"PATTERN SOURCE
The B Module Pattern Source consists of four independent
digitally stabilized spatial oscillators, each producing three synthetic
monochrome video signals, and a voltage controlled video mixer. The
most important application of this unit is the creation of simple or
complex, moving or stationary, video patterns."
EDIT: Though one of the issues I recall with the Videolab was it was very hard to get it to STOP scrolling but it’s been a few years since I played with one…
I know exactly what you’re seeing even in the posted video, and I agree with @creatorlars that looks pretty darn good. There will ALWAYS be noise. ALWAYS. Given that we are shoe-horning relatively high frequencies into a format never designed to handle them, using unshielded patch cables, it’s to be expected. Even if you start off with a clean looking signal, process that through a couple of high-gain amps (Staircase/Stairs? or an extreme FM of any vertical waveform) and the noise will be right back there.
The following should be considered more philosophical than technical 
: Folks rave about the ‘Fat Sound’ of analog audio oscillators. We’re looking at that equivalent in the Analog Video Synth world. Why this is considered ‘noise’ in the video world relates to the difference in the senses of hearing and sight, but that’s probably really outside the scope of this thread.
Embrace the noise!
I’ve been thinking on this more, and I definitely see video synthesis in part as “revealing the hidden world of signals” and using the visualization of those signals to make art. Especially in the case of an analog specific generator like a VCO. So when I see an SD vs HD capture for example, and this reveals more noise, I am thinking “wow, I can see more of what’s there! it’s like getting my glasses cleaned.” It’s more like a microscope with a sharper lens pointed at the same thing, than a different resolution. That’s what quantifies good noise vs bad noise for me. The bad noise is a different signal getting in the way of visualizing the target one – PSU crosstalk for example. The good noise is whatever is part of the character of the analog generator itself (where it’s not a question of Yes/No to noise, it’s always a question of how much, as Bill points out above.)
I feel the same way about the noise. I think of it as grit. Like the “grit in the oyster”, the “pearl in the shell”. As you said, embrace the noise! Use it as a texture. Control it - perhaps with filters (like Contour or the Gen1 filter module, if you can find one) or key/fading - or let it loose.
It’s all part of the process. I don’t just mean the creative process. I think of it all as one big computational process. Not all computation is digital. Video certainly needn’t be. However, with feedback, you enter the world of cellular automata. This takes the process to a new level.
That’s why my goal, using Gen3 and HD, is to use TBC2 and an HD camera pointing at my monitor. I might also feed the system output back into the TBC2 directly. It’s amazing what a little frame delay can do there.
So obviously HD oscillators will play an important role in that. In my experience, they’re ideal for “perturbing” the feedback process, and the “grit” goes a long way.
I hope that makes sense.
Hey! You’re paraphrasing my Artist’s Statement 
I refer to the Bad Noise as ‘Egregious Noise’ and it can manifest as serious static along vertical lines or a giant deformation rolling in sync with 59/60Hz line freq. There’s an in between here - as @nerdware pointed out in the scrolling ramp edge on Diver (see Diver thread) It doesn’t affect my workflow but if it did I would find it hard to hide.
If you want an example of Bad Noise, I’ll see if I can provide that. 
Makes sense to me. The automata-like behavior may be seeded by the noise!
I post process nearly all my non-live video now and that goes a long way to alleviating analog noise and any digital stepping in the final output. Live, I still run SD out scaled with noise reduction, but I don’t find noise much matters live so long as the content is compelling. I’ve certainly had comments like ‘the repeating geometric shapes were over-represented’ but never about noisy signals. 
Sorry we’re veering out of the OP DWO3 question.
Hope everyone had a fun and restful weekend. On Friday I fired up the studio and this is what I saw coming out of both DWO3’s:
The output of the four oscillators was variable, but they were all unquestionably out of spec.
This looked like a sync issue to me, the bending at the top of the screen is characteristic of sync problems.
Disassembled the rack completely, stripped it down to one DWO3 and one ESG3, both powered by 5 amp DC Distro. Doepfer PSU3 powered down.
Much better. Still noisy to my eyes, but probably not out of spec.
Rebuilt the rack, re-routed sync. Previously the chain was ESG3 → FKG3 → DSG3 → VU009 → VU009 → VU009 → DWO3 → DWO3. The new chain is ESG3 → DWO3 → DWO3 → DSG3 → FKG3 → VU009 → VU009 → VU009 → 75 ohm BNC terminator. Results nearly as good as the stripped down configuration, still noisy to my eyes.
Here’s the VU009 output for comparison. All three VU009 outputs are consistently less noisy than either of the DWO3’s.
Here’s a video showing 1) DWO3 #1 V-sync, DWO3 #2 H-sync, VU009 H-sync. This is a better video than previous posts, with less compression.
Throughout the process, in H-sync mode (vertical bars), the noise is worse on the right side of the screen. In V-sync mode (horizontal bars), DWO3 shows noticeable jitter. I didn’t capture VU009 in V-sync mode, but it didn’t show any jitter.
Overall, the noise is variable from day to day. I’m wondering what the X factor is. Temperature? Spirits of grief and mischief? EM interference? Should I move the Wi-fi router farther away? It’s not in the rack, but only a few feet away.
The current status quo is probably within spec, and suitable to the artistic practice of some, but I would like to bring the noise down as low as possible. Certainly the DWO3’s should look as good as the VU009’s, if not better.
Thanks
in the video I only see the hard edge of your v-sync waveform flickering slightly. i dont see the frequency showing any jitter
V sync waveforms with hard edges will always show slight tearing. as the "wraparound’ of the oscillation will not automatically line up with each line start. this is something that is ultimately very hard to perceive with scrolling. but apparent in sync. similar to @wiatrob 's attention to the diver edge statement above.
DWO3 is likely a little noisier for 3 reasons: the FM response bandwidth is higher, it uses a linear-to-exponential current source for frequency modulation (V/Octave scaling rather than V/Hz scaling), it is a triangle core oscillator with a sawtooth shaping circuit, instead of a sawtooth core oscillator. All 3 of these factors are a recipe for adding more noise to frequency stability, and part of why designing an analog triangle VCO core for video is such a challenge. Where the triangle core shines is with it’s ability to create multiple phase accurate output waveshapes without a waveform discontinuity caused by the sawtooth core’s hard edge (which is always going to have its own time constant/interval.)
NOTE: I could be making assumptions on the VU0009’s circuit architecture, which I haven’t seen. I’m just commenting based on what seems likely.
So I think it is much more to do with the architecture of the oscillator core (sawtooth vs triangle, expo vs linear, FM bandwidth, etc) where you are seeing a difference. In the LZX designs, we use ramps when we want a precision spatial reference, and VCOs are for modulation and FM textures. So we’re going to lean into the differences between the two approaches when designing the modules. In other words, the VCO focuses on what the ramps can’t do, and the ramps focus on what the VCO can’t do.
The vertical jitter you’re seeing is, as Nick described, because the hard edge of the waveform is hitting the center of the scanline. That’s another artifact of how video synthesis with oscillators reveals the nature of the video signal itself. With the triangle and sine outputs, you won’t see the same artifact. If you need a frozen vertical bars, key out a ramp based shape, or use an hsync clock divider based approach.
It looks like you resolved the noise issue that explained the differences in performance between the VCOs, is that right? That was my main concern.
As far as what you should do with this stuff, make some art and don’t be afraid to finger paint! You might read up on all the pigments and fluidity of some new paint series in advance… but you still need to spend many hours getting used to how the pigments mix with each other and how it flows on the brush, and what painting techniques work best with it, before everything is put into context. Watch the demo videos on the modules or wait for a demo before purchasing, if you’re unsure if the results are what you’re after. There’s nothing we’re seeing in this thread that isn’t also in Johnny’s overview video, or previous LZX oscillator demos.
I know, it’s not your first instinct to explore before analyzing – it’s not really for me either. I was the kid who read the full software manual before installing the computer game, and spent more time in the level editor than actually playing the game. But that’s part of what I love about analog so much – there’s a sparkle of chaos in every signal. You’re not defining the outputs, you are driving them.
As for a previous comment about VU002 being apparently less noisy than Stairs, I think it also comes from the fact that bandwidth is probably not as good on VU009 than DWO3, as the hard edge on VU009 isn’t that hard, so this area is a little blurry, which may hide the noise. Someone once said “with great bandwidth comes great responsibility” (or maybe it was power 
), what I mean is that reduced bandwidth will help in hiding high frequency noise/ringing at the cost of a blurrier output, which I guess is one of the reasons behind the efforts put into the Gen3 power solution: reducing power induced noise as much as possible as HD will reveal what wasn’t always visible in SD.
It is PLL (CD4046) based, as it is derived from Castle Clock Oscillator and T. Henry X-4046.
Was wondering why the sawtooth output on DWO3 was twice the frequency of other waveforms, though looking at triangle to sawtooth waveshaping techniques, inverting the triangle using the square yields a clean sawtooth waveform, at twice the frequency of the initial waveforms. Then the square can be summed again with the resulting sawtooth to get the initial frequency, though it would ask for precise trimming to avoid discontinuity, and even then, there would still be a small glitch, which I guess would be particularly visible when using it to FM a H synced VCO.
On the other hand, full wave rectifying a sawtooth does produce a triangle, but once again, there is a glitch at the top/bottom of the triangle, depending on how “hard” the edge of the sawtooth is, but even with a really fast edge, it would still be visible.
So triangle core VCO definitely makes more sense for video oscillators. It’s fun cause those glitches are not much of an issue in audio, as it will only produce a bit more harmonics, which may or may not be heard, however, even a tiny glitch in a waveform used for video will be visible.
Since VU009 uses a passive sync thru configuration, I think it is better to place them at the end of the chain as you did, so it doesn’t effect all the other modules which have buffered sync I/O.
All I can think of after reading these recent posts is the classic eurorack trope of “get both”! (Edit: that I personally want both!) Having both tri and saw core oscs in an audio eurorack case is useful for their respective strengths, and it appears to be true for video as well. Though of course the core is only one factor to an osc’s character, the nuances are fun to explore.
Yes, that’s right. Our triangle core uses a comparator with complementary outputs to drive complementary inputs on the OTA. So it forces the square wave out to be in phase with the triangle out perfectly, regardless of frequency. Since we have Triangle & Square in phase, we can use the square wave to invert half of the triangle to create a sawtooth wave shape without any waveform discontinuities.
In the original VWG (Video Waveform Generator) module, we doubled the triangle frequency using a rectifier based approach after sawtooth shaping, so that the triangle and sawtooth would come out of the front panel at the same frequency. After thinking about it since, I like having the base triangle frequency and 2x saw instead. If you think about the waveshapes more visually, sawtooth is like a linear sequence, and triangle is like a “ping pong” sequence, where the direction reverses at Clock divided by 2. So I didn’t want to lose the ability of thinking of it this way – it makes a big difference when you are combining patterns, that half of the triangle = all of the sawtooth.
For Prismatic Ray, we dropped the sawtooth output entirely, so for DWO3 I wanted to bring it back. The sawtooth waveshaper from a triangle core is not the same as a sawtooth core, though – that is to say (for anyone reading), that VU002 should have a place in the rig beside DWO3, even though both have sawtooth output – as are different types of cores with different characters.
So triangle core VCO definitely makes more sense for video oscillators. It’s fun cause those glitches are not much of an issue in audio, as it will only produce a bit more harmonics, which may or may not be heard, however, even a tiny glitch in a waveform used for video will be visible.
It entirely depends on the module’s identity, in our world. We are not needing to quantize all our frequencies to musical scales, for example. It is HV phase lock and frequency ranges up to 1MHz+ that are the only core requirements for a video oscillator. Everything related to waveshape outputs and resting noise floor are where the character is, or “what it looks like.” So for DWO3, the module’s identity tries to align very much with an idea of “basic dual oscillator for a video synth”, and in this case we wanted all basic waveshapes present (sine, square, tri, saw), and we wanted to focus performance on a very wide FM range and on the waveshape accuracy and waveshape output phase accuracy. We wanted a module that was cost effective and practical in any system. We also wanted to emphasize what was different from this module compared to DSG3 (ramps based shape generator.) If using it as a dual LFO, you would not think “ugh I am wasting this module’s capabilities”, was another priority.
So I think, VU002 has a distinct identity as sawtooth core VCO, and there are plenty of reasons to have that in a rig also. Bajascillator is another example of a different VCO core from DWO3 as well, and is the only one of these providing multiple outputs at different phases. If you want a wide palette of oscillator textures, all of them is a good answer.
Thank you all for your expert responses. I don’t pretend to understand all of it, but I try.
Bottom line is, I was able to get both DWO3’s to the same level of performance. Still hoping to eke that little bit of quality out of it. The stripped down configuration was slightly less noisy than the current status quo.
I had considered that the “jitter” I was seeing in DWO3 V-sync mode, but not in VU009, was due to a sharper rising edge on the DWO3. Thanks for that confirmation.
Had also considered the fact that DWO3 is a way wider band oscillator than VU009. It’s trying to do so much more. Maybe that’s not really at issue. Didn’t know about the other factors, FM bandwidth, exponential scaling, triangle core. That’s all great to know.
Regarding finger painting… another bad analogy. But I get it. Here’s another crap analogy: cooking. My approach is a little bit of the precision baker and a little bit of the “f*** the recipe and use your senses”.
I’ve experienced so much pent-up desire for this stuff. Simply could not wait for demo videos to come out, need to be in line for that first production run. Otherwise, who knows? There may not be any other opportunity. The uncertainty of market availability is a real factor. One thing that’s a lot less uncertain is the enduring, and even increasing, value of these instruments. So again, even if something doesn’t meet my needs, it’s no big deal because I can easily find someone who will put it to good use.
On a “problem” scale of one to ten, I’d rate this as a two or three. Once I got the VU009’s at the end of the chain, things got better. But I was a bit surprised by that, because of course in the pro video world, a loopthrough, unterminated signal doesn’t cause any issues unless the daisy chain is super long. Buffered outputs are great to have, but usually not necessary in a video studio.
Thanks again
When thinking about sync connections, it’s probably best to consider your whole modular synth as the “device”, more like the components inside a self built PC, than each module having external genlock in/out like a broadcast studio rack. In other words, the sync I/O is meant for syncing modules, adjacent to each other, with short cables. I’m not sure why going thru the VU009’s loopthru caused signal integrity issues (or if that was the root issue), but I’m glad you got it working well now.
I think the cooking example is a great analogy. For example, there will always be some random factor of chaos when someone cooks food – even if you measured every ingredient on a gram scale, timed every operation with a stopwatch, there would still be countless little variations that you learn to navigate through experience. And if you’re wanting to try a new ingredient you’re unfamiliar with, you would likely experiment with it in a couple different recipes before saying whether you like cooking with it or not. Or maybe it’s just the wrong ingredient, and not for you. I don’t know! I love analogies though.
I guess in the end my point is less “embrace the chaos! turn your brain off and just play!” and more “travelling the wasteland rim requires an experienced driver who’s used to navigating rough terrain.” You aren’t going to have the same perspective if you are just doing test drives on the roadway – you need to get those new wheels out in the badlands for a few weeks, to see what they do best.
Honestly I have absolutely no idea what I’m doing when I patch videosynthesis. These discussions about how accurate things tend to pop up every now and then. The most inspiring posts for me were patch diagrams so I have a starting point to explore. I would probably never find out if a module would have an error and I honestly don’t care because I feel super overwhelmed with what I already have achieved with two rows of gen 3 and Syntonie modules.
Beside from that comparing analog with digital, feedback and noise is way much more fun and beautiful to explore. Same goes for Audio Modular…
I treat it all like circuit bending. Internal feedback really makes things go wild, so I’m alrways eager to try that with each new Gen3 module. This thread makes HD VCO feedback sound most promising. Thanks.
1000% how I am taking this ride in video and audio patching!
Veering off topic here, but…
Absolutely. I intend to continue my contributions to the community through video tutorials. Patch diagrams need to be part of that. I haven’t yet found a diagram software that does what I want quickly and easily. Not interested in command line, it has to be GUI so I don’t invest time learning a new language. Output needs to be custom styled and formatted for display on a video screen. Things like positioning and line width are super important. I could do it all in Photoshop or Illustrator, but that is very slow because there is no semantic meaning involved. I.e. if I move a box, the connecting wires are not going to stay connected. Does anyone have any suggestions for an application like this for Windows?
Thanks, I am looking for Windows software that I can run locally. I’m allergic to browser-based online tools. Because I need to own my data.
I’ve been using AVSnap but it has a fatal flaw where it’s not rendering the nodes to PDF, so no way to export a graphic and preserve the content.
