Bitcoin Forum

Other => CPU/GPU Bitcoin mining hardware => Topic started by: cicada on August 16, 2011, 02:22:49 PM



Title: remotely controlled power switch?
Post by: cicada on August 16, 2011, 02:22:49 PM
I've seen quite a few posts where people are controlling their farms remotely, often 10s or 100s of miles away, asking about automating reboots, cycling power with automated timers, etc.  And sometimes miners just lock up hard, and there's nothing to do but flip the power switch.  

It occurs to me the ideal solution would be computer/remote/network controlled relay switches, allowing you to power-cycle your miners from anywhere you please.

I've done some research on solenoids, solid-state optical relays, etc, that look like they could be pretty easily wired and driven from a serial port or even USB, however my EE skills are severely lacking.  Something pre-packaged in a <$50 range would be ideal, but I'm not finding any 'consumer friendly' options.  I'm willing to pull out the old soldering iron if I can find a good guide to follow, but time is precious.

Is anyone using such a thing they could recommend?


Title: Re: remotely controlled power switch?
Post by: MiningBuddy on August 16, 2011, 02:30:38 PM
Sounds like you want to be using an IP remote power switch, similar to what most data centres use.
I'm using this one in my garage: http://www.openxtra.co.uk/p/usb-net-power-8800uk and it works really well for its cheap price.


Title: Re: remotely controlled power switch?
Post by: cicada on August 16, 2011, 02:38:57 PM
That's the ticket, but at 6A max current I don't think it'd hold up too well against my 4x gpu rigs.  Looks like they've got a few IP controlled switches with built-in web pages, etc, but they're a bit pricier..


Title: Re: remotely controlled power switch?
Post by: MiningBuddy on August 16, 2011, 02:43:19 PM
That's the ticket, but at 6A max current I don't think it'd hold up too well against my 4x gpu rigs.  Looks like they've got a few IP controlled switches with built-in web pages, etc, but they're a bit pricier..
6A is fine for what I need here in the UK with our 230v.
But yes, at the end of the day you get what you pay for and at £35 this is really the bottom end of things.
Have a look around ebay, you can usually pick them up second hand for decent prices from data centre clearances.


Title: Re: remotely controlled power switch?
Post by: Detritus on August 16, 2011, 03:38:35 PM
x10 appliance modules handle up to 20 amps.
The cm17a firecracker module interfaces them to a PC.


http://www.linuxha.com/bottlerocket/ (http://www.linuxha.com/bottlerocket/) provides a script to turn modules on/off.


Title: Re: remotely controlled power switch?
Post by: cicada on August 16, 2011, 03:57:30 PM
Something like this SSR controlled power strip would work pretty well too, fairly cheap outside of the labor involved, but choosing your own SSR means you decide the max-current and voltage specs..

http://www.instructables.com/id/A-USB-Power-Controled-Plug-Strip-With-Isolation/

This $23 SSR + crap I have laying around could get the job done:

http://opto22.com/site/pr_details.aspx?cid=3&item=120D25


The only problem I see with this is it introduces a dependency on whatever device is supplying the DC current to the switch..  one more possible failure point..


Title: Re: remotely controlled power switch?
Post by: kirax on August 16, 2011, 10:42:02 PM
Something like this SSR controlled power strip would work pretty well too, fairly cheap outside of the labor involved, but choosing your own SSR means you decide the max-current and voltage specs..

http://www.instructables.com/id/A-USB-Power-Controled-Plug-Strip-With-Isolation/

This $23 SSR + crap I have laying around could get the job done:

http://opto22.com/site/pr_details.aspx?cid=3&item=120D25


The only problem I see with this is it introduces a dependency on whatever device is supplying the DC current to the switch..  one more possible failure point..

Wouldn't all of this seem to be better served by a simple USB relay to put the pins of the power switch on the mobo high for the 10 seconds needed to tell it to shut off now, dammit? Then you don't need it rated to run the actual load of the full power draw, it is just like having a USB (or ethernet) interface to the power switch. No need to interrupt the power to the rig, really, it's just wiring the power button to a switch that you can control remotely. There are commercial solutions that do this already, and a friend of mine that does micro hardware is looking at building one for me.

(As a random example: http://www.controlanything.com/Relay/Relay/CAT_RELAY4. I found cheaper when looking a while ago, but that is a random common example company.) 


Title: Re: remotely controlled power switch?
Post by: cicada on August 16, 2011, 11:22:13 PM
Wouldn't all of this seem to be better served by a simple USB relay to put the pins of the power switch on the mobo high for the 10 seconds needed to tell it to shut off now, dammit?

Man, I got options.

That would be pretty easily doable with an arduino or similar, and gets rid of the relay-failure problem.  Its also a hell of a lot safer, we miners get enough high-current play in the breaker box!

X10 isn't a bad idea either, I forgot about those :)


Title: Re: remotely controlled power switch?
Post by: kirax on August 17, 2011, 01:45:57 AM
Wouldn't all of this seem to be better served by a simple USB relay to put the pins of the power switch on the mobo high for the 10 seconds needed to tell it to shut off now, dammit?

Man, I got options.

That would be pretty easily doable with an arduino or similar, and gets rid of the relay-failure problem.  Its also a hell of a lot safer, we miners get enough high-current play in the breaker box!

X10 isn't a bad idea either, I forgot about those :)

Interestingly, my friend that was looking into building one of these mostly uses the arduino, so it might even be based on that :p


Title: Re: remotely controlled power switch?
Post by: twobitcoins on August 17, 2011, 03:02:11 AM
I use kit 108 from kitsrus.com, purchased assembled from electronics123.com (http://www.electronics123.com/s.nl/it.A/id.291/.f?sc=8&category=35).

It has 8 relays controllable by serial port.  I wire them into the motherboard power switch connector.  One box can reboot 8 machines.  Of course you need two boxes connected to two separate computers if you want to reboot the machine controlling the relay box.


Title: Re: remotely controlled power switch?
Post by: kirax on August 17, 2011, 03:25:56 AM
I use kit 108 from kitsrus.com, purchased assembled from electronics123.com (http://www.electronics123.com/s.nl/it.A/id.291/.f?sc=8&category=35).

It has 8 relays controllable by serial port.  I wire them into the motherboard power switch connector.  One box can reboot 8 machines.  Of course you need two boxes connected to two separate computers if you want to reboot the machine controlling the relay box.

Right, don't hook it one of your miners, or you have a chicken and egg solution, unless you maybe had a really clever relay box that if it didn't get a keepalive pulse on pix <whatever> from that box, then it reboots it... but that's a mess. I already have a file server with not enough PSU for a real video card that I would hook mine to, personally. Or, for a bit more, I'm sure you can find one that has an IP and a web interface to skip needing a computer to go with it...


Title: Re: remotely controlled power switch?
Post by: teflone on August 17, 2011, 11:18:24 PM
Low tech could just use a timer...



Title: Re: remotely controlled power switch?
Post by: rearwheels on August 18, 2011, 02:14:06 AM
Low tech could just use a timer...



Low tech would be placing a cd-rom drive which the tray will open and hit the reset button.


Title: Re: remotely controlled power switch?
Post by: cirz8 on August 18, 2011, 06:50:09 AM
Low tech would be placing a cd-rom drive which the tray will open and hit the reset button.
Someone please do a thing like that and make a video  ;D


Title: Re: remotely controlled power switch?
Post by: BkkCoins on August 18, 2011, 07:35:07 AM
Another option: PIC-WEB for $30 Euro and a small relay or transistor for each mainboard controlled.
You'd likely have to write some small bit of code to toggle port pins but the main stuff is there already in the TCP-IP stack and web server stuff.

http://www.olimex.com/dev/pic-web.html

They have a bunch of other ethernet connected  boards too but I think this is the cheapest one.

And useful for other projects later too.


Title: Re: remotely controlled power switch?
Post by: cirz8 on August 21, 2011, 10:52:00 AM
Low tech would be placing a cd-rom drive which the tray will open and hit the reset button.
Someone please do a thing like that and make a video  ;D
Having two rigs with each cd-rom drive tray pointing at each others reset button and a simple monitoring script, then just watch those rigs resetting each other as you discover bugs in the script  ;)


Title: Re: remotely controlled power switch?
Post by: Swishercutter on August 21, 2011, 11:17:19 AM
Wouldn't all of this seem to be better served by a simple USB relay to put the pins of the power switch on the mobo high for the 10 seconds needed to tell it to shut off now, dammit?

Man, I got options.

That would be pretty easily doable with an arduino or similar, and gets rid of the relay-failure problem.  Its also a hell of a lot safer, we miners get enough high-current play in the breaker box!

X10 isn't a bad idea either, I forgot about those :)

I built one with an arduino.  I even wrote a GUI so that you could log into a control computer (Teamviewer) and press power/reset, it even indicates when power turns on.  The pins on the power switch just ground when turning on so you can just use a transistor or a relay (I used a transistor). 

Also, the design a couple thermistors also so I could monitor ambient temps remotely also (not yet completed in code but very simple).

I am sure you could eliminate the control computer and just wire in a network interface also but it was much easier in my case to just write the program and build the arduino shield.

I have yet to have a miner lock up to the point where I actually needed to cycle the AC to get it to restart, physical reset with the buttons always works.

The plan was to release the code and the design on the forum I just have not got around to it yet.  It is a very simple "Processing" code using the Arduino mega with "Firmata" firmware.  The most expensive part is the arduino at $65.  If you ever felt the need to also cycle the AC power the circuit could be very easily modified to switch SSR's or Mechanical relays.


Title: Re: remotely controlled power switch?
Post by: CubedRoot on August 21, 2011, 04:20:47 PM
Heres a power switch that has its own web interface that lets you control each power outlet individually.

It has 4 outlets, and a LAN port for network.  You just setup a port forward on your router to this device, and you can login to the power switch from outside the network (like from work, hotel, or a hotspot).

Amazon.com is selling it for around $74, which I think is a great deal!!

Heres the link: http://amzn.to/lhspnX


Title: Re: remotely controlled power switch?
Post by: thirdlight on August 21, 2011, 06:04:42 PM
I have this & it works well. Rated at 6A per outlet (10A overall). It does get quite hot, and the web interface only works 100% with IE. And that price is not far off half of what I paid!  ;D


Title: Re: remotely controlled power switch?
Post by: stellan0r on August 21, 2011, 06:23:53 PM
in Germany we have these babies, have to try out one:

http://www.anel-elektronik.de/deutsch/Produkte/produkte.html (http://www.anel-elektronik.de/deutsch/Produkte/produkte.html)
http://shop.varia-store.com/product_info.php?language=en&info=p990_ANEL-Elektronik-NET-PwrCtrl-POWER.html&XTCsid=00d20aa61240065c2e58471667a61cee (http://shop.varia-store.com/product_info.php?language=en&info=p990_ANEL-Elektronik-NET-PwrCtrl-POWER.html&XTCsid=00d20aa61240065c2e58471667a61cee)

http://www.gude.info/index.php?lng=1&section=products&product=epc1200 (http://www.gude.info/index.php?lng=1&section=products&product=epc1200)
http://www.gude-shop.de/7758_Expert_Power_Control_NET_D4x_SchukoDer_4fach_Remote_Power_Switch_mit_integriertem_Strommesser_fuer_TCPIPNetzwerke/dd0693ad49c2800d478eb6fa9bf4b846/ (http://www.gude-shop.de/7758_Expert_Power_Control_NET_D4x_SchukoDer_4fach_Remote_Power_Switch_mit_integriertem_Strommesser_fuer_TCPIPNetzwerke/dd0693ad49c2800d478eb6fa9bf4b846/)



Title: Re: remotely controlled power switch?
Post by: cicada on August 21, 2011, 06:36:54 PM
I have this & it works well. Rated at 6A per outlet (10A overall). It does get quite hot, and the web interface only works 100% with IE. And that price is not far off half of what I paid!  ;D

That's pretty cool.  10A is just short of what I need though, I'd have to get two...

[edit] actually, my rigs pull about 6.5A.. won't work for me :(


Title: Re: remotely controlled power switch?
Post by: kirax on August 21, 2011, 11:25:17 PM
A friend and myself are looking at a solution to this, so far it looks like it connects to a computer via USB, wires to up to 14 (!) computer power switches, and supports a keep alive on the computer it is wired to via USB in case it locks up, and will reboot that one as well. Price ranges so far looking like it could be around $50 USD (Or about the same BTC) + shipping from here in Canada. Other possible options include custom cases for it, like mine will likely be in a grey skull so I can make bad jokes about "by the power of greyskull!" We are working out details and software right now, let me know if people are interested. 3D printers are awesome. Send me a PM if you have interest :p


Title: Re: remotely controlled power switch?
Post by: cicada on August 25, 2011, 09:46:35 PM

I was finally able to really research this some more and it seems like it's pretty trivial.

I'm looking at building mine this weekend with a Jennic ZigBee IC I happen to have.. it's way overkill for this but hey miners are good at utilizing whatever happens to be around in a pinch :D

I'm almost convinced that this could be done with serial cable + transistors alone just by pulling a serial pin high to trigger the transistor.  ~$5 in parts ;)


Title: Re: remotely controlled power switch?
Post by: Swishercutter on August 26, 2011, 02:10:46 AM

I was finally able to really research this some more and it seems like it's pretty trivial.

I'm looking at building mine this weekend with a Jennic ZigBee IC I happen to have.. it's way overkill for this but hey miners are good at utilizing whatever happens to be around in a pinch :D

I'm almost convinced that this could be done with serial cable + transistors alone just by pulling a serial pin high to trigger the transistor.  ~$5 in parts ;)

This would work fine...build a serial cable and set it up.  It worked fine with an arduino and transistors so there is no reason a serial cable would not.  Be sure to connect a common ground thats about the only thing that I could see throwing things off.  The MSI boards come with some nice connectors for the board pins, you can actually hook the transistors directly to the connector and then plug it in without having to worry about mis-pinning (connector will only go on one way).  I'll take some pics of what I built so people can see...I should probably post my arduino/processing sketch for my gui control also, been busy and haven't got around to doing a writeup.


Title: Re: remotely controlled power switch?
Post by: mrb on August 26, 2011, 08:26:53 AM
I designed a USB device that can remotely power cycle 45 computers. Total cost in parts is about $100, or $2.20 per computer. This is much more computers and much less expensive than any commercial solution. It is based on an AVR microcontroller, and SSRs connected to the power switches via standard cheap cat5 cables.

Watch my blog for a future announcement. I am probably going to sell kits.


Title: Re: remotely controlled power switch?
Post by: Swishercutter on August 26, 2011, 11:16:30 AM
I designed a USB device that can remotely power cycle 45 computers. Total cost in parts is about $100, or $2.20 per computer. This is much more computers and much less expensive than any commercial solution. It is based on an AVR microcontroller, and SSRs connected to the power switches via standard cheap cat5 cables.

Watch my blog for a future announcement. I am probably going to sell kits.

You don't need SSR's for the  front panel switch...its just DC unless you are switching the supply in which case I don't think the necessary Triacs/Optotriacs can be purchased for 2.20/unit.  AVR's are cheap too though.

One arduino mega would switch 54 (also avr based). More if you did some sort of multiplexing.

At any rate simple transistors are fine.


Title: Re: remotely controlled power switch?
Post by: abracadabra on August 26, 2011, 03:05:34 PM
Been looking into this for a few days

http://forum.sparkfun.com/viewtopic.php?f=14&t=12563&start=0
http://dataprobe.com/ipio8-web-relay-control.php
http://www.tekkies.co.uk/v2/index.php?option=com_content&view=article&id=24&Itemid=4
http://tuxgraphics.org/electronics/200606/article06061.shtml


I like this one:

http://tuxgraphics.org/electronics/200911/home-automation.shtml





Title: Re: remotely controlled power switch?
Post by: mrb on August 26, 2011, 03:46:34 PM
You don't need SSR's for the  front panel switch...its just DC unless you are switching the supply in which case I don't think the necessary Triacs/Optotriacs can be purchased for 2.20/unit.  AVR's are cheap too though.

One arduino mega would switch 54 (also avr based). More if you did some sort of multiplexing.

At any rate simple transistors are fine.

4 things:
- I want SSRs because I want the electrical circuits to be isolated, for safety. A motherboard or PSU going up in smoke and sending voltage spikes through the power switch wires shouldn't fry my AVR or the 44 other remote-controlled computers...
- If using transistors, the user would have to know which pin of the power switch is + and which is -, increasing setup complexity for non-technical users.
- My SSRs are cheap ($0.75 in my quantities)
- One Arduino Mega would be twice the cost of my AVR board (about $50 vs. $25)


Title: Re: remotely controlled power switch?
Post by: cicada on August 26, 2011, 03:51:20 PM
- My SSRs are cheap ($0.75 in my quantities)

What SSRs are you getting so cheap?  When I was looking for these it seemed impossible to find something that could handle >6A for any reasonable price..  how much load are yours rated for on the AC side?


Title: Re: remotely controlled power switch?
Post by: mrb on August 26, 2011, 04:00:00 PM
- My SSRs are cheap ($0.75 in my quantities)

What SSRs are you getting so cheap?  When I was looking for these it seemed impossible to find something that could handle >6A for any reasonable price..  how much load are yours rated for on the AC side?

They are not on the AC side. They control the ATX motherboard power switch (so 5VDC).


Title: Re: remotely controlled power switch?
Post by: Jezzz on August 26, 2011, 04:11:22 PM
Been looking into this for a few days

http://forum.sparkfun.com/viewtopic.php?f=14&t=12563&start=0
http://dataprobe.com/ipio8-web-relay-control.php
http://www.tekkies.co.uk/v2/index.php?option=com_content&view=article&id=24&Itemid=4
http://tuxgraphics.org/electronics/200606/article06061.shtml


I like this one:

http://tuxgraphics.org/electronics/200911/home-automation.shtml


Pfft.  I couldn't disagree more.


Title: Re: remotely controlled power switch?
Post by: cicada on August 26, 2011, 04:12:04 PM
They are not on the AC side. They control the ATX motherboard power switch (so 5VDC).

Ahh, sorry misunderstood since you were talking about isolation.  There's really little danger of voltage spikes over the power switch pin - when was the last time you saw or heard of a fried PC power button? :P

A diode on the transistor base pin would keep it pretty safe I think, but diode+transistor is pretty similarly priced to your SSRs unless you're buying them in bulk anyway.

Good luck to you, it may be slightly overengineered for the purposes here, but I wouldn't dare kit whatever I'm about to cobble together here with transistors, you're in a better spot for it ;)


Title: Re: remotely controlled power switch?
Post by: abracadabra on August 26, 2011, 04:17:12 PM
Been looking into this for a few days

http://forum.sparkfun.com/viewtopic.php?f=14&t=12563&start=0
http://dataprobe.com/ipio8-web-relay-control.php
http://www.tekkies.co.uk/v2/index.php?option=com_content&view=article&id=24&Itemid=4
http://tuxgraphics.org/electronics/200606/article06061.shtml


I like this one:

http://tuxgraphics.org/electronics/200911/home-automation.shtml


Pfft.  I couldn't disagree more.

You got any better ideas :p ;)


Title: Re: remotely controlled power switch?
Post by: mrb on August 26, 2011, 04:29:48 PM
They are not on the AC side. They control the ATX motherboard power switch (so 5VDC).

Ahh, sorry misunderstood since you were talking about isolation.  There's really little danger of voltage spikes over the power switch pin - when was the last time you saw or heard of a fried PC power button? :P

A diode on the transistor base pin would keep it pretty safe I think, but diode+transistor is pretty similarly priced to your SSRs unless you're buying them in bulk anyway.

Good luck to you, it may be slightly overengineered for the purposes here, but I wouldn't dare kit whatever I'm about to cobble together here with transistors, you're in a better spot for it ;)

Power switches don't fry. But it doesn't mean the 5V standby voltage they are connected to remains perfectly stable or safe, especially given the power supplies failures reported in this community...

If you can have safety and user-friendliness for $0.75, take it :)


Title: Re: remotely controlled power switch?
Post by: Swishercutter on August 26, 2011, 09:04:52 PM
About those 4 things:
- I want SSRs because I want the electrical circuits to be isolated, for safety. A motherboard or PSU going up in smoke and sending voltage spikes through the power switch wires shouldn't fry my AVR or the 44 other remote-controlled computers...

You can use any optoisolator, doesn't have to be an optotriac.   

- If using transistors, the user would have to know which pin of the power switch is + and which is -, increasing setup complexity for non-technical users.

If they cannot figure out which pin is the pos/neg on the switch they probably shouldn't be trying to pin anything to the Mobo, took me about 2 seconds.  Plus, I checked my MSI vs. an Asrock and they had the exact same pinout for the Mobo power switch pinning...not saying all are, just the 2 I checked.

- My SSRs are cheap ($0.75 in my quantities)
Sounds like you are buying the optotriacs...small components of SSR's that are use to trigger the bigger Triacs in a SSR circuit.  If you were building an SSR that would switch the PSU current you would be looking at around $5/SSR minus the board (and snubber).

- One Arduino Mega would be twice the cost of my AVR board (about $50 vs. $25)
Yep, probably.  But you can use it for other things if you no longer need a resetter. 



Also, have you tested this.  Every time I try to run DC with a Triac it will turn on but it will stick because it requires a negative voltage swing to trigger turn off.  So basically you will trigger it and it will hold the power on.

A quick google search verified what I thought.  Triacs (SSR's) do not properly trigger DC.
http://answers.yahoo.com/question/index?qid=20081024022745AAwQ8nF
http://www.electronicspoint.com/triac-optocoupler-moc3043-switch-12v-dc-t11292.html

So looks like you will have to either use small relays with transistor drivers, DC optoisolators, or just transistors as I was saying before.


Title: Re: remotely controlled power switch?
Post by: kirax on August 26, 2011, 10:11:01 PM
About those 4 things:
- I want SSRs because I want the electrical circuits to be isolated, for safety. A motherboard or PSU going up in smoke and sending voltage spikes through the power switch wires shouldn't fry my AVR or the 44 other remote-controlled computers...

You can use any optoisolator, doesn't have to be an optotriac.   

- If using transistors, the user would have to know which pin of the power switch is + and which is -, increasing setup complexity for non-technical users.

If they cannot figure out which pin is the pos/neg on the switch they probably shouldn't be trying to pin anything to the Mobo, took me about 2 seconds.  Plus, I checked my MSI vs. an Asrock and they had the exact same pinout for the Mobo power switch pinning...not saying all are, just the 2 I checked.

- My SSRs are cheap ($0.75 in my quantities)
Sounds like you are buying the optotriacs...small components of SSR's that are use to trigger the bigger Triacs in a SSR circuit.  If you were building an SSR that would switch the PSU current you would be looking at around $5/SSR minus the board (and snubber).

- One Arduino Mega would be twice the cost of my AVR board (about $50 vs. $25)
Yep, probably.  But you can use it for other things if you no longer need a resetter. 



Also, have you tested this.  Every time I try to run DC with a Triac it will turn on but it will stick because it requires a negative voltage swing to trigger turn off.  So basically you will trigger it and it will hold the power on.

A quick google search verified what I thought.  Triacs (SSR's) do not properly trigger DC.
http://answers.yahoo.com/question/index?qid=20081024022745AAwQ8nF
http://www.electronicspoint.com/triac-optocoupler-moc3043-switch-12v-dc-t11292.html

So looks like you will have to either use small relays with transistor drivers, DC optoisolators, or just transistors as I was saying before.

Reading circuit design and embedded hardware threads is good for me, it makes me remember how I keep meaning to learn more about the actual base electrical principles of the computers I work with so much... But haven't yet :p


Title: Re: remotely controlled power switch?
Post by: Swishercutter on August 26, 2011, 10:26:55 PM
About those 4 things:
- I want SSRs because I want the electrical circuits to be isolated, for safety. A motherboard or PSU going up in smoke and sending voltage spikes through the power switch wires shouldn't fry my AVR or the 44 other remote-controlled computers...

You can use any optoisolator, doesn't have to be an optotriac.   

- If using transistors, the user would have to know which pin of the power switch is + and which is -, increasing setup complexity for non-technical users.

If they cannot figure out which pin is the pos/neg on the switch they probably shouldn't be trying to pin anything to the Mobo, took me about 2 seconds.  Plus, I checked my MSI vs. an Asrock and they had the exact same pinout for the Mobo power switch pinning...not saying all are, just the 2 I checked.

- My SSRs are cheap ($0.75 in my quantities)
Sounds like you are buying the optotriacs...small components of SSR's that are use to trigger the bigger Triacs in a SSR circuit.  If you were building an SSR that would switch the PSU current you would be looking at around $5/SSR minus the board (and snubber).

- One Arduino Mega would be twice the cost of my AVR board (about $50 vs. $25)
Yep, probably.  But you can use it for other things if you no longer need a resetter. 



Also, have you tested this.  Every time I try to run DC with a Triac it will turn on but it will stick because it requires a negative voltage swing to trigger turn off.  So basically you will trigger it and it will hold the power on.

A quick google search verified what I thought.  Triacs (SSR's) do not properly trigger DC.
http://answers.yahoo.com/question/index?qid=20081024022745AAwQ8nF
http://www.electronicspoint.com/triac-optocoupler-moc3043-switch-12v-dc-t11292.html

So looks like you will have to either use small relays with transistor drivers, DC optoisolators, or just transistors as I was saying before.

Reading circuit design and embedded hardware threads is good for me, it makes me remember how I keep meaning to learn more about the actual base electrical principles of the computers I work with so much... But haven't yet :p

It all runs on smoke...never forget that.  Once you let the smoke out it doesn't work anymore. ;)


Title: Re: remotely controlled power switch?
Post by: Pipesnake on August 27, 2011, 03:20:51 AM
I designed a USB device that can remotely power cycle 45 computers. Total cost in parts is about $100, or $2.20 per computer. This is much more computers and much less expensive than any commercial solution. It is based on an AVR microcontroller, and SSRs connected to the power switches via standard cheap cat5 cables.

Watch my blog for a future announcement. I am probably going to sell kits.
ETA?

I'll buy a kit.


Title: Re: remotely controlled power switch?
Post by: BkkCoins on August 27, 2011, 04:06:27 AM
I don't know why you'd start a new board from scratch for this. I already posted above about the PIC-WEB board available for $29 euro. It has a TCP-IP stack built in with sample web server code, is easy to program with C tools and free PIC tools. It has a ready to use expansion connector that you could hang many SSR off of to control power switches. Seems to have about 28 pins there but I didn't check each one to see if there are special conditions on some.



Title: Re: remotely controlled power switch?
Post by: mrb on August 27, 2011, 05:18:46 AM
Pipesnake: ETA is mid-September.

BkkCoins: There is only 15 usable digital outputs on the PIC-WEB's EXT connector ($50 in the US). Plus you would have to design a separate board for the SSRs and their resistors ($15). In the end it would be twice more expensive per controllable computer ($65 for 15 computers) compared to my solution ($100 for 45 computers).

About those 4 things:
- I want SSRs because I want the electrical circuits to be isolated, for safety. A motherboard or PSU going up in smoke and sending voltage spikes through the power switch wires shouldn't fry my AVR or the 44 other remote-controlled computers...

You can use any optoisolator, doesn't have to be an optotriac.    

Find me, on Digikey, any type of opto-isolator comparable to my SSR: $0.75 or less for quantities of one hundred, and that operates with an input current of 1mA. I would be delighted to be shown there is one.

- If using transistors, the user would have to know which pin of the power switch is + and which is -, increasing setup complexity for non-technical users.
If they cannot figure out which pin is the pos/neg [...]

I may sell assembled kits and don't want to tell my clients "GTFO if you don't have a multimeter". I am like Steve Jobs, I care about ease of use :)

- My SSRs are cheap ($0.75 in my quantities)
Sounds like you are buying the optotriacs...small components of SSR's that are use to trigger the bigger Triacs in a SSR circuit.  If you were building an SSR that would switch the PSU current you would be looking at around $5/SSR minus the board (and snubber).

Nope. These are plain standard SSRs. Just go look them up on Digikey, there are plenty in this price range. Once again, I am controlling 5VDC, not 120VAC.

Also, have you tested this.  Every time I try to run DC with a Triac it will turn on but it will stick because it requires a negative voltage swing to trigger turn off.  So basically you will trigger it and it will hold the power on.

A quick google search verified what I thought.  Triacs (SSR's) do not properly trigger DC.
http://answers.yahoo.com/question/index?qid=20081024022745AAwQ8nF
http://www.electronicspoint.com/triac-optocoupler-moc3043-switch-12v-dc-t11292.html

So looks like you will have to either use small relays with transistor drivers, DC optoisolators, or just transistors as I was saying before.

I have not seen this pb. When I stop applying voltage, the SSR output circuit opens...


Title: Re: remotely controlled power switch?
Post by: Swishercutter on August 27, 2011, 07:17:18 AM
Pipesnake: ETA is mid-September.

BkkCoins: There is only 15 usable digital outputs on the PIC-WEB's EXT connector ($50 in the US). Plus you would have to design a separate board for the SSRs and their resistors ($15). In the end it would be twice more expensive per controllable computer ($65 for 15 computers) compared to my solution ($100 for 45 computers).

About those 4 things:
- I want SSRs because I want the electrical circuits to be isolated, for safety. A motherboard or PSU going up in smoke and sending voltage spikes through the power switch wires shouldn't fry my AVR or the 44 other remote-controlled computers...

You can use any optoisolator, doesn't have to be an optotriac.    

Find me, on Digikey, any type of opto-isolator comparable to my SSR: $0.75 or less for quantities of one hundred, and that operates with an input current of 1mA. I would be delighted to be shown there is one.

- If using transistors, the user would have to know which pin of the power switch is + and which is -, increasing setup complexity for non-technical users.
If they cannot figure out which pin is the pos/neg [...]


I may sell assembled kits and don't want to tell my clients "GTFO if you don't have a multimeter". I am like Steve Jobs, I care about ease of use :)

- My SSRs are cheap ($0.75 in my quantities)
Sounds like you are buying the optotriacs...small components of SSR's that are use to trigger the bigger Triacs in a SSR circuit.  If you were building an SSR that would switch the PSU current you would be looking at around $5/SSR minus the board (and snubber).

Nope. These are plain standard SSRs. Just go look them up on Digikey, there are plenty in this price range. Once again, I am controlling 5VDC, not 120VAC.

Also, have you tested this.  Every time I try to run DC with a Triac it will turn on but it will stick because it requires a negative voltage swing to trigger turn off.  So basically you will trigger it and it will hold the power on.

A quick google search verified what I thought.  Triacs (SSR's) do not properly trigger DC.
http://answers.yahoo.com/question/index?qid=20081024022745AAwQ8nF
http://www.electronicspoint.com/triac-optocoupler-moc3043-switch-12v-dc-t11292.html

So looks like you will have to either use small relays with transistor drivers, DC optoisolators, or just transistors as I was saying before.

I have not seen this pb. When I stop applying voltage, the SSR output circuit opens...



If it works for you it works then...build it, sell it.  I was just offering solutions which are easy and cheap I am not trying to sell kits, but I am planning an Instructable on the topic.  You could make a simple connector which has to be pinned in just like if you were pinning a case switch, my point is that if someone cannot properly pin a Mobo power switch they probably shouldn't be trying to setup a multi computer remote reset. A multimeter should be in every large scale miners possession...if not you do not have the proper tools.  They are like 5 bucks for a cheapo one at Harbor freight that would be good enough for checking the pins.  If you actually buy a decent one it becomes more useful for things like checking PSU voltages and other things.

As far as difficulty finding which pin on the Mobo pinout is the power switch(-) its the one of the 2 that is ground...that easy, all the power switch does is ground a +5v pin.  You can test this with the computer off, just connect one lead of the multimeter to any mobo screw and touch the other lead to the power switch pins til it reads 0 ohms and that is the switch negative.

Which part number "SSR" are you using.  An optotriac (which is a small ssr in itself but is commonly used for driving large triacs) like the MOC3023 runs .33c per unit at the single unit price.  Something like the MOC8016 should work (didn't look too hard at it but I think it would work) is $5.60/10 units...but you still have the polarity issue.  If you sink the drive current to the AVR (active low) from the optoisolator the current to drive it shouldn't be an issue.