Isolating transducers from the PICAXE

cravenhaven

Senior Member
Having just destroyed my LCD display because of a meltdown in my external current sensor, I am looking for ways to fully isolate the sensor from the rest of the circuit. The problem was caused by the sensor sending 12v onto the 5v feed line which is of course the same 5v that feeds the PICAXE and LCD. The problem is that the output of the sensor is an analogue voltage that is proportional to the current sensed and voltage applied.
The current sensor is the ACS714 from Allegro: http://www.allegromicro.com/Products/Current-Sensor-ICs/Zero-To-Fifty-Amp-Integrated-Conductor-Sensor-ICs/ACS714.aspx
I managed to damage the sensor by attempting to incorrectly attach a second battery to my main battery, and the resultant current spike melted the sensor and hey presto.

My plan so far is to isolate the sensor input to the PICAXE by using a voltage follower opamp, but how do I provide 5v to the sensor that is exactly the same level as that provided to the PICAXE and LCD without having it directly connected?.


Untitled.jpg
 

Dippy

Moderator
It would be a good idea to post the circuit you used. I can't see how you did this if you followed the Data Sheet example.

That chip is already isolated so I can't imagine anything other than shaky-fingers or carelessness (in the nicest possible way) causing your meltdown.

If it was overcurrent that caused the internals to melt then you should fuse the High Current (12V??) side.

Is it the 5V supply to ACS714 and the Analogue out that you want to isolate?
i.e. did this spike shoot up the supply or signal line? (Or both?)
Isolated DC-DC supplies are available widely.

I'm currently (no pun) designing an ATE rig which measures current for 400V 3phase and I rely totally on the device isolation.
Even though it's a different sensor device , you're getting me worried :)
In my design I have to isolate for V measurements but not for current.
 

cravenhaven

Senior Member
I think it was overcurrent melted the sensor and in the process connected the 12v from the measured circuit onto the 5v sensor supply.
The circuit monitors my camper battery charge and drain and had been operating successfully for some months and was not being fiddled with in any way.
I was reconnecting another battery but accidentally tried to connect it around the wrong way (+ to - and - to +) and the huge current surge melted the sensor chip. The end result was that the 12v path became disconnected but a 12v-5v path was created.
You are quite right that I need to fuse the high current side and that is exactly in the plan, but as you can see, the device isolation is only effective while the chip is intact.

batmonv3.jpg
 
Last edited:

Dippy

Moderator
Speaking from a distance it's difficult to nail this, but here are a few suggestions.
It sounds like you need to protect the DC line as well as the signal.

1. Buy a bigger Current sensor that can handle carelessness :) Fit a fuse.
2. Use an isolated DC-DC module to power ACS. I use one made by XP power in my design. Add a buffer plus V/F then optocoupler to PICAXE
3. Protect the 5V supply to ACS using a fat zener (or similar) plus tiddly fast fuse. The signal line only needs a resistor and you can clamp the max. with a diode to +5V. This limits signal at PICAXE ADC to a nominal 5V and the resistor limits the current.

Or maybe something like this?
http://uk.farnell.com/avago-technologies/hcpl-7510-000e/op-amp-isolation-ic/dp/1085055

In your case your buffers should take the strain for the signal overload, so just limit/isolate the 5V supply.

PS. It would really help if you numbered the ACS pins correctly, rather than just your connection numbers.
 

Goeytex

Senior Member
Why is it important to have the supply "exactly" the same as Picaxe and LCD?. A few millivolts difference between supplies can be compensated for.

However, since the failure was caused by an operator mishap and will not likely happen in normal operation, I would look to put overvoltage/overcurrent protection on the supply rail instead of trying to completely isolate the sensor.
 

SAborn

Senior Member
I have used the alleros for some time now with outstanding results, and never had a problem (but also never tried to dump several hundred amps through them in reverse)

Nothing is totally idiot proof (i mean that nicely) and as Dippy has said a fuse is your best option, if its a unidirectional allegro than maybe a large diode in the 12 volt line, or even a 220R resistor in the 5 volt supply should work, and if you have a 5 volt reg in circuit then the reg should control the over voltage, should you make the same mistake again.
 

cravenhaven

Senior Member
It sounds like you need to protect the DC line as well as the signal.
That's the reason for my post, I think I have the signal line protected now though a suggestion as to which OpAmp to use would be welcomed. The link to the Avago chip is appreciated but I dont think I need kV protection so probably the zener/resistor front ending a rail-to-rail opamp will be the go.

Why is it important to have the supply "exactly" the same as Picaxe and LCD?
The output of the sensor is a ratio of the supply voltage, just as the ADC reading on the picaxe is a ratio of the supply voltage. It seems to me that if the supplies are independent then the errors in the readings could become significant and unable to be compensated for.

I would look to put overvoltage/overcurrent protection on the supply rail instead of trying to completely isolate the sensor.
That is what I am looking for. The problem as I see it is that the protection must not alter the actual voltage, or at least it needs to be predictable.

a 220R resistor in the 5 volt supply should work
I looked at the ACS chip specs and they quote a 10mA current draw, so to minimise the voltage drop wouldnt 220R be too high?.

and if you have a 5 volt reg in circuit then the reg should control the over voltage,
I do have a 5v regulator (LM2940) and it is additionally protected by a 100R resistor in the 12v power line. This resistor got quite hot during the 'accident'.
 
I would look to put overvoltage/overcurrent protection on the supply rail instead of trying to completely isolate the sensor.[/I]
That is what I am looking for. The problem as I see it is that the protection must not alter the actual voltage, or at least it needs to be predictable.
Check this overvoltage protection circuit: http://en.wikipedia.org/wiki/Crowbar_(circuit).
Just be sure the power used has enough capacity to blow the fuse, - if not, the TRIAC might act as fuse.
 

Reloadron

Senior Member
You may want to give this a read as to analog isolation. However, I fail to understand how this happened. The current sensor you were using is a hall effect sensor. The sensor is powered by 5 volts. This can be the same 5 volts that powers your uC as well as the same 5 volts that serves as Vref for the uC. The sensed current path through the sensor is an isolated path, there is no galvanic connection. Therefore there is no way that the sensor, when correctly powered can output greater than the 5 volt rail powering it. The link for analog isolation mentions but a few of the dozens of ICs out there designed for the purpose but in your application it should not be required.

<EDIT> If the sensor had a total melt down then all bets are off. :) I read how it happened with a sudden polarity reverse. </EDIT>

Ron
 

inglewoodpete

Senior Member
Unless you are looking for the "perfect" circuit, apart from adding a fairly hefty fuse, it may be simpler to make a spare board and carry it with you. I think you'd admit that the mistake you made is not one you'll repeat in a hurry.

And remember that two LA batterys (or just one) in a short circuit can explode and shower everything with suphuric acid. It happened to a work colleague of mine: luckily, he had a garden hose nearby but still got nasty burns.
 

SAborn

Senior Member
The cruel reality is, there is only one way to make something idiot proof.
Remove the idiots.

(the comment is not meant personally, just a general fact)
 

cravenhaven

Senior Member
I think we all have to recognise that mistakes happen regardless of how clever and careful you may be. My quest was to find a way to prevent damage to my circuit from external causes no matter how they are caused. If one of the many components within the camper should fail dramatically, or the insulation on a wire should wear through and short out, will the fuse blow in time to protect the electronic components after all it is just a bit of wire that melts when it gets hot. Much like a hall effect current sensor really!!.

I read in a very old post on this forum that when building circuits to be used in automotive applications one should ensure that every external connection has a fuse on it. That may well have saved my circuit, however it seems that there may be other ways to protect a circuit from external faults.

Inputs can be protected by sacrificial opamps or clamping diodes, but I'm still not sure how to effectively protect the power supply when the sensor requires power from the circuit. I guess I will resort to a separate power supply and try to compensate for the voltage variations.
 

cravenhaven

Senior Member
I think we all have to recognise that mistakes happen regardless of how clever and careful you may be. My quest was to find a way to prevent damage to my circuit from external causes no matter how they are caused. If one of the many components within the camper should fail dramatically, or the insulation on a wire should wear through and short out, will the fuse blow in time to protect the electronic components after all it is just a bit of wire that melts when it gets hot. Much like a hall effect current sensor really!!.

I read in a very old post on this forum that when building circuits to be used in automotive applications one should ensure that every external connection has a fuse on it. That may well have saved my circuit, however it seems that there may be other ways to protect a circuit from external faults.

Inputs can be protected by sacrificial opamps or clamping diodes, but I'm still not sure how to effectively protect the power supply when the sensor requires power from the circuit. I guess I will resort to a separate power supply and try to compensate for the voltage variations.
 

Dippy

Moderator
Yes, we can all make mistakes and failures elsewhere can happen - hence FMEA :)
I think "Idiot Proof" is a little strong.


There is no reason why you shouldn't have a separate 5V reg designated to power the ACS714.
As you say; any slight variation can be compensated for with a bit of arithmetic.

It may be an idea to protect your overall PICAXE etc. circuit with a diode in the supply to prevent damage to due to polarity reversal.
A fully isolated DC-DC for the whole PICAXE etc circuit would cost about a tenner as an option.


As far as protection goes, it is the designer's responsibility to cover every angle.
Sure enough, you couldn't have envisaged the Failure Mode of this device without specific testing and hence why manufacturers do exactly this for FMEA.
But it should be pretty obvious that a small package like this cannot handle huge currents.


I see that the manliest of the ACS714 range is rated +/-30Amps. This would hint at the use of a 40Amp fuse.
I'm surprised your PCB tracks didn't burn before the chip.
I assume you didn't have a PCB failure which caused the power to jump?
This, of course, assumes that this wasn't a genuine failure of a semi-duff part.

Maybe if you envisage higher currents (as worst case) you should have chosen a beefier package.
As I mentioned before you could also fit some crowbar protection plus fuse/CB on the power line to the ACS.
Is all this gubbins in a sturdy box?

Anyway, lesson learnt and good luck with the Mk2.
 

Captain Haddock

Senior Member
No matter how well you 'idiot proof' something sooner or later Darwins theory of evolution will throw out a better grade of idiot that hadn't been catered for.
Occaisional failiures are inevitable, good luck with the new version.
 

Goeytex

Senior Member
.. My quest was to find a way to prevent damage to my circuit from external causes no matter how they are caused.
A noble but daunting task since it covers everything from initial testing during R & D to final product. From user error to lightening strikes. There are so many things to consider and allow for that the time and cost will be quite high.

If one of the many components within the camper should fail dramatically, or the insulation on a wire should wear through and short out, will the fuse blow in time to protect the electronic components (?) ... after all it is just a bit of wire that melts when it gets hot. Much like a hall effect current sensor really!!.
Wires can be protected by using good wiring practice, such as wire protectors, grommets, etc. This does not necessarily have to be done with extra protection circuitry. A fast acting fuse combined with a crowbar type circuit or a TVS diode should be more than adequate. A polarized connector would have prevented your mishap and any future mishaps of that type. Adding sacrificial op-amps, an additional supply, and completely isolating the sensor seems a bit of overkill. (Unless of course you are flying the camper to the moon, and in that case you will need triple or quadruple redundancy as well.)

I read in a very old post on this forum that when building circuits to be used in automotive applications one should ensure that every external connection has a fuse on it. That may well have saved my circuit, however it seems that there may be other ways to protect a circuit from external faults.
And you probably also read here that in automotive applications automotive grade regulators should be used in conjunction with TVS diodes, etc. ..... Yet it seems you did not take any of this advise to heart as far a your project goes.

There are lots of ways to protect circuitry , no doubt. But how much time and money do you want to spend to "prevent damage to my circuit from external causes .... no matter how they are caused." ??.

I would suggest that you may be overreacting a bit to your unfortunate mishap. Before pursuing the "other ways" that you have eluded to, I think you would do well to consider standard and accepted practice, which will cover the vast majority of potential problems.
 
Last edited:
Top