Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Fast Switching Circut On/Off

Status
Not open for further replies.
The signal is VGA 1024 x 768 15khz at MAX would this suit the chip?

VGA 1024 x 768 is called "XGA" and typically requires 65Mhz of bandwidth. I'm not 100% sure if that is total, or if that is for each line. I am almost positive the MC14551 will *NOT* pass that without causing massive distortion in the video quality, but you can try it. You will probably need a better chip. I will look for a generic demux with that kind of bandwidth in the meantime.

Also, what is exactly the 15Khz refereeing too? Horizontal refresh?

I am currently not using a relay setup its manual at the moment.

So... you haven't tried it yet?

Chances are, with that quality of video standard, you will get a good amount of distortion using relays. Especially if they are not small signal type. You also need to use coaxial feed line for all of the connections, you can't just use plain wire, the video will look abysmal. As I understand it, anything in the 10's of Mhz and above is where you need to start paying attention to layout and termination and such. Such a project would be best done with a real printed circuit board having separate power planes for digital and analog, which gets you into multy-layer PCBs and $$$.

You might be able to get away with just a double sided PCB if it's small and you use the one side as a the ground plane. I'm not sure though, someone else will have to chime in on this one for me.

How would we delay the chips so its not possible for them to have 1 card half hooked up? eg (the delay between switching cards)

The MC14551 automagically employs "break before make" logic for all it's switch elements, ensuring no half hooked in connections. The maxim chip does something similar, no tricky delays and such required.
 
VGA 1024 x 768 is called "XGA" and typically requires 65Mhz of bandwidth. I'm not 100% sure if that is total, or if that is for each line. I am almost positive the MC14551 will *NOT* pass that without causing massive distortion in the video quality, but you can try it. You will probably need a better chip. I will look for a generic demux with that kind of bandwidth in the meantime..

Hrm thats odd, atm i am using a vga male conector with standard wire soldered to the monitor nothing special

Also, what is exactly the 15Khz refereeing too? Horizontal refresh?.

Yes thats the Hoz refresh.

Well if thats going to be a big thing i guess i could always cheat and use this KVM switch i found in my bag of goodies...... :S
Its triggered by a momentary dpst switch.

However i am unsure as to how we would make this work.

However i am unsure how we would make this work in the circut.
 
Hrm that's odd, atm i am using a VGA male connector with standard wire soldered to the monitor nothing special

That is odd, but then again, so is 15khz horizontal refresh rate for XGA. I suspect the refresh rate translates to lower bandwidth.

How short are your wires though? Using standard non-RFI shielded wire will get you into trouble to be sure, but is not as noticeable if it's really short lengths. Otherwise why do you think they make VGA cables by shielding the RGB lines in the cable, then shielding the cable as a bundle once again? They don't do it because they want to, it is very much a necessity.

Case in point, when I was quite a bit younger I tried to go from my Genesis to my TV with plain wire of about 10 inches because I din't have the proprietary cable made for the job. It didn't work at all, I couldn't see crap, and that was just plain composite video signal. Multiply the cabling requirements by about 50, and that is close to where XGA stands.

Well if thats going to be a big thing i guess i could always cheat and use this KVM switch i found in my bag of goodies...... :S
Its triggered by a momentary dpst switch.

You have a KVM that just uses a simple switch??? Because that would be far better than doing all the rest of the crap above honestly. We could just throw a relay in it, then it's a short trip from there to do the NR-monostable thing. I looked around and found nothing of the sort under $100. What are the specs exactly? Can it handle XGA, or is it only rated for SVGA? Most of those things are only good up to SVGA, unless they are premium grade equipment. Maybe crack it open and picture us some images of the circuit board, that is, if you don't mind killing seals.
 
Hey There,

Sure can here is the inside

**broken link removed**

Ritmo 2 ports KVM Switch Box
This 2 ports Manual KVM SWITCH is the most econimical solution for controlling 2 PC's. This powerful tool perfect for desktop placement and won't take up critical workspace.

Feature:
* 1920 x 1440 Max Resolution
* 250Mhz Video Bandwidth
* 30-100Khz Horitzontal Frequency
* 43-120Hz Vertical Frequency


So i think this means its not usable in this application because it doesnt support 15khz ? (guess i can always hook it up and press the button and see what happens !

Also forgot to mention all my cables insulated and run through Ferrite Core.
However the plan with using this was to put some panel mount VGA connectors so i can use the correct cables in and out of the monitor!
 
Hey Little bit of Math for everyone :)

Specs for the Monitor
https://www.electro-tech-online.com/custompdfs/2013/01/wg-k7000.pdf

Video Bandwidth 12MHz typical
Horizontal Scan 15.1kHz to 16.8kHz
Vertical Scan 47 Hz to 63Hz

400Pixels x 240lines




modeline '240x240@58,795' 4,83 240 252 276 310 240 243 246 265 -hsync -vsync
modeline '256x240@60,436' 5,3 256 272 296 336 240 244 247 261 -hsync -vsync
modeline '256x256@59,496' 5,36 256 268 292 330 256 257 260 273 -hsync -vsync
modeline '256x264@58,317' 5,35 256 268 292 330 264 265 268 278 -hsync -vsync
modeline '288x240@59,885' 5,84 288 296 328 368 240 243 246 265 -hsync -vsync
modeline '296x240@59,941' 5,95 296 304 336 376 240 243 246 264 -hsync -vsync
modeline '304x240@59,305' 6,2 304 320 352 396 240 243 246 264 -hsync -vsync
modeline '320x200@59,535' 6,2 320 336 368 410 200 216 219 254 -hsync -vsync
modeline '321x240@59,014' 6,45 321 336 368 414 240 242 245 264 -hsync -vsync
modeline '320x256@59,917' 6,68 320 340 372 416 256 257 260 268 -hsync -vsync
modeline '336x240@59,749' 6,83 336 352 384 433 240 243 246 264 -hsync -vsync
modeline '352x256@59,697' 7,28 352 368 400 450 256 257 260 271 -hsync -vsync
modeline '352x264@57,257' 7,35 352 365 405 452 264 265 268 284 -hsync -vsync
modeline '352x288@51,116' 7,4 352 368 408 464 288 289 292 312 -hsync -vsync
modeline '368x240@59,196' 7,47 368 384 424 478 240 243 246 264 -hsync -vsync
modeline '384x288@51,219' 7,85 384 400 440 496 288 289 292 309 -hsync -vsync
modeline '392x240@59,898' 8 392 408 448 504 240 243 246 265 -hsync -vsync
modeline '400x256@52,419' 8,08 400 416 456 519 256 268 271 297 -hsync -vsync
modeline '448x240@60,01' 9,16 448 464 512 576 240 243 246 265 -hsync -vsync
modeline '512x240@59,973' 10,68 512 544 600 672 240 243 246 265 -hsync -vsync
modeline '512x288@50,939' 10,68 512 544 600 672 288 289 292 312 -hsync -vsync
modeline '632x264@56,751' 13 632 664 728 824 264 265 268 278 -hsync -vsync
modeline '640x240@59,96' 13,22 640 672 736 832 240 243 246 265 -hsync -vsync
modeline '640x288@50,955' 13,1 640 672 736 832 288 289 292 309 -hsync -vsync
modeline "240x192@60" 10.06 480 512 576 640 192 216 219 262 -hsync -vsync
modeline "240x252@60" 10.08 480 498 560 640 252 253 267 278 -hsync -vsync
modeline "256x192@60" 10.74 512 542 592 684 192 216 219 262 -hsync -vsync
modeline "256x224@60" 10.6 512 548 596 688 224 234 238 263 -hsync -vsync
modeline "256x240@60" 10.6 512 548 596 688 240 241 242 262 -hsync -vsync
modeline "256x248@60" 10.80 512 528 608 688 248 250 258 262 -hsync -vsync
modeline "256x256@60" 10.84 512 546 594 688 256 258 262 263 -hsync -vsync
modeline "280x240@60" 11.82 560 576 672 752 240 241 244 262 -hsync -vsync
modeline "288x224@60" 12.02 576 606 670 768 224 231 234 258 -hsync -vsync
modeline "294x238@60" 12.34 588 614 684 784 238 239 240 262 -hsync -vsync
modeline "304x224@60" 12.72 608 642 700 810 224 233 236 262 -hsync -vsync
modeline "320x224@60" 13.32 640 673 737 848 224 233 236 262 -hsync -vsync
modeline "320x240@60" 13.32 640 674 736 848 240 241 242 262 -hsync -vsync
modeline "336x240@60" 13.66 672 697 761 866 240 242 245 264 -hsync -vsync
modeline "338x240@60" 13.66 676 704 768 866 240 243 246 264 -hsync -vsync
modeline "384x224@60" 8.04 384 405 442 511 224 232 235 262 -hsync -vsync
modeline "384x240@60" 8.04 384 408 440 512 240 241 251 262 -hsync -vsync
modeline "384x256@60" 8.04 384 409 441 512 256 257 258 262 -hsync -vsync
modeline "392x240@60" 8 392 408 448 504 240 243 246 265 -hsync -vsync
modeline "400x256@60" 8.42 400 424 464 536 256 257 260 262 -hsync -vsync
modeline "448x224@60" 9.43 448 468 524 600 224 233 236 262 -hsync -vsync
modeline "640x480@60" 13.15 640 671 729 836 480 483 489 525 interlace -hsync -vsync
modeLine "768x224@60" 16.632 768 825 897 1056 224 232 235 260 -HSync -VSync

Other res are 800x600@60 & 1024x768@60

bandwidth = 1.05Y*1.3X*R (not exact)
That taken into account
Video Bandwidth Range - 4.7mHz - 25.1Mhz Not often Used 39.312 mHz


I Will test the KVM with the monitor tonight to see if it can pass the low resolution
 
I see what they did, they just used surface mount FETs as generic switch elements, I wonder if they even bothered to do break before make? One thing to note is they intended that KVM box to pass two video output devices to one display, so it may not work the other way around. If they didn't put the FETs back to back (like a typical analog switch) then the control of signal is very much one directional.

The only way you will know if any of this stuff could be a problem is if you just try it and see. Low chances of breaking anything, so go nuts. (not my fault if anything does break)

(Edit: Actually, Judging from C,B,E, in the silk screen, they are probably actually BJTs, not FETs.)

So i think this means its not usable in this application because it doesnt support 15khz ? (guess i can always hook it up and press the button and see what happens !

Most of the time, these kinds of things have upper limits, but no real lower limit. So it should work fine as long as the box it's self just passes the H-sync and doesn't try to recreate or process it in any way.


Edit2: You know what, is there anything on the bottom of the board near the video ports? I just realized that VGA signals are + and - swing, I wouldn't think that BJTs can pass that kind of a signal. I was expecting them to be FETs, but that has the opposite problem, not being able to block that kind of signal. Something I'm not getting here...
 
Last edited:
OK, I did some screwing around, tried to make a 1-of-2 signal selector using just BJTs, to more or less recreate the KVM. *CLICK THIS* for the simulation.

It doesn't work very well, murders your video signal, massive distortion. I don't see it as a workable KVM muxing/demuxing circuit, but you get the idea. Here is a screen shot right after a switch over. The scope view is what is on video out. The waveform on the left half is with "Video 1 In" selected, and the waveform on the right is "Video 2 In" selected...

1-of-2 selector.png


I might be able to do better if I knew the exact circuit/schematic for the video section of the KVM. But the only other way I think they could do it is if they used the BJTs as shunts, to short the signal out rather than pass it through. But that is even worse than the above method I would think.

Edit: I suppose they could also be using them as amps, and just removing the bias or something to redirect the signal.
 
Wow nice work mate, thanks heaps for your dedication to the project.
Ill borrow a friends camera a hidef one and ill take some decent pictures for you. can e-mail them if you like.

Well i can report that that KVM is eith S&*(#T or plain FU^*()*)ED

I believe its the khz barrier. i plugged it in normally (the correct way around to make sure it worked)
and i got a massive hi pitched scream, while no load was on second the computer fired the black was grey and the colors were okay but it did not like it at all!!!.

I believe i have a circuit board kit laying around somewhere i can hand draw it and develop it in the solution if we get somthing working well in a bread board.

I spoke to a gent today who said i can draw it on a computer (better not be paint!) and send it to a company and they can make it for about $15


What did you use for that simulation i could have drawn my circuit on that!
 
Well i can report that that KVM is eith S&*(#T or plain FU^*()*)ED
...
I believe its the khz barrier.

It could be the Khz. Most of the time with amplifier circuits you use capacitors as the input/output, this would block the signal the closer you get to DC causing distortion, and even total loss of signal. Even so, 15Khz horizontal refresh is any thing but DC and is plenty low enough to not hit the upper limit. So I wouldn't count on the problem being slightly out of spec signal. I think the problem it's more likely to be that they used BJTs, in either a select-able unity gain amplifier configuration, or as pass/block elements. Either way, such circuits are one directional. So if you were trying to make one video out device go to two displays... that would be the real problem.

Like I said, if I had the exact schematic, I would know for sure what they did to switch the signal. I'm 80-90% positive the business end of the channel selection circuit is all the transistors near the VGA ports, and probably some stuff on the back side of the board. Their arrangement is a dead giveaway. Probably crude common emitter amplifiers, either unity gain or a very small gain to keep nonlinear amplification noise to a minimum.

EDIT: Probably *NOT* common emitter amp, I got mixed up. More likely common COLLECTOR follower if anything. Could also be common base amp.

I spoke to a gent today who said i can draw it on a computer (better not be paint!) and send it to a company and they can make it for about $15

They probably want gerber files, it's the industry standard as far as I know. That means PCB CAD software like Eagle or such.

I can't remember if Eagle it's self is freeware or not, may have to ask someone on the forums if they have a copy and would be willing to make up the PCB for you. We would need to design a circuit first though.

There is also PCB123, which I'm fairly sure is freeware. I don't know if it outputs gerber files. I know you can order boards right out of the software though. It also has a built in 3D rendering to show you about what your finished PCB will look like.

I believe i have a circuit board kit laying around somewhere i can hand draw it and develop it in the solution if we get something working well in a bread board.

Might be able to etch a board, you would have to have a REALLY steady hand, and it really should be a double sided board. One side for the parts, one for a solid ground plane.
 
Last edited:
OK, I made a circuit that I think is possibly workable. *HERE* is a simulation of it. And here is an image...

common collector based 1-of-2 selector.png


I used common collector buffers that can be enabled or disabled by switching the power to them on or off, and thus, turn on and off that particular buffers output. Then I made inverter channel selection logic, some crude RC-diode based "break-before-make" logic, followed by some Schmitt trigger buffers. The biggest problem is that the channel that is supposed to be off has ~150mV of bleed through. I am currently thinking about ways to correct for this, but it is a minor problem when compared to the distortion that would be caused by other methods. Chances are if your display does not see a valid signal it will simply display black or something, but I can't be sure.

Finally, two very critical points.

(1) First, this is a very simple simulation, very rough draft. I have not validated it in a more accurate simulator. It's possible that a real simulation will expose bugs that indicate the whole design is impossible or highly impractical.

(2) And second, even if it does end up working well in a better, more accurate simulator, in the end it will all come down to whether or not we can use single/double layer PCB designing, or if it has to be some kind of super happy special fantastic PCB.

So there is no telling at this time if this circuit is truly workable for doing what you need or not.
 
It may be that the KVM switches between signals by simply shunting the inactive one to ground, like this?
VideoSplitter.gif
 
Hi Alec_t. I was hoping someone that knew what they were doing would jump in here and tell me how it is.

It may be that the KVM switches between signals by simply shunting the inactive one to ground, like this?

I had briefly considered a shunt topology, not long enough to try and figure exactly how though. I erroneously rejected the idea because it "felt icky". That was probably a mistake considering how well your way works, and that it uses about the right amount of parts. Given no better options, I would think that you more than likely have the correct answer to how his KVM works.

Is the output signal properly centered around ground? Looks like it is. My research suggests that this is a pretty critical parameter. The majority of the signal is positive, but special parts of the signal dip just below ground. I could see problems with this if the signal had even a small offset, but was otherwise reconstructed perfectly. But I could be wrong.

What do you think about layout requirements? Do you think either method can be done with just his PCB kit? Or do you think that this is going to have to be high end radio board design? I personally think he could possibly pull it off with the kit, *IF* all the connections are short and he uses the back side as a ground plane. But the signal he is working with is almost certainly in the low VHF range. So it probably needs to be treated fairly well.
 
Is the output signal properly centered around ground?
In the sim a sine input gives an output centred around ground (within 8mV). The asc file is attached if you want to play.
Don't know what the PCB kit is or the OP's abilities are. Layout would clearly matter at video frequencies. I've little experience there.
 

Attachments

  • VideoSplitter.asc
    2.8 KB · Views: 110
Last edited:
Hey All sorry for my late reply!!!

Was auditing time at my work, and well that stuff is just no fun at all!.

I did some reading on the above circuit and the amount the mode lines switch it may cause distortion and hr parts would be rather costly at the end of the project.

However i struck gold last night, my work was clearing out a heap of KVM switches, i tryed 20 different ones all distorted the picture and caused alot of noise on the line.
(scrambled picture)

However one of the KVMS worked perfectly! a kvm0201 Level one.

The only issue with the KVM is it requires you to send keystrokes to it to switch monitors.

So i was thinking about programming a Pic to send some commands to it via PS2.

Left ctrl Left ctrl 1 (selects Input 1)
Left ctrl Left ctrl 2 (selects Input 2)

I have not tested this yet but i believe it will default back to input 1 when loss of power, so i guess i can use a relay to trigger the pic to send Left ctrl Left ctrl 2.

Unless there is away to make a serial device to do this for me?
Any thoughts guys?
 
Last edited:
Was auditing time at my work, and well that ---- is just no fun at all!.

S-word is bad bad news, you will get moderated for sure. Doesn't offend me at all, though I still have to flag it. Kids come here from time to time, so it makes sense that we got to keep that off the site. Not personal or anything.

i tryed 20 different [KVMs and they] all distorted the picture and caused alot of noise on the line.
(scrambled picture)

Not too surprising, most KVMs are made to go one way. If I remember right you are actually going the other way. It's like paddling up river, so yeah, you could expect to get distortion.

So i was thinking about programming a Pic to send some commands to it via PS2.

Yep, should be easy as pie. Though I have never personally dealt with PS2, I understand it's not much different from serial port.

* Googles it... *

Nope, not really much different. Here you go.... silver platter.
**broken link removed**
 
Thanks for that has been edited slip of the keys there.
The KVM is going to normal way not backwards. 1 monitor to two displays the only difference is instead of being separate computers its just the one PC.

I have been looking all night no luck yet. Might have to go hire a book from the library and read up on pic programing then pull apart a keyboard and see how it all works.
 
So... when you say "monitor"... You ARE talking about a video OUTPUT device, right? Like, the graphics card? Because in the world I live in, the world of the personal computer, monitor = display, not the video output. However, I can't imagine you are attaching a display to a KVM, then to two more displays. That would be quite an absurd and pointless configuration. So I assumed for your project, monitor = video output.

Maybe I am wrong, but I made the assumption that you have one display for the user, and one as a service display. With one device that generates the picture? Or is it that the system just has two displays running concurrently, and the normal use of the system automagically switches between them?

Normally, a KVM is used when you have several towers/computers, and you want to access them all from one set of peripherals. If I am correct in my understanding, you are actually doing the reverse of this. You are using the equivalent of one tower, but piping the output to two screens. The problem with this is that the flow of information is going the wrong way. UNLESS of course your KVM has TWO monitor outputs. BUT, I just looked yours up, and it doesn't.

If my above conjecture is correct, then the fact that it works is very interesting. That would make it bi-directional.
 
Last edited:
The Arcade Machine only has one monitor "Display"

The device that controls it computer has to Video Outs (2 different Graphics Cards) One is a native 15khz (Video out 1)the other is a software 15khz with hacked drivers that display crazy mode-lines(Video Out 2). These modelings are software controlled and requires windows to be loaded.

The KVM is used to switch between Outputs.

Output 1 is Native 15khz used for boot of the pc to access bios etc etc if anything goes wrong configuration etc etc.
Once windows is loaded Output 1 is disabled.
(i need to send keystrokes somehow to KVM to switch to Video out 2)

Just need to work out how i can send these commands!
Sorry i for my bad explanation1
 
AH... I see now. I was confused because the words "monitor" and "display" are synonyms for most of us.

With the KVM running the normal way, it's not much of a surprise any more. Though it is some what a surprise that you had to go through 20 to find one that worked. Must be some freaky weird signals or something. Oh well, if it works now then it works.

PS2.
The commands will be easy. You can probably just use a USART peripheral in a PIC if you want to. Or if your windows machine has a serial port, you might be able to hack it out of that with something like a dropping resistor setup to get away from RS232 voltage levels. Otherwise, you can use a parallel port. Lots of tutorials out for manipulating parallel port pins via windows software.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top