Good day
I have a project where I need 200 relays that control chemical solutions.
From what I can see, it seems the most I could control are 8 relays per board. 25 Arduino to control all the relays seems overkill.
Is there another way? I'm planning on using a Raspberry-PI for some other tasks, so that can help out.
Regards
there are a bunch of way's to expand I/O, for example you could use 25 shift registers, which all go back to 3 wires on the arduino
You could use one of these boards fully populated daisy chained with 2 more with additional shift registers, each capable of sinking higher currents at higher voltages to drive relays.
Serial interface with the Pi to receive turnon/turnoff information.
(LEDs were intended for cars, with built in resistor, more importantly for this application had individual mounting hardware, and the TPIC6B595 shioft register can take higher voltages like that)
Relays are not the most reliable of devices, so I'd be seriously concerned about any system that requires 200 of them to continue working for any length of time.
dc42:
Relays are not the most reliable of devices, so I'd be seriously concerned about any system that requires 200 of them to continue working for any length of time.
I beg to differ on that. I worked at a refinery where there where literally thousands of DC relays in service for decades and the only failures I ever saw was on the contact sides where there was a short circuit on the relay output circuitry causing the contacts to burn up, not hardly the fault of the relay. Relays like any other component come in various quality, buy a good one and it's just as reliable as buying one of your reliable caps. 
Lefty
Hmmm. 200 relays? That will take a lot of power if they are all going at once. What devices are the relays powering? Power transistors or MOSFETS might be a viable alternative. 8)
You can use latching multiplex chips like the 74hc259 to do the job.
Seems like a poor choice to drive relays.
Both are only 5V parts and not much drive current to speak of.
TPIC6B595 is good for high voltages, such as 12V relays, and is intended to be a current sink.
6A595 can drive more current
There's also a 6C595
and 6D595
http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00163738.pdf
Now, if you want to read up about "real" relays simply google "post office 3000"
We used them by the thousands in my apprentice days (nuclear industry) and one of the many tasks assigned was to periodically polish the contacts using chame sticks and set the contact pressures using bending tools and gramme-gauges.
In later life we used hermetically sealed relays for ESD systems and I cannot recall any ever failing. Yes they were expensive but reliability and cost tend to go hand in hand.
Mr_Blah:
What devices are the relays powering? Power transistors or MOSFETS might be a viable alternative. 8)
The relays switch power to tubes of chemicals sequential, only 3 of the tubes will be powered at a time and at most each relays will only be powered 2 or three times a minute.
5000v 10-700 micro amps.
As things stand though what we are trying to achieve will not scale up, so might try something else.
You'll need specialist and probably expensive relays to switch 5000V. As the power requirement is 3.5W maximum, I'd look at using 200 simple inverters instead, with one microcontroller generating the timing signals for each group of around 2 to 8 inverters.
5000V? Really? Or is that just the voltage the contacts can withstand?
Such as
which has a 24V, 10mA coil.
Seems like a poor choice to drive relays.
Both are only 5V parts and not much drive current to speak of.
Well, I would expect you and the op to understand that you probably can't directly drive a relay of any size directly from the 259 chip, just like you can't from an arduino.
CrossRoads:
5000V? Really? Or is that just the voltage the contacts can withstand?
Such as
http://www.alliedelec.com/search/productdetail.aspx?SKU=70213404
which has a 24V, 10mA coil.
As I thought, expensive. In which case, a simple inverter to generate the 5000V for each tube would cost less, unless the 5000V needs to be regulated. It would also avoid the hazard of distributing 5000V power throughout the system.
If you do go with the relay solution, you could consider driving them from TPIC6B595 shift registers. [EDIT: as in reply #2.]
A project controlling many strings of christmas lights for an idea of the wiring and such.
If its 5kV DC in air, that cannot easily be switched off! You just create an electric discharge that'll melt the relay contacts and start a fire. (*)
With AC you have a bit more chance, but that would probably require specialist evacuated, extra-large, or oil-filled relays or some such.
(*) often you'll see that switches and relays rated for say 10A at 240VAC are rated for less current at say 24VDC - the issue is
arc formation as the contacts separate - with AC the current stops 100 or 120 times a second, allowing the arc to dissipate, with
DC there is no break in current and with enough current the arc will grow to fill the gap, limited only by the voltage and power available.
The voltage to hold an arc is far far less than that required to strike an arc.
The contacts of the reed relay that crossroads linked to is rated at 5000VDC, 10mA. Maybe it's oil-filled, or something similar.
dc42:
The contacts of the reed relay that crossroads linked to is rated at 5000VDC, 10mA. Maybe it's oil-filled, or something similar.
And most likely well worth the $172 each.
retrolefty:
And most likely well worth the $172 each.
That's why I think separate 5kV inverters might be a better solution, unless there are strict requirements for regulation or rise and fall time.
We can only speculate until 2wice tells us what the relays actually are.
