LIMIT ADC INPUT WITH ZENER DIODE ??

[MRSSPOCK]

Hi.

I have a project involving a load cell.

The actual load to be measured is about 20kg, however in order to provide some mechanical protection (i.e. to prevent the load cell getting loaded past its elastic limit, due to operator error), I decided to install a 50kg load cell instead.

The load cell has a 10V excitation voltage, and 2mV per V output.

Since I have already sacrificed some resolution by opting for a 50kg rather than a 20kg device, I want to have the input to my 5V ADC making the most of what resolution I have left. (i.e. I DON'T want to amplify the signal very slightly to keep the output within a safe value for my 5V ADC, and thereby losing any of the ADC resolution available).

To cut a long story short, I want to amplify my signal so that with a full load of 20kg applied, I have 5V on my 5V ADC input pin.

I want to amplify my input signal of ((20kg / 50kg) * 2mV per V) of 0.008V by 625 = 5V at full 20kg load

Since in reality the load cell has a safe overload of 150% and a ultimate overload of 200%, in the worst possible case the ADC would see 25V with the suggested amplifier arrangement! (if the amplifier was capable of outputting 25V, which it isn't, but yet it can easily still output more than 5V).

Nevertheless, my question is, can I simply include a 5V zener diode on the input to the ADC, so that any spikes above 5V would be clipped off?

It seems too simple a solution, so I'm guessing I can't do it?

Would it mess up the linear behaviour of the load cell output amplified signal?

If so, is there an easy way to provide protection for the ADC input?

It can only tolerate 5.5V

Thanks

 

[benta]

Just use a single-supply rail-to-rail opamp powered at 5 V. Much simpler. The Zener diode idea is not optimal. For overvoltage protection, you'd normally divert to the positive rail using a diode.

 

[MRSSPOCK]

Okay Benta.

I will have a look at that.

Could you please point to a link showing an image of such, if such is readily available?

I will have a look also, but just in case I misinterpret what you mean.

Thanks

 

[MRSSPOCK]

@benta Sorry, I should have added, I already have a Burr Brown instrumentation amplifier, the output from which I would like to limit. Does your suggestion for the OP AMP intend to do away with my amplifier completely? This Burr Brown amp has the reference voltage provided to supply the bridge. I would like to make use of this device if I can.

 

[itsmoked]

I'm not following the 5V op-amp suggestion. I can't see it.


But, I certainly agree with using a diode from the input signal to the 5V rail. This will limit the input voltage to 5.7V which will likely not actually hurt the input. If you're chicken to do that then use a Schottky diode as they have only ~0.3V forward limiting the input to 5.3V. They can be slightly problematic since they leak reverse current which varies greatly with temperature.

This is sort what we're describing:

Keith Cress
kcress -

http://www.flaminsystems.com

 

[MRSSPOCK]

@itsmoked Thanks.

I will give that a try later.

 

[benta]

"I'm not following the 5V op-amp suggestion. I can't see it."

Well, given that the OP was very selective about describing the circuit, instead of giving complete information (like the Burr-Brown opamp/reference), the answers are correspondingly useless.

 

[danw2]

FYI, strain gauge signal conditioners are commericially available:

Link

 

[btrueblood]

I once had a similar problem, but with temperature readings, where I wanted to be able to measure high readings, but wanted fine resolution of the lower, in-control readings.

I fixed the issue by incorporating two precision voltage references, operated by some spare microcontroller outputs. The high reference voltage was commanded "on", and fed to the ADC reference voltage pin when necessary to make the high readings. Once the readings were down to the lower range, the ADC reference was switched to a lower value allowing the same ADC to provide higher resolution measurements.

Dunno if that helps, Mrs. Spock. But logical it is...

 

[Compositepro]

It would have been a better approach to choose a load cell with the desired range that has mechanical stops inside for overload protection. This is simpler and provides much more robust overload protection. Are you sure that the load cell that you have does not already have this feature, and therefore your concerns are unnecessary?

 

[Sparweb]

What about the ADC's resolution? At 10-bits you only get 1024 levels. At 16-bits you can resolve 65,536 levels. I have a load cell with a 14-bits and it's plenty.
It also doesn't have an amplifier. Instead it has a carefully programmed statistical sampling software on-board.


STF

 

[MacGyverS2000]

But at 16-bits, you can lose the three to four lower bits of precision without hardly trying... a poor board layout can do that in a heartbeat.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

The HX711 is an example of a bespoke Load Cell Amplifier. It has a programmable gain stage, and 24-bits resolution. With those specs, you could (perhaps) just skip past the whole issue.

 

[MRSSPOCK]

Thanks for all your replies.

I'm only getting back to this now.

Regarding D2 in the image below, am I correct to presume it is an option?

I mean, am I correct to presume that D1 on its own, (along with the resistor), is adequate to protect against a positive voltage spike?

If I am wrong to think that, could someone please explain the function of D2.

I don't understand it.

Thanks

 

[itsmoked]

D2 is indeed for negative excursions. If you can have excessive positive spikes usually negative ones are also possible.

Most integrated semiconductor devices have protection diodes like the ones in the above circuit. Check the data sheets. If they are present then you don't need external ones as long as R1 (above) is large enough to limit the current to below what the internal diodes are rated for.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[MRSSPOCK]

@itsmoked

Thanks for the clarification.

Since the positive spike I am predicting will happen, will in reality be as a result of excess load on the load cell, (beyond its planned window of operation), I suppose I am probably using the word "spike" incorrectly, as like you said, it suggests a negative spike might also be in the vicinity. I'm quite confident there won't be a negative spike since this signal is the output from my INA125 amplifier, and as I have it wired, it can't output a negative.

Your solution is a nice simple modification for me to add on, since the circuit is already finished.

It may well be that the excess load never happens, but its good to just make provision and forget about it.

Thanks again.