Some of that can be mitigated by power loss to the Pi dropping every thing to off. ie you need to force things on.
A hardware timer (555 or WHY) that has a high output to enable the control lines that has to be regulary reset by the Pi. If it doesn't get reset and times out everything is dropped off. The reset needs to see a transition (lo to hi) or a short high pulse of no longer than a given duration and both edges detetected.
Mains detection. There is a chip Allegro Microsystems ACS712ELCTR,
2.1 kV isolation, produces a voltage output proportional to the current flow. Available in 5, 20 and 30 A versions.
True enough but "control failure" ought to include "fail safe", ie things have to be forced on. Yes the Pi could crash and keep the GPIOs on, that's where the hardware timer comes in.
Don't you need more than one, output of safety contactor and feed to kiln?
Also that is only an on/off indication. The Allegro chip would enable you to monitor the currentdrawn and thus power consumed and maybe detect overload with more sensitivity than normal protective devices.
As I said somewhere up this thread - I'm aiming for an improved level of safety, but not quite up to nuclear power station (or railway signalling) standards..
The kiln runs unattended most of the time - I'm just looking to trap the most likely failure mode - where the commercial controller gets 'confused' and calls for heat over & above what's required, or where the main switching relay to the heating elements fails 'short'.
I did this (mutter) years ago with three dil relays - two of which had big electrolytics across them, and their n/o contacts wired in series. The third c/o relay was pulsed on/off by the controller (back in those days a completely-inappropriately-specced Texas 9900 computer!) - and put 12v on each of the relays in turn. If the pulsing relay stopped, or stuck 'on' or 'off' then one of the relays would eventually drop out (time-delay provided by the big capacitor) and the 'safety' contacts would break - providing a 'the computer's fallen over' signal...
555 would do the same thing - but not as mechanically satisfying
Clever! Actually, I'm less concerned with the 'failure to heat' (i.e. element open-circuit) situation, as I am with the 'failure to stop heating' (= relay stuck on or thermocouple failure). 'Not heating' is a fail-safe situation - and isn't a big worry for me
Precisely . Even more years ago, I was involved with animal feed mills - auto-formulation & mixing / routeing - that sort of thing. Most of that problems we encountered in those applications were down to the heavy-duty hardware doing 'silly things' - and the code eventually became full of 'idiot-checks' - based on real-life scenarios.
Same with the automated fork-truck systems. Things like the truck knowing which aisle it was in (in the high-bay warehouse) by measuring the length of a reflective strip stuck to the racking. Worked well until the driving wheel skidded on some fool's banana skin (really!) and the truck tried to turn right into the next aisle - except it was _already_ in the last aisle.
Anyway - as I say, this isn't NASA - just trying to make the kit reasonably safe, and know that it can fail 'gracefully' if it decides it doesn't like the information it's seeing.
I'm not over-bothered about the actual feed to the kiln. If there's volts there then the kiln will heat up and all is good. If there isn't volts then it won't heat, and I could spot that in the 'safety' software, or wait until the kiln controller notices and complains. Either way - it's safe.
True. But it's not really a big deal at the moment.. The most likely / critical failure of the kiln control is the 'heat' relay being stuck 'on'... as that can lead to unfortunate things happening (and, in the worst case, my workshop becoming a big bonfire!) That's why I'd like the safety monitoring circuit to know that it _could_ drop out the big relay if it felt so inclined...
Thanks for the perspectives - it's all useful Adrian
d will halve its life, very roughly, causing failure.
the optoisolator would be far smaller, but drive it with series R&C not jus t C. Pick R to limit peak i, then C to control mean i.
If your plan is to switch it twice instead of once, it will last something like half as long. Ie your scheme, if that's how it works, would cause fail ure.
900 times is a lot. An SSR wouldn't care. Another way to boost reliability could be to split the heating load so the rate of temp climb is much slower when it's thermostatically cycling. It also means less current being switc hed.
I would not be happy using software & active electronics as basic safety pr otection. I don't know why you don't just connect the power through a bit o f metal inside the kiln that melts if it goes above working temp. Then if i t's running a reducing atmosphere your objects could come out brass plated :)
Are you going to use this in or attached to or anywhere near your house? If so you're mad to build a fire waiting to happen. Other people's madness is no defence.
Just stick your brass strip at the right height in the kiln so its mp is reached when the main body of the kiln is at 950.
The Allgero chip is £3 something plus VAT from Farnell. It's not an optoisolator but a hall effect device and the load current passes through it. Its output voltage is proportional to the current. Not easy to do with an opto-isolator.
Or are you refering to the £6.95 single channel ready to play optoisolated mains to 3.3 / 5 V on/off interface board on ebay?
Sort of. Under my scheme the safety contactor would be switched on/off in a 'self-test' mode once per firing schedule, and then once on and once off during the actual run.
Given that theoretically this doubles the wear on the contactor, but that it only gets switched twice per firing (which lasts about 7 hours, and takes place maybe 5 times per week) I can live with it.
Understood - but, as I said, I happen to like the click-clack of contactors - and I'm less confident with ssr's.
Another way to boost reliability could be to split the heating load so the rate of temp climb is much slower when it's thermostatically cycling. It also means less current being switched.
All good in theory - but irrelevant in this application, where it's important that the firing schedule consists of certain predetermined ramp/temperature/hold combinations...
Fair enough - you're welcome to your opinion. I've looked at the possibility of having a 'hardware only' safety facility - and I'd rather do it the way I describe. Thanks Adrian
What make and model of contactor is it? Many din-rail mounted contactors are designed to allow one or more auxiliary contacts to be added. They simply clip on at the top and are automatically operated by the plunger mechanism of the contactor.
Nothing wrong with software and active electronics - lots of industrial stuff does that. However a RPi is not the way to go for that. For an amateur, it'd need to be something like a PIC microcontroller, coded in assembler so that the code can be substantiated - simple code, with no interrupts, no parallel threads, completely deterministic, with watchdog and brown-out detection and fail-safe output.
ing like half as long. Ie your scheme, if that's how it works, would cause failure.
You'd be better off having no protection than what you describe, it would l ast twice as long before it fails.
I think you've misunderstood. Splitting the heating power makes no differen ce to the firing temperature profile. What it affects is how fast the heati ng cycles when the kiln is set to any fixed temperature. So by splittig the load your hardware can last far longer.
It's also safer since one contactor failure only applies part load, keeping teh kiln at a safe temp.
Understand where you're coming from - and the bit about 'keeping it simple'. If I was doing this as part of a commercial product I'd probably go that way (I was writing control systems in Texas 9900 /
99000 assembler back in the '90's) However - in this case, the kiln is currently running quite happily under it's own commercial controller, and the RPi is there, currently providing a monitoring function (logging time vs temperature, drawing a graph of same, counting/timing relay cycles, estimating firing costs.... and I was planning on adding the safety-monitoring to the overall program.
I do understand that an RPi will be doing 'other things' at the same time (this is part of the attraction as it allows remote monitoring via Remote Desktop and (eventually) web - which is why I was planning on using a hardware watchdog to keep an eye on the RPi.
Agreed that, in safety, the simpler the better - and I did look at using a completely hardware solution using a dedicated thermocouple amp ic switching a relay - but, in this case, the RPi's there and has more capabilities...
Appreciate your input - but, believe me, I've had all these discussions (with myself!) already Adrian
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.