Luma encoder DIY?

Im wondering if it would make sense to design a luma encoder DIY module.
The simple way is just to mult a signal to all three inputs of the C2, but my initial thought is that there might be a cheaper/smaller/simpler build possible? Perhaps a dual channel luma encoder would be even more cool.

Would have only one Jack input per channel and therefore only need for one buffered input per luma channel etc…
It of course halts on the assumption that the ICs ( AD724?) doing the actual video signal magic can be found in a single channel/luma version instead of the now used RGB color version.
With my limited knowledge, i searched the analog devices homepage, but it seems there ar only RGB and more advanced chips being produced by them. But perhaps there exist other suitable manufacturers or even older ICs still lying around that could be of use here…

Any hints/thoughts?


You mean for doing grayscale video output?

The very easy mod would be to internally the jacks: connect the tip of input 1 (red) to the switch of input 2 (green) and the input 3 (blue)

I tested this, it works.

a variation would be :
connect the tip of input 1 (red) to the switch of input 2 (green) connect the tip of input 2 (red) to the switch of input 3 (blue)

Hey Martijn.

Yep i mean doing grayscale out.

Not a bad idea. this would mean that when connecting only one input to the encoder it will give grayscales. Might do this on one or two of my encoders

But what im after, is if its possible to do a cheaper and smaller board. After all the build price of PCB with video IC, frontpanel, and all components costs around 90€. (+ vat) And it is a big board for ‘just’ a luma encoder… if it could be done cheaper and simpler, smaller, this could also make a basis for small spinoff stand alone devices, small systems to be used along the vidiot and whatnot… The C2 board is one of the bigest, if not the biggest of the Cadets. And also the most power hungry of the boards.

I have no clue what goes on behind the scenes on the Vidiot, but it has a luma out. Might take a look to see if it is effectively just the C2 with the inputs connected together, or if its another circuit.


If you want something like a C2 but cheaper than 90€ and smaller, the best way to do this (if you’re already thinking of a from scratch approach) is redesign it to use SMD, make your own pcbs/panels (via cheap Chinese PCB manufacturers), and source your parts smartly.

First, the size. A lot of size can be cut down by making it SMD. I think you can easily find SMD equivalents of all the components, including the video op amps. You might even be able to design the PCB to sit parallel to the front panel, making it very shallow (might need two pcbs in that case, and you’ll need to use different jacks and switches of course). The trade off is it’ll be a lot harder for beginners, but not too much harder for those with some experience.

Next, cost. If you shop around and look for the best deals on components you can save a fair bit of component cost. Places like Mouser and Digikey are rarely the cheapest place for simple passives like resistors and capacitors, when ordering small quantities. Another thing you can do is order for multiple boards at one to get quantity discounts. At some places, ordering just 10 LM6172’s gives you a not entirely insignificant discount per part. And if you make your own PCBs via a Chinese fab, you might end up paying less than what you pay LZX for PCB and panel, and get 5 or 10 copies. If it’s not cheaper overall, it’ll at least be cheaper per board. The trade off here is this will all take more time to shop around for parts and boards and design boards, and LZX’s boards have been proven good and any flaws have been already found. Making your own boards can be an iterative process where you don’t get things right until the second, third, fourth time.

1 Like

for greyscale you can try combining h+v sync with the greyscale signal and blanking. you don’t need colorburst for b/w, but im not sure actually if not having colorburst will freak out some gear that may expect it.

1 Like

Most of what you need for a luma output is already in C2… three times over! :slight_smile:
You need black/white level clipping (C2), blanking (C2), 1V to 0.714mV gain translation (C2).

Then, instead of the AD724, you need a video opamp 2X gain buffer/ 75 ohms driver, with CSYNC insertion. These are the parts that are not part of the C2 circuit. If you post a schematic image I can verify if it’s correct or not, if you want to make a modified C2 circuit.

1 Like

Oh, thanks Lars. Sounds like this should be easier than expected… i Figure.
Not that i am anywhere near able to design this in the near future. But its good to have something to aim for… :slight_smile:

1 Like

but maybe I can help you :slight_smile:


A luma out module is a great idea. :slight_smile: Also, if you have this, you have the luma channel for a component output! Would just need RGB-to-YPbPr matrix circuit as a pre-conditioner (between black/white clippers+blanking and the output buffers) and you could do component out. Or you could use 3x of them to do RGB out (even if you have sync on red and blue, should be fine.)

1 Like

OK, I’m in the process of making this. I could use a svelte little B&W encoder as well.

I found this sync insertion circuit (below). It appears to be using an op-amp like a switch by powering it on and off with a sync pulse, allowing the sync signal to pass during the blanking interval. Then it occurred to me the signal is blanked already in the C2 circuit.

So far my circuit is one channel from the C2 schematic, up to the blanking switch. Is this Rube Goldberg circuit here what I want, or can I just use a 2x gain summing amp to sum CSYNC with the output of the blanking switch from C2?


1 Like

Something like this?


I think I’ll just go test it and report back. Also, is that cap just for coupling the chips? Do I still need it if I’m going out to an op-amp?

Report: the summing amp “works” but it’s noticeably darker than the luma output from the 724 and it’s also filthy.

Conclusion: I’m going to make a tidy little green-input-only version of C2, tie the other inputs to ground and use the onboard luma output from the 724.

This basically:

Or, I guess I could just tie all the inputs of the 724 to the single clipped signal.

– OK, fixed a couple small errors in the above schematic and threw together a prototype. @LauLindqvist – I just ordered a bunch of PCBs, so it might be a while until I send stuff out to fab again, but I’ll send you one of these eventually if it works.


How come you didn’t go with more SMT? Could have made the board even smaller I bet.

I have those thru-hole parts in stock. My priority is the shortest route to a functioning module. This is intended as a prototype, but if it works the way I want it to, it will also be the final version. I may do another rev only to move the banana jacks pads in a bit so that everything can mount to a panel evenly.

1 Like

Ah, I figured that was probably the case. Makes sense.

1 Like

Ay yi yi, I got some protos fabbed and there were a couple fatal errors I missed. Ordering some new things soon though, so the revisions of these will be in that batch. Luckily I caught the mistakes before wasting an encoder IC. :sweat_smile:

1 Like

@creatorlars, would you mind taking look at this? Before I start tracing the signal or troubleshooting my layout, I just want to verify that there aren’t any glaring errors in my circuit. I omitted the AD724 NTSC/PAL switch circuitry, and I think I have the correct circuit for NTSC?

The goal is to have a simplified, monochrome version of C2 with an attenuator and luma output. My C2 is a little deep for my travel case, I never use RGB anyway, and my wired-up banana and BNC jacks were starting to bother me.

I’m getting a composite signal out, but it’s unstable. Ramps display correctly, but get torn weirdly when they’re processed. Oscillators seem to display properly when they’re h-synced at high frequency, but they freeze and stutter when v-synced at audio rate. My working theory is that my clock circuit is wonky somehow.

  • The “14-pin header” is correct on my PCB even though it looks wrong – my way of avoiding 14-conductors is using 16-pin headers and installing the key backwards, but the cable/signals are connected where they need to go.

  • Power to the ICs is all correct/measured/verified.

  • No hot chips.

  • The crystal cap is a 1% 10pF MLCC.

  • The crystal reads “3.58 mHz” on the schematic, but it is in fact a [ABRACON +/-20ppm 3.579545MHZ].

Also, as this is basically a direct clone of the C2 – I have no intention to distribute it. Just for personal use, and will send one to the OP if I can get it working.



1 Like

duh. my clipping circuit output is on the wrong side of the last diode. kludged. fixed. working. @LauLindqvist – DM me if you want a board. I’ll gladly send you one for inspiring this thread. you’ll just have to cut a trace and put in a kludge wire. also, the format is banana/BNC, but it can be adapted for 3.5mm jacks with a few wires.


Sorry, just looking at this. You don’t need the AD724 at all!!

  1. Buffer your video after blanking switch.
  2. Sum CSYNC (attenuated to 300mVpp) with your clipped and blanked video.
  3. Drive it thru 75R resistor at 2X gain with a final driver opamp.

Luckily I had a 724 already, so I’m not out $10. The circuit works and the module is installed in my case already, so I probably won’t do another one, but I’m going to post a revised schematic just for posterity.