6

When using load cells (or strain gages generically), is the Excitation Voltage spec provided as a strict minimum -- or is it just a recommendation which one can go below as long as one is OK with a smaller full-scale output?

The details:

  • I wish to use a certain standard load cell (full bridge), but in a portable application -- powered with a single cell Li-ion battery (3.5 V to 4.2 V).
  • The load cell spec simply states "Excitation: 10 Vdc (15V max)". (Full specs)
  • Instead of using a boost converter, if I were to just excite the load cell at 2.5 V for example, I'm curious if there will be any performance diminishment (aside from the full-scale output being 1/4 the magnitude).

In other words, is it more ideal for a battery-operated project (3-4V) to:

  • Option A: Use the 10V-recommended load cell at reduced excitation voltage (e.g., 2.5 V)

  • Option B: Or use a boost converter first and apply larger excitation voltage (> 10 V) ?

boardbite
  • 4,892
  • 11
  • 47
  • 73
  • The excitation voltage must be VERY well regulated or as usual is used in the output circuitry or measured alongside the signal. You do not need to regulate it to 2.5V if your ADC can measure it at the full cell voltage and use it directly as the excitation voltage with minimum power waste and noise and maximum signal. – KalleMP Nov 25 '16 at 07:52

3 Answers3

3

Load cells are ratiometric, so the output voltage will be proportional to the excitation voltage. Operating at a low excitation voltage will reduce the fullscale output from the gage, as you have identified. This isn't really a problem other than the fact that the differential output voltage of a load cell is so low to start with.

Assuming that you're using a full bridge load cell (all four resistors in the Wheatstone bridge are actual strain gauges), you can probably expect an output of 3mV/V. At 2.5V of excitation, you'll have +/- 7.5mV fullscale output. Even though the output impedance of most load cells is low, noise can still be a problem. Note that with a 2.5V excitation, you'll be giving up 12dB of SNR.

In your application, I'd stick with the lower excitation voltage. A boost converter will add noise that may be difficult to reject with your sigma-delta ADC.

(As an aside, exceeding the maximum load cell excitation isn't recommended as it can cause self-heating and resultant errors due to inconsistent changes in resistance.)

HikeOnPast
  • 7,306
  • 2
  • 25
  • 20
  • Noted. Regarding the boost converter comment: if I would be using a high-precision reference to also power the excitation, shouldn't that sufficiently mitigate the influence of regulator noise? I am also curious how you calculated the 12 dB SNR -- being somewhat new to this subject -- but I'll try to figure that one out myself! – boardbite Apr 12 '13 at 12:53
  • Could you please explain better "Note that with a 2.5V excitation, you'll be giving up 12dB of SNR."? – RHaguiuda Feb 18 '16 at 18:59
1

The the load cell is a Wheatstone bridge configuration. You'll get a difference in voltage regardless of supply voltage. The more important question is what will your instrumentation tolerate? To run off a small supply like that, you're going to need a CMOS rail-to-rail instrumentation amplifier. With a smaller voltage difference, more gain will be required.

Matt Young
  • 13,734
  • 5
  • 34
  • 61
  • Yes, I'm using AD7730, a 24-bit ADC with PGA, and was planning to measure directly, without any in-amp. In any case, I would be OK with a smaller full-scale output. And in "Option A", the 2.5V excitation would be provided from a precision reference capable of sourcing sufficient current. – boardbite Apr 12 '13 at 01:53
  • 1
    If it were my project, I would just try it. Intuition says it will work just fine. – Matt Young Apr 12 '13 at 02:28
  • Indeed; a LOWER voltage shouldn't damage the cell, so I'll try it. And I could always check if the linearity persists, by taking a few datapoints using various excitation voltages first. – boardbite Apr 12 '13 at 12:55
1

I've used pressure sensors from Kulite that have been recommended to run at 10V. I contacted them because I wanted to run at 3V with a precision PSU and they said "no problem" but they weren't prepared to "reduce" what the offset accuracy would be.

At 10V they said it's offset was +/-5mV of nominal. Full-scale was +/-100mV and they confirmed that running from 3V would have a full-scale of +/-30mV but, the offset would still be +/-5mV. I could live with this and maybe you can.

A lot of bridges are run with AC excitation in order to avoid DC errors in amplifiers and maybe you could ask them this too because, if it's capable of running on AC then it truly is ratiometric. One word of warning, the temperature compensation techniques used in some bridges may not be linear - I've heard this mentioned but not in connection with strain bridges I've used so it's definitely worth speaking with the supplier.

Regarding noise, what is your bandwidth? If it is low then digital or analogue filtering should help a lot. Regarding excitation, put some kind of current limit into your excitation - you don't want a short on the leads to the bridge dragging the whole supply down.

Andy aka
  • 434,556
  • 28
  • 351
  • 777