100base-Tx Ethernet to TTL????

[()blivion]

Hello, I'm looking for ideas on what the easiest way would be for a hobbyist to bridge your basic home networking "100base-Tx Ethernet" voltage level system to a TTL voltage level system WITH OUT A TRANSLATION IC!? (IE, building it from discreet components)The idea being for use with any generic system, but emphasis would be on use with microcontrollers or DSP's that don't have 100base-Tx support built in.

Before you answer, take some notes with you...

1) Yes, I know that it would be plenty cheep and definitely easier to just use some form of translation IC. For the sake of argument, pretend they never existed.

2) Yes, I am aware that there exists MCU's and DSP's with built in support for 100base-Tx with few external parts. Pretend they don't exist either.

3) I'm not trying at this time to deal with anything from higher levels of the OSI model, I'm just trying to work with layer 1. Particularly the modulation aspect.

4) I would consider 10base-T if it proves impossible to do 100base-TX with my criterion.

Any CONSTRUCTIVE thoughts on this?

 

[3v0]

Start with the transformer from an old ethernet card. The input to the transform is a differential pair so the output should be a lower voltage differential pair. You can check that with a scope.

To convert the differential voltage to TTL use a RS485 converter or build one.

From there you can play with the modulation.

But seriously why would you want to do this.

HTH

 

[RichTheDude]

If I assumed that interface IC's did not exist and DSPs/microcontrollers did not have inbuilt Ethernet PHY's I would start to look at the 7 Layer OSI model (you have already stated your interested in level 1 - which to be honest is not much use in terms of getting data from A to B).

To get data from A to B you are least going to have decipher the 14 byte MAC header, to determine if the destination MAC address is the devices MAC address. Firstly this is going to require decoding the signal presented on the media (look at https://en.wikipedia.org/wiki/Fast_Ethernet). You are then going to have some kind of state machine to detect packet headers, verify the destination MAC is for you, and then present this data to some kind of external bus (be it a fifo or a memory mapped external bus). This would be a very crude system (no CRC, no upper layer protocols such as TCP to ensure data actually got there in the correct order, it would be liable to packet loss!).

This would be an undertaking to do in an FPGA yet alone discrete components, most Ethernet PHY's also run at an 125 MHz clock - so you are outside of the realm of 74 series logic.

Decoding the bytes may prove a fun educational experience, but honestly even FPGA designs have either dedicated silicon on-board to do Ethernet (up to a byte level [PHY & MAC]) or a dedicated IC external to the FPGA. Designing such a device from scratch is a very large project, and not something I would personally even consider.

Sorry to be a bit negative, but I think being honest is always the best.

 

[()blivion]

Thank you 3v0/HTH for your input.

I figured I would need to use a transformer for the first stage, Making something like the RS485 converter is more than likely the information I'm after. What I'm most interested in figuring out is how to convert from the three voltage levels @ 25Mhz back to your familiar 5v or 3v binary TTL, with emphasis on generic uses. It would be ok if it ended up being 125Mhz TTL, but I would prefer to do a parallel 25Mhz bus to drop the speed requirements in exchange for using more pins/parts. This would put 74 series into play as well as low tolerance discreet parts.

"RichTheDude", I think you misunderstand the nature of this discussion. I hear what your saying though, and if I was making a router or a network interface card that information would be relevant. I am however only interested in converting from the UTP medium to another medium at this point. The system I am building is simply a building block in a larger system and not the final product. I will likely have a later stage that will handle the higher level stuff should I need it. So worrying about how to handle addressing and any other such data processing is pointless at this time.

As for what I'm doing, I have a few projects in mind. One is a LAN cable to IR LASER to LAN project. So I can hook two PC's or a router and PC up line of sight, without the need for wires. And as stated, this would not need any packet decoding or processing. It would simply take the LAN Tx twisted pair signal, convert it to LASER, receive it on the other end, convert it back to LAN Tx, and send it down the other twisted pair. And since it's full-duplex it would have two of these, one for each pair of wires. It should be clear from reading this that there isn't a need for MAC decoding or anything like it. What the wire pair sees is what the other end gets, plain and simple. And I am aware that it could be done differently.

(Inb4, try modulating the LAN voltages directly.)

Any more thoughts?

 

[3v0]

Can we ask why you want to do this. There may be an easier way to get the result you want.

 

[()blivion]

@ post #5
I'm not sure of your question...

1) Why would I want to make this converter my self, and not buy chips/MCU's with support? That's answered for you in post #4 and in the following..

2) Why would I want to make a RJ45-LASER-RJ45 converter? To extend a network of course. LASER is ideal for what I want to do. Say... to get across a public street where running wires would be illegal without permits and such. Or in a situation where WiFi won't cover the range and speed required, but I have a usable line of sight for LASER. What ever really... you decide. I'm going to bed.

See you tomorrow.

 

[3v0]

In post #4 you indicated what but not why.

Wifi should have enough speed and distance for personal use esp with a cantenna or similar setup. My shop is on the other side of the street from my house and a plain jane wireless router works fine.

 

[KeepItSimpleStupid]

Just change the antenna system for the wireless router. Make both antenna's directional. Try the Cantena. It requires a not much more than a coffee can and a connector.

 

[()blivion]

I've built many a cantenna, Bi-quad, helical, Yagi, and parabolic antenna in my time every one. I'm no stranger to EE, just a low post count on these forums mostly do to a lack of needing help. And despite the fact that people have made WiFi connections over 200 Miles, unless your SNR is > 20 your not really getting the max 54mbps of the standard. Radio is also subject to RFI where as LASER is not. Also, (not that it's a major problem) but... even with the 256bit AES of WEP2 WiFi is still totally hackable. However a well focused LASER that's custom modulated (and possibly 256bit AES encrypted) is much harder. I could go on and on since I have been thinking about this particular project for 2+ years.

At the end of the day since this is just one project I have in mind for the underlying circuit. Doing it a easier way that does not involve this circuit would undercut the other projects. I would prefer to open doors, not close them. Plus doing this the way I want, is as much about doing something new and different as it is about achieving the goal I describe. So trying to find a better way, one that's been done before is the opposite of what I want. I'm just looking for tips and opinions and possibly schematics on how YOU would do this with the restrictions above. So, for the sake of this discussion, just pretend that there is no better way. Because something that has been done many times before is no fun.

But if you can't or don't want to do this, then I can understand. Not every one is going to get it like I do.

Anyway, back on topic... Here is a link to a schematic that is more or less along the lines of what I had in mind, But it's got more fundamental problems than I care to admit. Makes it seem like the project is doable and not doable all in the same picture. But it's a start.

http://iq-technologies.net/projects/pc/023/

 

[KeepItSimpleStupid]

You might want to read this article: https://www.electro-tech-online.com/custompdfs/2012/01/Wireless_Transmission_of_Fast_Ethernet.pdf

 

[()blivion]

Sweet PDF bro, thanks for being useful. I thought of everything in there already though, other than bending of light through different air density's. Good thing I don't have to go < 10 kilometers. To bad that even if I could find a source to sell one of those systems to me it would still cost to much. Otherwise I would go find one right now. Looking to do this on the cheep, with common salvage if at all possible because that's how I roll.

"we designed the TerraLink transceivers to be eye-safe at the transmit aperture"

Yeah...... Ummmmm.... I don't know if I'm going to care about that XD .... I have a 200mw LASER from a printer that I might try, but I'm going to not focus the crap out of it, should make it easier to aim. I live in the woods BTW, I don't think the deer and coyotes are going to mind. Now if I only knew where that CO2 IR laser tube went off to.... I think the pot heads might have tried to make a bong with it (-_-) LOL

Back on topic though, If you or some one you know actually owns one of the systems in that PDF, try to get a schematic or something. I'm sure it uses some form of DPS, MCU, or PLD to do all the work. But I'm also sure that the meat and potatoes of something like this can be done with less.

 

[3v0]

I know ghostman11 used to have an optical link between buildings on his farm. He may have some advice. But he has not been on since mid December.

 

[RichTheDude]

()blivion, I have worked on a few laser based system now - my suggestions are:

1. Develop a circuit which can through from Ethernet to a serial stream of TTL and vice versa (this is the hard bit to do with such restrictions). Until you have this working, it does not make sense to put any effort into other parts.

I severely doubt the circuit you have posted btw. I have never seen a 400 MHz quartz crystal (RF gurus please correct me if am wrong!, plus I thought filtering of the correct overtone would be required...). I have seen oscillators of various forms to well over 1 GHz, but not crystals. Also I have never had much look pushing any 74 series above about 80 MHz. It smells very fishy to me.

2. Make your laser diode and receiver combo. To do this you will need a logic controlled laser driver, a receiver (most likely a Si PIN photodiode at these speeds, phototransistors will definitely be far too slow). There are laser driving IC's out there (hint hint https://www.ichaus.de/iC-HK, iC-HKB) , and at these speeds turning the laser OFF to ON will most likely be too slow. Most communication systems intensity modulate a laser diode, i.e. logic 0 is a current just above the lasing threshold and a logic 1 is a higher current (but below the maximum for the device of course ).

For reception start looking up fast photodiode transimpedance amplifiers.

Also make sure your laser diode has a collimator, the beam will diverge considerably even in a short distance without one.

Possible problems : insufficient receiver bandwidth, system jitter, and slow driving circuitry or a laser not meant for communications.

 

[()blivion]

Edit: RichTheDude, I really truthfully do appreciate that you are actually trying to talk electronics with me and not questioning my ideas or other wise stalling and being difficult for no reason what so ever. The following comments, though possibly hurtful, are not an attempt to cut you down, be little your intelligence, or otherwise be hostile toward you. They are intended to only point out that you were trying to join the conversation while obviously not paying attention to it. Though I can understand making minor bad judgment calls or missing things because of influences like being up way too late. Still, if your honestly trying to help me, then you should try to pay more attention. If this offends you I'm sorry, but these are all facts as much as it is that I thought "discreet" was actually the word "discrete". I accept my failure with honor and have corrected that mistake.

"Develop a circuit which can [put] through from Ethernet to a serial stream of TTL and vice versa"
This is precisely why I'm here, check the post title again. If I was doing something else I prolly would have labeled it differently don't you think?

"I severely doubt the circuit you have posted btw."
That makes two of us. Which is why when I posted it I also said... "it's got more fundamental problems than I care to admit." I was fully aware of it's flaws. And I was never saying that circuit as is would work, merely using it as a reference.

"Make your laser diode and receiver combo."
Done ages ago on the bread board.

"There are laser driving IC's out there "
There sure are, and I still don't care.

"most likely [use] a Si PIN photodiode"
Yep, and make it all reverse biased and such.

"at these speeds turning the laser [fully] OFF to [fully] ON will most likely be too slow."
Yep, QED

"Most communication systems intensity modulate a laser diode"
kinda Like mine did/does/will.

"Also make sure your laser diode has a collimator"
Laser pointer lens modules + CD burner LASER diodes = Close enough for me.

Edit: ===================== Speaking to RichTheDude ends here... General commenting to the community begins =====================

OK, fundamentally the problem at hand is still just how to get the 100base-Tx three symbol 25Mhz information off the twisted pair medium back to a plain Jane TTL 1's and 0's STREAM where it's nice and easy to work with. Not having an adequate discrete solution to this hasn't changed. It's a tricky problem to address with the restrictions of doing it all from scratch and ignoring preexisting parts specifically built for doing this task. This is because we are working in an area that's not quite analog, but not quite digital. It's a grey zone. But I'll start things off with some basic ideas...

We start with the twisted pair medium, we use a differential amp and maybe (probably) a transformer to prepare the information to make it no longer be differential nor contain any common mode line noise. The amp will have to be able to do at least 25Mhz as that's what the 100base-TX standard uses on the wire. After this we need a way to return the three MLT-3 encoded voltages back to TTL. This could be done with two properly biased transistors, An NPN for +1 volts and a PNP for the -1 volts. Lack of triggering either transistor would indicate the 0 volt level of the MLT-3 encoding. So either transistor being activated would deactivate the third (ie logical NOR) This would provide us with three lines at TTL levels. Now some logic to convert this to a serial data stream is needed. This is where things get tricky... Everything up until this point doesn't need clocks or logic gate IC's or any of that fancy jazz, it can all be done with discrete parts. (I consider some 8-pin IC's discrete parts) But the actual information is basically Manchester/NRZ encoded. So it looks like it's going to need edge detection, a high Q clock, and a shift register circuit. That or some kind of magic...

Thoughts on this so far?

Lets continue. Now if all I was after was JUST the LAN-LASER-LAN conversion, I could from this point build a voltage controlled RF oscillator whose oscillations bring my LASER diode into and out of the lasing threshold. Then use the three wires obtained above to shift the frequency of it to higher, lower, and neutral frequencies. Or use some third frequency and leave neutral for debugging, aiming, and/or error correction. Finally use some form of RF filtering and FM detection on the receiver end to demodulate the RF back to the three wires above. From there it's easy to go back to differential, and thus back to 100base-TX UTP. And none of that would require any form of processing or decoding, AND the circuit would be fairly simple as far as I know.

If any one would like to work on JUST this LASER bit, I'd be down for it for the most part. It would be a good first step toward the goal stated in the thread title. And I could still use the device when finished. Even if it's not exactly the main/first topic at hand.

(Edit: THIS COMMENT IS NOT DIRECTED TO ANY ONE PERSON IN PARTICULAR.)
Any grammar, spelling, or punctuation errors are not my fault but are in fact the fault of the English language, If I have made a mistake and it bugs you... PM me and I'll fix it ASAP. Also, post count does not equal the size of ones manhood, nor is an indicator of ones IQ. I've been EEing for a while now. I'm NOT a nub or moron. Just thought I should point this out once more, I don't have a lot of posts because I don't often need help with stuff like this. I CAN do it all by my self.

 

[RichTheDude]

I was pretty tired when I writ that last night, but I was trying to give you advice from the rather considerable experience I have within this field (about 5 years, my own company with 3 people, journal papers, IEEE conference papers) you can take it or leave it. I also rarely post here (is 50 some odd posts a lot on a forum?), and the only time I have asked for help was about the law of pots in audio tone controls. Also discreet and discrete are two entirely different words if you wish for the forum to become full of grammar/spelling Nazis.

At no point have I said your a moron, "nub" or whatever, there is no need for defensive hostility. Can we stop playing the "compare the manhood" game and get on with electronics?

 

[()blivion]

Here is something interesting on the subject at hand. Found it on a 3com 3C905 NIC I had laying around.

https://www.electro-tech-online.com/custompdfs/2012/01/dp83223.pdf

Only wish I knew exactly how it worked.

 

[anas.ansarie]

Oblivion ...

Sorry m interuppting you ... but m trying to find the same exact thing ... conversion of ethernet differential signals to TTL using discrete components without affecting its integrity.

i wanted to ask you ... have you come up with something like any circuit or design ?

Actually i have build transmitter and receiver but now the problem is interfacing them with Ethernet Phy layer ..

Can you please help me Oblivion ?

Thanks

Anas

 

[cr0sh]

Have you looked at the Ronja project (Twibright Labs)?

http://ronja.twibright.com/

It may not be exactly what you are expecting, but it is a homebrew FSO system, and may give you some ideas...

 

[()blivion]

have you come up with something like any circuit or design ?

Sorry Anas, all I have are ideas at the moment. Keep your eyes on this page, if I figure something out I'll let you know. One of the paragraphs in post #14 has a decent description of a circuit that could *JUST* convert the MLT-3 encoded voltages back to TTL voltages. The only problem is that you would end up with three wires and there is no gain control. I could prolly whip up a schematic for it. I haven't actually tested it though, so no guarantees it will even work.

Have you looked at the Ronja project (Twibright Labs)?

http://ronja.twibright.com/

It may not be exactly what you are expecting, but it is a homebrew FSO system, and may give you some ideas...

Very nice link, thank you.

And yes, I do think the project on that page is just a little "over done" and could be reduced in complexity by an order of magnitude with the right design. As much trouble as that project entitles, one might as well just use the a fore mentioned prebuilt parts. I say that being fully Hippocratic of course. (^.^)

 

[cr0sh]

And yes, I do think the project on that page is just a little "over done" and could be reduced in complexity by an order of magnitude with the right design. As much trouble as that project entitles, one might as well just use the a fore mentioned prebuilt parts. I say that being fully Hippocratic of course. (^.^)

Something to realize about this project is it's history:

https://en.wikipedia.org/wiki/RONJA

It was developed and "released" to the world in December of 2001 as the first open-source FSO system; it had probably been in development for a few years prior. There is a few interesting things about this project: One, it was developed in a country where not too many years before was a part of the former Soviet Union, and two, they released it as open-source rather than commercializing it. I speculate that these two factors could very well be related; as such an FSO link allows for high-speed networking, that can't be easily "jammed", nor can it be easily "eavesdropped".

It is also telling that such an open-source implementation wasn't released by a group in a country with more freedoms (especially freedom to get the necessary parts - as you note, getting such parts is fairly easy - today. But back then, in the Czech Republic? Probably easier now, but back then, close to impossible without more than a bit of cash changing hands). Also interesting to note is that, while Twibright Labs have shown how to create such links, there hasn't been much (if any?) interest in groups elsewhere developing the technology (yes, it has limitations - but it also has its strengths). Part of this is the easier (and more developed) use of WiFi for such free-space links, as well as the limitation of the technology (and its inherent complexities).

Maybe your developments in the space will make it easier to implement such links?