an inductor coil works - could be useful, it was for me. Makes me think that the success of the 'thin driver' idea may really be an example of good "coefficient of coupling". My intuition suggests that the best configuration would be having the depth the same as the width - so a single coil driver with a 2mm core that is 10mm wide in total, the optimim thickness of the coil would be (10 - 2)/2 = 4mm Using the pickup winding calculator, an optimum config for an 8ohm coil using .23 (.25 inc insulation?)wire might be something like: core 2 - 6mm (probably not crucial) coil thickness 3.5mm 161 turns for 8ohm... cheers Col

Hmm, after some searching - including this thread, it seems like this idea is getting even closer to the ebow. I like how the ebow has driver and pickup very close to each other (are they stacked?), although it seems probable that the only way this can be achieved without emi feedback being a problem is by their one string only approach. I'm also starting to wonder if having a seperate pickup for the driver will not really help matters as the crosstalk to the isolated output pickup will still be audible. I have another idea though - will try to get it set up and tested soon.... Col

OK, here's a just concieved, un-tried idea for the group.... The biggest (only) problem I have with my setup is the fuzz/fizz/crosstalk call it what you will.... The most likely cause for this is a combination of highly sensitive pickup with > 10000 turns combining with a powerful driver coil of ~160 turns... producing a parasitic transformer.... any low level crosstalk gets quickly amplified as it is fed back through the amplifier.... It seems like one way of dealing with this might be to use a very low impedence pickup - this would be less sensitive, and if made with only a few hundred turns of wire, the transformer effect would be much less of an issue (I think?). The problem here is that a very low impedence pickup will sound crap, and even if it can be made to sound good with some circuitry, what we really want is sustain with our favourite pickups and their sound... So what if we have a purpose built pickup just for the driver - none of it's sound goes to the output. As it is not intended to produce a sound, it needn't sound 'good' in the way traditional guitar pickups do - no need for thick warm tone etc. All it needs to do is get a signal from the strings. The driver, custom pickup and circuit would be completely electrically isolated from the existing passive pickup which would provide the guitar output. Although there would still be crosstalk between the driver and the passive pickup, there would be no feedback loop between them to amplify or other wise accentuate this crosstalk. It may be that depending on the guitar, a piezo pickup could be used, possibly even just mini stick-on piezo transducer rather than a 'proper pickup'... or even a little mic... (theoretically, it may even be possible to use the same coil for pickup and driver. The difference between the signal we feed the driver and the actual signal at the driver leads would be the signal produced by the strings in the coils field. This could be isolated by phase correction and negative feedback and send back to the circuit input. However, in practice, the complexity of the phase correction circuitry required and the manufacturing tolerences of the coil in order for it to match this circuitry would probably be prohibitive) Any thoughts ? (makes me think that some of the fancy phase correction circuitry in the commercial units might have been for reducing cross-talk ?) cheers, Col

A core 17mm wide works well. You can detect a roll-off of response compared to a 2 mm core e.g. my 2mm cored single coil driver could sustain 5th fret harmonics well, whereas the 17mm dual core turns them into fundamental (it can do 12th fret harmonics fine and some 7th fret ones). This effect is also related to the circuit response which I have changed for better fundamental response. The overall impact of this response roll-off is minimal and anyway, 17mm is much wider than necessary, it would be simple enough with the right magnets to produce a dual core of 10mm - this would have a response that in practice wouldn't have any disadvantage over a 2mm core. [Devils advocate mode] I thought that the speed that the magnet can change polarity had more to do with the core material rather than the winding arrangement. Is it not possible that the issues that occur with deeper coils with more turns are related to a much higher inductive reactance and therfor a greater roll-off of high frequencies - all that zobel network malarky... The 'thin' driver will still have a phase response that varies with frequency, but maybe it isn't as dramatic or doesn't have as big an impact because of the lower reactance of the coil... When i put my magnet on its side and put a 2mm thick drive core on each side, the magnetic attraction along the top of the cores was considerably greater than in single core configuration. Just like a horse-shoe magnet, there is less air between the N and S, so the magnetic circuit flows better... when the strings are close to the driver as they need to be anyway, it seems to be a much more efficient configuration. Would you be able to measure the milliAmps that your circuit is drawing while sustaining (and while on but with the strings damped) ? It would be very useful for comparison. There's not much point in basing a discussion on subjective analysis . If you are running with no AGC and only using 20 - 40 mA, then I need to re-think my driver and/or circuit. On the other hand, if your setup is drawing 90 - 130 mA, then it will eat through batteries quickly no matter what your perception... It sounds like a different take on the 'woofer/tweeter' idea that I posted a few pages back. I'll try to explain that idea better in case it sparks some ideas for anyone else. What I was thinking is that the signal the guitar produces ranges roughly between 80 and 10000 hz. Thats a broad band of frequencies to efficiently process with a single magnetic driver.... So the idea was that IF drivers could be made more efficient by targeting a smaller band, we could have two drivers say one for 80Hz to 1k and the other for 1k to 10k. These would both have to target all six strings ( there is not such a significant difference between the frequency bands of the top three strings and the bottom three strings - they have more overlap than seperation). For this idea to work, the two coils would probably be quite different - different wire guages, different number of turns, different core dimensions etc. There also would not be so much of a 'humbucking' effect - although I can't see the need for that anyway (its a driver not a pickup) - but there may still be some potential for emi reduction due to the magnetic circuit being tighter... This approach would also help to sidestep the small ammount of high frequency damping caused by a 'wide' dual core. It would however be a lot of work - building and testing different coil/core configurations for frequency response, efficiency etc... designing a 'crossover' circuit... not to mention the cost of stocking up on all the wire guages etc. EMI in terms of squeeling feedback only starts to happen on my setup when you get past about half way between the neck and bridge. The real problem is the 'crosstalk grunge' as described on the sustainiac website. I think its also worth noting that I can drape the long driver lead pretty much anywhere - over the circuit, accross the pickup etc without it having either a positive or negative impact on crosstalk noise or EMI feedback. btw, I'm not just sitting here be negative I'm trying to come up with solutions as well cheers, and keep on sustaining (as much as I detest them, we really need a recursive acronym for this project - it would be so apt) Col

Interesting idea. would look cool Question 1, what is the difference between this and a dual coil with one full bar and one full 'air core' (or perspex). Question 2, would it really be worth all that extra juice - half of it not driving the strings ? From my limited but practical experience, a dual core with full bar in each side 'seems' to be more efficient than a single core driver, it also has a pretty good frequency response even with a big gap between the two cores and gives good noise reduction while being simpler to construct... doesn't look so cool though... My only concern is how to get rid of the crosstalk between pickup and driver, and I'm not sure this will have any effect? Col

After some research, it turns out that the cap from pin7 is for power supply filtering and that 10u is a good value - 100n is a bit on the low side. My success with a 4.7n was because the problem I had was high frequency oscellation - a 4.7n wouldn't help with lower frequency noise. My guess is that a larger 10u cap would become much more important if we tried to use phantom power. Not sure about a Zobel network, I looked that up, and most of the examples I saw were a lot more complicated. I think (whatever it is) it has something to do with compensating for some of the effects of a speaker coil ?...... ok, you're right its a zobel network.... heres some info i found: It may be that we could have better values for the zobel components as we are using a very non-standard 'speaker coil' cheers Col

the driver goes on the right hand side - it replaces the speaker symbol that comes after the LM386 chip Col

well, I tried a 10u in there, and it made sod all difference compared to a 4.7n so apologies for ignorant nonsence

Heh, I guessed it was mostly a demo - just don't want anyone to build that layout and expect it to work Yes, it's an interesing point, my limited understanding is that anything above the frequency response of that cap will be attenuated - kinda guesswork based on the pin being called 'bypass' and looking at the chip schematic in the datasheet. What I initially thought was that if the cap on pins 1 & 8 is the same as the bypas cap, then your only trying to put gain on the frequencies you've already setn to bypass... But thinking it over some more, I guess that the cutoff slope is very gentle.. and its never as simple as it first appears So, I will have a play around with that cap - and try your 10u value For what its worth, I did have some problems with parasitic oscellations at one point, and I put a 4.7n cap there - fixed it completely. Col

This is not verified or a complete circuit... No **** :D It seems that you have some pins mixed up? pins go from the blob at the top left downwards 1, 2, 3, 4 then from the bottom right pin upward: 5, 6, 7, 8 ...In this layout, you have the output connencted to nothing, the bypass pin connected via a cap to 'out+', ground pin unconnected etc. The interesting thing though is that you have some chunky caps in there - 100u seems ok for an output cap, and 10u from pin1 to pin8. However C4 seems a little large and (assuming this is the intention) a 10u from pin7 to ground is a very large value! If this is what you have in your circuit, that would explain why you have to run with such a high gain to get any response - a normal value for that cap I guess would be somewhere between 0.001u and 0.1u. (If this is what you use, maybe it's what gives you less noise, I doubt it though) Can you verify what values you are using ? and which pins ? I'm doing some stripboard layouts, the idea being that although the will be a little larger, and more noise prone, they will allow much easier modifications. Rather than one for the whole board, I'm doing the 3 stages seperately. Once I have double checked them for errors, I will post them... or maybe I should wait until they are built and verified... cheers Col

There's no real 'magic' to designing a layout. If you're building stuff that works with radio waves or microwaves, then layout can be critical, but for simple low power audio circuits, it's just not as important. You should have a go, it's good fun to see how much you can squeeze into a small space . You don't need fancy software either, I just use a pencil, eraser and pad of squared paper. The fetzer ruby is a small simple circuit, a great choice one for a first attempt at a layout. Generally, try to keep the input away from the output. A good start is to put components in roughly the same positions as they are in the schematic.... Col

I'm going to try building a circuit on perfboard now. The updated kinda worked. Unfortunately I couldn't get a satisfactory configuration with just unity buffers at the input as I'd hoped, so I'm back to using a gain stage. The variable 'DynamicRangeInverter™' curve seems to be working though, so there is more control and some of the settings are cool, but with coolness comes more noise :-| So, I'm hoping that getting away from that nest of wire ariels, parasitic capacitance and dodgy connections I call a breadboard and onto a good compact soldered board will help... will have to just wait and see I've also unearthed a pickup from an old junk box. Its a 'reflex blues' active bridge humbucker that I bought decades ago on impulse at a sale, only to find that the preamp was missing from the box - it's never been used .. If I get no joy perfboarding the circuit, I might have a go at setting up this pickup for some tests... It's dc resistance is 2.3k compared to the 16.4k of my DuncanJB In other news, I'm wondering if trying to use a transformer to isolate the circuit might be a bad idea - adding more coils to try and fix coil related issues seems a little daft... So I've been looking at 'opto-isolators' as a possible alternative - does anyone know anything about them ? @pete I had another go at running my setup with no buffer for the straight through signal - it worked ok, so I'm not sure what went wrong before. The tone was noticably weaker though, possibly just a side effect of all the bare wire connecting everything up... Anyhow, I have a few more things to try Col

National and Maxim both do a range of low power mono 'filterless' class-d amps (all have maximum supply of 6v) - good examples are: MAX9700 and LM4673. These really do have minimal part counts... Maybe run the whole kaboodle @4.5v from 3 AA cells (assuming the LM13700 can be run from 4.5v - it can from 5v). It should be possible to get some unity gain buffers to run within 4.5v without clipping the peaks (i think?) col

Not sure, although I'm mot really so interested in the input impedence of the poweramp stage - I'm sticking with AGC just now. And The LM386 seems to work well, and its easy to get. I really want to try class D at some point, but for now the goal is to complete the AGC, install the system properly and try to debug the low level fuzz on clean sounds (or at least understand it better). Heres another reason for pro SMD - voltage! Many of the newer techs work on much lower voltages, so it may be possible to use 2 AA cells instead of a 9V battery - these could provide a similar or longer life, and are much cheaper to replace (most likely no good for an 'active' buffer though due to the limited head room). check these out. Its like an smd riser card - space for a chip, and the supporting components in smd form. we could probably do the whole circuit on a couple of these - with one ore two little trimpots and caps on the main board. That site also has a guite on how to use their tech - and its cheap stuff. Much of the class d tech needs more supporting components - filters etc. heh - my last question there is a difficult one, any good answer would be complicated and difficult to understand let alone write - hows your complex number theory ? ..... A simple (and probably incorrect )way to think about the input impedence is that a low one will try to suck more current out of the stage that feeds it - if the output impedence of the preceding stage is higher then it can't supply enough current, and everything goes pearshaped. If that stage is a buffer, then the buffer gets the current from battery - if its the guitar, then it 'loads' the pickup which knackers the tone and response. I kinda get it, and for normal everyday stuff like transistor and op-amp stages, there's loads of info available so its simple enough to get a rough idea whats going on. The LM13700 is a different beast - current amplifier - I can't find any rule of thumb about how to work out the input impedence (i think it's dynamic anyway?). IF the input impedence is high enough and you didn't want to buffer the main output, you could scrap the buffers completely in my latest design, and have just the AGC and Poweramp stages.... (btw, the LM13700 is available as an SMD) cheers Col

I think you just opened up a can of worms try googling "SMD adapter" - there are LOADS commercially available. e.g. here or here - some of these can be used with multiple footprints of chip, others can convert them to standard dip sizes... cool stuff I guess the trick would be to buy one or two to test the water, then buy some pcb diy kit and make a batch of home brew ones. Anyone wanting to build a circuit could go with a commercial adapter... edit: seems like farnell(and others) do a range of smd to sil adapters (like the ones you posted a link to). An 8 pin adapter is less than £2. Great for breadboarding, and also for helping the layout in the final perfboard build Col

Could be a lucky combination of guitar hardware, strings, driver construction, pickup design, circuit... etc. However, you've noted that you get some fizz when you play in harmonic mode. So you're still getting the same issue as others.... maybe its slightly worse on my setup and/or onlastgoodbyes... baybe its a matter of perception - one persons 'barely noticable' can be anothers 'completely unacceptable'.... Q how well does your bridge/trem attract a magnet ? (I just checked mine on a hunch, and even with a pretty strong magnet there is very little attraction - so possibly I would have less of a problem if my trem used a more magnetic alloy that would sink some EMI) Also, there is the pickup - a strat pickup has somewhere between 6000 and 8000 turns... a paf humbucker has about 5000 per coil, a JB like mine has twice the resistance, so maybe as many as 8-10,000 turns per coil. My guess is that this would make it more susceptible to the parasitic transformer coupling between pickup and driver, or at least magnify the issue. At some point, I will get around to switching the JB for a weaker pickup (don't have any single coils though), to see if that makes a difference... it maybe that non-magnetic bridge/trem though... I've been thinking about this as well. There are potential goods and bads... Breaking the circuit into seperate components will make it easier to squeeze it in to a guitar that doesn't have much space. The circuit can be designed so folks can start with just the pre-amp and power-amp, then add an AGC at a later time. If phantom power is used along with surface mount tech housed in the same unit as the driver, its possible that we could have just a slimline driver with a cable - very few mods within the guitar (maybe even none). It may be that putting the poweramp circuitry within 'X'mm of the driver can cause EMI related problems directly between the driver and the LM386, or its support circuit. The construction becomes more difficult - building, mounting and connecting a single board, and driver (assuming you can find space for the board) is easier and less error prone than having multiple circuits with all the required connections, mounts, ground points, and possibly the need for these to be shielded... News on my circuit. I Am in the middle of some updates, that if successful will be a good step forward: variable width of the 'Dynamic Range Inverter' curve. variable peak position. Also, a simpler circuit with fewer components. (it might use less power as well - not a lot though). Right now the knob for the inverters curve interacts with the output level (a lot), this means that any change in the curve requires a change in the level knob - making setting the thing up quite tricky (a bit like when the tone knob on a stomp box also changes the volume level - its difficult to A/B two settings). So I'm trying to work something out with a dual ganged pot and some resistors... This will put the circuit complexity back to where it was... just can't win... hehe Anyway - 'IF' this works out, I'll most likely leave it at that for the time being and put the thing on some perf board. So - fingers crossed cheers Col

just like in the grolsch ads "it's not ready yet" afaik, this circuit only exists on my bread board. It is currently in a state of development. The schem posted here is already out of date, and it would be crazy of anyone to create PCBs for a circuit at this stage of development. If you want to try it out, the best thing to do is by a plugboard/breadboard, the components, and plug them all together. Then as the development continues, you can chance parts at required. As soon as the circuit is in a more finalised state, I will design and post a perfboard layout. If you're lucky someone else at that point might make a pcb - who knows. when you say SCH document, do you mean for ExpressPCB ? cheers Col

cling film (clear film?). epoxy won't stick to it, and its very thin, cheap and easy to wrap things in. col

hmm, interesting... what happens if the spacers are steel ? This process will soon have the page count flying up @pete Thoughts on EMI 'fizz' Been thinking about this some more, and I'm not so sure now about the earth being a problem... I tried again holding the driver above the strings, and the problem seems to be completely dependant on where the driver is in relation to the pickup. If it was an earth related issue, surely the fizz would be there wherever the driver is placed? Col

Magnet thickness is 3mm (a little over), material is alnico cores I think are some sort of steel - called 'flat-iron' in the shop, but it said 'drawn cold steel' on one of the shelf labels, so who knows. Looks like I need to reduce the width. what would it look like with a 8mm gap between the cores, with a 6mm wide ceramic magnet with a 1mm spacer on each side - spacer material is up to you . Top of magnet is 4mm from top of cores. cheers Col

Heh, first thing i did was try all permutations - even tried parallel configs just out of interest (they sucked as expected) reversing the polarity of the magnet and/or switching the coils didn't _seem_ to make any noticable difference. (I could have made a mistake, so I will be trying all that stuff again (and again and again ) Well...the thin core is more of an idea than the thin driver is a theory Actually, I still think the thin core is a valid argument, but the rail style magnetics are a bit different. The two polarities close to one another focuses the field as it is attracted between the poles not in a ring around it...effectively making a tight field between the rails and so "thin" for all intensive purposes. 14mm seems quite wide though compared to what Tim produced, maybe your driver is even thinner. I'd be interested on how both of you arranged your magnet/s in this design. I could try modelling the idea in FEMM so we could see how the field compares to a single coil design. Yep, my coils turned out just a little too fat for the 5mm wide magnet I wanted to use, so I ended up using a 13mm magnet from a humbucker pickup. Its in roughly the same config as onelastgoodbye one. The core is not laminated... Might be that using a narrower magnet and having the coils closer together improves the EMI cancellation ? (FWIW, I tried moving the driver around near the pickup, and got similar screaches and squeals to the ones a single coil produces.) Coil 3 mm thick core 56x10x2mm 1mm of core protruding above the (bobbinless) coil... (maybe this is another reason for the increased efficiency and noise when compared to my original driver - the original one had 2mm protruding at the top) Although I'm not sure how useful a FEMM model is to us becuase it can't incorporate the electromagnetic component of the feild ? Yep, I'm starting to think its related to shared earth. Thats why I want to get a small 1:1 isolation transformer to experiment with. Also, you may want to try adjusting the output capacitor...you seem to be using a 220 uF, I have tried 470 uF down to 100 uF and found the 100 uF to be far superior in driving the high strings though you do get a typical and consistant octave feedback harmonic morph on lower strings/notes (up to about the A note, 2nd fret, g string). This alone would account for my better response of high strings if you haven't tried it yet. I tried that with the Fetzer/Ruby - all it really does is help balance the signal by filiering out the lower frequencies. My circuit balances the signal without needing to filter out the low stuff, and harmonic bloom can be encouraged in ways that don't kill the fundamental of the low E and A strings. It's possibly more efficient because of some of the differences in construction compared to my other driver rather than because it's a dual coil, although having both North and South of the magned up near the strings may make a difference - certainly focuses the feild better... kinda like a 'horseshoe' magnet. Imagine if we end up with a slimline DIY sustainer that is cheap and relatively easy to build, can be installed with no guitar mods, works well with an even response, and lasts for 30 hours playing time on a 9v alkaline battery... That would really sock it to Sust@in1ac and Fern@nd3s cheers Col

Ok, done some more testing. I've checked each of my new coils as a single coil driver. They both work, but are both noisier than my original 8ohm driver ! I'm not sure if this is because they are only 4ohm, or (more likely) because i didn't wrap tape around the core before winding, maybe makes is a little more efficient, but possibly more noisy ? heh, most likely though is because at the moment for testing convenience, each side of the double coil has a lead almost twice as long as the one on my original driver. So, back to my results... I'm getting some fizz still, but the general level of noise from the dual core driver is much less than that from one side used as a single coil driver. More interestingly, with the new driver, and the lates incarnation of my circuit (a few changes since i posted the schematic) I can get a good even sustain from all strings at all frets (some slightly weaker, and some more likely to bloom to harmonics) with a current draw of around 20 - 22 mA Yep 20 - 22 !!!!!!!! - I hope this is a true reading, I think the battery on my meter is running a little low, however, I've had to turn the gain right down for the new driver, so it's probably about right. Batteries last a lot longer with lower drain. i.e. a 9v battery drained at 22mA will last more than twice as long as one drained at 44mA, and as it ramps up to say over 100mA all bets are off unless you are using some really hefty expensive battery technology If this is a true figure, we might get as long as 20 hours out of a single battery... thats running flat out. With no signal, and when playing hard, the drain drops to between 13mA and 18mA, so an actual battery life might even be as high as 30 - 35 hours playing time (depends how low the battery voltage can go before the circuit stops working) It may be that putting the circuit on a well layed out perfboard, and minimizing the lead length reduces the fizz to an acceptable level. Altenatively, I want to try isolating the driver circuit using a 1:1 transformer - this could be as small as about 12mm cubed - which may be acceptable... don't know if it works of course, so I'm going to keep a look out for a salvagable tranny to experiment with (don't want to spend about £8 + postage just on the off-chance) So its not the yippee i had hoped, but definately a big step forward. cheers Col

EMIBucker? news: I built it, the coils work (one is 4ohm, the other 3.9) After my first trial, the results are inconclusive. There still seems to be a fizz in the background, and I think it's EMI fizz. I will do some more testing to see if there is a reduction compared to my single coil driver. It does seem to be a more efficient driver though, works my E string better and even with 14mm between the rails, there is not obvious note cancellation (again, more testing is required) - this adds more credence to the idea that wide driver cores will not be a problem. col

Cool, might give that a whirl next time Col

I think pva will always shrink when(if) it dries no matter what you mix in there. Thats one advantage of epoxy - no evaporation = no shrinkage I would steer well clear of 5 minute epoxies - they get hot, smell bad and can really damage your skin... The slow setting stuff seems a more useful - it doesn't seem to heat up - if you get it on your hands, you don't get the burning reaction that you can get with the rapid stuff... and of course you get about 2 hours before it becomes unusable. I don't think this is going to happen. You need current to generate decent magnetic force in the coil, and the relationship between current, voltage and resistance means that you need loads of volts to get a reasonable current through such a high resistance. Pumping a high voltage into a big resistor generates loads of heat which is very inefficient. The big advantage is that using a pickup as a driver wouldn't have the same EMI/transformer effect that massively distorts any bleedthrough from the driver to the pickup in the 8ohm drive setup, but I don't think you could use a battery for power (maybe a car battery in a backpack ), and you also have the problem of designing a high voltage driver circuit and guarding against overheating.... (I wouldn't want to use a normal guitar poweramp for fear of damaging the output transformer ?) If we can deal with the EMI problems, I think the 8ohm version has the advantage. I wonder if it might be possible to go the other way and use a lower voltage supply ! many of the newer power amp ICs can run from 5 or 6 volts. What about a class D based system running from 4 lithium AA cells (just dreaming of possibilities) Col