Eric Archer's 9v korg MS-20 style filter w/video content

[Nocentelli]

http://ericarcher.net/devices/diy-lpf/

A great sound, worked first time on the breadboard once I got the pinout of the rail-splitter sorted. I've got it hooked up to an expression pedal for the frequency control, and I'll post a vero once I've got it verified. Here's a really rough demo of what it can do (please excuse the poor sound quality and playing...):



I'm also keen to try out an envelope detector to trigger the sweep, so if anyone has suggestions for a simple-ish way to get the 0-9v CV triggered by the guitar's envelope, I'd love to hear them.

[~arph]

How does it behave with your standard voltage divider to get the 4.5V?
Plus, how about using a LM386 as a rail splitter.. Just hook up power and take 1/2Vcc from pin 5.

[Nocentelli]

I'll try both and report back: Any suggested values for the resistors in the voltage divider? I've seen everything from 10k/10k to 1M/1M used in related circuits, and have no idea of the merits or drawbacks of either.

[jonasx26]

I'll try both and report back: Any suggested values for the resistors in the voltage divider? I've seen everything from 10k/10k to 1M/1M used in related circuits, and have no idea of the merits or drawbacks of either.

R.G. explains it here: http://www.geofex.com/circuits/biasnet.htm
Depends on what load the voltage reference sees. If its mostly direct connections to non-inverting op amp inputs (as in inverting amps) and/or high value biasing resistors to op amp inputs, a passive divider with reasonable resistor values (10k-100k maybe) will probably work. Measure your virtual ground with a multimeter and see if it's stable. Or even better, look at it with a oscilloscope while changing the load.

Buffering the virtual ground with an op amp follower is a easy solution that allows you to use high resistor values and reduce current draw.
Most of the time an op amp buffer will be able to source/sink enough current (something like 20mAish) to maintain a stable virtual ground reference.
Capacitive loading of the virtual ground can be more or less of a problem with certain op amps. Sometimes it can be fixed with frequency compensation inside the feedback loop.
I've never used a rail splitter IC but they are likely designed for unity gain stability and thus better suited for the job.

Google "single supply op amp design" for some design reference PDFs. Also some great info on rail splitters in the Art of Electronics.

[Nocentelli]

Here's the first draft of a vero, I'm still checking it over and will hopefully verify by actually building it when I have time within a day or two: I think it's ok, but obviously check against the schem if you plan to build from this. I did the layout with the rail splitter because I've got one and it's in the original schem

VC LPF_ericarcher_nocentelli vero-UNVERIFIED.gif (33.14 KiB) Viewed 1273 times

Note:
I'm having a few issues with the input caps: They seem to be reversed compared to the orientation I'm used to seeing (i.e. neg leg facing input). I've avoided the issue so far, since I've just got 1uF film caps in there at the moment, but there is significant pop when bypassing the circuit, either with the toggle switch that bypasses my breadboard directly, or with the footswitch looper I use to bypass connections to the breadboard in/out... The single bjt boost pedal that follows the filter also now pops, which it didn't do before. I'm guessing a pull-down resistor at input and/or output might help: I wonder if this would this affect the preferred orientation of the input caps?

I'm also going to continue breadboarding to see if i can rectify the caps issue, see how a voltage divider with resistors works, try out a few different buffers, and maybe try to squeeze a simple jfet buffer onto that vero. Updates to follow.

[jonasx26]

I'm having a few issues with the input caps ..

The orientation of the coupling cap looks strange to me. I'd reverse the polarity of the input caps (C5+C9) so that the negative side goes to the input. Maybe C6 as well.
To me that would make more sense since most effects are AC-coupled and ground-referenced.
Try adding pulldown resistors to the input as well. The input impedance is already way low, so you could get away with something like 100k-470k pulldowns without making it much worse..
Adding a input buffer is a good idea indeed. It just might solve your popping issues to some degree as well.

I wonder if this would this affect the preferred orientation of the input caps?

Easiest way to figure out is to put the cap in the circuit and measure the voltages on its leads. The negative side should be connected to the lowest voltage.
It gets trickier when both sides of the cap is at the same voltage, which happens a lot when everything is centered around 4.5V ..
In those cases a non-polar cap is preferable. However, most of the time you can get away with having the cap "the wrong way around" or with both sides at the same DC-bias.
But it will shorten the capacitors life, and is generally considered bad design.

You could make your own non-polar capacitor by connecting two polar electrolytics in series. Negative to negative or positive to positive, doesn't matter which.
Two 22µF in series would make a non-polar 10µF (approx.)

I like the video btw I'm also tinkering with a voltage controlled filter. Fun stuff.

[Nocentelli]

Thanks for all the useful advice. I had already measured the voltage on either side of the caps which both gave readings of zero on the input side, and 4-point-something on the circuit-side, so I'll reverse the input caps: For some reason, I started over-thinking it, and got myself in a knot over the virtual ground/vref and the way the schematic labels this as 0v - I thought if the rest of the circuit was using 4.5v as (virtual) "0v", then 4.4v was closer to "0v" than ground (zero volts on multimeter).

I'm pretty sure the pops were there with or without the buffer and 2M2 pull-down resistors, but I'll try some lower value pull-downs along with reversed electros of the correct value: If that doesn't help, I'll try constructing some NP electro pairs and see if I can bodge them into that layout.

Thanks again for the help.

[jonasx26]

Haha, yes. Ground is so relative and abstract. I've been doing so much single supply stuff that negative voltages confuse me, or at least makes me think twice.
Sure does make circuit design classes, where op amp supply rails always are +/-12V, unnecessarily challenging..

I usually think of the metal casing of the effect to be ground and "zero" volts. And everything else is measured from that reference, including Vref.
When measuring caps its all about the voltage difference between the cap terminals. You could measure voltage directly across the cap to get a clearer and more 'independent' reading.
Negative/positive readings would then indicate which node is lower/higher and a zero reading would mean both sides are at the same DC bias.

Bypass popping can be very difficult to get rid off. I've got best results with non-true bypass CMOS switching. Quiescent DC bias seems easier to control. And it is all about DC levels at the switching nodes.
You could use the same technique as when deciding cap polarity to try to redce the bypass popping. Switching noise is at its lowest when the points that you switch between (FX input and FX output) are at the same voltage.
Try different values of pulldown resistors/caps/no buffers/buffers until you get the same voltage at the input and output.
Unfortunately, you never know what will be plugged into your pedal. And that seems to make it more difficult to get a consistently pop-free pedal with "true" bypass setups.
I haven't fully figured out how to bypass silently (and true..) with mechanical switches. The contacts bounce and move around alot and never makes super-clean perfect transitions.
It's very difficult to remove the pop altogether. Relays makes it easier. CMOS/Jfet bypass easier still. Best bet is to make sure DC voltages at switching nodes are the same with no signal applied.
Also, circuit layout can be crucial. Where your grounds go, where ground currents flow.. .. also, LEDs can make alot of noise. Best to experiment without them.

Thanks again for the help.

Glad you found my rambling useful! Hope I don't come off as arrogant and know-it-ally. I've heard swedish people can seem that way when speaking english. Weird..
ANYWAYS. Looking forward to see/hear video/audio of your progress. All the best.