Universal Interface Converter

Status
Not open for further replies.

chintan_jagani

New Member
I plan to build a universal interface converter. This project will consist of making a kit that would have various physical interfaces related to computer hardware, such as RS-232, RS-485, USB, firewire etc., at its input and output. At a time we could select any one of the interface combinations e.g. RS-232<=>USB, RS-232<=>RS-485 and the kit would control proper signal conversions. Would anybody help me in realising this project. How can I minimize the hardware requirements?

I am still a student and in initial phase of project analysis correct me if I go wrong.
 
I suspect your project is MUCH more complicated than you think!, you are likely to require custom IC's to make it possible!. You have a bewildering range of fast and slow, and simple and complicated - this isn't at all trivial!.
 
You're going to run into serious problems unless you can constrain the scope of your project. The main one being that these interfaces just *arent* compatible in software. Just a rough run through of some interfaces and some rough numbers. Simple electrical conversions aren't enough...

RS232: 150baud - 921KBaud, +-12V signaling, a half dozen or so control lines. point-to-point.
RS422/RS485: ~1Kbps - 50Mbps, *No predefined bit format*, +-3V 2 wire differential serial signal, can have multiple devices on the same wire.
USB: 1.5Mbps/sec - 480Mbps/sec, large amount of firmware and software required, implementation differs on whether this is supposed to be a host or device connection.
IEEE1394: 100Mbps - 800Mbps, ditto on the firmware, all devices are "peers" on the network

The other external (digital) interfaces I can think of:
PS2: keyboard/mouse protocols, 2 wire 5V synchronous bidirectional -hey you can use that mostly useless USRT here
IEEE1284, Parallel port: ~17 data and control lines, can do a couple MBytes/sec
SPDIF: ~3Mbps, not much protocol, .6Vpp

Then there's the really difficult ones:
Ethernet - 10MBit - 1000MBit (actual wire frequencies vary - 12.5MHz - 250MHz multilevel signals)
VGA - 100+MHz analog * 3 + I2C
DVI - 3.3Gbps (550MHz signals) + I2C
 
Son,
Minimize the hardware requirements. That's not remotely possible. It's a grand idea, but nobody would even remotely consider paying real money for this capability. Point solutions to interface problems RULE the day. I'm sorry but that's just the way it is.
 
What kind of minimizing is desired? Do you want to minimize for combining all of the possible conversions? Or do you want to minimize per conversion pair?

Will this kit be modular? For example, you could have a basic board with programmable components (microcontroller + FPGA), with interface modules for the different PHY. You will need to manage multiple modules with this approach. Also, there are potential signal integrity issues with the higher speed interfaces.

If you want this universal adapter to handle all possible conversions, WITHOUT REPROGRAMMING, then the project is too large. If you allow reprogramming, then you will need to manage the collection of conversion firmware modules.

Another issue is how to convert between vastly different speeds. The USB solution is to make the high speed connection wait. But what do you do if the slow connection does not have flow control? Will you buffer it (require external RAM)? How much data loss can you tolerate?
 
Last edited:
do it and you will have customers. i would be interested to buy such device myself. currently my laptop bag is about to explode again (biggest one Dell has and the sewed edges and zippers are not going to hold for long) because it filled with number of different cables, adapters and converters and i don't want extra bags. about the task - this is not impossible but it is far from trivial. if i had to do it, i would probably start looking at SBCs or better start with something like PC104, make it work and later figure out how to shrink it.

different interfaces have different bandwith so buffering is important (need memory). basic converting just from RS232 to RS485 or 422 is simple and need no brain (or memory) - IF the speed is the same and RS232 is not using handshaking. we use this all the time to extend RS232 cable beyond RS232 spec (pair of 485 converters back to back and you can go up to 4km for example). Parallel port will be more involved, ethernet and USB much more. using bare bone PC such as PC104 and maybe linux will cover most aspects of the project, one will need to focus mainly on programming. it would be really nice to provide optical isolation but so far even RS232/232 isolators I've seen only isolate half of the signals (4 out of 8, handshake is normally left out). this is one of the big annoyances (others are rather slow speed and only point to point connection) so I'm quite happy with ethernet. USB is ok, doesn't impress me much, because it has so many limitations. it's handy for local devices only where having no user assigned address is not disadvantage and where cable can be short. but this is what it was designed for anyway...

btw. there was an IBM commercial about such device (universal converter) which will work anywhere but in Europe (to be used in Europe it would need - a converter. so much of "universalness"...). this of course was just a joke but lot's of people called IBM asking to buy such product. after while they announced that due such high demand they started considering making it for real but eventualy gave up... now you are the new hope, young jedi...may force be with you...
 
Thank you all of u for replying.

I will consider ur suggestion and build only much useful combinations as I have to complete my project in 4 to 5 months. Are the serial adapters so difficult to build or they are something different? If at one end if I keep fixed terminal as computer than would not I get software drivers for eg. for emulation process of USB, matching of speeds,etc.. I got to see a product on net which had a DB-25 or some other port at one end and had other ports such as USB, PS/2, etc. ports at the other end.
 
Status
Not open for further replies.
Cookies are required to use this site. You must accept them to continue using the site. Learn more…