An new twist on a LDR based envelope filter

[rring]

I just changed the schematic ....it correct now on the LM358 input reversal. the correct rev so far is rev1.1 (as shown on the current schematic)

[rring]

here is my surface mount layout and the silkscreen:

http://circuitsaladdotcom.files.wordpress.com/2014/05/minimur.pdf
http://circuitsaladdotcom.files.wordpress.com/2014/05/minimussilkscreen.pdf

[Seljer]

I just changed the schematic ....it correct now on the LM358 input reversal. the correct rev so far is rev1.1 (as shown on the current schematic)

IC1A is also at fault 

[rring]

well here comes rev 1.2..... duh!

[rring]

Its fixed...sorry about that. For what its worth the filter op amp is correct -I swear it! The new rev is 1.2

[rring]

To answer an earlier question, the filter topology is a  Fliege filter (the low pass form). It is a second order filter with control of Q and filter knee. The interesting thing is the pass band always has a gain of two no matter what the Q is (the peak at the knee goes higher if the resonance is set high). I used it because its relatively simple and easily adapted to single supply operation.

[Seljer]

Its fixed...sorry about that. For what its worth the filter op amp is correct -I swear it! The new rev is 1.2

Haha  

I checked the filter in SPICE and it filters correctly .

Thats a nice filter if I ever saw one


But as they say: "the proof is in the pudding", I'm going to have the build this to find out!
Heres my layout fixed of the aformentioned issues, I even managed to get rid of one jumper . Board size is 4.5 x 6.5cm .




link to transfer http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=51000&g2_serialNumber=2 (300dpi)

[midwayfair]

Yay, a new CS design!

A few inevitable questions:

Do you think TLC2272 will work for the op amp? It's rail-to-rail with more bandwidth than the TL072. How important is the rail-to-rail output -- would the OP275 perhaps work? It's got really great bandwidth characteristics, but not the rail-to-rail output. What I'm focusing on is finding usable op amps in DIP through-hole packages.

Do any of the concerns with the TL072 go away with a higher supply voltage? After all, the original Mutron ran on split rail. I assume it would help with the headroom problem, but not sure about the bandwidth or oscillation. If we're talking about using $4 op amps, I wonder if it's not just easier on some people to power it on 18V.

Thanks for linking to the LDRs used. It looks like their turn-off time is a bit slow, but there's also no Decay control in this design. Since the Silonex 7532 and Smallbear's 9203 have similar on/off resistance, I'm going to try those (though they're a bit faster). But I was wondering if you had some insight into the switching range really needed to get a working effect, since you also have the vactrols limited with 470K. Would it simplify the design to use, say, the NSL-32 (1K on/500K off, about $2-$4 each, but uses a superbright LED inside) or even the cheap ones at Tayda (~1-5K on/1M off) and omit the limits, or are you using the high dark resistance and parallel limiting resistors to set a sort of "taper" to to the sweep?

[rring]

I figured the RC time constant of C5 and R10 would dominate the time response. Honestly, I just chose those photocells because they were small and not too expensive. Their R range was pretty good. Also C4 limits attack somewhat- if you made that bigger it will respond faster.
The main thing I needed to achieve electronically was a fairly linear change in R over change in current. This is so the peaking control ratio stays the same over the range of the filter. To your point, I think just about any LDR combo will work as long as you get the sent point current correct with R5. In this circuit your at about 10k frequency knee when the photocell is @ 1K. The way I have it set the best you can get to is about 1.5 to 2k ohms when you slam the output of IC1 to the rail. You may find a better performing LDR combo so if you do let me know! The general theme was- can I grab any ol' photcell and get it to work. I was hoping the op amp wouldn't be so critical but I am sure there is a more common junk box type that will work- just got to find it.

With regard to the op amp choice. You may be right about the TLC2272 - I don't have one to test and I have not done a detail analysis to determine exactly what is going wrong with the TL072. It must be related to loop gain and phase shift and if you look at that filter topology, it pretty scary because you to op amps driving each other in a loop. Maybe someone can breadboard just the filter portion and use a POT to shift the LDRs to test the TLC2272?

As I mentioned before: the TL972 comes in DIP is a buck and should work great

[deadastronaut]

Yes that extra op amp could be used for up/down reversal of the envelope following maybe.

cool, 

[midwayfair]

Yes that extra op amp could be used for up/down reversal of the envelope following maybe.

If we rig up a version with only a single LED and three discrete LDRs, I think you can just switch R6 to +9v instead of ground.

[rring]

You can't do that because the peaking control needs independant control of current to be shifted as desired from the other two frequency sweep LDRs- this means it needs it's own LED(for LDR3)

[StephenGiles]

Very nice, consider a second filter in parallel using say 0.047 or even 0.068 capacitors, which may add a vocal feel to the sound.

[midwayfair]

You can't do that because the peaking control needs independant control of current to be shifted as desired from the other two frequency sweep LDRs- this means it needs it's own LED(for LDR3)

Oh, bugger. Okay.

[Mustachio]

This Looks and sounds awesome rring! Gonna give this a try when I get some time. I love envelope filters! 

Nice playing in the demo's too!

The smd version looks fun! I just got a bunch of 0805 resistors and caps to play around with. I've only done smd IC's by hand prior, for the Pearl Octaver from Gaussmarkov .  I noticed you included a stencil for doing the solder paste and bake method! That's Awesome , I've never tried it that way but it looks great. Do you use a regular toaster oven for that ?

In the past when experimenting rolling my own vactrols I found using 4mm green LEDs with built in fresnel lens worked great , I believe the ones I used had a slightly different forward voltage and a wide viewing angle which I think helps even out the light distribution across the LDR. On the LDR side I used the common KE10720.

[rring]

Are you confusing my "liquid tin" coating for a solder wash. I soldered the board by hand - I just get liquid tin (from amazon in a little bottle) to plate the copper boards with a thin tin coat- you just wipe a little bit on..it takes about ten seconds. It protects the copper from corrosion and makes for easy soldering.

This layout has lots of open space - a good choice someone who dosent work with SMT much to give it a try. Also, I intentionally used as few a variety of and the most standard values for parts possible. There are only thirty some parts - it goes quick by hand.

The main reason I did not use leaded parts was - I just didn't want to drill all the holes! Otherwise leaded parts would have fit just fine on that board.

As for the toaster oven yes I have used that before with leadless packages - it works well - you just have to be careful taking the board out.

Yes the DIY vactrols are very workable. My choice of LED was based on the nice flat face to bond the photocell. I noticed the amount of light required is not very much in my testing. I made about ten of the things and all were within 5% of each other over a range of current, except one which was way off and an outlier. So just as your expereience with making them conveys, I think its a very practical solution.

[rutabaga bob]

VERY nice!  I would like to make one using standard Vactrols.

[jatalahd]

Good work, it sounds amazing!

Earlier in this thread there was a question about the purpose of D2. I just would like to add that with C4, D2  forms a commonly known Clamper circuit. When the diode is placed with that polarity it makes a positive clamper and opposite polarity makes a negative clamper. If there exist a very high resistance path to ground for C4, then in this case the negative swing of the output of IC1A quickly reverse-charges C4. This creates a voltage across the C4 terminals and due to high resistance path to ground this voltage stays as a bias for the output of IC1A. Hence, the signal at the junction of D2 and D3 is always above ~ 4.5 V and the maximum signal voltage at this junction can momentarily be 9 V + 4.5 V. This voltage is also the base voltage for D3, which does not conduct when the voltage of C5 exceeds the voltage at D2. What puzzles me though is, does IC1B go to saturation if the voltage over C5 is over 9 V ...  

edit: Ah, the clamper base level is a diode drop below 0 V and not 4.5 V as I claimed. C4 first takes the average signal level to 0 V and only after that the clamping takes affect and rises the bottom level around 0 V. Small details, but just wanted to understand what is happening here. 

[StephenGiles]

Good work, it sounds amazing!

Earlier in this thread there was a question about the purpose of D2. I just would like to add that with C4, D2  forms a commonly known Clamper circuit. When the diode is placed with that polarity it makes a positive clamper and opposite polarity makes a negative clamper. If there exist a very high resistance path to ground for C4, then in this case the negative swing of the output of IC1A quickly reverse-charges C4. This creates a voltage across the C4 terminals and due to high resistance path to ground this voltage stays as a bias for the output of IC1A. Hence, the signal at the junction of D2 and D3 is always above ~ 4.5 V and the maximum signal voltage at this junction can momentarily be 9 V + 4.5 V. This voltage is also the base voltage for D3, which does not conduct when the voltage of C5 exceeds the voltage at D2. What puzzles me though is, does IC1B go to saturation if the voltage over C5 is over 9 V ... 

I'm very interested in the front end of this circuit up to R4, and bearing in mind what you are saying jatalahd, how would you make the voltage at the junction of D3 and R4 only go high on the leading edge of the note played - or does it do this already? Would a sample and hold circuit be needed perhaps?

[rring]

Because of the varying signal, the op amp not able to make it to the rails all the way, the finite Z of the op amp, bleeding of the charge on the caps through R4 and the attack POT, slow charging of C5 via the attack pot and of course diode voltage drops...I don't really observe the behavior described in the last couple of posts.

The simulated behavior and observed behavior on the scope is:  the clamper circuit, combined with the half wave rectifier, forming a half wave voltage double with diode voltage drops. It has a slow rise time and it
bleeds away static charge relatively quickly with no signal. So you never see actual true voltage doubling and never any voltage level exceeding little more than 4 to 6 volts when you are beating your guitar like a gorilla with the sensitivity all the way up.

All in all its not very exciting but it works. A BAT46 schottky or similar will work better(lower voltage drop) and then you can lower the gain of IC1A by increasing R3 some.

The IC1B can only source up to approx:  +l9V at the output,  but the current mode feedback requires that (I Output*R5) = Voltage In (at IC1B's positive input).
This maximum current is (V+supp - LED V Drop)/R5. So the output of IC1B will always be the (LED V drop) greater than the voltage at the + input of the op amp. The maximun output will be reached  when the + input of IC1B is somwhere between 6 to 7 volts.