[WIP] 2-to-1 RGB Luma Switcher

Hi everyone,
I have a project in the works that I hope to finish soon and start a group order once it’s tested and finalized. @rempesm and I have been working very hard on many ideas that focus on RGB and component video processing as companions to existing systems.

The first project I hope to bring all of you is a video-rate luma switcher which will switch between two sets of RGB signals.
Users may:
(1) Choose to switch each of the 6 luma inputs independently or all three as an RGB-set utilizing just the Red-Selector input and the “normalled” inputs of the G- and B-Selector inputs.
(2) With the use of the threshold knob, users may set the voltage at which the inputs will be switched from 0 to 1v.
(3) Lastly, without using any of the selector inputs; the threshold knob may be used to switch manually between inputs.

This will be an 8HP module including one faceplate and two circuit boards in a DIY format.
I have been granted permission by @Z0NK0UT to advertise this project on the forums, so please join me in the development process.

I will be sending away for this version along with several other future prototypes within a day or two. Feel free to offer any suggestions, corrections or kind words of encouragement. :slight_smile:



Still waiting on a few more parts. Can’t wait till I can actually test them. :smiley:
I am also still open to any and all suggestions. I don’t see much needing to change atm, but I would like to add a cutout to the top board so the power ribbon is easier to access. There’s plenty of clearance, but it is a tad inconvenient like this.

11-15-20 update - Here are some sick demonstrations from rempesm:
Corrections and modifications being made soon.


Well done on a cool looking addition to a video synth setup @Fox & @rempesm & for collaborating together :+1:t3:
Count me in on a group order and for distributing a few around Germany or Europe if neither of you are located here & it helps everyone out.

My only recommendation is to make sure the silkscreen “+” sign for the 5 electrolytic capacitors is placed to the side of the caps’ footprint so it’s still visible after they’re soldered in, just in case some troubleshooting is required.

Ok, I’ll be hated for this :wink: but a version with the smt Analog Devices quad IC presoldered at the JLC factory and the rest of the components left as is, through-hole.

PS. I love the headers to link the two PCBs. Why didn’t I think of that or see it elsewhere :ok_man:t2:


@VanTa, @horacio2222
Does this module design sound interesting to ye guys? If so, we could do a small 1 day workshop at the Pedal Store in Neukölln once it’s finalised and ready to ship.


Id be down for in Berlin workshop. Count me in.


Two quick notes:

I mentioned this idea as a passing thought to Fox like a week ago so all credit to him for just running with it and getting the design finished this quickly! There aren’t very many switching modules that can handle wide bandwidth signals (SSSRLabs Matrixarchate comes to mind) let alone are optimized for LZX format and feeding 3 Faders with a square wave into the CV input feels a little like overkill.

Fox’s description is great but I want to emphasize you can plug any signal into the Select jacks, from DC offsets to ramps to oscillators to external Luma video and define what voltage level causes the switch with the Threshold pot.

Generate a diamond gradient and you can choose where along that diamond shape Channel B punches through Channel A. In the external Luma video case, this means you could choose whether the darkest or lightest parts of the image or anything in between triggers the switch from A to B.

All sorts of interesting compositing possibilities come up so I’m really excited to see this released!


@transistorcat, @syntonie, @reverselandfill
What do ye think of the idea & any thoughts regarding the PCB layouts or circuit design?


Really nice idea!

I was talking of a similar principle with @reverselandfill, basically a CD4053 with buffered IO and 0-1V control over the switches, as I’ve been using this configuration to switch composite signals, thought it would be interesting to have some kind of 2 channel keyer controlled with a 3rd signal.

The selection of the inputs with the threshold knob is really neat!

1 Like

I’ve been thinking of doing something pretty much exactly like this, but with another set of outputs so you have three full cross switches.


Ha, let all the different builds come to reality!
I was thinking to inject noise into our circuit, as that looked cool with the test version. (it was actually a bug with the cd4053)
As a sort of noisy keying effect when used with the video input.


The lo-fi or circuit-bender Edition :grin:

love the functionality of this module!
with a tbc2 and chromagnon on the way eventually at least one of these will be super helpful in a system. If I can get someone to build it for me I’ll be set as DIY is not really my thing.

one question
is there a reason the inputs are labeled “R1” and the select section is “RED SELECT”

I guess I’d like to see it be one of the other so

1 Like

Thanks! That would be incredible of you.
Added “+” signs. :wink:
As for the first run of prototypes, I’ll just be using thru hole 6172’s but I have done a lot of auto populated boards through JLC so it is not outside of the realm of possibilities. Worth mentioning is that they also have 6172’s on hand as well as some Analog Devices parts.

Thanks for the additional info, Rempesm. Switching can be done via whatever signal users have on hand, including mult’ed video. I can even see the possibility of manual transition sweeps using a ramp input and the threshold knob. :smiley:

That is precisely what we have here. Inputs are buffered, with a few 2x gain 6172’s, wired through the CD4053 and buffered again at the output. My threshold voltage by default is 0-1v, but with the swap of a few resistors , users can configure it to be anything from -5v to +5v, -0.5v to +0.5v or whatever. 0-1v is easily the most useful. I added a couple of DNP resistor footprints like the scaler has for this purpose.

We are also looking at an alternative to the CD4053 which I am going to mention to reverselandfill below.

Another set of outputs? Do you mean three sets of inputs or am I misunderstanding?

When a complete prototype is in hand, we’ll see what noise looks like and what we can do about it. I think that the addition of hysteresis to each comparator should probably get rid of any glitches. rempesm breadboarded a similar circuit with a different analog mux that introduced glitches around some transition edges. The glitches were only present near certain thresholds from what I understand and in all honesty, the effect was not a bad thing at all! Actually something I can imagine wanting to employ in certain patches.

We were looking at another IC found in the Linear Tech application notes that was made specifically for this purpose. The reason this version uses a CD4053 is because it came to mind first and because the other chip is limited to a +/-5v supply. By the time I even looked at the alternative, I already had this drawn out. :stuck_out_tongue: Anyhow, that other chip is probably the very best part for this project for many reasons. First of all, it has input buffers and output gain stages internally without the need for external resistors or op amps.


Not sure if I’ll be doing any complete sales this time, but someone would probably be happy to help you there.

Thanks for the faceplate CC. I had several different ideas on how to label it and apparently they all started to bleed together. The reason I have R0 instead of RED0 is because it aligns with the other projects I have in the works.
Does fit with Ch. A, B and OUT? Ch. OUT kinda bugs me but idk what else to use.


We had glitches because we powered the 4053 with +12v and GND. the glitch emerged from the fact that the LOW could not get to zero. so we powered the cd4053 with +5v and -12v

Edge glitches are something I had not looked at yet. but our focus point was create dirty video glitches, so it is another approach :slight_smile: :face_with_monocle:

this is what it looked like: (patch with the freq doubler and camera feedback)
the grainy stuff come from the switcher

so that is why I rather want it back in a more controlled form. :slight_smile:


The idea was to have three sets of two inputs and three sets of two outputs, and three sets of key inputs (adjustable level like yours).
When the key is above the threshold, the outputs swap.
This could be patched into a full RGB switcher (like your design) but also as a panner (one input routed to either output), or a single luma four input switcher (also like yours), or a one-input four output router and some combination of weird setups

1 Like

Oh thats cool! Looks like marbling. Have you considered starving the chip using the voltage drop of series diodes? Connect +12v to the VDD pin and connect a diode’s anode to the chip’s GND and VEE pins, and the cathode to ground. Try several diodes in series. It could either be interesting or it could be way off. :stuck_out_tongue:

oh cool cool. How big would that be? Whenever I design a board I am always worried about the HP, or at least worried what people would think if I “took up too much width.”

1 Like

I’ll try that!

ps: I love that we are all building/designing here, open on this community forum
learning from each other, helping…


I was hoping to fit it into a slightly-too-dense 8hp (Which would be fine enough since there are few controls), failing that i feel like 12hp for an RGB module is not too much to ask

1 Like

Looks amazing. Definately in for a order!

1 Like

Hello guys,

It might be worth checking out the ADG333A IC - it’s a through-hole video-rate quad SPDT switch with separate controls for each switch and it runs off +/-12V. It’s absolutely perfect for this sort of thing, and avoids the need to generate a +5V supply rail. Plus it has a higher bandwidth than the venerable 405x ICs so sharper image quality.

I used it in my Quarterizer DIY module (just announced… sorry for the gratuitious self-promotion :slight_smile: ) which is a 8HP four-channel amplitude classifier inspired by the one from the Sandin IP. See www.visiblesignals.net for it and my other modules, including the RGB Matrix which has a dual-bus crossfader/keyer that is similar to the module you’re designing here (but it’s a fader rather than a hard switcher).