NewAI Part GeneratorFree to get started
FREE TOOL

Load cell calculator

From the mV/V rating and excitation, this load cell calculator works out the output voltage, then what an ADC such as the HX711 makes of it: counts per kilogram and grams per count. Enter the ADC’s input noise to see the resolution you can actually hold.

Solve for
Examples
Load unit

In the load unit above.

mV/V

From the cell’s datasheet or calibration sheet.

V

Measure it at the cell (E+ to E−) if you can.

Above the tare, in the load unit above.

Optional: platform and fixture weight sitting on the cell.

V/V
bits
V

Leave blank when the excitation is the reference. HX711: AVDD.

ADC input range

From the ADC datasheet. HX711 and ADS1232 use ±0.5·Vref / gain.

Noise entered as
V

Type 50n for 50 nV. HX711 at gain 128: 50 nV at 10 SPS, 90 nV at 80 SPS (typical).

Ideal noise falls by √N; the settling time grows.

Bridge output at this load

5 mV

Net counts at this load
2,147,484 after tare
Full-scale bridge output
10 mV
Sensitivity per kg
1 mV
ADC input range
±19.53 mV
Share of ADC range used
51.2 % at capacity
One ADC code at the input
2.328 nV
Counts per kg
429,497
Load per count
2.328 mg
Input noise, RMS
50 nV
Input noise, peak to peak
330 nV
Noise in counts, RMS
21.47
Noise-free resolution
330 mg
Noise-free counts over capacity
30,303
Noise-free bits over capacity
14.9 bits
ADC noise-free bits over its range
16.9 bits
ADC effective bits over its range
19.6 bits

Noise sets the resolution, not the count size: the last 7.1 bits of each reading are noise.

Ratiometric: the ADC reference is the excitation, so counts do not change when the supply drifts.

Bridge output against load
02m4m6m8m10m12m024681012Load (kg)Bridge output (V)Capacity 10 kg5 kg

Expected readings across the range

% of capacityLoad (kg)Bridge outputNet counts
0 %00 V0
25 %2.52.5 mV1,073,742
50 %55 mV2,147,484
75 %7.57.5 mV3,221,225
100 %1010 mV4,294,967

How to calculate load cell output and ADC counts

Output = rated output × excitation × load ÷ capacity, so a 2 mV/V cell at 5 V gives 10 mV at full load. On an HX711 at gain 128, the input range is ±19.53 mV across 2²⁴ codes, one code is 2.33 nV, and a 10 kg cell reads 429,497 counts per kg. Input noise, not count size, sets the limit: about 330 mg noise-free.

Load cell formula: output voltage and ADC counts

A load cell is a full strain gauge bridge calibrated at the factory. Its rating already includes the gauge factor and the mechanical design, so the load cell equation starts from mV/V.

Vout = RO × Vex × W ÷ C
RO
rated output in mV/V, so Vout is in mV (divide by 1000 before comparing with an LSB in volts)
Vex
excitation voltage
W
applied load
C
rated capacity, in the same unit as W

What the ADC sees

A bipolar N-bit ADC spreads 2ᴺ codes over its input range. The HX711 and ADS1232 both specify a full scale of ±0.5 × Vref ÷ gain, and on those parts Vref is normally the excitation:

LSB = Vref ÷ (G × 2ᴺ)     counts per unit load = (RO × Vex ÷ C) ÷ LSB
G
PGA gain
N
ADC bits
Vref
ADC reference (AVDD on an HX711)

Noise-free resolution

For noise, use the ADC’s input-referred RMS figure at your gain and data rate. Peak-to-peak noise is about 6.6 × RMS, because ±3.3 standard deviations hold 99.9 % of Gaussian samples. TI’s ADS1232 measurements land close to that, with a ratio of 6.5 at gain 128 and 10 SPS.

noise-free step = 6.6 × en ÷ (RO × Vex ÷ C)     noise-free bits = log₂(FSR ÷ (6.6 × en))
en
input-referred RMS noise
FSR
full ADC input span, Vref ÷ G for these parts

Load cell calculation examples

Example 1: 10 kg cell on an HX711

2 mV/V × 5 V = 10 mV at 10 kg, or 1 mV per kg. At gain 128 and AVDD = 5 V, the input range is ±19.53 mV, so the cell uses 51.2 % of it. One code is 2.33 nV, giving 429,497 counts per kg and 2.33 mg per count. With 50 nV RMS at 10 SPS, peak-to-peak noise is 330 nV, so the resolution you can hold is 330 mg, about 30,303 steps across the capacity (14.9 bits).

Example 2: 200 kg cell on an ADS1232

A 200 kg, 2 mV/V cell at 5 V has the same 10 mV full scale but only 50 µV per kg. The ADS1232 at gain 128 and 10 SPS has 17 nV RMS, so peak-to-peak noise is 112 nV and the resolution is 2.24 g, or 89,127 steps. TI measures about 85,000 counts for this same 10 mV signal over 60 seconds, within 5 % of the estimate.

Example 3: load from a raw reading

A 5 kg, 1 mV/V cell at 5 V on an HX711 at gain 128 gives 429,497 counts per kg. A net reading of 1,000,000 counts above the tare is therefore 2.3283 kg. Treat that as a starting point: calibrate with a known weight, because the rated output of both cells cited on this page is only specified to ±10 %.

Full-scale output by mV/V rating and excitation

Bridge output at rated capacity. Compare it with the ADC input range before you pick a gain. For a rating not listed here, the tool at the top of the page doubles as an mV/V calculator.

Rated output3.3 V5 V10 V
1 mV/V3.3 mV5 mV10 mV
1.5 mV/V4.95 mV7.5 mV15 mV
2 mV/V6.6 mV10 mV20 mV
3 mV/V9.9 mV15 mV30 mV

HX711 resolution for common load cells

Gain 128, AVDD and excitation 5 V, 24-bit, single readings. Noise is the datasheet’s typical input noise: 50 nV RMS at 10 SPS and 90 nV RMS at 80 SPS. For another cell or ADC, use the load cell resolution calculator above.

CapacityRated outputFull scaleCounts per kgPer countNoise-free, 10 SPSNoise-free, 80 SPS
1 kg1 mV/V5 mV2,147,484466 µg66 mg119 mg
5 kg1 mV/V5 mV429,4972.33 mg330 mg594 mg
10 kg1 mV/V5 mV214,7484.66 mg660 mg1.19 g
20 kg1 mV/V5 mV107,3749.31 mg1.32 g2.38 g
10 kg2 mV/V10 mV429,4972.33 mg330 mg594 mg
50 kg2 mV/V10 mV85,89911.6 mg1.65 g2.97 g
100 kg2 mV/V10 mV42,95023.3 mg3.3 g5.94 g
200 kg2 mV/V10 mV21,47546.6 mg6.6 g11.9 g

HX711 and ADS1232 at 5 V

Typical datasheet values. The HX711 datasheet gives noise only at gain 128, so peak-to-peak is 6.6 × RMS there; the ADS1232 peak-to-peak figures are TI’s measurements.

ADCGainData rateInput rangeOne codeRMS noisePeak to peakNoise-free bits
HX711, channel A12810 SPS±19.53 mV2.33 nV50 nV330 nV16.9
HX711, channel A12880 SPS±19.53 mV2.33 nV90 nV594 nV16
HX711, channel A64any±39.06 mV4.66 nVnot specifiednot specifiednot specified
HX711, channel B32any±78.13 mV9.31 nVnot specifiednot specifiednot specified
ADS123212810 SPS±19.53 mV2.33 nV17 nV110 nV18.4
ADS123212880 SPS±19.53 mV2.33 nV44 nV247 nV17.3
ADS12326410 SPS±39.06 mV4.66 nV19 nV125 nV19.3
ADS12326480 SPS±39.06 mV4.66 nV48 nV307 nV18

What limits a real scale

  • Resolution is not accuracy. The Vishay Model 60060 specifies a combined error of ±0.02 % of full-scale output (the equivalent of 2 g on a 10 kg capacity), and the MinebeaMitsumi U2D1 C3 series lists 0.016 % of rated output of creep over 30 minutes. An ADC finer than those figures shows repeatable numbers, but they are not more true.
  • Rated output carries a tolerance: ±10 % on both of those cells. Calibrate against a known weight rather than trusting the mV/V figure.
  • Zero balance uses ADC range. The same datasheets give ±1 % of full-scale output and ±0.1 mV/V (5 % of a 2 mV/V rating). Add the platform weight as dead load and keep headroom.
  • Make it ratiometric. Tie the ADC reference to the excitation so supply drift cancels, as in TI’s ADS1232 design example. Filter the reference and the signal with matched time constants.
  • Measure the excitation. An HX711 using its on-chip regulator sets AVDD to VBG × (R1 + R2) ÷ R1 from its external divider, with VBG = 1.25 V, so a module on 5 V may not excite the cell at 5 V. AVDD must stay at least 100 mV below VSUP.
  • Common mode matters. The HX711 needs inputs between AGND + 1.2 V and AVDD − 1.3 V, and the ADS1232 at gain 64 or 128 needs AGND + 1.5 V to AVDD − 1.5 V. A bridge excited from the same supply sits at half of it, which suits both.
  • Long cables drop excitation. Cells such as the Model 60060 offer two sense wires so the electronics can measure the voltage at the cell.
  • Averaging helps less than √N in practice. TI’s ADS1232 example improves from about 85,000 to 135,000 counts with a moving average of 4 readings: 1.6×, not the ideal 2×.

Glossary

Rated capacity
The largest load the cell is specified to measure.
Rated output (mV/V)
Bridge output at rated capacity per volt of excitation.
Excitation
The voltage applied across the bridge (E+ to E−).
Ratiometric
Using the excitation as the ADC reference, so excitation changes cancel out of the reading.
Input-referred noise
ADC noise expressed as an equivalent voltage at its input, so it compares directly with the bridge signal.
Noise-free resolution
The number of steps the range divides into when one step equals the peak-to-peak noise. Peak to peak is about 6.6 × RMS.
Zero balance
The bridge output with no load, as a share of rated output or in mV/V.
Dead load
Weight permanently on the cell, such as a platform, which the tare removes but the ADC range still has to hold.

Related guides

Frequently asked questions

What does mV/V mean on a load cell?

It is the bridge output at rated capacity per volt of excitation. A 2 mV/V cell excited at 5 V gives 10 mV at full load and half of that at half load.

What is the load cell formula for output voltage?

Vout = rated output × excitation × load ÷ capacity. A 10 kg, 2 mV/V cell at 5 V with 5 kg on it gives 5 mV.

How does the load cell sensitivity calculation work?

Sensitivity = rated output × excitation ÷ capacity, which is the output per unit of load. A 10 kg, 2 mV/V cell at 5 V gives 10 mV ÷ 10 kg = 1 mV per kg.

How do I convert load cell mV/V to kg?

Divide the bridge output by the excitation to get mV/V, then divide by the rated output and multiply by the capacity. On a 10 kg, 2 mV/V cell, 5 mV at 5 V is 1 mV/V, which is 5 kg. Calibrate with a known weight for the final scale factor.

What is the HX711 input range?

±0.5 × AVDD ÷ gain. At 5 V that is ±19.53 mV at gain 128, ±39.06 mV at gain 64 and ±78.1 mV on channel B at gain 32. The datasheet’s text rounds these to ±20, ±40 and ±80 mV.

How many counts per kilogram does an HX711 give?

With a 10 kg, 2 mV/V cell at gain 128 and AVDD as the excitation, 429,497 counts per kg, or 2.33 mg per count. Those counts are not all useful: noise is far larger than one code.

How small a weight can an HX711 resolve?

With the datasheet’s typical 50 nV RMS at gain 128 and 10 SPS, peak-to-peak noise is about 330 nV. On a 10 kg, 2 mV/V cell at 5 V, that is about 330 mg before averaging, or 594 mg at 80 SPS. Real boards with long leads, shared supplies and vibration do worse.

Why do the last digits of my load cell reading jitter?

Because a 24-bit code is about 2.33 nV at gain 128 and the input noise is many times that. Noise of 50 nV RMS is 21.5 counts RMS, so the bottom 7.1 bits of each reading move at random. Averaging or a slower data rate reduces it.

Does more excitation voltage improve resolution?

In counts per kilogram, no, if the ADC reference follows the excitation: the signal and the reference scale together. In noise-free resolution, yes, because the signal grows while the input-referred noise stays about the same. Never exceed the cell’s rated excitation, and for an ADC limited to 5 V, divide the reference down if the bridge runs higher.

Can I run an HX711 or ADS1232 at 3.3 V?

Yes for both. At 3.3 V the HX711 accepts inputs from 1.2 V to 2.0 V (AGND + 1.2 V to AVDD − 1.3 V) and the ADS1232 at gain 64 or 128 accepts 1.5 V to 1.8 V (AGND + 1.5 V to AVDD − 1.5 V), and a bridge excited from that supply sits at 1.65 V. The ADS1232 needs at least 3 V for this to work at those gains. The bridge signal falls with excitation, so the noise-free resolution is worse than at 5 V.

More free PCB tools

See all free electronics tools →

Sources

Design the whole board, not one number

ProtoFlow is free AI PCB design software: describe a circuit and get a schematic with real parts, PCB layout with an autorouter, SPICE simulation and Gerber files. Native KiCad, Altium and EAGLE import and export. Free on macOS, Windows and Linux.

Get for Mac

Free to start. No card required.