finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 8, 2024 16:03:59 GMT -5
Linear Hall-effect sensors are quite cheap, costing (in small amounts) less than a dollar and I started thinking why they don't seem to have been ever used as electric guitar pickups, especially in hexaphonic setups. I found a patent from 1978 regarding this topic but I am not aware that it would yet have been implemented in any practical applications. I have already started my first experiments with different magnets and sensor placements but I would be very interested in hearing about possible earlier findings in order not to waste my time making mistakes which have already been made by others. Should anyone happen to have more information regarding this topic or be able to point me to some info sources, I would really be very thankful for any hints.
|
|
|
Post by JohnH on Jun 10, 2024 5:14:29 GMT -5
Sounds interesting. Can you add a bit more info on what Hall Effect sensors are like? How do they work? I wonder what output voltage (if that's applicable) and impedance is expected? I've heard of them but never learnt more.
|
|
|
Post by mikecg on Jun 10, 2024 6:46:08 GMT -5
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 10, 2024 9:51:58 GMT -5
Hi JohnH, this excellent link describing all the basics of Hall effect sensors was already posted by mikecg. www.allegromicro.com/en/insights-and-innovations/technical-documents/hall-effect-sensor-ic-publications/hall-effect-ic-applications-guideEdit: The basic principle is that a Hall effect sensor delivers and output voltage which is dependent on the magnetic field, which the sensor detects. There are digital Hall effect sensors, where the output jumps sharply to either 0 or 1 depending on the magnetic field but in this case I am speaking about linear Hall effect sensors, where the output voltage is proportionally dependent on the magnetic field at the sensor. If the sensor doesn't "feel" any magnetic field, its output is half of the operating voltage but any changes in the magnetic field (caused e.g. by a magnetized, vibrating string) are reflected as proportional changes in the output voltage.
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 10, 2024 10:01:25 GMT -5
Thanks for the links, mikecg. After having read this discussion thread I have, however, already decided to experiment with SS49E instead of A3503 due to the better speed, sensitivity and linearity. I already bought some SS49Es for initial testing and am currently waiting for the delivery of the bigger batch (TO92S as well as SOT 23).
|
|
col
format tables
Posts: 474
Likes: 25
|
Post by col on Jun 10, 2024 14:15:17 GMT -5
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 10, 2024 14:37:04 GMT -5
Thanks for the links, col. I already knew most of them (and the Reddit-link is from me, indeed ![;)](//storage.proboards.com/forum/images/smiley/wink.png) ). The last patent-link was new to me but I suspect that it would not work because one of the claims is that the pickup would "not be reliant on wire coils or additional magnets". I strongly doubt (and my experiments hitherto seem to confirm it) that the vibrations of a string alone, without it being in a magnetic field of an external magnet, would create a sufficient change in the magnetic field to be detected by the sensor. The patent application also seems to have been withdrawn - for whatever reason.
|
|
|
Post by JohnH on Jun 10, 2024 16:09:59 GMT -5
Interesting ideas! So here's a theory question: For a linear Hall-effect sensor, is it responding with a voltage dependent on the magnitude of the flux? or of the rate of change of that flux?
Traditional pickups respond to rate of change, which adds a treble rise with frequency, compensated by a smaller amount of metal moving in the higher frequency strings. It's why when doing pickup testing with Bode plots, it's best to integrate the raw output.
Also, is the Hall-effect sensor low impedance, like a microphone? In that case it's easier to apply high signal gain to it without too much noise, like a mic preamp.
I guess the best thing is to get one, (or six), and test what it does on a guitar, maybe using the magnetism from a normal pickup, and a preamp from a standard mixer. Has that been tried?
|
|
|
Post by mikecg on Jun 10, 2024 17:25:17 GMT -5
Hello JohnH, Yes, I understand that the Hall effect sensor responds to a static magnetic field, and so if that field is the result of a current flowing through a conductor, then it will be sensing a static field arising from a dc current. The sensor will have a 'linear' region where output voltage is proportional to magnetic field intensity, and frequency response will be from dc to some upper limit, typically ~20 kHz or above. The A3503 is a typical linear sensor, with an integrated opamp buffer, and so the output impedance of the buffer will be very low.
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 11, 2024 8:48:54 GMT -5
I haven't (yet) myself compared in detail the specs of the A3503 against the specs of the SS49E, which I am currently experimenting with but I trusted the information I got from the thread which I mentioned already previously. The Honeywell SS39ET/SS49E/SS59ET Series specs in this PDF document indicate the linear frequency range to go up to around 35kHz, which would surely be sufficient for this purpose. ![]() ![](https://i.postimg.cc/qvpK813K/Screenshot-2024-06-11-at-16-20-46.png)
So far I have experimented with neodymium buttons and ceramic bar magnets (also in a Halbach array setup) under the strings and the sensor but have achieved the most promising results by placing the sensors between the pole pieces of a single coil pickup and the string (like JohnH suggested ![:)](//storage.proboards.com/forum/images/smiley/smiley.png) ), slightly off-center from the pole piece. Using existing inductive pickup magnets would obviously simplify the setup of the construction but on the other hand, the distance from the strings could not be adjusted quite as small as it could be if the sensors were placed even nearer to the bridge. The distance between sensor and string producing best usable results seems to be quite small (around 0,5mm). I am not good enough in physics to be able to tell whether there would be some ways to "shape" the magnetic field so that this distance could be increased by maximizing the amplitude of the fluctuations of the magnetic field caused by the vibrating string. Although the SS49E already has a built-in voltage regulator, I used an additional external 78L05 to regulate the 9V battery output to 5V - just to be on the sure side and to avoid possibly affecting the characteristics of the sensor response by decaying battery voltage. I use an 1 uF capacitor as a high-pass filter in the output of the SS49E to eliminate the DC component from the output signal. Whether the outputs of several sensors can be connected in parallel and then fed through one common capacitor or whether each sensor would need a capacitor of its own, still remains to be seen. After I have some multi-string test results, I will probably "reduce to the max" the complexity of the circuitry. Perhaps even the external voltage regulator might be superfluous. Different sources indicate the maximum supply voltage of the SS49E to be between 6,5V and 8,0V so using the 78L05 for the time being eliminates at least one source of failure.
|
|
|
Post by ms on Jun 11, 2024 11:15:58 GMT -5
I use an 1 uF capacitor as a high-pass filter in the output of the SS49E to eliminate the DC component from the output signal. Whether the outputs of several sensors can be connected in parallel and then fed through one common capacitor or whether each sensor would need a capacitor of its own, still remains to be seen. After I have some multi-string test results, I will probably "reduce to the max" the complexity of the circuitry. Perhaps even the external voltage regulator might be superfluous. Different sources indicate the maximum supply voltage of the SS49E to be between 6,5V and 8,0V so using the 78L05 for the time being eliminates at least one source of failure. The output impedances appear to be low (emitter follower or maybe op amp); thus, multiple outputs should be added through resistors. The best way would use an inverting op amp as a summer. I would start with separate capacitors (DC blocks), but you might find a way to use just one.
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 11, 2024 12:41:17 GMT -5
multiple outputs should be added through resistors. The best way would use an inverting op amp as a summer. I would start with separate capacitors (DC blocks), but you might find a way to use just one. If I understand the summing suggestion correctly, the circuitry would look something like this - at least in the beginning: ![](https://i.postimg.cc/rs3JV3y3/Screenshot-2024-06-11-at-20-30-44.png) What might be good values for the resistances? The trimmer should probably be at least twice that value to allow some increase in gain.
|
|
|
Post by mikecg on Jun 11, 2024 14:07:31 GMT -5
Hello finn, I think ms had something more like this in mind: ![](https://i.postimg.cc/FFTRYPt9/summing-amp-for-hall-sensors-01.png)
|
|
|
Post by ms on Jun 11, 2024 14:21:23 GMT -5
The dc voltages will not be exactly the same, and so there will be some dc current through the resistors. That might be OK; if not, then use six caps.
I guess the first question is sensitivity: good enough or not? The Allegro A1309 has about 9 mv change for 1 Gauss change. I have not found any information on how much the field changes when a string vibrates, and so it might be necessary to measure it, and yes, the easiest way to do that is to build one and see!
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 12, 2024 12:27:16 GMT -5
I think ms had something more like this in mind Thanx for adding the properly biased op-amp. Is there a specific reason why you picked a 4.7µF capacitor instead of the 1µF I have used, so far? BTW: Which app did you use for the diagram?
|
|
|
Post by mikecg on Jun 12, 2024 13:57:10 GMT -5
Hello finn, The 4.7uF cap combined with the 6 x 10k resistors in parallel (1.66k) produces a low frequency 'roll-off' (high pass) that is -3db down at ~ 20 Hz. This ensures that the passband gain is reasonably flat for all normal guitar frequencies. If you use a 1uF cap you will experience some attenuation at bass frequencies. I use the 'MicroCap 12' circuit simulator. It is no longer supported by Spectrum Software, but is available for free download, here: archive.org/details/mc12cd_202110
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 12, 2024 14:08:36 GMT -5
Thanks, mikecg. I checked out MicroCap 12 but was unfortunately unable to install it on my mac (OS: Big Sur 11.7.10). BTW: Here is a picture of my current test setup. ![](https://i.postimg.cc/YS03CfKB/Piezo-test-setup-2.png) To my surprise, the sensor seems to work quite well also with the neck pickup, although the distance to the string is bigger.
|
|
|
Post by mikecg on Jun 13, 2024 6:23:38 GMT -5
Hello finn, Interesting! It looks like you have it wired directly to your guitar amp? Can you comment on signal level, and tone for each of the strings? I would expect to hear clean tones with extended highs. Hum may be an issue but a 'humbucker type' solution will be possible.
|
|
|
Post by ms on Jun 13, 2024 9:32:11 GMT -5
Hello finn, Interesting! It looks like you have it wired directly to your guitar amp? Can you comment on signal level, and tone for each of the strings? I would expect to hear clean tones with extended highs. Hum may be an issue but a 'humbucker type' solution will be possible. But here the voltage change is proportional to the change in field, but a normal pickup is sensitive to rate of change of the magnetic field. So it might be lacking in highs. to illustrate the difference, displace the string with a sinusoidal motion. with the hall effect pickup: the voltage is the same at any frequency up to the cutoff of the device. with the normal pickup: the voltage increases proportional to frequency.
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 13, 2024 9:36:36 GMT -5
In my test setup I have, indeed, connected the output of the sensor directly (through the 1µF capacitor) to the amp input.
The signal level seems to depend very strongly of the exact position of the sensor but at the loudest the level seems to be in a comparable region with the inductive pickups.
I have not yet tested every string individually but - as expected - the thicker strings seem to deliver a stronger signal than the thinner ones, because the bigger magnetized mass influences the magnetic field more.
I will probably not be able to make any reliable judgement regarding the quality of the tone as well as hum, hiss etc. before I have some fixed 6-sensor prototype attached to the guitar and can directly compare it with the inductive pickups.
|
|
|
Post by mikecg on Jun 13, 2024 10:48:35 GMT -5
Hello again finn, Some suggestions: Try to minimize the sensor load capacitance (100pF max), so try using a short cable link to the amp. Ideally, the diameter of the sensitive area of the sensor should be close to the diameter of the pickup pole piece. This may be an important guide to the selection of an appropriate sensor for this application. If the sensor sensitive area is much less than the pole face area, then sensitivity will suffer. An imbalance in sensor output due to string diameter can in theory be compensated for by the traditional method of adjusting the height of the pole pieces, summing the output of vertically stacked sensors on the thinner strings, or adjusting the individual string sensor output levels, electronically. As you point out, a more meaningful assessment of sensitivity and tone must wait for the construction and test of a 6 sensor prototype.
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 13, 2024 11:35:33 GMT -5
The dimensions of the sensitive area of the Hall effect sensors I have found so far are in the range of 0,25mm (1/100") so the placement of the sensor seems to be quite a delicate matter. ![](https://i.postimg.cc/VLBZ3p6v/Screenshot-2024-06-13-at-19-29-21.png) This "disadvantage" does, however also bring good possibilities to balance the volume of the individual strings by making minor adjustments to the sensor position.
|
|
|
Post by ms on Jun 13, 2024 19:16:22 GMT -5
Hello again finn, Some suggestions: Try to minimize the sensor load capacitance (100pF max), so try using a short cable link to the amp. Ideally, the diameter of the sensitive area of the sensor should be close to the diameter of the pickup pole piece. This may be an important guide to the selection of an appropriate sensor for this application. If the sensor sensitive area is much less than the pole face area, then sensitivity will suffer. An imbalance in sensor output due to string diameter can in theory be compensated for by the traditional method of adjusting the height of the pole pieces, summing the output of vertically stacked sensors on the thinner strings, or adjusting the individual string sensor output levels, electronically. As you point out, a more meaningful assessment of sensitivity and tone must wait for the construction and test of a 6 sensor prototype. The output of the sensor is a low impedance amplifier. Load capacitance of reasonable length cables should not be a problem.
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 19, 2024 2:25:25 GMT -5
Update: This is a reply which I got from a well-respected expert in the area of guitar electronics: Well... being a "guitarnut" I think that I will still continue experimenting, despite of that chastening forecast. I just happen to be a guy who doesn't easily take "no" for an answer
|
|
|
Post by JohnH on Jun 19, 2024 3:30:21 GMT -5
Yes gotta carry on! We need to hear sound samples!
|
|
|
Post by ms on Jun 19, 2024 6:17:58 GMT -5
Update: This is a reply which I got from a well-respected expert in the area of guitar electronics: Well... being a "guitarnut" I think that I will still continue experimenting, despite of that chastening forecast. I just happen to be a guy who doesn't easily take "no" for an answer I don't think the expert has it quite right. If you accept that you need to provide power for the sensors, adding additional electronics for equalization does not seem like a big deal. The highs are not completely missing; they are reproduced at their "natural" level, which is, of course, not what you want for the usual electric guitar. I see a potential problem with signal to noise ratio because of the equalization, however. I believe that the noise would come from the sensor itself, and so if it is a problem there would be little you could do about it. The frequency response should stop rising by about 5KHz to avoid increase in noise at frequencies you do not need. You will not hear it of course, since guitar speakers (and simulators) have little response there, but it is better not to have extra noise, especially in a situation where you might use a non-linear response. Have you measured the output dc voltage with the sensor and magnet in place? With no field it is in the middle of the range, but a strong field moves towards one limit or the other. If you use a very sensitive sensor in order to get good sensitivity to changes from string vibrations, it might saturate from the strong field (that is, the voltage might move too close to one of the limits).
|
|
|
Post by mikecg on Jun 19, 2024 8:32:48 GMT -5
Update: This is a reply which I got from a well-respected expert in the area of guitar electronics: Well... being a "guitarnut" I think that I will still continue experimenting, despite of that chastening forecast. I just happen to be a guy who doesn't easily take "no" for an answer Hello again finn, Firstly, let's not get carried away, just yet, by the opinions of 'well respected experts'! So, let us inspect his 'theories' in a little more detail: "The invention did not catch on". It would be interesting to learn, what precisely, this assertion refers to. Perhaps your 'expert' could provide some documented evidence? "The output voltage is proportional to the string deflection". No, the output is proportional to the change in field intensity intercepted by the sensor. That field, is a sample of the field in the magnetic circuit that comprises the string, air gap, sensor, pole piece, magnet, and return air gap path. The magnetic field in a cylindrical volume whose axis is centred on the string does not have a uniform intensity, and therefore the time dependent change due to string movement will be some function of the amplitude of the string movement, the average field through which that movement passes, and the diameter of the magnetized component of the string. "The highs are completely missing". No, the highs may be more attenuated, not "completely missing". "Then a low-pass filter with resonance boost". I think he may have meant to say high pass filter, and, of course, there would be no "resonance" to "boost"! Perhaps he is actually trying to imply that if you wanted to 'clone' the tone of a magnetic pickup, you would need to add some 'EQ' to simulate their resonant high frequency response. "The sensors draw a lot of current". Typically around 7 mA apiece - around 42 mA for six sensors. So, yes, that would mean that a decent 9V PP3 (~700 mAh) would only last for about 16 hours. However, if a hexa-phonic connection is envisaged, phantom power could be made available, so this may not be a show-stopper. I must agree with JohnH's most recent invocation: "Yes - gotta carry on!"
|
|
|
Post by ms on Jun 19, 2024 9:30:41 GMT -5
Hello again finn, Firstly, let's not get carried away, just yet, by the opinions of 'well respected experts'! So, let us inspect his 'theories' in a little more detail: "The invention did not catch on". It would be interesting to learn, what precisely, this assertion refers to. Perhaps your 'expert' could provide some documented evidence? "The output voltage is proportional to the string deflection". No, the output is proportional to the change in field intensity intercepted by the sensor. That field, is a sample of the field in the magnetic circuit that comprises the string, air gap, sensor, pole piece, magnet, and return air gap path. The magnetic field in a cylindrical volume whose axis is centred on the string does not have a uniform intensity, and therefore the time dependent change due to string movement will be some function of the amplitude of the string movement, the average field through which that movement passes, and the diameter of the magnetized component of the string. "The highs are completely missing". No, the highs may be more attenuated, not "completely missing". "Then a low-pass filter with resonance boost". I think he may have meant to say high pass filter, and, of course, there would be no "resonance" to "boost"! Perhaps he is actually trying to imply that if you wanted to 'clone' the tone of a magnetic pickup, you would need to add some 'EQ' to simulate their resonant high frequency response. "The sensors draw a lot of current". Typically around 7 mA apiece - around 42 mA for six sensors. So, yes, that would mean that a decent 9V PP3 (~700 mAh) would only last for about 16 hours. However, if a hexa-phonic connection is envisaged, phantom power could be made available, so this may not be a show-stopper. I must agree with JohnH's most recent invocation: "Yes - gotta carry on!" For small displacements, like string vibration, "The output voltage is proportional to the string deflection" is a very good approximation. "Then a low-pass filter with resonance boost". This is exactly what he means. Once you have taken care of the 6db per octave increase, you have simulate the response of the electrical circuit of the normal guitar pickup. You need to pass low frequencies up to some cutoff with an increase just before the drop off (or possibly not for some pickups). This is exactly what an underdamped second order electronic low pass filter does.
|
|
finn
Apprentice Shielder
![*](//storage.proboards.com/forum/images/stars/star_blue.png)
Posts: 25
Likes: 1
|
Post by finn on Jun 19, 2024 9:57:46 GMT -5
Have you measured the output dc voltage with the sensor and magnet in place? With no field it is in the middle of the range, but a strong field moves towards one limit or the other. If you use a very sensitive sensor in order to get good sensitivity to changes from string vibrations, it might saturate from the strong field (that is, the voltage might move too close to one of the limits). When the sensor is laid directly over the pole piece of the inductive pickup, the DC output seems to be either around 1.5 or 3.5 Volt, depending on which way of the sensor points towards the magnet. Of course pickups with stronger magnets might push the voltage to saturation limits. With my current setup, however, there should still be sufficiently of the linear range available for the voltage changes caused by the string vibrations because the linear range (with 5V supply voltage) is between 1 and 4 volt. ![](https://i.postimg.cc/tTSr9B01/Screenshot-2024-06-19-at-17-52-54.png)
|
|
|
Post by ms on Jun 19, 2024 12:19:05 GMT -5
When the sensor is laid directly over the pole piece of the inductive pickup, the DC output seems to be either around 1.5 or 3.5 Volt, depending on which way of the sensor points towards the magnet. Of course pickups with stronger magnets might push the voltage to saturation limits. With my current setup, however, there should still be sufficiently of the linear range available for the voltage changes caused by the string vibrations because the linear range (with 5V supply voltage) is between 1 and 4 volt. ![](https://i.postimg.cc/tTSr9B01/Screenshot-2024-06-19-at-17-52-54.png) Yes, it seems to be working just the way it should. Thanks.
|
|