Can CSC machine be programmed to do scattered winding?

[Hugh Evans]

I wish I could understand why my post sat dormant for nearly three weeks and is now causing so much offense... Especially considering that it all seems to revolve around a general statement that was not aimed at anyone, rather I find the available scientific data insufficient and am curious to better understand it. I haven't made any wild claims or promises, nor do I consider people who wind by hand inferior in any way. At the moment I feel that I am being mocked for my curiosity and musical/scientific background.

Since I have apparently offended you, I am sorry as that was not my intention. I have nothing against you, and I would appreciate being offered the same respect.

[David Schwab]

My hypothesis, although I hesitate to make it sound more formal than it is, is that the existing scientific models of pickups are insufficient. I propose that by measuring and analyzing pickups made with different winding patterns, tension, wire gauge, and other materials can be quantified and provide a useful model capable of producing pickups with predictable tonal characteristics.

We have been saying that here for years. And since we might not know all the little aspects that affect the tone of a pickup, or how they interact, mapping that all out seems insurmountable. Or on the other hand it might be much simpler that it appears, but I doubt it.

But who is this information going to help? Unless you are looking to copy another pickup, having someone's specs, or even "trade secrets" are of no use to anyone. You pick up a tip here and there, but unless you are trying to make an identical pickup, I don't see why any winder here would share more than they already do.

[Hugh Evans]

My assumption has been that being able to duplicate a pickup or predictably create one with specifically desired characteristics would be useful to more than a few people. I'm starting to get the feeling that I should keep my thoughts to myself on this topic, because it's definitely a raw nerve.

I suspect the key to this project will boil down to design of experiment (DoE) to minimize the number of tests. It does not strike me as insurmountable, but it will not be without challenges. Multivariate statistics will be rather heavy, but it should be possible to find separable variables and develop a complete model. Private industry might be more interested.

[David King]

I see the corollary to the pickup maker BS. You could be making the best pickups in the world (whatever that might mean) but the average customer has no way of knowing that, they can hardly play guitar let alone listen to nuance with their ears. If you don't have a way to telling them what they should be hearing you'll never get their attention. The Customers WANT the BS. That's what they are willing to pay for when you have no way to distinguish your product from anyone else's. It's not like one HB looks significantly different from another on the store shelf or in the guitar on the wall. No one's chrome plating is likely to be shinier.

[Lyrebird Steve]

I propose that by measuring and analyzing pickups made with different winding patterns, tension, wire gauge, and other materials can be quantified and provide a useful model capable of producing pickups with predictable tonal characteristics.

Yes. This is my aim as well.

The issue is in finding what quantifiable measurements are reliable indicators of the tones that you hear. Modelling how inputs to pickup winding influence these measurements is what I'm after

How they interact, mapping that all out seems insurmountable. Or on the other hand it might be much simpler that it appears, but I doubt it.

David, I spent many many years at University persuing a PhD in Neural Network forecasting/modeling. There are many commercially available software tools that will allow you to present a known set of input parameters to measured output results and the software will do all of the non-linear mapping for you. The key here is NON_LINEAR relationships. Most standard Stats packages struggle in this regard.

Here's the software i used back then

But who is this information going to help? Unless you are looking to copy another pickup, having someone's specs, or even "trade secrets" are of no use to anyone. You pick up a tip here and there, but unless you are trying to make an identical pickup, I don't see why any winder here would share more than they already do.

My ultimate goal is to be able to measure the existing pickups in a clients guitar, and ask them what tonal qualities they are after and model the desired result. From this we can reverse engineer the model to get the input parameters and wind using Automated CNC winders to achieve desired result. Hopefully resulting in satisfied customer.

Is this possible. I think yes.. Is it easy ... no.

[Hugh Evans]

I plan to handle the stats with JMP or SPSS, but in all honesty I expect both will be overkill. It will be interesting to see what patterns and relations emerge from the data. Since I know not everyone is into active designs one idea I have considered is to use my active system to determine the customer's desired parameters. The circuit allows for independent control of gain, resonant frequency, and Q... So if a store kept this system on hand, installed a set of my low-z test coils, and let a player adjust everything until they are satisfied with the tone: The parameters could be recorded and submitted so that passive pickups matching that tone could be produced. I hadn't considered this particular angle before, but that could be a very cool way to order custom pickups. Since it is based on the players ears, rather than being skewed by what they've been told will sound good, they would have a great chance of getting exactly what they want.

In my estimation, the most challenging part will be optimizing the experimental design to minimize time spent making test pickups. Some will sound good, but in order to have a full data set there will undoubtedly be some weird ones in there too. Being able to swap magnets will likely be a requirement, a large number of coil taps might also simplify matters, and most importantly a proper test rig must be devised. I've been talking over the test rig with a friend, because we both believe a simple pickup test system would be useful for anyone who is into winding... and doesn't want to drop $5k for test equipment that massively exceeds what is needed to analyze guitar/bass pickups.

[John_H]

...I've been talking over the test rig with a friend, because we both believe a simple pickup test system would be useful for anyone who is into winding... and doesn't want to drop $5k for test equipment that massively exceeds what is needed to analyze guitar/bass pickups.

Cool, I'm interested to see what you come up with. There are so many variables to consider, I don't see a simple answer. It's been discussed here before, and it seemed that the smallest detail could pollute your data. eg. pickup height and attack. Are you going to compile data for different gauge strings?

Maybe there's something here that will help if you want to make a driver coil to excite the strings on your test rig.>>Misc : DIY Sustainer : DIY Fever ? Building my own guitars, amps and pedals

[Steve Conner]

The biggest problem (IMO) is mapping measurable pickup parameters to subjective impressions.

The standard "scientific" model of a pickup consists of two parts:

A transfer function from string motion to induced coil voltage. There are actually two transfer functions, one for each orthogonal set of string vibration modes. The transfer functions are non-linear, depending on the shapes of the pole pieces and the spacing to the strings. A pickup with several coils has two transfer functions per coil.

(There is also a transfer function from pick attack to string motion. You might call this the impulse response of the string. It is a function of the point sensed on the string, due to the finite speed of sound in the string. This explains the out-of phase sounds got by combining pickups, "quack", comb filtering in humbuckers and so on. But it's debatable whether this should be considered part of the pickup system.)

A transfer function from induced coil voltage to terminal voltage. Analysing the pickup with a driver coil only tests this part of the system, so you are missing at least half of the fun. Personally I think Lemme's second-order model with eddy currents is adequate.

[Hugh Evans]

My plan is to go from simple to complex and stop when the model is sufficient to function reliably. While driving coils can be used, I prefer the more direct approach of passing frequencies through pickups with a constant current source. Measurements of real and imaginary parts, inductance, capacitance, and magnetic field strength will be a great starting point. My opinion is that driver coils increase test complexity, and I would prefer to avoid them if possible.

You're absolutely correct that interaction with strings is an important factor to the tone in a specific instrument. I have some ideas, kept as simple as possible, for the purpose of determining if data from the first method will sufficiently correlate (I have a suspicion it will.) Basically, it seems logical that variables such as strings and pickup location would constitute a nuance... But if the refined electrical model can achieve an estimate that is 95% of the final result, it's likely more than good enough. If anyone wants to bankroll a more involved study after the first round of tests I would certainly be open to it. However, I will say that at that point, a computer model might do a much better job.

Thanks for the input, and if anyone wants to bounce around ideas regarding how to deal with studying the other half of the equation... I'm up for it.

[Hugh Evans]

We'll see. I'm not yet convinced that looking at pickup height, attack, string gauge, and pickup placement need to be studied (although I like your idea of artificially exciting the strings.) Using that method and a big array of solenoids I can imagine creating a surface plot based on every permutation of fretted notes... That would be wild! The question in my mind on regarding the investigation of strings and pickup height and all that fun stuff, is that they are adjustable variables. So, if you know a pickup has the characteristics you need it is reasonable to assume that they can be adjusted in situ to produce the desired tone. My approach is that of making this as simple as possible, but no simpler.

[SonnyW]

The Customers WANT the BS. That's what they are willing to pay for when you have no way to distinguish your product from anyone else's. It's not like one HB looks significantly different from another on the store shelf or in the guitar on the wall. No one's chrome plating is likely to be shinier.

I agree. Well, of course maybe if I got hold of some kind of a machine to measure the luster of my chrome covers they could be said to be shinier All you have to do to prove your statement above, is to look at the prices for NOS "bumblebee caps" They are nothing more than old difilm (paper+mylar) caps made for TV's that cost 18 cents in 1952 and they are now easily going for $40-50 each, and the reproductions sometimes go for more than that, even over $60 each. Somebody is buying them. People all think they are oil filled too and 90% of the ones in guitars never were. It is tulip mania deja vu all over again.

[John_H]

We'll see. I'm not yet convinced that looking at pickup height, attack, string gauge, and pickup placement need to be studied.

I still think that you should have to maintain very consistent standards for the test data to be any good. Any of these if varied would make significant differences in the outcome. The playing field must be level.

[Hugh Evans]

Right, and a very consistent test method is exactly my plan. It completely removes the rest of the instrument and the player from the equation, otherwise the number of variables will be much too large to manage. Even trying to add wood to the equation would make the data more difficult to use, because its properties such as density, hardness, and strength can easily vary by ±20% with large sample sizes. If anyone has a good idea of how to handle additional variables, it's worth discussing... But as a starting point, putting a pickup into a chamber lined with mu-metal and measuring all of its electromagnetic properties seems hard to beat. It's a lot more data than what is out there otherwise.

Could it turn out to be nothing more useful than a science fair project? Yes. But I'm a fan of knowledge for the sake of knowledge.

[Spence]

Try making some pickups. That's always the most appropriate starting point.
Any other approach is absurd.

[Hugh Evans]

I have wound pickups... That's not the point of my study, though. What exactly is absurd about a controlled scientific analysis of the parametric effects of wire gauge, magnets, winding patterns, and tension on the tonal properties of pickups? If my goal was simply to master my winding technique I would absolutely agree with you...

I haven't had any problems with anyone else on these forums and, if possible, I would like to know what has you so worked up and seemingly hostile towards me. All I know is that I have nothing against you, and if there's a reason behind how you are treating me, it would be great to know what that reason is. I don't make a habit of having enemies, especially if its someone I don't even know... You've ignored my previous request for an explanation, and I don't plan to ask again. Forums are much better when people are civil towards one another. Can we play nice, please?