Wireless data transmission

Grumps wibbled on Thursday 14 January 2010 13:18

Air bourne optical link for the data? I thinking 3 or 4 TX leds with common drive spaced around the axis and one photoreceiver on the other end. Repeat in reverse if duplex is required.

Or use the brushes and perhaps use a balanced current driven signal with plenty of error detection. Carbon brushes might be better than small contact area wipers.

--
Tim Watts

Icicles - nature's way of pinpointing all the leaks in your guttering...

I'd be thinking of transferring the power by means of a transformer consisting of two coils sharing a common axis, about which one is rotating and the other isn't. I can't see why data wouldn't pass the same way. Perhaps not so effecient in transferring power, but no wear.

Sylvia.

Thanks for the idea. I was thinking of some form of optical interface too.

Ooh, I like that idea! Thanks.

Thanks. I know of one patent which covers this approach, so it's not going to be good for the finished product.

Interesting. There might be issues with channeling the light so that the receiver can always pick it up; 1Mbps @ 1500rpm works out at about

42000 bits each revolution, so any 'dead zones' would c*ck things right up.

Maybe the spindle that everything's mounted on could be hollow? That way you could have one TX LED and one detector mounted inside, with one part rotating relative to the other.

cheers

Jules

1 Mbps is pretty high..you need a carrier up at least 27Mhz, which would be my first thought, since its essentially cowboy territory. Anything goes below a half watt or so.

..but then you ned all that mod/demod kit.. Yuk.

probably time to grab a couple of WiFi chips and pics and get coding/soldering.

well that's just using the M part of teh EM wave instead of the E part. Frequencies and modulation schema are the same for either.

Surely the principle of inductive coupling is so fundamental that you should not have any IPR issues with the rotary transformer approach.

Bob

ing

er

Others have suggested magnetics. I will suggest electrostatics just to be different.

A pair of conductors with a small gap between make a capacitor. At some very high frequency, the Xc of this capacitor will be small enough that reasonable amounts of power can be transfered this way. The advantage could be that the mechanical part is simple.

ing

er

How much less than 1Mbps?

Digital model railways send power and data (much lower rate however) through rolling contacts between wheels and rails and wipers on the back of the wheels quite successfully. A simple XOR error check is used.

MBQ

691,200Hz at present. I may want to increase this (x2 or x4) later.

Any idea of the data rates that these use.

I can try.

This is my current favourite solution.

On a sunny day (Thu, 14 Jan 2010 13:18:25 -0000) it happened "Grumps" wrote in :

2 halves of a potcore, each with a coil in it (rotating transformer, by 5mm is a rather large gap, make it smaller. Or optical.

about 50 baud ;-)

If this is a one of, you can pickup a junked VCR for next nothing and use the rotory transformer from the video head assembly. It has good bearings and is very stable. I'm not sure about the frequency response or the power level it is capable. Opps, your 5mm spec. means you can't use the whole assembly. Oh well the ferrite pieces are there. Anyone have thoughts about it? Mike

How much space on each 'wheel' do you have and what sort of weight can it carry before causing some sort of imbalance... ?

I'm thinking of xbee modules... but you might need a smart sort of microcontroller on each do do the data caputre/transmit - but maybe you have that already?

Then again, I've just looked and they're 250Kb/sec max. however there are then things like this:

up to 2Mb/sec, but you will need some sort of smart device to push data into it and get it out again...

Gordon

Thanks, that looks a good little module. My system could easily support that weight. But I'm tending to the optical approach.

I've seen shaft horsepower meters (on 30" steamship propeller shafts) that powered a rotating load cell via inductive coupling at 60 Hz. They wound, I'm guessing from memory, 10 or 20 turns of wire around the shaft. A stationary "C" core pumped 60 Hz flux into the rotating secondary. They got the signal back out as capacitively-coupled FM. That all worked independent of shaft speed.

For an always-rotating shaft, stick a couple of flexprint coils to the surface of the shaft, adjacent to a stationary magnet or two. That will steal shaft rotation power.

Couple back out magnetically or capacitively, essentially short-range RF.

John

I once worked with a slowly rotating system using 20 mA current loop at 9600 bit/s through two sets of rings and the third pair of rings carrying 220 V at a few amperes.

Originally it used a half duplex (command/Ack) protocol, but since both the command or the acknowledge message could be lost, due to individual bit errors at that data rate, it created quickly quite a nasty backlog of messages due to retransmissions. I wrote a broadcast (no ACK) protocol with heavy error correction coding (ECC) and after that the system worked OK.

At that data rate several consecutive bits will be lost due to dirt etc. so you also would have to use bit interleaving to convert burst errors into random errors, that can be readily corrected by the ECC.

Exactly for the same reason ECC and interleaving is used for instance in CDs or digital radio links.

Adding 30-50 % of ECC bits with interleaving should help lot with the reliability and since the data rate is only about 1 Mbit/s and hence the wavelength is in order of 200-300 m, ring assemblies less than a few meters in diameter should not be a problem.