roller bearing guided small aggragtes on elliptical path

Go To Last Post
20 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I design several small aggragtes moving along an elliptical path in an endless turning sequence.

My probrem is , I am searching some robust ways to send data to these aggragtes. there are 4 physical segments each 4meters long, And each segment should send some segment specific data to the moving aggregate.

I can not use cable, because it is an endless turning loop

I can not use wireless because, there are several aggragates and wireless can not differ in which segment is the aggregate moving. attaching a position sensor to all aggrehates is very expensive (there are 420 aggregates)

What I imagined is that Each segment send some differential data through conductive roller bearings on the aggregates. The problem is that the bearing contact is very noisy and have serious prell effects.

what would you recommend for a very noisy and uncertain contacts . How can I send data to aggrgates.

I imagined 2 phase shifted signals . And the mover measures the time difference of the signal 100 times a second and sorts the results. The most equal ones are filtered and this will be interpreted as result. (using standard deviation).

Any ideas?

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Perhaps an optical system ?

How much data ? (bytes, update-rate)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How about inductive coupling like low frequency RFID?

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Can you post a photo of the system?

What is an aggragte? And how fast do they move?

If you are sending data to the aggragtes, what do they have for power and processing power already on board?

How electrically noisy is the environment?

How dirt & dust filled is the environment?

Mission critical or not, i.e. can an aggragte afford to miss a data packet now and then?

Need more info.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Each Head has seperate DC-DC converter. these converters are fed through a roller bearing contact. There are 2 roller bearing contacts
One contact:36Vac
Other Contact: GND

The DC-Dc converter converts this 36Vac to a 0..50A upto 8V adjustable current source. The value of the set current should be send to the moving unit.

Wireless is not effective because each segment has different current set values. I send a photo of the system . total length of the elliptic path is 25m.
Segment 1 : 1m Electrical Current value1
Segment 2 : 2m Electrical Current value1
Segment 3 : 6m Electrical Current value1
Segment 4 : 3m Electrical Current value1
Segment 5 : 6m Electrical Current value1

I thought abaout changing the frequency of the power bus of each segment and count the frequency on the slave.
For exampel I send 36Vac at 3000Hz this means 30A current set value.

Each moving unit is 10cm wide and can source 50A upto 8V. it means if a segment is 6m long I should modulate a frequency on 36Vac at 500A.

But it is a very corrosive enviroment. (chromic acid ).
So the bearing contact is not reliable for data transmission.

for power transmisson on a noisy prelling contact I rely on a high voltage buck converter. Input:36Vac output is 0..50A upto 8V , so if a prell or noise induced on the bearing, the buck converter will compensate it throgh internal stored energy margin and regulation down to 8V.

But for data transmission, I can not feel comfort.

An aggragte should not miss a packet. Because if missed. The product it is processing will be destroyed.

If wireless is implemented the range sgould be very tight up to 5cm.So that each segment can have its own wireless sender.

The parts move with up to 20cm/sec.

Best regards.

Attachment(s): 

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Cheap and reliable tend to be at opposite ends of the spectrum.

Similar to your idea of changing the power supply frequency, you could also inject a signal onto the power line. This would be similar to BSR or X10, where the controller adds its signal to the Mains power line. You would not need to modulate your signal with data, as the BSR / X10 system does, just change your frequency that you inject onto the power buss.

In a electrically noisy environment it would be easier if there were a fixed number of levels, rather than a full, analog range.

Additionally, if you have a micro on each unit to decode the signal and set the level, then no reason it can't have some A.I. If it knows the last level, or last two levels, then maybe it can predict what the next level is likely to be. It can set the next level and use the command as verification, if time or signal integrity is a problem. It doesn't sound like your system will need this, though.

An optical system with LEDs and a photodiode on each receiver would also be an option. The micro and photodiode go in a environmentally proof plastic box with a clear cover. You have a strip of LEDs, (all in series), mounted along the path. You then send a data packet as the units pass by.

Like all data comm's, you need to packetize the data, start byte, data byte, data byte, data byte, checksum or crc, at a minimum. And typically send it several times. In this case the three data bytes could all be the same value, the current level.

Extra LEDs on the transmitter flood the receiver area to make sure you have a good S/N ration.

Your call on two or three LED strips in parallel, for redundancy on the transmitters, and likewise how many photodiodes on the receiver.

Perhaps the units show a green LED whenever they receive a valid signal, and flash a red LED if they haven't received a valid signal in X seconds. Making it easier to spot a defective device.

With sparks from the Power/Ground rails, and switching 50 A's, you will need to pay careful attention to power supply and EMI filtering. What works great on the bench may not work so well in the real world without attention to detail.

You might consider the 38 kHz ( ? ) IR remote modules for the receivers, otherwise you end up building your own. The LED signal needs to be modulated so that it isn't getting fluorescent and other lighting interference.

If the environment is too dusty for reliable optical comm's, then you might also consider using ultrasound. Depending upon how many power levels you need you can transmit your data signal ping at any of several frequencies. Usually one operates the piezoelectric ultrasound modules at their resonant frequency, but they work at higher and lower frequencies, also, just with lower efficiency, which wouldn't be a problem in this case. You then measure the received ultrasound ping and measure its frequency.

You could, alternatively, pulse width modulate the pings, or use an on/off keying modulation scheme. Lots of options.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I also thought of using irfrared. But the enviroment has a severe acid cloud. It will be sucked by the ventilation, but anyhow it seems to me risky to rely on a shiny clean led surface.

I also agree that some range/band for signal data should be defined and only these in-range data should be processed.

I have the idea to use 2 signals they are 2 pwm signals that are phase shifted to each other.

Lets assume each PWM has a frequency of 500Hz.

I define some bands of phase shifts

Band1= 100us-200us
Band2= 400us-500us
Band3= 700us-800us

the slave controller measures the phase difference.
it throws out the ones that are not in band range.

The filtered data is sorted. And the most equal series from 100 in range measurements data set should be considered the right data. I believe I should use standard deviation formula to decide which set and number of series should be the same and can be considered as right data.

Example I have from 100 in band measurements:

42 times 410us-412us
12 times 400us-402us
5 times ....
2 times ....
......

It is the question how to find the minimum number so that it can be considered as right data. In this case I have 42 times 410us to 412 us range . So what is the minimal number of the most frequnt range so that it can be considered as right.
(ex: except 42 what if the maximum frequent equal data is 32, is it enough according to the theory of standard deviation?

Below : standard deviation has +/- %31.2. So if 32 values out of 100 are the same , I accept them as right data?!?!?

Or is this %32.1 also data set dependant?. And should be calculated dynamically?

Any ideas?

http://en.wikipedia.org/wiki/Sta...

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There was a programme on yesterday on the BBC called Bang Goes The Theory. It was about trains, and one segment was about the Docklands Light Railway (DLR) - a driverless train system in London - and how they monitor the position of the trains and how they communicate with them.

http://www.bbc.co.uk/iplayer/epi...

Not sure if it is accessible without a UK IP address, but if it is, then start watching about 4 minutes in.

Also check out:

http://dodger.home.xs4all.nl/tec...
http://en.wikipedia.org/wiki/Sel...

Could be of use, as from your description it is a similar situation.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes,
Inductive data transmission seems the best, robust to acid cloud .(sender and receiver can be fully encapsulated by a thin plastic film).

I am also investigating a transformer like stucture where the one half of the E core is 1cm apart from the other half. Just think abaout 2 seperate coils and they have 1cm air distance. If I modulate a suitable signal on the first coil , I can get information on the second coil.

But in my case I should consider the mechanical tolerances. 1cm (10mm) air gap will swing with a tolerance of +/-2mm it can be 8mm or 12mm according to the tolerances of the mechanical construction.

Have anyone tried something like that , maybe a PCB coil with a PWM and 10mm apart another PCB coil, o good proper modulation scheme (ex, phase shift modulation or frequency shift keying, etc.). It will be more immune than a physical contact in acid cloud .

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Rfid does that.

Quote:
Have anyone tried something like that , maybe a PCB coil with a PWM and 10mm apart another PCB coil, o good proper modulation scheme (ex, phase shift modulation or frequency shift keying, etc.). It will be more immune than a physical contact in acid cloud .

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I should begin searching spi to rfid ics.

MCU spi to rfid ------ rfid to spi MCU

I always thought that rfid is not a bus system but a active - passive transmission (libraries, bookstores etc.)

But if both sides can be active (as MCUs) I should try to implement it.

Also importants details : Can Rfid do a multicast or broadcast to all slaves? Ex: I tried to figure out ASI bus. But after I discovered that it can not do multicast, i thought it would be a headache to send to all adresses in a loop.

Any ideas for spi to rfi ics and multicast.

I believe I can limit the range to 15 to 20cm so that each segment can have its own range.

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
But the environment has a severe acid cloud

That is why you have several LEDs doing the transmission, to increase your signal strength.

The fact that the road is foggy doesn't mean that your headlights don't penetrate the fog at all, just that your range is diminished.

I would think blasting an optical signal a few inches would be much easier than trying to link via moving transformers.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Rfid or Optical? .this is the question.

I can built a dark encloser from extruded plastic for the sender and receiver. So that no sunlight pass through. The problem there is if any contanimation or dust of acid is collected in this enclosure. I must examine .

Any recommendations for rfid to spi chips and 38Khz irda transmitters so that I can try both on a prototype where acid cloud will be more severe than normal condition. IMHO best way is to test it under these conditions.

Any name/vendor for RFID to spi and IR transmitter to spi.

best regards.

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Rc522 nxp.

I don't think irda is 38kHz. Ir remote controls are. The 38kHz modulation is easy to achieve.

Multicast depends on your protocol.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I examined the MFRC522

The active unit sends data to a passive card, which can be read . I can not figure out how I can connect the passive card (IEC 14443) to spi to read the data that is written by MFRC522.

Is it possible to connect the passive card to spi in a glue logic. Or are there some cards that sotre the data in eeprom that can also be read by spi externally.

The long way is to write the data to passive cad and the slaves read from these passive cards through a seperate MFRC522.

the card is 10mm apart from the elliptical path. it will be updated by host rc522 and read by slave rc522 that is attached on moving slaves. But i dont know if host can write and other slaves can read from the same card. There must be also collision avoidance etc.

Best and shortest way seems to read the cadrd data through spi and install cards and readers on the moving unit.

Any recommendations to read data from passive card thgough other slave rather than the host that has written into it.

How abaout multicast. Can all cards have the same adress. If so, how will they reply back to verify data?

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The intention was not to use a rfid card! You can use a small micro to emulate it.

The words aggragt and prell do not translate well. From what i can understand you have things like train carriages that carry an object to be electroplated. Depending on the surface area and the plating bath, you vary the current. Each carriage may have a different piece, thus each carriage needs to know what current for each bath. If this is known beforehand, then when each carriage is loaded, the current profiles can be loaded onto an onboard micro. Then all the carriage needs to know is what bath it is at. This could be done with rfid cards - one at each bath to identify the bath and another that gets loaded with the current profile and gets read first up. If the process data is measured, it can be written to the rfid card with the current data thus forming a process 'traveller'

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Sorry, my last post is lost. I send it gain

aggregate : set of machines (I mean carriage)

prell (word from other language): contact debounce or untight contact, discountinuty in contact resistence reliability.

Any clue , starting point how I can simulate rfid card through spi?. Any starting point?

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This is a very interesting hack. Not only power, but even the oscillator frequency is derived from the antenna.

But in my case my rfid card emulation mcu is powered and clocked by external source , so it will be harder to convert the code to my needs. Especially the synchronization of my mcu crystal freq against abtenna frequency.

Are there any full active rfid cad emulation mcu example.

My sender rc522 will have a PCB trace. My coil card has also a pcb trace but the decoding of the RFid on the card side is not clear to me, if there would be any active IC on the card side just like rc522 on the market, it would be a great help. Maybe I can try to misuse the rc522 so that it functions as a rfid card emulator.

the master rc522 sends rfid data and the card emulation rc522 replies back.
Is my logic straightforward or forked to uncertainity?

________________________________ We dream of a world where current does not need the voltage to flow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I've not tried to emulate a rfid card, so my comments could be totally wrong. From my limited knowledge, the sender gives a carrier and modulates it. To receive this, you're not worried about the carrier, just the modulated data. A diode and comparator would be needed.
To send data back, the sender gives carrier and the emulated card shorts it's coil with modulated data to respond. Not having looked at the code for the link i gave you, i dare say it is basically what it does. Elektor had an article on the sender side using a tiny2313 a few years ago doing mifare - this might give some hints. Get out the 'scope and do some experimenting.

I think TI have some products that implement 'wireless' inductive charging where they do comms similar to rfid.
Here it is:
http://www.ti.com/lsds/ti/power-management/wireless-power-receiver-solut...