Questions about DWO3

Helpdesk
1

Hi,

Got my DWO3s a few days ago and logged enough time to think I might be able to ask intelligent questions.

First and foremost, how do I get vertically oriented bars, scrolling left / right? Should be the Range selector at the right-most position, right? I’m not seeing the frequency stabilize anywhere near horizontal scan rate. Am I doing it wrong? Video format is 1080i59.

No problem getting horizontally oriented bars, scrolling up / down.

This is pretty important to me. If it’s not possible, then I will be very disappointed. My dim memory of the Hearne / EAB Videolab II from 30 years ago was that this was super easy, the oscillators had coarse and fine controls.

Second, what’s the process for getting a Reset that just does a reset, without blanking the oscillator? The blanking effect is cool, but not what I want at the moment. I’m assuming the Reset input is a gate? Positive voltage blanks the oscillator off. When voltage goes back below a threshold, the oscillator starts up again at phase = 0. Right? And that threshold is… 0.5 V?

This would seem to indicate that in order to just reset the clock, without blanking it off, I need a trigger pulse. Hmm. I actually don’t have that in my system. Plus, I suspect that there’s always going to be some amount of blanking, because a trigger pulse has to have some duration. There’s no such thing as instantaneous. I just want the oscillator to sync to an arbitrary incoming signal without blanking. What’s my move here?

Third, in the video below, we can see some noise, a subtle jitter. The video compression does soften it up somewhat, but it’s there. DWO3 directly into ESG3. It’s pretty obvious in full screen / HD.

Also, this is a square wave and it’s really soft. I guess that is just something I have to live with in high definition formats. The internal components of these modules just aren’t fast enough to resolve a transition from zero to one anywhere near the HD pixel clock. :cry:

Thanks,

Aaron

2

To scroll an oscillator left or right, to the best of my knowledge, you’re going to need to take the output of a scrolling audio rate oscillator and run it through the waveform visualizer of a Diver. Then, take the h ramp from the diver and that should achieve the horizontal scroll that I think you’re describing.

2 Likes
3

VU008 Phase Shifter can shift vertical bars with ramp, saw and triangle waveforms left and right.

It’s also possible to put an audio rate scrolling oscillator into Memory Palace and rotate it so it appears to be scrolling left <> right.

4 Likes
4

Forgot about the VU008 for that. Gotta try it!

1 Like
5

That’s not possible with an analog oscillator core – you’d need something extremely precise, like a crystal clock controlled DCO for left-to-right scrolling with a source that is asynchronous to your video sync. We get away with vertical scrolls because that frequency is much, much lower.

My dim memory of the Hearne / EAB Videolab II from 30 years ago was that this was super easy, the oscillators had coarse and fine controls.

IIRC, the VideoLab oscillator used a trick related to delaying the sync phase, or a kind of variable “weak sync pll” to achieve this. I’d like to add sync phase modulation (Horizontal & Vertical) with a dual clock/sync controller module later on. Here’s an old test of something like that:

Second, what’s the process for getting a Reset that just does a reset, without blanking the oscillator? The blanking effect is cool, but not what I want at the moment. I’m assuming the Reset input is a gate? Positive voltage blanks the oscillator off. When voltage goes back below a threshold, the oscillator starts up again at phase = 0. Right? And that threshold is… 0.5 V?

You understand it correctly! What you want can be achieved by sending a narrower pulse. You could use a module like Keychain to key out a narrow pulse for this purpose (use the saw/triangle outs of one oscillator into Keychain, then vary the pulsewidth via threshold – or see above, re: sync/phase controller module, which could also include an adjustable monostable. Contour could also be used to generate a short pulse at an edge.)

Third, in the video below, we can see some noise, a subtle jitter. The video compression does soften it up somewhat, but it’s there. DWO3 directly into ESG3. It’s pretty obvious in full screen / HD.

That looks perfectly normal to me. There’s going to be some “sparkling”, especially in HD.

Also, this is a square wave and it’s really soft. I guess that is just something I have to live with in high definition formats. The internal components of these modules just aren’t fast enough to resolve a transition from zero to one anywhere near the HD pixel clock.

Correct, that’s all as expected.

This is how it goes with analog oscillators and video synthesis. DWO3 is the most technically precise and high performance VCO core we’ve designed, and also the most affordable. I would use ramps based modulations/transforms (like DSG3) or digital sources that count pixels in cases where the analog VCO is not helping you achieve what you want.

2 Likes
6

After I posted, my memory started coming back to me. It’s not a “fine” control at all, it’s a phase offset and drift, like you said. I guess that just goes to show how advanced the Videolab was.

Without that functionality, I’m at a loss to see how I can achieve a drifting horizontal movement. Diver and Memory Palace are not available to me. VU008 is introducing extra black bands into the sawtooth, and just doing weird folding things with sines and triangles. That looks cool, but it’s not what I’m trying to achieve.

Sync phase modulation definitely goes onto the wish list.

1 Like
7

Another note on using hard keying to control reset. The reset input has a 0.5V comparator (fixed threshold hard key generator) at the input. So if you mix an offset in with your ramp or triangle waveform before patching it to the input, you can vary the pulse width. You can do this just with Proc, without a need for Keychain.

1 Like
8

It would be like a normal video sync generator (HV pulse outs) but with modulation controls to scroll or reposition the phase relative to the input sync. So the neat thing about it, is that you could use it to reposition/scroll anything with a sync input (for example, it could be used as a phase/scroll controller for DSG3 as well.) Essentially Diver/MemPal like on the control side,

After I posted, my memory started coming back to me. It’s not a “fine” control at all, it’s a phase offset and drift, like you said. I guess that just goes to show how advanced the Videolab was.

It’s a very cool circuit! It uses the ICL8038 for the VCO core like the Sandin IP oscillator did, if I recall correctly. Having the sync phase controlled by a crystal clocked syncgen (like “scrolling momentum”, above) is a lot more stable. You can see from that video that the reset periods in the older cores were very long. DWO3 syncs in 200-300ns (as opposed to 2000ns-3000ns from previous cores), so it will be capable of seamless repeats in a new way.

3 Likes
9

Turn your feedback camera 90 degrees and rescan it! That was the trick, before things like Diver were introduced.

2 Likes
10

… to the LZX modular standard. Obviously Bill Hearn solved this artistic / technical problem ~40 years ago.

I’m not trying to be snarky, just want to get back to where I was in 1993. Don’t get me wrong, I fully appreciate all of the MANY advances you’ve made. I lust after Diver on a regular basis.

Even just Swatch by itself is a huge advancement. Today I learned that I can use a Function Generator to process the positive and negative parts of I and Q separately. Never would have imagined this.

Thanks

2 Likes
11

Of course! I am talking about early LZX days here. We decided to address sync phase modulation outside of the core oscillator module’s feature set. The “scrolling momentum test” was a proof of concept. Later that idea made it into Diver. But we’ve never really done the concept justice as an independent module.

Just to be clear: the Videolab’s ICL8038 circuit doesn’t do asynchronous horizontal scrolling with it’s VCO core either. It uses a sync PLL so that the oscillator core is “genlocked” to the sync, but with a weak pull on one side or the other. Super clever, I agree. My goal with all this is not innovation so much as preserving and expanding the legacy ecosystem.

3 Likes
12

Is this normal? Because to me it looks doubleplus ungood.

The other DWO3 is also noizy, but not nearly this much.

13

Dammit, the Vimeo compression just smears the heck out of it so you can’t even see the issue

14

Here’s a still image showing the horizontal noise, super problematic. Hopefully this won’t get mangled by Discourse … I uploaded a PNG but it looks like it got automatically JPEG’d

00_12_17
00_12_171920×1080 129 KB

15

Looks great to me, especially for HD. If you’re looking at straight, non modulated bars near the lowest frequencies in a range like this on an analog VCO, with HD output, with no modulation at all, there is definitely going to be some texture and nonlinearities. What we’re seeing here is the “good” kind of texture created by subtle thermal properties of discrete parts. The “bad” kind of texture involves diagonal, jagged edges and scrolling ripples caused by power supply bleed.

There will always be some of both in the analog path, but the goal is to get the bad kind low enough so that it doesn’t show up in the recording. Without any texture at all, it might as well not be an analog VCO core – which is why we use a crystal clocked 10-bit DAC for ramp generators. A generator that’s running on a pixel clock is always going to line up with a digital capture in a way that creates the illusion that the image is perfectly crisp or “frozen” (rather than being in constant motion, as it is in reality.) An analog VCO at high resolution on the other hand is going to have some sparkle when it turns into pixels – it’s on it’s own time base, one constrained by the physics of the transistors. That’s part of the appeal of it – it breathes in a way that a DCO can’t.

Our role in the creation of video art tools will always be to provide the underrepresented alternative to real time software visuals, and try to balance the engineering challenges against the cost of making the gear accessible. We’d never presume that an analogue VCO is superior to other approaches – just that it’s worth making it exist, for those who want it. So while you see noise here, I see life! I’d use VCOs when I want to create an organic, flowing pattern with smooth curves. I’d use a digital generator or Touch Designer if I wanted to save and recall precise settings, and modulate things at mathematically perfect ratios.

You can think of it like FM synthesis vs Subtractive synthesis. FM synthesis is prohibitively difficult to do with analog circuitry. The results of an analog FM synth can sound cool, but the math doesn’t add up to some of the clear bell tones you get from a digital FM engine. If clear bell tones is what you want, digital FM is definitely the way to go. On the other hand, if you stack multiple analog oscillators into an analog filter, the slight variations in texture are what create a squelching, strong bass sounds that the FM synth just won’t be able to reproduce on its quantized timebase. Different tools for different jobs.

Analog video VCOs will create patterns you can’t get any other way, especially under complex modulation. But if you want pixel perfect frozen patterns with exact counts, where things start and end on specific pixels and can be saved/recalled, the analog VCO is not the right tool for that job.

17 Likes
16

Maybe I’m missing something here, but can’t you just use a finely tuned unsynced mode on DWO3 for scrolling bars in either direction? Not as precise as maybe you’d like yet, but would get the motion.

17

At audio rate frequencies – vertical ranges, where the waveform scrolls top to bottom at multiples of vsync frequency, this is exactly what you can do. That is range #6 on the dial.

At video rate frequencies, slight fluctuations in FM current in an analog triangle VCO core cause the frequency to not be stable enough to “drift” left to right, when at an asynchronous multiple of the hsync frequency (15KHz - 30KHz). That’s why a sync input is necessary when working at those frequencies, and also why you see some slight movement in the waveform even when it’s synchronized.

We could probably reduce slight movement by filtering the FM CV input signal, but then you would be making it more difficult to achieve video rate FM response. So we have to pick a sweet spot between response time and FM fluctuations at higher frequencies.

When working with oscillators, I like to consider the high frequencies (the vertical bars) as the “stroke” of the pen. The waveshape matters a little, but not a lot – that’s the dot size of your pen and how fuzzy the lines are. The waveshape of the midrange frequencies (the horizontal bars) matters much more, since their waveshape is what creates the pattern. Since FM should happen at a lower time constant on the modulator, compared to the source, you usually modulate the high frequency oscillator with the midrange frequency oscillator in order to create 2D FM patterns that are very characteristic of analog video VCOs. The waveshape of the modulating oscillator (the midrange one) matters a lot more than the high frequency one in this case, because it is the one actually drawing the waveform (using the “pen stroke” created by the higher frequency one.)

Yes, analog VCOs are going to feel “fixed” in their relationship to horizontal and vertical space. That’s very much part of the medium itself. I love 2D FM VCO patterns for this reason, they reveal how the space on the screen is related to the timing of the analog signal in their structure.

7 Likes
18

OK, you’re the expert. But there is a wide variation in the amplitude of this noise between the two DWO3s. The first one is acceptable to me, this second one is not. Maybe it’s within spec, but from an artistic standpoint, I can’t use this except as an LFO. I am going to try moving things around in the case to see if anything changes. If not, I may sell it. One great thing about these boutique instruments is that they not only hold their value, but historically they actually appreciate. So even if a module doesn’t meet my needs, in the long term there’s not much financial risk.

It’s great to hear the background info about why the DSG3 ramps look so clean and the DWO3 waveforms look so… analog. Thanks for the education, as always.

19

You need to quantify this with some numbers somehow, or else I have no basis for understanding what you mean or if there is an RMA situation. Can you try some things?

Can you take a comparative side by side screenshot of both VCOs?
If you swap the VCOs’ locations in your case, does it help?
What if you swap the power cables?
What if you swap the sync cables or change the sync order?
What’s the amount of headroom on your power supply right now?
Have you seen if powering the module alone makes a difference in what you are seeing?
Do the waveshapes and frequency ranges work consistently across both units?

20

Thanks, I will conduct some more thorough tests.

The power supply is a 5 amp DC Distro. At the moment the available overhead is 3800 mA.

All waveshapes and frequency ranges look normal, consistent across both units.

Also, would you be so kind as to update the product page and docs to reflect the DWO3 power draw of 230 mA according to the rear panel?