Posted: Wed Sep 03, 2008 10:10 am Post subject:
Using the MC1377 to generate video
I've been researching the Sandin IP Video Synthesizer's schematics for several months now -- mainly trying to find more modern work-arounds for the obsolete ICs used, in an effort to at some point build some video input/output/manipulation modules that will fit right in with the rest of my synth. (Using LFOs to modulate video parameters as well as audio, for example.)
I've recently stumbled upon a local supplier of the MC1377 RGB-to-NTSC Encoder Chip, which I believe could be the answer to easily solving the biggest challenges in dealing with analog video signals, which are all the timing/sync/burst pulse mess. I don't even pretend to understand all of that. It seems like I could use this chip along with some circuitry to buffer and scale the R, G, and B inputs, to create a kind of video output module with signal inputs for these three colors.
I've attached the datasheet. If those of you with more knowledge than I could take a look at the example applications and offer any feedback as to the feasibility of the idea, I would appreciate it! Also, I'd like to prototype the example circuit (minus the audio modulation portion), but can't seem to identify a couple of the components, such as the color bandpass transformer, 3.58Mhz crystal (i think?) or the items uncommon to synth circuits (which are what I'm familiar with.)
No ideas? No interest? In another topic, is anyone interested in some video-related Synth DIY PCBs/projects? I'd really like to see a set of modules that can get a video signal in and out of a standard modular synth easily, with an emphasis on integration with existing CV generating modules.
Joined: Nov 30, 2007 Posts: 317 Location: sometimes
Posted: Thu Sep 04, 2008 8:46 am Post subject:
I don't know much about video synths but I have suggested using RGB in the past for the exact reason you mention, the sync signals are separate from the video signals. There is the downside that you won't be able to affect all 3 channels at once (I really have no idea what's possible with analog video synthesizing/processing anyhow) but I still think its worth looking into.
If you have an old CRT around it might be worth using RGB straight into the monitor (with appropriate sync signal generation).
I'm not really sure what all the possibilities are, either...
But it seems like the R, G, and B input pins on the MC1377 are each expecting an un-sync'ed 1.0Vpp signal. What I'm hoping is that VCO's could potentially be connected to these inputs via attenuators (with an overall scaling of a synth-level signal down to 1.0Vpp) to generate moving colors at the output. CV over R, G & B gain is then easily realized with an offset control + VC resistor summed with the inputs. Besides R, G & B you can probably have voltage control/offset over Luma too, as well as separate outputs for Composite NTSC, Luma (aka Monochrome/B&W), etc. All of this would be an output module, generating all the sync signals necessary for a composite output. The Sandin was similarly designed, except it used an encoder board from a color video camera which is no longer available (and very expensive at the time!) I think the MC1377 can do the same thing as all of the Sandin's sync generation/encoding related circuitry.
With that working, the second step would be to build a monochrome video "input" module, which would take a video input signal and strip it of a sync signal, and buffer it for processing. Naturally this strips away color information, but we could add a standard RGB/VGA input and eventually some type of Composite-->RGB converter in the input too.
The idea, I guess in the Sandin context, was you could buffer a video input (or multiple ones), send them thru various comparators and multipliers with VC parameters controlled by standard oscillator circuits, and then patch into the RGB output, basically using the system as a type of VC analogue colorizer/FX unit.
In the effort of more "video synthesis" type stuff, maybe a microprocessor-based VC digital pattern generator could be implemented, or oscilloscope-based visual generators.
Joined: Nov 27, 2005 Posts: 846 Location: Bristol, UK
Audio files: 1
G2 patch files: 1
Posted: Thu Sep 04, 2008 1:43 pm Post subject:
I'll put my hand up and say I'm interested!
I have had a look through the sandin docs a couple of times and would *love* to try things out - but, who knows when/if I'll find the time!
I'll do my best to have a check of the datasheet you've posted..
I know very little about video electronics
(I've got a VideoElectronics book on order from Futurlec - but its already taken a while and they still ain't got in in stock, so....)
Wow, I didn't know they had it either... Mouser sells an NTE replacement as well. I have a couple of those as well as a couple I bought from the local shop (Tanner's.) I also have a selection of video/wideband op-amps and OTAs to play with.
I'm thinking the whole endeavor as far as cloning or reinventing Sandin modules goes will be very easy once we find a way to do the basic input and output modules. It's generating the sync and creating a final signal that can be displayed properly that is the challenge, I think.
Let me know if you have any thoughts on the datasheet... we'll keep the discussion going as I try to breadboard some stuff over the next few weeks.
Joined: Jun 20, 2007 Posts: 158 Location: Buffalo, NY
Posted: Thu Sep 04, 2008 7:23 pm Post subject:
You might want to look into the LM1882 sync generator. You'll need to program a few registers in it to set up the parameters but it does generate hsync and vsync. _________________ Nothing succeeds like excess.
Joined: Nov 30, 2007 Posts: 317 Location: sometimes
Posted: Thu Sep 04, 2008 8:17 pm Post subject:
Sorry about that! A CPLD is a programmable digital logic chip, kind of like a whole mess of CMOS chips in a single chip. You can design circuits for them graphically using schematics or hardware description languages like VHDL or Verilog. CPLDs are kind of like small FPGAs if you've heard that term (Altera and Xilinx are the big names in this industry).
Anyhow the CPLD would basically be a fancy clock divider to derive the H and V sync signals. I already have code around to do this, but its fairly simple if you know what you're doing.
I'm in for V-synth modules!
I know a couple of guys who has researched a bit into this already, I'll check with them.
I need PAL though, i'm in Europe and NTSC (Never Twice the Same Color) isn't that popular here!
I suppose I am one of those guys you refer to...
I started digging in the concept of video-synths a few years back when I worked at Swedish National Television, SVT. As we were going digital I salvaged quite a lot of old analog video-gear from the dumpster, but most of it is gone now. I gave a few distribution amps and a serious syncpulse generator to the other guy you probably are talking about...
My first idea was to make a PAL vide encoder/decoder to fit into my modular synthesiser project. I decided to hack a video processor kit from Velleman to make things easier. The Velleman processor is made up of a decoding part (CVBS/S-video -> RGB + Sync) and an encoding part (RGB + sync -> CVBS/S-video) and some extra cirquitry to tweak RGB levels, hue, saturation etc. By making the connection between decoder and encoder parts availabla on a patch panel this would be a nice video interface. My idea was to use that sync generator that I gave away to drive the encoder and sync oscillators. But you could as well take any stable video signal into the decoder and have sync.
I started building the kit and put it into a modular panel with a bunch of BNC connectors, but then my life took another turn, I moved several times, changed job etc so the project stalled. The half-built video module is still in its rack cage just above my desk and making faces at me...
Back then I used to go swimming several days in a week, and used to come up with all sorts of bizarre inventions because my brain wanted something to do while my body was swimming a couple of km:s.
One of these ideas was an x/y-to-raster-converter with voltage control over x, y, hue, saturation and luminance.
When I did my mandatory military service I worked as a technician at the air photo surveillance. In our mobile photo- and analysis labs we had an analog image processing device connected to a hi-res video camera mounted above a light table. With this we could zoom in onto the photographic film and manipulate levels, invert, change gamma curves, thresholds etc to bring out details from the film. This was at the end of the cold war, when we had fighters and reconnaisance planes on alert 24/7, so I have seen some interesting soviet stuff on that monitor... Anyway, it's a pity I din't make copies of the schematics and service documents of the machine. (Nothing classified anyway...)
As always when I post something, it tend to be loooooong.
But to make a long story short; Analog video is equally cool to analog sound synthesisers, and i should REALLY do something about that video module in the rack.
I have the Sandin IP plans at my bedside table and look into them now and then...
Thanks so much for sharing that. Rotwang -- your work is truly inspiring... it very much sounds like we are on the same page. I was not aware of the Velleman 4600. This sounds like a PERFECT place to start. This appears to use the MC1377 as well, and this with separate ins/outs and voltage control over the R, G, B, contrast, etc. parameters is exactly what I was envisioning for an "input/output" module. I only wish there were full schematics available for the K4600. Is anyone aware of similar device schematics? I'm going to research the other ICs it uses as well... as well as track down one of those kits for myself.
I have a handful of 1881s and 1889s as well as some 592s and a handful of other video chips. I am cleaning out my bins right now and should be able to provide a full list.
I was in a similar position with you recently as far as old analogue video gear goes. A local university had a film department in the late 70's that closed down in the 80's. I acquired several pieces of analogue video gear, including an ENORMOUS rack-mounted NTSC sync generator. I haven't had a chance to open it up yet, but I imagine it will become incredibly useful. I also have an ancient video FX/mixer unit and a Paltex Editor. The FX unit doesn't have any ICs inside and looks very old.
Also have an Ikegami tube-based camera that receives R, G & B from separate tubes and includes an RGB encoder board. I had an idea to modify this camera heavily and use it for some film projects in which analogue modules and sequencers were used to manipulate the camera feed live during a shoot, before capturing on tape (I have a BA in film but haven't really done much with it yet.) There are all sorts of R, G, B adjustment trimmers inside that It would be fun to voltage control.
The good thing about using the MC1377 is that it goes from NTSC to PAL with the flip of a switch. If we get a design together I'm more than happy to go buy a huge lot of MC1377 at the local supplier and distribute them, perhaps with a PCB.
Yes, the Sandin used two ICs (the 1445 and something else...) and I can't seem to find ANYTHING that resembles a drop-in equivalent (of course I also don't know what I'm doing.) Someone smarter than me could probably explain how to make the NE592 do what the 1445 does. I think the two ICs are used for pretty much the same purposes throughout the Sandin, so that's what makes me think that once some substitutes are found, and the I/O problem is efficiently solved, making very accurate clones of the various Sandin processing modules should be extremely easy. No need to clone the Sandin's Oscillator... just scale everything to work on the same CV range as any synth module.
Sorry for the ramble, just a lot to discuss. Keep it coming. Gonna go googling for that 4600 kit now...
Posted: Sun Sep 07, 2008 3:06 am Post subject:
Re: Using the MC1377 to generate video
I've recently stumbled upon a local supplier of the MC1377 RGB-to-NTSC Encoder Chip, which I believe could be the answer to easily solving the biggest challenges in dealing with analog video signals, which are all the timing/sync/burst pulse mess.
I'm probably that "other guy" Rotwang and dar303 mentioned. There are how ever other names that knows alot more about this. Dan Slater is one, Grant Richter, of course we have the old guys Bill Etra, Dave Jones and Steve Rutt.
Apart from the heritage from Rotwang (tack igen!) among the funnier stuff I still have are a Fairlight CVI, a Atari Video Music and a unbranded brittish colour corrector/synthesizer. I also have a small Grass Valley video mixer.
A videosynthesizer could be several things. One is basic generation of shapes and colours. Another is colour and light processing of input video signal. Yet another could be a video feedback system using a monitor and a camera.
I believe that analogue videosynthesis is easiest built using similar modules as musical modular synthesizers. There a few modules that has little or no use in audio that we need while generating a video signal. But other than that most modules are already available.
Below are a few toughts on an idea on a system of video voices. It's several ideas combined so I hope it makes sense.
One idea to generate video is to set up a video voice used as a shape generator. A video voice could basically be thought of as a moveable dot, just like a dot on a oscilloscope. This dot can change shape to a line or some other shape.
The video voice would accept three inputs; X, Y for position and L for brightness or gain. Preferably those would be generated from a DC signal say 0-10V. I'll talk about an idea HOW to do all this a bit later.
So, we have a dot that we can place on the screen using ordinary XY. What if we want a different coordinate system? What if I want polar coordinates where the positioning is done using angle and distance from center? Easy, or at least not at all impossible. We put a little module that uses phaselocked oscillators to translate θ = angle and r = distance from center into X and Y and feed those into our video voice.
By controlling position thru the polar>cartesian module we can use the same video voice used earlier but makeing it do completely new patterns: circles, flowers, spirals etc.
-Set a synced LFO to the r and another one to θ and you have your polar shape generator.
-Add a DC-bias to X and Y using a simple mixer and you can move that flower on the screen.
-Add a DC-bias to the signal that is fed into r in that polar>cartesian module and you have a control over size of the flower.
-Trim the θ to for instance 2V = 2pi and you would be able to spin the flower using a ramp oscillator.
We could probably make a usable 3D position translation module that takes X Y and Z as in depth and translate those into 2D X and Y. Based on a patch on the Nord Modular list many years back I did this http://www.youtube.com/watch?v=ZlZH1p0EUBs
It is "balls" that are moved around in six positions in a 3D world. It's all done using virtual modules in a Nord Modular synth so it it's definitely possible to do with a analogue modular musical synth.
I also have a Pong game on video using the same modules.
A analogue computer would very handy for computation on positioning ...
So we can do alot with positioning a white dot. How about adding colours?
Our voice only has brightness so to make it more fun we can split the L into R G and B. Rotwang has the correct numbers on the amount to be feed into each basic colour.
However RGB is what the monitors expect and that's what we feed into a RGB>PAL or RGB>NTSC circuit but that's not how we percieve colour.
For humans colours are best thought of Hue, Saturation and Brightness. But to be able to use HSB with video equipment we must translate these into RGB using something like the schematics Rotwang described. I had some circuit idea now lost on another computer but I remember that I made it kind-of work on the nord modular so I'll try to dig that up.
For the positioning to work it needs a reference coordinate system, it needs video H and V sync. The X and Y position of the dot is actually a delay measured from the top left position on the screen.
So to show a dot at position X and Y we turn the L (brightness) on at time X1 and time Y1 and then off time X2 (our desired dot size later) and time Y2 (at least one line higher). We have to deal with interlaced lines using PAL but I'm too lazy to think about that now
We could achieve the delay above of the X and Y using a real delay synced with the video sync which would be expensive or we could use a comparator on a ramp signals which is cheaper and makes DC control much easier.
One common way to deal with positioning based on a sync signal is to setup two ramps:
V; a ramp falling from top to bottom, sync it to each frame
H; a ramp falling from left to right, sync it to each line
All video mixer I've seen schematics on uses comparators for positioning of say a wipe or a border.
I have not given this time enough but it should be fairly straightforward to setup two comparators each for H and V so that we turn the L on at level X1 and off at level X2 and the same for Y1 and Y2, see simple ASCIIgram.
We need both X and Y active to turn the L on otherwise we would have vertical a line on the screen.
Each voice would need "house sync" H and V then the ramps are probably best generated for each voice module. So the inputs of a voice would not only be X and Y but actually X1 and X2, Y1 and Y2.
X2 would be X1 + an offset where that offset sets the X size. The same goes for Y. Some thinking and experimentation needs to go into this.
Maybe some one has an all analogue video mixer schematic that we could be inspired by. My GVG mixer has analogue control but digital generation of patterns and shapes.
POSITION AND COLOUR
How do we combine a positioned dot with a colour generator?
If we can generate L at a certain time as a function of X1X2 and Y1Y2 we can of course generate the three channels R G and B at the same time.
Probabaly it would be easiet to input L from a voice as a controller of the primary colours RGB. Those should then be the input to a composite video encoder. RGB values could be modulated by a HSB>RGB module to get proper colour control.
What if you want five voices in one video synth? I'm not shure how to best mix several shapes but there are atleast two options.
Either we mix the RGB signals from each voice before they enter the composite encoder or we have a composite encoder for each voice and then mix the composited signal.
The first one allows for easier control of sync. So that is probably the best way to do it.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum You cannot attach files in this forum You can download files in this forum