Let's design a tap tempo tremolo.

[Taylor]

Since this is a digitally derived LFO, I'm guessing the square wave is a true 50% duty cycle square.

[Processaurus]

... it is possible to bring the LFO back into line with what you're playing by giving it a single tap.

I think that's a brilliant feature! 

Absolutely, that would be really exciting, as far as syncing the phase to music.  Maybe for the deluxe program, instead of the first tap reseting the waveform immediately  it sorta speeds it up to be in phase on the middle of the next beat so it doesn't glitch the sound with a sudden transition on the ramp up or sinewave?

Is averaging of more taps something that everyone would really like to see?

T.

It would help scientifically for better timing (when editing audio into loops I've had good luck starting with several measures and then dividing that down, as far as getting a truer timing of a single measure, that can stay in time with the original tempo longer, before it starts to drift out of phase).

But asking a musician to be able to tap their foot on time isn't exactly unreasonable , and it sounds like a can of worms to get into averaging.  Measuring the last two taps is wonderfully simple and straightforward.

[G. Hoffman]

All `yall had to do this just as I was finishing up my MV-53 based tap trem, didn't you?


I'm about to plug it in and see if it works, but it appears to be good.  Everything but sound is testing right, so I guess its time to give it a listen, right?

I normally only ever lurk here, and i haven't check in over a week because I've been working on this thing.  And now you tell me there is going to be a pre-made board?  Thanks a lot!



Gabriel

[MoltenVoltage]

All `yall had to do this just as I was finishing up my MV-53 based tap trem, didn't you?

I'm about to plug it in and see if it works, but it appears to be good.  Everything but sound is testing right, so I guess its time to give it a listen, right?

I normally only ever lurk here, and i haven't check in over a week because I've been working on this thing.  And now you tell me there is going to be a pre-made board?  Thanks a lot!

Gabriel

I may be more than a little biased, but I think you will find MV-53 to be a superior choice for the following reasons:

* The waveforms are analog (from an on-board Digital Analog Converter) and as clean as you can imagine, including the sine waves - no PWM filtering required, and the sample rate is many times higher (1.23 MHz)

* MV-53 has a duty cycle control to skew the waveforms "on the fly" using 11 different musical ratios that can "swing": 1/8, 1/6, 1/4, 1/3, 3/8, 1/2, 5/8, 2/3, 3/4, 5/6, 7/8

* MV-53 lets you switch between potentiometer and tap controlled tempos "on the fly"

* MV-53's tap control has tested accuracy of better than 1/1000th of a second

* MV-53 outputs two simultaneous waveforms (one is inverted)

* MV-53 allows the user to program sequences of waveforms and playback rates to add much more interesting rhythmic elements to your effects

* MV-53 stores 9 different user-defined programs, and also has a "free running" mode

* MV-53 allows a real-time LED interface indicating current waveform and playback speed

* Its very easy to defeat any MV-53 features you don't want to use

More info is available here, as well as a video demo:
http://www.moltenvoltage.com/mv-53.html

[G. Hoffman]

All `yall had to do this just as I was finishing up my MV-53 based tap trem, didn't you?

I'm about to plug it in and see if it works, but it appears to be good.  Everything but sound is testing right, so I guess its time to give it a listen, right?

I normally only ever lurk here, and i haven't check in over a week because I've been working on this thing.  And now you tell me there is going to be a pre-made board?  Thanks a lot!

Gabriel

I may be more than a little biased, but I think you will find MV-53 to be a superior choice for the following reasons:

* The waveforms are analog (from an on-board Digital Analog Converter) and as clean as you can imagine, including the sine waves - no PWM filtering required, and the sample rate is many times higher (1.23 MHz)

* MV-53 has a duty cycle control to skew the waveforms "on the fly" using 11 different musical ratios that can "swing": 1/8, 1/6, 1/4, 1/3, 3/8, 1/2, 5/8, 2/3, 3/4, 5/6, 7/8

* MV-53 lets you switch between potentiometer and tap controlled tempos "on the fly"

* MV-53's tap control has tested accuracy of better than 1/1000th of a second

* MV-53 outputs two simultaneous waveforms (one is inverted)

* MV-53 allows the user to program sequences of waveforms and playback rates to add much more interesting rhythmic elements to your effects

* MV-53 stores 9 different user-defined programs, and also has a "free running" mode

* MV-53 allows a real-time LED interface indicating current waveform and playback speed

* Its very easy to defeat any MV-53 features you don't want to use

More info is available here, as well as a video demo:
http://www.moltenvoltage.com/mv-53.html

Oh, I'm not dissatisfied with it.  However, if I could have used someone else's circuitboard, instead of having to design my own, it would have made life much easier.  I'm kind of lazy that way, sometimes!

By the way, any hints on getting rid of the clicking I'm getting in Square and Sawtooth modes?  I'm assuming it is to do with my circuit board design or some such, but any hints would be appreciated.  Also, was I just wasting space on my PCB by putting current limiter resistors on the program inputs?  Other than the clicking it worked perfectly on fire up (shocked the heck out of me!  That never happens to me, and I was particularly shocked as I didn't breadboard the circuit), but the current limiting resistors are taking up a fair bit of real estate that I would like to use for other things.

{edit}  OH! And I think you might want to change the depth control in your tremolo circuit to a audio taper pot - the sweep is kind of sudden.{/edit}


Gabriel

[MoltenVoltage]

We didn't notice any clicking on our prototype when using only the Tremolo, but we did have that problem on the VCF circuit when using square or sawtooth waves.  Take a look at the pop filter we used on the VCF schematic:
http://www.moltenvoltage.com/downloads/MV-53_-_VCF_Schematic_-_Molten_Voltage_v1.bmp

Also take a look at how we did it with the MV-52 circuit - the 100uF cap to ground right before the optocoupler did the trick.
http://www.moltenvoltage.com/downloads/Application_Note_1_for_MV-52_-_v1_1_-_Molten_Voltage.pdf

Try placing a cap to ground in front of the optocoupler (after the 4.7K resistor).  100uF is a good place to start.  If you still have issues, try a larger value.  If it is dead silent, try going down in value, as the higher the value cap, the more it rounds off the corners of the square wave, and the less "choppy" it will be.

You don't need any resistors that aren't shown in the schematic:
http://www.moltenvoltage.com/downloads/MV-53_Tremolo_Schematic_v1.jpg

When you say "current limiting resistors on the program inputs", I assume you mean the Program Pins (4, 5, 6, 7, 9, 10, 11, 25, & 26).  These have internal weak pullups so no need for a pullup resistor.  They get pulled to ground when engaged, so no need to limit the current either.  If you don't want to use any program features, just leave them all unconnected - then the chip will remain in "free running" mode.

Thanks for the tip on the depth control!

A real fun way to use the LFO (especially if you like to experiment) is to put it in a separate box like in our video, then have a jack that lets you connect to your LFO-based effects.

[ElectricDruid]

Thanks to Taylor for pointing out that the 'beat' in a sine wave is at the top of the waveform. It *is* obvious, but I *had* forgotten to think about that. I'll have a play with the waveform tables to take that into account.
It'd be nice to be able to smooth the transition when you tap a single tap to reset the phase of the waveform and keep it in sync, but to be honest, I don't think it's do-able. At least, not on this chip.

The square wave width is variable using the Pulse Width CV, so it's only square if you feed 2.5V into that input.

As to MoltenVoltage's comments, mostly I agree entirely - they've got a bigger chip and bigger feature set. I've done other LFOs on the 16-bit dsPIC series chips, and you can get better quality output. My tap tempo LFO is small, simple, and cheap (The chips cost £1 / $1.50). Code is available for tweaking, but it's only an 8-bit chip. Ultimately, they're rather different beasts. You decide whether you need the extra features and quality of the Molten Voltage chip and whether that makes it worth it for your application.

In one detail, my little PIC does keep up with their chip. The timing accuracy is much better than 1/1000th of a second - in fact 1/19500th of a second.

Thanks to everyone for all the support with this project,
Tom

[MoltenVoltage]

Thanks to Taylor for pointing out that the 'beat' in a sine wave is at the top of the waveform. It *is* obvious, but I *had* forgotten to think about that. I'll have a play with the waveform tables to take that into account.
It'd be nice to be able to smooth the transition when you tap a single tap to reset the phase of the waveform and keep it in sync, but to be honest, I don't think it's do-able. At least, not on this chip.

The square wave width is variable using the Pulse Width CV, so it's only square if you feed 2.5V into that input.

As to MoltenVoltage's comments, mostly I agree entirely - they've got a bigger chip and bigger feature set. I've done other LFOs on the 16-bit dsPIC series chips, and you can get better quality output. My tap tempo LFO is small, simple, and cheap (The chips cost £1 / $1.50). Code is available for tweaking, but it's only an 8-bit chip. Ultimately, they're rather different beasts. You decide whether you need the extra features and quality of the Molten Voltage chip and whether that makes it worth it for your application.

In one detail, my little PIC does keep up with their chip. The timing accuracy is much better than 1/1000th of a second - in fact 1/19500th of a second.

Thanks to everyone for all the support with this project,
Tom

I couldn't agree more that there are trade-offs between the two chips, but I must take issue with your accuracy claims.

Please correct me if I am wrong, but I cannot imagine you have actually rigorously tested and proven that accuracy like we have.  Mathematically, ours should be much more accurate (nearly 1/40,000,000th of a second), but can never be because of physical limitations.  Further, changing the placement of the 0.1uF bypass cap by 2 millimeters can cut tap accuracy in half (down to 1/500th of a second).  I know you are giving away your code, which is very generous of you and great for the DIY community here, but I really think you need to have your facts straight before making specific performance claims.

Again, if you have tested and I am wrong, I apologize.

[JKowalski]

Would it really make a difference,

It's not like you can tap your foot to even a 1/2 second accuracy, much less 1/1,000 or 1/40,000,000 

Although I suppose any claim is worth arguing over in the marketing world. 

[ElectricDruid]

I'm not making any enormous accuracy claims. All I was referring to is that the timer that times the time between taps is clocked at 19500Hz, so it's accurate to better than 1mS - it's a crystal clock, after all. And no, I haven't tested it - what need? I agree with JKowalski; the big timing error is the human doing the tapping anyway!

Tom

[MoltenVoltage]

Even though 1ms accuracy isn't that big of a deal playing live if you have a "flexible" drummer, accuracy is an issue if you are playing along with other tracks while overdubbing.

Assume a typical song of 120 bpm (2 beats per second : 500 ms between beats)
If this is a 3 minute song, it is 180 seconds (180,000 ms) long.

Now assume you want the effect tempo at 1/4 notes, you are at 125ms between quarter notes.

Even your effect gets out of time only 1ms per second, by the end of the song (180 seconds), you are well over a quarter note out of time (180ms).

Instead of being on the downbeat, you are well past the snare beat.

Even if you wouldn't usually use an effect for a whole song, you can calculate the degree to which you will be out of sync based on the length of time the effect will be on.  The obviousness of this effect depends on the type of effect you are using, but a VCF type effect or chopper tremolo will be easy to hear if its out of time, while a soft flanger might not.

While you might not be able to tap within a 1/2 second of the beat, I've met plenty of pros that can nail it.  I guess the degree of accuracy you require depends on who the end user is going to be.

[ianmgull]

Even though 1ms accuracy isn't that big of a deal playing live if you have a "flexible" drummer, accuracy is an issue if you are playing along with other tracks while overdubbing.

Assume a typical song of 120 bpm (2 beats per second : 500 ms between beats)
If this is a 3 minute song, it is 180 seconds (180,000 ms) long.

Now assume you want the effect tempo at 1/4 notes, you are at 125ms between quarter notes.

Even your effect gets out of time only 1ms per second, by the end of the song (180 seconds), you are well over a quarter note out of time (180ms).

Instead of being on the downbeat, you are well past the snare beat.

Even if you wouldn't usually use an effect for a whole song, you can calculate the degree to which you will be out of sync based on the length of time the effect will be on.  The obviousness of this effect depends on the type of effect you are using, but a VCF type effect or chopper tremolo will be easy to hear if its out of time, while a soft flanger might not.

While you might not be able to tap within a 1/2 second of the beat, I've met plenty of pros that can nail it.  I guess the degree of accuracy you require depends on who the end user is going to be.

You're assumption is that first, the tempo will not be reset/recalibrated at all during the song, this negates the necessity of a tap tremolo to begin with. Second, that someone would actually be able to tap with 1ms of the tempo of a "flexible" drummer (that's 1/1000 of a second, nobody got that much rhythm). I have to agree that these differences seem like somebody trying to push their product a bit too hard.

[newfish]

...and if you're accurate to 1/1000th of a second, all the time,  *everytime*, you may as well be a machine.

[jkokura]

This is a bunny trail and defeats the purpose of this thread I think. While I agree MV, it's important to test and the accuracy and quality of your chip is excellent. I'm sure many of us would like to see it used. However, we're trying to design a Tap tempo tremolo circuit here. If you'd like to offer your input based on the last 5 pages, perhaps come up with a schem, including your chip, we would appreciate that. If I understand correctly, you've already done that to a certain extent, however I looked at your video with the crazy external tap tempo box and I was overwhelmed.

We're trying to design a simple trem, with some important features, and if you and anyone else would like to contribute wonderful, but arguing about the veracity of calculations in your chip versus someone elses... Go somewhere else and do that. Nobody really needs to watch that argument take place, and it doesn't help us get a PCB together with a project file that anyone can etch, get the parts for and build to show off.

Sincerly, and with all due respect.

Jacob

[jkokura]

By the way, I think the MV52 has more potential for the application we're thinking of also.

Jacob

[Taylor]

Ok, I got side tracked by some other things, but the PCB is under way. Hopefully will have proto boards in a couple of weeks, followed by boards for you guys in about a month.

This'll be fun, I haven't had a really nice trem since I sold all my pedals a few years ago, in a baffling haze, thinking that computer processing had hardware effects obsolete. Now I know better.

[Taylor]

The PCB is complete. I'm trying an interesting grounding scheme on this one, using separate ground planes for the audio and digital section, and then connecting them at a single point. Should hopefully keep things sounding nice and clean with no ticking.

For my personal taste, a 1590BB box is about right for something with 2 stomps and 4 or 5 knobs. In fact, I just like gigantic metal boxes. But just because I know how much you guys love fitting everything into tiny boxes, I designed the board to fit into a 1590B! It should work either with Electric Druid's TapLFO chip or JKowalski's modified code, and I included an option to add the "Wave Distort" knob (duty cycle that works on all waveforms) if using the modified code.

I decided to go with this chip rather than the others for a few reasons. The main thing is that it is "open". The code is open to be changed by the community, which is in line with the spirit of this forum. Also it's cheap. I will be talking to Electric Druid and JKowalski to orchestrate the availability of these chips for people who aren't able to program their own PICs. I will also hook you guys up with part numbers for the crystal oscillator, etc. so this project will be buildable for all.

[.Mike]

I used this crystal from Mouser (although I bought mine from Newark):

http://www.mouser.com/Search/Refine.aspx?Keyword=73-XT49S2000-20

If I recall correctly, I chose them because they were shorter than most of the other crystals I saw. You can see them in this image and this image.



MIke

[jkokura]

I'm in for two PCB's if you're getting them made, or if you're going to be making a Trace you can etch than make an extra two for me Taylor!

Jacob

[Taylor]

These will be pro double sided boards with silk screen, plated pads, solder mask, the whole deal. So much nicer than building on home etched boards! I have a hard time going back to homemade boards now that I'm spoiled on pro boards...