Maximum PC

Just wanted to add some things...

It seems to be working fine with 1024x768 resolution. Anything higher, and I see "speckles" in video playback. Much higher, and it just doesn't show anything on the monitor.

You might try increasing the VDD voltage from 6 to 12 volts. That should reduce the ON resistance of the 4066. Not sure if that is a problem or not. Also I am not sure of the bandwidth requirements for DVI.

Ron

And that might be something to try later... but this problem I'm having doesn't even involve the 4066's. You can see from the pic I provided that by just sticking a few solder joints in the middle of a DVI cable, the signal is extremely degraded.

My bad. I will venture a guess though. Looking at the pinout of a DVI cable they seem to pay special attention to shielding the TDMS Data (The Colors). I don't have an extra DVI cable lying around to hack up handy but I wonder if the lack of shielding is an issue? That may make some sense as the higher the resolution settings the worse the signals seem to degrade. Now as to a workaround I am not sure what to suggest.

I have a pile of DVI cables at work. Monday if I have time I could hack one apart and see how they are shielded by pairs in the cable.

Ron

I wonder if its the big breadboard itself (the white one in the middle of the pic, not the soldered ones off to the side) that causing the problems? might be a lot of capacitive effects from the way its made? all those spring clip channels and such laid out side by side..

as an experiment, can you bypass the main breadboard and just wirenut the input and output wires together and see if the interference goes away (or reduces)?

I'll try that, thanks.

Ron -
Wikipedia has a great color-coded dvi pinout. There are four pins that are solely devoted to shielding, but I'm a bit of a noob on how exactly shielding works. If you do get around to hacking open a DVI cable (I'd do it myself, but I need all the ones I have) I'd be interested to know how those shielding wires are integrated into the mix.

I think CharBroiled is into something:



My guess is that overall maybe a lack of shielding and the result of capacitance are hampering the high frequency response of the configuration.

Something you can try is a resolution that is partially messed up. Then gently touch and move different conductors and see the result on the monitor.

Oh yeah, the quote is from here:

http://www.audiolofftreport.com/dvi_hdmi.html

I will cut up a DVI cable to see how it is made internally.

Ron

Definitely, I'll get to eliminating the breadboard today.

Just wanted to give a little update... yesterday I posted the dvi problem on an electronics forum and have been getting some good responses there too. They suggested the same thing as char as far as the breadboard.

http://www.electro-tech-online.com/show ... post656507

Ok, I've done some thinking, and here's what I have:

Obviously, what I have going on so far isn't going to work unless I add either a buttload of shielding or some kind of miracle occurs. My hand-soldered dvi connectors (complete with fragile-as-heck pins) are falling apart. And those stupid radioshit solder boards aren't faring much better.

What if I were to just buy a bunch of dvi cables that had a female connector on one end and chop it in half? I'd be left with a bunch of wires that were not only shielded, but soundly interfaced with the connector.

Now, take those wires and stick them into something like this:
http://datasheets.maxim-ic.com/en/ds/MAX4886.pdf
and for the low-frequency data:
http://datasheets.maxim-ic.com/en/ds/MAX4929E.pdf

Will that work? The only thing I'm unsure about is what I will need to connect the dvi cables wire's to these chips as well as what all that logic stuff means.

That would likely work very well but there is a problem:



Problem here is the package (TQFN). You would have to design a board to mount the chip(s) on. These are not like the DIP package you see on the chips you have. They have no pins sticking out.

The option with a Maxim / Dallas chip like that is to use their Eval kit:

http://datasheets.maxim-ic.com/en/ds/MAX4886EVKIT.pdf

The kit supports both the chips but carries a $75 price tag.

On another note I did drag home a pair of DVI cables today. I have not cut one open yet. Should you need them to experiment let me know and they are yours.

Ron

Thanks for the reply!

I requested samples of those two chips, so I should be able to take a first-hand look at them and macgyver some type of hookups.

I can't ask you to send me dvi cables, you've been too kind already. I don't need any more of the male ones, and I'll be cutting the female ones open anyway if I go that route.

Until I receive my sample chips I'll begin work on the usb switch. I've already done something crude with relays that worked, so I don't expect problems. Heck, 4 pins... that's a piece of cake...

Haha, I had a nice laugh when I saw how small those sample chips were!

That is why I mentioned the kit. LOL Sure not like the soldering of years ago huh?

Ron

Okay, so basically I've conceded that at this point I really don't have the necessary tools to construct a proper DVI switch, which would undoubtedly require the ability to construct and drill printed circuit boards.

Instead, as mentioned, I've decided to press on with the USB switch. I worked on a prototype today, and it seems to work rather well. What is pictured below will switch a single USB device between two computers. It only requires one 4066 because the power will be constantly connected, thus leaving only two data wires to be switched between computers (hella lot easier than 18, huh?).

I had a lot of fun testing this--that is, once I got it working. Turned out that a 6v supply voltage was causing too much of what I guess is termed "on resistance". I tested across closed switch pins and measured ~160 ohms. That was causing trouble for the USB data. I tripled the supply voltage to 18v (still 4 less than the 4066's limit) and that resistance went down to ~50 ohms. It's working like a charm.

What's left is to lay this out a bit neater and then solder it. There will be three of these in total. I will also throw in a 5.1 sound switch and whatever else floats my boat. I can run a network cable (which obviously contains 8 usable wires) to each computer to easily control power, reset, lights, etc. This, I guess you'd call it a VM switch, is still extremely useful given you do have two monitors but would like to use only one mouse and keyboard.





Also, I have some pics of the agonizing process of soldering leads to the DVI connectors. They're so small, and so fragile, it's not even funny.

The point of this entire exercise is you actually came away with a good bit learned. Not to mention drifting right into a new project.

Unlike years ago today's homebrew experimenter faces obstacles like the tiny components and mounting things.

Anyway, you gave this one hell of a good shot!

Ron

I wouldn't give up on it just yet... if you can get something fabbed up, try this out?

http://www.pad2pad.com/

You get free software that you can download and lay out your board and then (for a fee) they will make it for you. If it is small, I think the rates are pretty fair and would be interesting if you were able to put it all together. Depending on where this goes, I would buy a board or two (the more you buy, the lower the per unit cost is).

Just a thought.

Kudos though, this is some pretty sweet stuff.

Hrm. I just looked at their prices, and they are higher than I remember...

::shrug::

Oh well.

Have you thought about getting an etching kit? I have one around somewhere...

Oh, without a doubt I have. The only trouble is that I don't have a good way to drill the holes, and with ones that small I think I'd need a drill press. Also, I don't have easy access to a laser printer.

Wow. This is a really nice project. Can you please post schematics for the USB switches. Thanks.



Try asking a local school with a tech shop or a motor garage if they have a drill press you can borrow. For the laser printer you could try a cyber cafe, a library or maybe some local businusses. They shouldn't mind.



Try these smt tutorials. They should help a lot with small chips and any other projects you may have.

http://www.kevinro.com/newdocs/learningcenter/surface.htm
http://www.ladyada.net/make/x0xb0x/fab/ioboard/index.html

There are a couple other ways to do it. You might not need the laser printer. The drill press could be a big issue.