Questions about SMX3

Sorry to be a pest. I’m running through a lot of signal tests, learning a lot about what I can do.

I am confused by the behavior of SMX3. With nothing patched into an input, rotating the knobs gives a slight DC offset. But not a full offset like I would expect if a 1V signal was normalled in. And on one of my SMX3s, the gain isn’t centered on zero. Bottom line, if I want to pass a single ramp transparently through either SMX3, I have to set the pots to completely wacky positions that make no sense. Are these out of spec?

Yes, that is normal. The important thing here is that you have a true nulling position in the center (this was often hard to achieve with the passive attenuverter controls used up thru Expedition series.)

But not a full offset like I would expect if a 1V signal was normalled in.

That’s right, inputs are normalled to 0V, but there’s always going to be some slight input bias in an op amp circuit like this.

And on one of my SMX3s, the gain isn’t centered on zero.

How do you mean? One of the knobs, or the entire module?

Can you be more specific? What knob position produces unity gain on the output? There is going to be some variation, but it’s a question of scale before I can know if there’s an issue with your modules.

Testing this is complicated by the fact that I don’t have a source of truth. DSG3 seems to have a very slight setup of a few IRE. I think DWO3 is more accurate to the 0 to 1 range. And calibrating ESG3 still eludes me. My studio setup doesn’t make it easy to switch ESG3 to RGB mode, so I’m winging it, trying to calibrate the encoder in YPbPr mode.

The variations I’m seeing seem to be specific to RGB channels of each SMX3, not to individual inputs.

With nothing patched in, and all knobs at zero, I’m seeing variation of up to 5 IRE in the outputs of SMX3 channels. That’s probably within QA spec, but I don’t like it. I can’t assume that if all attenuverters are at zero, I’ll get a voltage of zero. All the more reason to always be looking at a waveform monitor. Trust no one.

So to fix that DC offset issue I need to adjust the knobs of the unconnected inputs. That’s distressing, but even worse, polarity is reversed. Setting the attenuverter to negative to introduce a positive offset makes my head hurt.

On some channels, the gain is centered a little above zero. I.e. increasing gain causes the darkest portion of the image to get slightly crushed. Not really a big deal when working with synthetic waveforms, but can be an issue when processing external footage.

This is probably just another case of my expectations being too high. But I had to bring it up. I am super confused by having to set null inputs to non-zero positions simply to preserve unity gain.

1 Like

Oh, and to remove the DC offset, I’m having to set one of the unconnected input knobs to as much as -1.

With nothing patched in, and all knobs at zero, I’m seeing variation of up to 5 IRE in the outputs of SMX3 channels. That’s probably within QA spec, but I don’t like it.

That’s correct, 5% of nominal is within spec. It’s usually more like 1-2% or less, but the decision to go 2X gain is likely exaggerating the amount of variation.

It will be much less frustrating to leave all of the attenuverter controls centered, and worry about RGB offsets / scale at the RGB encoder module where you have dedicated controls for amplification, attenuation, and offset on each color channel. Dial in your desired color balance after the patch has been set up. If you are setting up a very complex mixing patch (like a staged pipeline), you can insert Proc modules on each subsystem/image generator’s output, so that you can adjust color balance before the downstream sum.

If you are used to parameters in digital systems, 0% always means Zero, because it’s a number. When dealing with analog systems that have 1% to 5% variation range to the controls, it’s best to not think about a number and instead monitor your output by looking at the screen. Watch the screen for “the point at which red is fully saturated” as you turn the knob and stop when you get there, for example. The controls have been designed to have over range where appropriate, so in cases where you get 101% on a control on one module, and 104% on another… the important thing is that the range covers that 100% mark. And that when you’re in the null spot, no signal from the input jack is bleeding thru to the output jack.

So while I get what you are saying, and there is always room to improve (like I said, 1% - 2% would make me happier than 5%), please realize that if these modules had everything precision zeroed out as if they were for the broadcast test signal generator market, they would be a lot more expensive – and without much to show for it, from the artist’s point of view.

Think of your LZX output like raw mixdown from your band’s recording session or logged footage from a video shoot. Before pressing it into a record, or cutting your video, you need to do mastering or color balancing across takes. I presume this is done like any other video project, with color grading and compression tools in the NLE.


To put it another way, and this is of course, just my perspective – aim not for 0% color balanced perfection on the output of your video synth. Aim for the same range of 5% that the circuitry has. Like you would if turning up an amp or fader on an audio mixer. Get there with the ESG3 knobs. Hit record. Record some takes. Worry about the precision color grading when you’re mastering your demo reel or piece for publishing. You should be doing that anyway! Recording direct to the 4-track and then copying tapes without mastering is a valid way to release music, but there’s a reason that’s usually called “tape music” or “lo fi.”

I apologize for all the audio metaphors, it’s just the world I am from. Hopefully it communicates!


Yeah, it’s hard for me sometimes to deal with the fuzzy nature of all of this. Control freak mentality.

But I must say that adjusting things at the encoder stage goes against my whole nature. And I think the audio mastering analogy is very poor.

Anyway, relying on the encoder to fix problems limits interactivity and experimentation. If I tune the encoder to get a certain result, and then decide I want to change some upstream parameter or connection… I’ve just wasted time tweaking the encoder. Now I need to tweak it again. Much better to have that encoder set to neutral, it’s not introducing yet another variable to an already complicated system.

It’s great to have those controls directly on the encoder, especially in small systems. But it shouldn’t be a requirement.

The replacement ESG3 I have doesn’t have detente positions, which is a blessing and a curse. Great for slight adjustments. Really bad for predictability and repeatability.

In an ideal world, I would be recording the output of the modular system and not touching it at all in post. I’m in front of a computer GUI too much already. So when I’m ready to capture, I really want WYSIWYG. The first piece I finished had a couple hundred shots in it, and only two of them had any color adjustments. They ALL had scale adjustments to get rid of the evil blanking bar on the left.

But it’s cool, you’ve designed these instruments after over a decade of development, and I have mad respect and admiration. You can’t please everyone. I hope I’m not coming off as ungrateful, nothing could be farther from the truth. I’m just trying to utilize this medium to the fullest using the skill set I have, which includes a methodical, logical process. Experimentation, messiness, unpredictability, that’s all great, but I want to start from a position of security and then branch out into the unexpected.

Thanks as always

1 Like

It’s important to realize that a problem only exists in your preconception about how the instrument should work. ESG3 isn’t an isolated device, like a broadcast RGB encoder in a TV studio rack – it is a component piece of a single device, like a building block. It becomes the second half of SMX3 when they are patched. The single source of truth for the patch is the video output. Meaning that the correct position of the ESG3 knobs is whatever position nulls out all the bias tolerances or gain stackup of the patch as a whole – which should be somewhere within 5% of center for each control.

There’s nothing messy, experimental or unpredictable going on – that’s not what I have tried to communicate – 5 percent tolerance or better is a technical specification like any other. The variation comes from differences in the properties of discrete transistors.

You are obviously diving in deep – but my perspective on your frustrations is that I think it’s more about learning to “think analog” instead of trying to “think digital” when it comes down to how you’re evaluating the tools. You don’t have to undermine your technical understanding to do that – it’s just a different kind of instrumentation, where you factor tolerance stack ups into your understanding of the controls. That’s one of the disadvantages of analogue, if you’re trying to construct a logic/deductive approach to using the system (which is great.) I get the appeal of having everything extremely precise, or having a mastered recording coming directly out of your instrument’s output, but in anything equivalent I can think of, that’s never the case – analog audio and videography always involves taking raw material and then mastering it.

In any case, the answer to this particular frustration is right in front of you. Touch those ESG3 knobs often and get used to adjusting them constantly as you work, especially if you’re doing something requiring color balancing! Those are the most constantly touched controls in the instrument, for me. Let the periodic color adjustments be a moment to take a breath and think about what you’re doing in the rest of the patch.


Thanks for your insights. Everyone’s artistic process is different. Certainly there’s an element of “serenity prayer” in everything. I will work on my expectations. Coming from the broadcast industry, 5% tolerance is completely unacceptable. But this isn’t the broadcast industry, it’s Eurorack. Totally different. I need to get that in my head.

And I appreciate the constraints within which you worked to develop the LZX modular standard. You have educated me (and anyone reading) all about that. You need to keep voltages low in order to keep speeds high. But the downside is that any small deviation in voltage is going to have an outsized effect. A 0.05 V DC offset doesn’t really matter in the audio domain, where signals are 5 to 20 V peak to peak. Compounding that, we can’t hear a DC offset, but we sure as heck can see it, especially in the blacks.

Having said all of that… generally speaking, I’m not touching those ESG3 knobs. Sorry if I’m doing it wrong. In my mind, the encoder is a conversion module, not a paintbox. I know that’s not the instrument you designed. And your design does exceed the expectations of the vast majority of users. I’m just an outlier, in too many ways to list here. it’s always been that way. Even in grad school, I was the only student in my cohort who really wanted to know how the video signal worked.

Anyway, I don’t want to use up too much of your time. Thanks for everything.

1 Like

I feel the same way about ESG. While patching I frequently check to see what a little bump will produce and also will bounce back to “detent zero”. The color pops you can get are just insane! I guess I can see the viewpoint of not adjusting, but if the concept was to not touch them they would be someplace other than on the front panel I guess?


Absolutely. I think of audio mastering vs audio recording analogies again here. Audio mastering equipment is several times more expensive and has specific demands because the application is different. Broadcasting a video signal at 110 IRE is not OK for live broadcast from a TV studio, for example. That’s going to millions of homes. But recording it somewhere between 95 and 100 IRE in the studio for a video project, or performing live where you have to deal with projectors and stage lighting? All situations where you will just want to tune the output to taste. It’s why cameras have iris and focus controls and why guitars have volume and tone knobs – and I see the “video instrument” in the same context as cameras and guitars, in it’s relationship to the studio and the artist.

ESG3 is specifically a combination of front end proc amp, sync generator and RGB encoder. So it is both paintbox and encoder. Maybe we will consider a more stripped down encoder-only variant in the future. But I would find that frustrating to use. The same as an audio synth needs a main volume control, a video synth needs main RGB Gain/Offset controls.

I’m not trying to invalidate anything you are saying, and I hope I don’t sound that way. You are welcome to whatever your views are on video instrumentation, of course, and there are many perspectives to have. You have the designer of this particular instrument right here, and I am just trying to explain how it works. So you’re not doing anything wrong. But it does put me in the position of needing to not just answer your technical questions, but also write some essays about how and why it works the way it does, instead of how you expected it would.

From my perspective, you bought the thing! You might as well at least try using it how I’m suggesting, before discarding the idea that your encoder can be a performative part of the instrument. :slight_smile:


You prompted Lars to explain a lot of fundamental thinking behind this system with this thread and the other about the oscillator, so thanks!


Sorry, I think I gave the wrong impression. I absolutely have used the ESG3 in the way you intended. It’s great, I just prefer to work differently.

I’m not disregarding your input, or your design intent, now or ever. I think I speak for many when I say I hang on your every word. This is why some users took it upon themselves to document (“condense”) every one of your posts across multiple platforms.

Sorry if you felt pressure to write essays. It is good for the community, but you have so much responsibility already. Thank you for your willingness to engage directly with users so thoughtfully, and so thoroughly. That is very rare in my experience.


I appreciate the time everyone has put in to share knowledge and perspectives. For neophytes like myself it’s fascinating to learn more, and there are likely lurkers who don’t even have an account who also gain from reading these discussions. Much respect to the time, work, and exploration everyone has put into this realm! :metal: