Ebow Exposé Part IV - RESURRECTED!

[anotherjim]

Could also be my driver coil is too big and loading down the LM386 output too much?

Generally, a higher impedance load reduces current consumption.
The choice of 220uF for the 386 output capacitor "might" indicate an 8ohm load since the low-frequency point is 90Hz which suits guitar range, but 220uF could just be a safe bet "plenty big enough" value and the coil impedance is actually higher than 8 ohm.

It may be that working the 386 with a reduced load alters the effectiveness of the 4.7uF in the circuit as its impedance will be less significant. That could be re-scaled with a lower-value cap. If you take it as a simple RC filter, 8ohm and 4.7uF is the same as 80ohm and 0.47uF.

[Paul Marossy]

Could also be my driver coil is too big and loading down the LM386 output too much?

Generally, a higher impedance load reduces current consumption.
The choice of 220uF for the 386 output capacitor "might" indicate an 8ohm load since the low-frequency point is 90Hz which suits guitar range, but 220uF could just be a safe bet "plenty big enough" value and the coil impedance is actually higher than 8 ohm.

It may be that working the 386 with a reduced load alters the effectiveness of the 4.7uF in the circuit as its impedance will be less significant. That could be re-scaled with a lower-value cap. If you take it as a simple RC filter, 8ohm and 4.7uF is the same as 80ohm and 0.47uF.

Right, but doesn't that also depend somewhat on the wire gage? I was thinking bigger wire = more current. I'm pretty sure that the wire I used was several gages larger than what's in the actual unit. In the DIY Ebow I made a week or so ago, using the internet schematic that is wrong, the LM386 seems to be very sensitive to what is happening at the input & output coils as far as the DC resistance and wire gage is concerned. I apparently killed one LM386 on my first attempt at trying to find an output coil that worked (ended up with 32 ohms, but I jumped from 8 ohms directly to 32). I also used wire gage that was too heavy, but my only goal was to see if it could actually work. I ended up with something which didn't really work - it got the string vibrating but it was not even close to being a usable device. I'm sure it could be refined somewhat but I still don't think I can get it to be a usable circuit.

I watched that YouTube video of the Russian(?) guy who made a working one with a 3D printed housing, but he was not using the internet schematic. Looks to me like he made the version of Ebow that pre-dates the Plus Ebow, and it only had an on/off switch. Looks like it had a feedback loop, otherwise it should have had one more resistor, two more capacitors and a diode. No idea what he used for the coils but he made it work by having them extend out of the bottom of the 3D printed housing. Probably had to do that because the coils weren't close enough to the strings otherwise.

On that diode, I was thinking last night that maybe it's intended to be a sort of limiter, or maybe to create asymmetry in the waveform? It must be there for some reason.

[anotherjim]

Diode is just one-way non-return valve. In the forward direction, the diode switches on and the cap across it is bypassed, in reverse, the only path is the cap since the diode will switch off.
This is definitely going to be asymmetric.

In normal mode, with the whole circuit DC current flowing in the forward direction back towards the battery negative, the diode never switches off and any AC current also gets a path thru the 4.7uF cap across the diode. You see a small waveform across the diode/cap because of the small DC volt drop of the diode junction and that the 4.7uF cap does not have zero impedance at the frequency. Rob mentioned earlier that the diode will offer some reverse polarity protection if someone tries to fit the battery the wrong way. What if the older E-bow without Harmonic mode also had this feature?

Switched over to the coil return only in harmonic mode, the negative half cycle of the output waveform has to go via the cap. For the positive half cycle, the diode switches on and bypasses the cap. This only happens like this in Harmonic mode.
When the waveform is negative going, the cap introduces an RC high pass filter (Actually RLC but ignoring the coil inductance for now). If the 386 is outputting square waves, a high pass filter will make it look more like a curved sawtooth with even harmonics taking over from the odd harmonics of the squarewave. However, the power in the coil will be reduced with the narrower waveform of a sawtooth so that may be the reason for treating only one half cycle of the wave.

[Paul Marossy]

Diode is just one-way non-return valve. In the forward direction, the diode switches on and the cap across it is bypassed, in reverse, the only path is the cap since the diode will switch off.
This is definitely going to be asymmetric.

In normal mode, with the whole circuit DC current flowing in the forward direction back towards the battery negative, the diode never switches off and any AC current also gets a path thru the 4.7uF cap across the diode. You see a small waveform across the diode/cap because of the small DC volt drop of the diode junction and that the 4.7uF cap does not have zero impedance at the frequency. Rob mentioned earlier that the diode will offer some reverse polarity protection if someone tries to fit the battery the wrong way. What if the older E-bow without Harmonic mode also had this feature?

Switched over to the coil return only in harmonic mode, the negative half cycle of the output waveform has to go via the cap. For the positive half cycle, the diode switches on and bypasses the cap. This only happens like this in Harmonic mode.
When the waveform is negative going, the cap introduces an RC high pass filter (Actually RLC but ignoring the coil inductance for now). If the 386 is outputting square waves, a high pass filter will make it look more like a curved sawtooth with even harmonics taking over from the odd harmonics of the squarewave. However, the power in the coil will be reduced with the narrower waveform of a sawtooth so that may be the reason for treating only one half cycle of the wave.

I see, that makes sense. Thanks for explaining that! Your guess is as good as mine as far as whether or not the Ebow version before this one had that diode in it. I tend to think not, but maybe they had one for short circuit protection. Maybe they came up with this quirky arrangement simply because they have very little room to spare on that tiny PCB.

[Rob Strand]

Here's some deeper dive analysis of Harmonic mode,






So after all that I ask myself if it's only resonating the inductance with smaller output cap to get more gain, why have the added complexity of the diode and DC shifting!

[Paul Marossy]

Here's some deeper dive analysis of Harmonic mode,

Very interesting stuff! Seems to me like the diode clamp is to create an asymmetrical waveform. What happens if that is not present? Will that still drive the string? Or will it drive the string and who knows what the result is? The "Ebow" that I recently made from the internet schematic doesn't come anywhere close to this circuit... it's a wild mess that makes little sense.

So after all that I ask myself if it's only resonating the inductance with smaller output cap to get more gain, why have the added complexity of the diode and DC shifting!

Is it enough to just change the frequency of the oscillator? Or does there need to be a corresponding change in the waveform to get the string to go one way or the other? The harmonic mode has those "extra" humps in the waveform... seems like that must be a factor. What keeps the string from just resonating at its normal frequency? Seems like that goes back once again to the place where the anti-nodes occur on the string. I think all these things work together. I did a crude model of just that diode/cap arrangement with the coils inductively coupled to guitar pickup equivalent circuits along with that simple battery impedance tweak, and I could easily get the harmonic mode waveform at the pickup but I couldn't achieve the regular mode waveform. It's quite likely that I just don't know how to really model that properly. Or maybe it needs to be far more complex so we can add the string wave characteristics? I don't want to really take if that far, that's really just a rhetorical question. That LTSpice is pretty amazing for a free program.

I think at this point we have an "Everything You Ever Wanted To Know About The Ebow But Were Afraid To Ask". Or maybe it's everything you didn't ask about.  Well, at least I've satisfied my own curiousity about it. Sure have learned a lot thru this whole process, and it was fun coming up with these testing jigs and whatnot. 

[Rob Strand]

Very interesting stuff! Seems to me like the diode clamp is to create an asymmetrical waveform. What happens if that is not present?

Once the caps charge up there's no asymmetry, in fact the diode is essentially removed from the circuit.   It serves only to shift the DC level.   When the Ebow is first placed on the string it might take a few cycles for the cap to charge up, so have will have a short transient.

We expect the diode to do nothing, and pulling the diode sure shows that in the simulation.   You can see the DC voltage across the out cap is now back to normal (Vcc/2).

Will that still drive the string? Or will it drive the string and who knows what the result is?

I'm pretty sure it will drive the string.   The point about the 4u7 cap forming a resonance with the output coil inductance is not so much will it work, but will generate the higher harmonics as advertised.   The idea behind it should help at least!

On your rebuilt "version 2" transmit coil the inductance is higher so you will need smaller cap to resonate at the same frequency.    Since both the inductance and resistance of the coil are different you might find it can still resonate in a similar way.

The "Ebow" that I recently made from the internet schematic doesn't come anywhere close to this circuit... it's a wild mess that makes little sense.

I guess the on-line schematic doesn't have all the frequency shaping tweaks.  It doesn't support harmonic mode so it is less fussy as well.    However, over all it's still a feedback system like the Ebow.   I wonder if those DIY versions make the string vibrate at the fundamental, wheres as the normal mode for Ebow oscillates at the second harmonic (probably due to the frequency shaping).

Is it enough to just change the frequency of the oscillator? Or does there need to be a corresponding change in the waveform to get the string to go one way or the other? The harmonic mode has those "extra" humps in the waveform... seems like that must be a factor. What keeps the string from just resonating at its normal frequency?
Seems like that goes back once again to the place where the anti-nodes occur on the string. I think all these things work together.

The way I see it at the moment is the 4u7 cap resonating with the output coil is what promotes the string to vibrate at a higher frequency.    From the earlier posts, the string is a high Q resonator so it doesn't case so much about phase.  You just make the gain high at high frequencies and it will cause the string to vibrate at the higher frequency instead of the lower one (since the string has many possible modes of vibration).

I did a crude model of just that diode/cap arrangement with the coils inductively coupled to guitar pickup equivalent circuits along with that simple battery impedance tweak, and I could easily get the harmonic mode waveform at the pickup but I couldn't achieve the regular mode waveform.

The waveform shape is one thing but what frequency the string vibrates at comes from a completely different cause.  The string is a resonator which can vibrate at many different frequencies.  The spice model for the string was simplified in that it only modeled the fundamental mode.   If you put a more complex and physically realistic string model in spice then it will have the capability to vibrate at different frequencies, and once you do that the changes you make in the circuit will show up as a physical effect.  For example, boosting the HF gain will cause the string to vibrate at a harmonics.

If you wanted to use spice to simulate placing the Ebow at different *positions* on the string and exciting the string around node points then you would need to add a lot more to the string model.    How does LTspice know what the spacing between the coils is and where you are placing the Ebow?   You would need parameters for all that.   It takes a lot of effort to get all that to happen.

It's quite likely that I just don't know how to really model that properly. Or maybe it needs to be far more complex so we can add the string wave characteristics? I don't want to really take if that far, that's really just a rhetorical question.

It's not a simple matter to capture all aspects of the string.   If it was easy, I would have tried it out already     

That LTSpice is pretty amazing for a free program.

Yes, really cool.  It's like a big circuit calculator.  I used to do all a lot of this stuff with algebra and hand-written programs.   Circuit simulators have been around for a long time. They started off with only text output devices like terminals and printers.  You could get free programs not long after PC's were around, first DOS text, then DOS with graphics, then the various Windows versions.    Around the late 1990's to early 2000's there were heaps of free ones.   Since LTspice arrived it's demotiviated a lot of that effort.

[Rob Strand]

Here's a very crude model of a string with multiple resonances.
The loss resistance 10k ohm is the same for each resonance.
The anti-resonance at 250Hz looks a bit dodgy - I haven't thought about it enough  .

The main point is it has multiple resonances.



FYI, here's an LCR version with multiple resonances.

[Rob Strand]

I did a quick simulation with the following set-up:
- A string with three modes, frequencies f1, 2*f1, 3*f1.
- Ebow set to normal mode
- Ebow schematic, parts for original circuit as we see it so far.
- Feedback resistor set so no self oscillation
- Strength of string coupling set so unit starts string oscillation in about 0.5 second

I then increased the input filter cap (the 33nF cap) so it starts to filter the higher frequencies.
As the input filter cut-off becomes lower it promotes the string to oscillate at the lower
frequency modes.   The aim is to show that modifying the response can cause different
string modes to occur (no intent to show more than that).

Result: it does just that.

Here's the schematic:



From the table you can see how large the filter cap needs to be before the string changes to the
next lower frequency mode.

Here's the waveforms when the input cap is 1, 3, 6 times the normal value.   kc was chosen
to be a reliable value for each frequency based on the table on the schematic.  Notice
that it doesn't take much to stop the 1200Hz mode, that means the input cap and
inductance values are just making it.



The next step would be to get the string mode to change when switching from Normal mode to Harmonic mode.

As far as the Rebuilt Ebow goes, it should be clear that if the coils aren't the same as the original the resulting changes in the frequency response could screw-up Normal-mode or Harmonic-mode.   We might be able to tune the input cap (33n) to help but switching between the two modes may or may not work with different coils.

[Paul Marossy]

I did a quick simulation with the following set-up:
- A string with three modes, frequencies f1, 2*f1, 3*f1.
- Ebow set to normal mode
- Ebow schematic, parts for original circuit as we see it so far.
- Feedback resistor set so no self oscillation
- Strength of string coupling set so unit starts string oscillation in about 0.5 second

I then increased the input filter cap (the 33nF cap) so it starts to filter the higher frequencies.
As the input filter cut-off becomes lower it promotes the string to oscillate at the lower
frequency modes.   The aim is to show that modifying the response can cause different
string modes to occur (no intent to show more than that).

Result: it does just that.

Very interesting. So I guess the different waveforms are just kinda incidental and not necessarily intentional?

Here's the schematic:



From the table you can see how large the filter cap needs to be before the string changes to the
next lower frequency mode.

Here's the waveforms when the input cap is 1, 3, 6 times the normal value.   kc was chosen
to be a reliable value for each frequency based on the table on the schematic.  Notice
that it doesn't take much to stop the 1200Hz mode, that means the input cap and
inductance values are just making it.



The next step would be to get the string mode to change when switching from Normal mode to Harmonic mode.

The capacitor values & corresponding frequencies is useful info. If I ever feel like revisiting this project, or maybe try building a proper one from scratch this info will be handy. I'm wondering what the Russian(?) dude with the DIY Ebow used for his coils. Those worked but they probably needed to be a little bit stronger so he didn't have to make them extend thru the bottom of his 3D printed housing.

As far as the Rebuilt Ebow goes, it should be clear that if the coils aren't the same as the original the resulting changes in the frequency response could screw-up Normal-mode or Harmonic-mode.   We might be able to tune the input cap (33n) to help but switching between the two modes may or may not work with different coils.

Yeah I think that is my situation with the re-built one. It has trouble with one mode but not with the other. I had socketed that fb cap and had the 20-turn pot in place of the fb resistor and I just messed with it until I found the best compromise. Was far from perfect, but my only real goal was to get it working again on some level. Did better than I expected I would.

[Rob Strand]

Very interesting. So I guess the different waveforms are just kinda incidental and not necessarily intentional?

I'd say incidental to a point.   Many oscillators use clipping to control the gain and it's amazing how crappy the waveforms can be.   The resonant/filtering network, smooths it all out.  In the case of the Ebow the smoothing would be done by the string.  However, the thing about the Ebow is when you place the drive coil over the pickup the crappy output waveform is getting injected directly into the pickup so you will hear some of it.

The capacitor values & corresponding frequencies is useful info.

There's quite a few caveats:
- Those cap *values* only apply to the assumed 150mH coil.  If you change the inductance of the coil you need different cap values to do the same thing,  C_new = C_old * L_old / L_new;  where L_old = 150mH and C_old are the values from the table.   
- A second thing is my string is 400Hz (with modes 800Hz and 1200Hz).  The different strings and fretted notes cover a wider range.
- If you hold the ebow close to the string (ie. increasing the coupling kf) you might get slightly different results.

I guess the main point is you can get the string to change what mode it resonates at but how to do it is buried in a zillion specific details - that's where the engineering of the Ebow has already been done.

I'm wondering what the Russian(?) dude with the DIY Ebow used for his coils. Those worked but they probably needed to be a little bit stronger so he didn't have to make them extend thru the bottom of his 3D printed housing

The Russian guys I saw were using those old buzzer/speakers which used to be inside the cases of PC's.  You can buy these.  They are electromagnetic devices, not the piezos types which look identical from outside.   I posted some stuff in the "other" thread.

Yeah I think that is my situation with the re-built one. It has trouble with one mode but not with the other. I had socketed that fb cap and had the 20-turn pot in place of the fb resistor and I just messed with it until I found the best compromise. Was far from perfect, but my only real goal was to get it working again on some level. Did better than I expected I would.

I think overall a tight enough box has been put around the problem to get the jist of what needs to be looked at.   We have some idea of causes and effect.

[Rob Strand]

I did a quick simulation with the following set-up:
- A string with three modes, frequencies f1, 2*f1, 3*f1.
- Ebow set to normal mode
- Ebow schematic, parts for original circuit as we see it so far.
- Feedback resistor set so no self oscillation
- Strength of string coupling set so unit starts string oscillation in about 0.5 second

FYI: While the results look convincing, there's a bit more to the why it works than I expected,
particularly for the case with kc=3.   I'll put it on the back-burner for now.

[soggybag]

This is very fascinating. I have a a couple Sustainiac sustainers. if you're not familiar, it's like the EBow but is a pickup that installs in the neck position. You can't I've the sustainer and the signal from the bridge pickup drives the sustainer which drives the stings like the EBow.

It has a couple modes, like the EBow, a fundamental mode, harmonic mode, and a mix mode that mixes the two. I could post some pictures if that would be interesting?

[Rob Strand]

This is very fascinating. I have a a couple Sustainiac sustainers. if you're not familiar, it's like the EBow but is a pickup that installs in the neck position. You can't I've the sustainer and the signal from the bridge pickup drives the sustainer which drives the stings like the EBow.

I only know if from my early digging around on this Ebow stuff.  I have a PCB pic but it has heatshrink over it.

It has a couple modes, like the EBow, a fundamental mode, harmonic mode, and a mix mode that mixes the two. I could post some pictures if that would be interesting?

Be interesting to see the structure they use and if they use the same tricks.    One thing's for sure, the devils in the details with these things: coil inductances etc.

Do you know if normal mode on these devices vibrates the string at the fundamental or the second harmonic (like the Ebow)?

[Paul Marossy]

It has a couple modes, like the EBow, a fundamental mode, harmonic mode, and a mix mode that mixes the two. I could post some pictures if that would be interesting?

Be interesting to see the structure they use and if they use the same tricks.    One thing's for sure, the devils in the details with these things: coil inductances etc.

Do you know if normal mode on these devices vibrates the string at the fundamental or the second harmonic (like the Ebow)?

I have a Fernandes Sustainer System on one of my guitars. Was broken, given to me by a dude at Guitar Center in 2009 I think it was. I fixed it and put it into my old hot rodded '92 Ibanez EX370-FM. It's kinda like the Sustainiac pickup but without the mix mode. The guitar has to be rewired so that all of the pickups go thru their circuit board. When you activate the sustainer, it bypasses all of the pickups and uses only the bridge pickup. I like it because I can get those Phil Keaggy volume pedal with an Ebow kind of sounds on any combination of strings and play chords too. Sometimes I just leave it on while doing riffs or whatever, kinda fattens up everything and it's like being on the verge of feedback all the time.

Anyway I never thought to look at the output of it before. I just looked at the guitar output with the scope. I let the open B string sustain and in the "regular" mode it appears to be tripling the fundamental, and in "harmonic" mode it appears to be around 1.5kHz... so that's about 6x the fundamental? I didn't think it was that high! Both of the waveforms are basically a sine wave. Quite different than the Ebow. I wonder how the Sustainiac compares. I found the patent for it some time ago, it's got quite a few pages and my head melted when I started reading through it.  This was around the time when there was an enormous DIY sustainer thread going on at ProjectGuitar dot com forum, ca. 2007. That's how I obtained that destroyed Ebow PCB, from someone over there. That thread apparently doesn't exist anymore. 

[Rob Strand]

Anyway I never thought to look at the output of it before. I just looked at the guitar output with the scope. I let the open B string sustain and in the "regular" mode it appears to be tripling the fundamental, and in "harmonic" mode it appears to be around 1.5kHz... so that's about 6x the fundamental?

Interesting.    Getting it to consistently pull out the 6th harmonic with a basic feedback loop is going to be tricky.   Perhaps the drive-coil and pickup positions need to be factored in.  From what I can see the drive-coil is always towards the neck and the output coils are bridge and/or middle.

My naivety regarding guitar sustainers is starting to show because I didn't realize there were so many alternative products to Ebow.

I found the patent for it some time ago, it's got quite a few pages and my head melted when I started reading through it.

There's a very long patent on sustainers.  The idea looked a lot like the DIY Ebows, perhaps with a phase switch.   I need to read it again to see if it mentions any of the things we have learnt from the Ebow.

[Paul Marossy]

Here's one more interesting tidbit. This morning I was curious to see what kind of waveforms that the previous version of the Ebow produced (the black one with white logo and red LED). No big surprises but I think it explains a bit about the Plus Ebow which I have been focused on. I'm curious to know if it also has that cap & diode arrangement. I wish I had an X-ray machine just about now. 

You can see the results in this short video

[soggybag]

Here are some pictures of my Sustainiac. I'm searching for the wiring diagram. Besides driving the sustainer the board also handles switching. When the sustainer is activated the bridge pickup is on - regardless of the switch position - and the guitar will work even when the battery is dead.

[deadastronaut]

soggybag : https://www.sustainiac.com/install.htm

lots of different wiring types..depending on guitar pups etc.....

[Rob Strand]

Here are some pictures of my Sustainiac.

It's going to take a bit of effort going through that one, SMD, lot's of external connections.

I'm searching for the wiring diagram.

I had a wiring diagram from part 2 of this blog. Superficially it seems to match the PCB.

https://settechitarre.wordpress.com/tag/diy/

but deadastronaut's link is more comprehensive.

The Sustainiac patents give some info on the circuits and the thinking behind them.  They are very long.  I can't even remember which one I found earlier on, maybe the 2000 version.
US4941388 - Sustainaic, Hoover 1990
US5932827 - Sustainaic, Osborne & Hoover 1995
US6034316 - Sustainiac, Hoover 2000

The IC's on the PCB all look fairly common.  The NDS8958 is a Fairchild P-channel + N-channel MOSFET - probably used to drive the coil.   Plenty of transistors and diodes I haven't looked up.