pasqualino
Rookie Solder Flinger
Shifting gears and now "honing my axe" and building a quick Tweed F1
Posts: 10
Likes: 1
|
Post by pasqualino on Oct 22, 2020 17:40:53 GMT -5
I'm modelling everything like I always do, but in this case I attempted to stay on topic with the pickup/guitar modelling. The pickups are of great importance because you can't effectively simulate a stomp box if you use the std "ideal" AC source from SPICE or Micro-Cap. The simulation is worthless without having a proper model for the source series L and R and loop shunt C of the coils. Your case-study of a post is testimony to this. I was wondering if it's possible to take a data sheet for a given pickup and plug new calculated parameters into the model. This may have to be done empirically, but it would be nice if this data were available straight from the manufacturer or even if it could be calculated from the data sheet. To do it empirically, you might as well have them installed and pick out every note and take measurements. What's the alternative, buy a pickup, blast a optimally loud chirp at it, measure the response and then return the purchase? I'm presenting an over complicated description and didn't mean to. I just want to have "reasonable" values for the model parameters of any given pickup. The AC resistance is easy to model in Micro-Cap since you only need to plug the equation for resistance as a function of f into the Freq field of the resistor. What equation is the real question. The L, C and DC resistance are relatively frequency independent, but the AC resistance is an order of magnitude greater than the DC resistance because the loss in the magnets is is significant. It's an entropy thing... Okay, I'm doing it again, sorry. There has to be a straightforward way to get the parameters for the AC resistance. The structure for the pickup is consistent and we can easily calculate L, C and R DC from the data sheet. I've never seen specs for anything other than R DC, Turns, wire gauge, insulation etc. Nothing regarding magnetic losses though. Maybe they exist and I just missed it. Now that I'm thinking about it, the distance between the strings and magnets in the pickup will vary and that's probably a square law thing regarding the intensity of the vibration of the magnet. Both intensity and speed (frequency) of vibration would contribute to magnetic loss. So I take what I said at the beginning of the previous paragraph back, there probably isn't a way to easily determine the parameters for magnetic loss unless the manufacturer tests the magnets in the lab and hands you a number or at least curves for changes of frequency at a few fixed values for intensity and visa-versa. Probably a case of over analysis, I'll just fudge some reasonable equation in the freq field of the resistor model in Microcap and make adjustments. When in doubt, just guess a ballpark solution and iterate, right? I'll shut up now.
|
|
|
Post by antigua on Oct 24, 2020 2:26:38 GMT -5
I'm modelling everything like I always do, but in this case I attempted to stay on topic with the pickup/guitar modelling. The pickups are of great importance because you can't effectively simulate a stomp box if you use the std "ideal" AC source from SPICE or Micro-Cap. The simulation is worthless without having a proper model for the source series L and R and loop shunt C of the coils. Your case-study of a post is testimony to this. I was wondering if it's possible to take a data sheet for a given pickup and plug new calculated parameters into the model. This may have to be done empirically, but it would be nice if this data were available straight from the manufacturer or even if it could be calculated from the data sheet. To do it empirically, you might as well have them installed and pick out every note and take measurements. What's the alternative, buy a pickup, blast a optimally loud chirp at it, measure the response and then return the purchase? I'm presenting an over complicated description and didn't mean to. I just want to have "reasonable" values for the model parameters of any given pickup. The AC resistance is easy to model in Micro-Cap since you only need to plug the equation for resistance as a function of f into the Freq field of the resistor. What equation is the real question. The L, C and DC resistance are relatively frequency independent, but the AC resistance is an order of magnitude greater than the DC resistance because the loss in the magnets is is significant. It's an entropy thing... Okay, I'm doing it again, sorry. There has to be a straightforward way to get the parameters for the AC resistance. The structure for the pickup is consistent and we can easily calculate L, C and R DC from the data sheet. I've never seen specs for anything other than R DC, Turns, wire gauge, insulation etc. Nothing regarding magnetic losses though. Maybe they exist and I just missed it. Now that I'm thinking about it, the distance between the strings and magnets in the pickup will vary and that's probably a square law thing regarding the intensity of the vibration of the magnet. Both intensity and speed (frequency) of vibration would contribute to magnetic loss. So I take what I said at the beginning of the previous paragraph back, there probably isn't a way to easily determine the parameters for magnetic loss unless the manufacturer tests the magnets in the lab and hands you a number or at least curves for changes of frequency at a few fixed values for intensity and visa-versa. Probably a case of over analysis, I'll just fudge some reasonable equation in the freq field of the resistor model in Microcap and make adjustments. When in doubt, just guess a ballpark solution and iterate, right? I'll shut up now. Testing a stomp box model by including the pickup load is an interesting idea, it makes a lot of sense if you're especially particular. If you want a bunch of raw data, I have a lot here www.echoesofmars.com/pickup_data/viewer/ , and here's a CSV of that data www.echoesofmars.com/pickup_data/viewer/pickups.csv , it has R(DC) L C values, and a handful of other things. I think LTSpice or some other spice implementations let you input a sound wav, so if you wanted to both approximate the load, and feed it a "guitar wave form" from a flat response guitar pickup, that's probably doable. If a pickup has an inductance under 1.5 henries, the response get's pretty flat, especially a Strat or Tele pickup, since there are hardly any eddy currents with the AlNiCo pole pieces.
|
|
pasqualino
Rookie Solder Flinger
Shifting gears and now "honing my axe" and building a quick Tweed F1
Posts: 10
Likes: 1
|
Post by pasqualino on Oct 24, 2020 11:22:24 GMT -5
"While it impresses me, I like to bear in mind that like any simulation the results are only as good as the model. The characteristics of pickups and other guitar electronic components are simplified to linear mathematical models of inductance, capacitance and resistance by using inductors, capacitors and resistors. As Antigua indicated, these are often distributed things and not really fully correct when lumped (it would be terribly complex to even begin to model them as distributed components, so I am not suggesting we ever try). "I'm an EE too, although I haven't worked analog since the early 90's when I was at Raytheon doing submillimeter (microwave) RF couplers and mixers. In order to test these components you had to "cal out" the losses and other effects of the test setup which included connectors and cables. You did this by connecting the driver of the analyzer to the rcvr through a very expensive "ideal" load or pass through. These only worked ideal loads only worked within certain bands, so you would need to recal for different spans within the same band, (let's say testing 2 to 20Ghz, you might have to break that up into s spans within that band.) These microwave systems and components are most certainly distributed systems. a capacitor looks like an inductor if it's 1/4 wavelength away, a short looks like and open and don't get me started with reflection and standing waves. It took a whole to wrap my head around a distributed sysem as opposed to a lumped component system. That part of Anigua's post caught my attention too. It's nice to not have to worry about wavelengths when doing audio. That being said, I'm searching hi and low for a good model of AC resistance from losses in the magnets. Right now I'm taking the DC resistance and taking a constant I pulled out of my fanny (4.7k ohms) and multiplying by ln(f). Micro Cap allows you do this for AC analysis (Bode) if you put an equation into the freq field of the resistor model (it will zero out the AC effects if you try to put it in the component model value field, if the freq field has a value, MC will use that and ignore the component's DC value. ) MC also lets you put sliders in for fixed value components, so you can tweak them and watch the freq response change before your eyes. I believe MC is free now, the creators are all old and decided they made enough $$$ on it and gifted it to us. That's the story, but I haven't verified it 100% though. Someone from Brookhaven National Labs hooked me up with it because I'm working a contract there. First time I have used it since the early 90's. I do FPGAs these days. Dig image and video processing and once in a while some audio. The audio is somewhat boring since it's straight replication (or an attempt) at what's taken in by the data acquisition circuitry outside the FPGA. "The notoriously difficult part of the electric guitar when viewed as a system, often completely ignored, is the effect of the volume of the amplified sound in the room..." [...Jimmy Page, Ramble On] That was made possible by the volume feeding back the acoustic signal through the guitar body - not via the strings or electronics. To model this would involve including another signal source, with another gain factor, and coupling this signal additively into the overall guitar signal through a resonant and frequency/phase-response altering body material (yes, I say again that the body material influences this especially when the volume goes way up). "Stuff like this fascinates me, now that's most certainly a distributed system!
|
|
|
Post by ashcatlt on Oct 24, 2020 13:48:24 GMT -5
I feel like unless you’re designing a pedal for exactly one guitar, the real specifics of a given pickup aren’t particularly important. We have a reasonable idea of the characteristics of a wide range of pickups and we design for whatever portion of that range we feel appropriate. Are you actually trying to develop a pickup modeling system like that Line6 guitar or the Roland system? That would be cool!
|
|
pasqualino
Rookie Solder Flinger
Shifting gears and now "honing my axe" and building a quick Tweed F1
Posts: 10
Likes: 1
|
Post by pasqualino on Oct 24, 2020 14:16:52 GMT -5
Yeah, plugging the gtr into the stomp box and then into the amp when using the old-school, pre-op amp day's circuits to try to reproduce a classic sound. Every time a great effect or amp hits the market and the mfrs (ha, mfrs could abbreviate two words here) come out with a 'new and improved" version it is never improved, they just make it worse.
I'm never going to get Jimi Hendrix's classic Fuzz Face sound, ain't happening. But if can come close to reproducing the configuration (it's well documented as I'm pretty sure you well know) and tweak things I might be able to get something really nice. Then I plan on isolating stages to reproduce the model of what I put together.
One of the things about the old-school stuff is that the lack of isolation makes any component or combination of components either already on the threshold of their tolerance or aging out of it affecting the overall system. This adds support to the case that a very accurate model for the rig and the input stage of the amp is needed to make the simulations of any stomp box I'm trying to design or any existing one I'm trying to tweak worth a damn. Blademaster2 commented on this as did you. The simulation is only as good as the model(s).
i reproduced your simulation in Microcap and messed around with that R(f) equation to try and model the losses in the magnets. I'd like to see how it bumps up against measured results. Right now I went with 5K + 47k*ln(f) for the ac resistance. I'll post a viddy of the demo up on my website and throw the link up here. I was also tweaking the cable capacitance value which gave interesting results. There's actually an optimal value. I was expecting the less capacitance the better, but it's not the case.
We've all been discussing lumped vs. distributed, isolation of stages falls somewhere in their. It makes the system more "lumped" if that's a thing. I'm all about isolation because it supports determinism. On the other hand, it may not be possible to get a good preproduction of that classic sound if the stages are isolated. This makes me cry.
There really is artistry when working with tube amps and stomp boxes with Ge transistors and diodes. A good ear trumps a good mind when all is said and done.
|
|
|
Post by JohnH on Oct 24, 2020 15:20:04 GMT -5
I can see there is interest in exploring the interactions of real pickups with some of the quirky classic effects circuits such as GE fuzz-faces etc. There has been terrific work centred right here on GN2 over the last few years in understanding the electrical characteristics of pickups, by testing and also modelling. The simplest way to get a somewhat pickup-like model into a SPICE circuit is with three components, being an R an L and a capacitance C which should include the self capacitance of the pickups plus that of the guitar wiring and cable. Such a model is easy to add to your LTSpice sim of the effect, and you can learn a lot from it, beyond just modelling the signal as a constant source. But it doesn't properly capture the losses in the pickup. Ive got a spreadsheet which is for modelling the whole of a guitar circuit and its output, from strings through pickups, pots, cable, up to the input of an amp (or stompbox). So when the testing info started to be available here, led by Anitgua, I used it to work out some more developed RLC models based on the test data, to represent pickups more accurately. So the spreadsheet uses 6 components for each , still just RLC's (3xR, 2xL and a C) so easy to code with and easy to build as a simple fixed network into a Spice model. You could use some of my 6 part models and put them as the front end of your larger models. Just add a simple resistor for volume and tone pots and caps for the tone cap and cable capacitance. I have a few dozen worked out and they cover all the generic pickup types and quite a few classic ones, all calibrated against the testing data to match response usually within a db over a wide range of loading conditions. There are other more complex or sophisticated models possible that have been explored by others for help in understanding the workings of the pickup, but if your focus is downstream of the pickup, then the 6-part models may be the easiest thing you could use. GuitarFreak: guitarnuts2.proboards.com/thread/3627/guitarfreak-guitar-frequency-response-calculator
|
|
im21
Rookie Solder Flinger
Posts: 3
Likes: 0
|
Post by im21 on Feb 4, 2021 18:35:22 GMT -5
Hi! I have a task to simulate guitar string in LTSpice. I have a electric scheme of how it should look like, but I don't know which capacitor and inductor values to put in. I will be really grateful for any information!
|
|
|
Post by JohnH on Feb 5, 2021 15:19:06 GMT -5
interesting, is it like a college assignment or something?
Hi Im21, welcome to GN2. If you are trying to make it ripple and vibrate like a physical guitar string, then that's cool. It's not something that we've done here though.
i expect there are textbook theories for it, but i wouldn't know.
Id start by picking a target frequency, say 250 hz, then pick a C and an L value based on the usual resonant frequency of a simple LC circuit. Then build your model with all values like that and see what it does, getting a transient voltage plot vs time from three nodes. Do you get a phase lag between nodes? what frequency does the whole circuit resonate at?
|
|
|
Post by antigua on Feb 5, 2021 17:58:09 GMT -5
|
|
im21
Rookie Solder Flinger
Posts: 3
Likes: 0
|
Post by im21 on Feb 6, 2021 8:52:46 GMT -5
Thanks! Yes, it's a college assignment, nothing too serious but it has to be done Professor drew this scheme and told us to see how string picking on different spots affects on sound (we have to move power source to different circuit wires and use FEM). That’s everything he told us and this is the first time we’re using this software… Do we have to use string parameters (length, mass, …) to calculate C and L? Also, I have a guitar and Focusrite Scarlett 2i2 so I was wondering if there is any software in which I could get graphic results which I could compare with ones in LTSpice?
|
|
|
Post by antigua on Feb 6, 2021 13:05:19 GMT -5
Thanks! Yes, it's a college assignment, nothing too serious but it has to be done Professor drew this scheme and told us to see how string picking on different spots affects on sound (we have to move power source to different circuit wires and use FEM). That’s everything he told us and this is the first time we’re using this software… Do we have to use string parameters (length, mass, …) to calculate C and L? Also, I have a guitar and Focusrite Scarlett 2i2 so I was wondering if there is any software in which I could get graphic results which I could compare with ones in LTSpice? I wish I understood the guitar string LTSpice model better, but it's over my head. I know that aquin43's model includes things such as the location where the string is plucked. The idea of incorporating it into FEM is even more confusing to me, but I'm very interested in seeing the end result. Try sending a message to aquin43.
|
|
|
Post by ms on Feb 7, 2021 10:00:40 GMT -5
aq's model uses delay line elements to simulate the movement of a pulse on the string. The assignment here appears to approximate a delay line from each direction from the picking location using Ls and Cs; that is a kind of transmission line. You want the speed of propagation to match that of a guitar string. The L and C per unit distance determine that. So you have a little bit of playing around and simple calculation to do to get something that works.
|
|
kitwn
Meter Reader 1st Class
Posts: 93
Likes: 22
|
Post by kitwn on Mar 15, 2023 1:00:30 GMT -5
An old thread I know, but still entirely relevant today.
There's an excellent series on using LT Spice on Max Robinson's "Fun With Tubes" website. In particular, part 6 "Installing, and using, potentiometer symbols and models" does exactly what it says, removing the need for dividing a potentiometer into a pair of resistors and allowing a plot to be made for a range of settings.
This is an example of an active tone control using two instances of the potentiometer model described with the plot showing the pots at 0-100% in 25% steps.
Kit
|
|