HDC1080: Typical Accuracy

Eelco Simonse

Prodigy 125 points

Part Number: HDC1080
Other Parts Discussed in Thread: HDC2080, HDC3020

The HDC1080 datasheet specifies a 'typical RH accuracy' of +/-2%.

How is this 'typical accuracy' defined/determined?

If I order many sensors and use them in the right way, what's the chance that they are within +/-2% in new state?

Is this figure defined by statistical Gaussian distribution? 1 Sigma, 2 Sigma, 3 Sigma, or...?

over 1 year ago

0 Ren Schackmann over 1 year ago

TI__Mastermind 43715 points

Hi Eelco,

The sensors are expected to meet 2%, but there is no max spec. If you need a max spec, please see HDC2080.

thanks,

ren

0 Eelco Simonse over 1 year ago in reply to Ren Schackmann

Prodigy 125 points

This does not really answer my question. If there is only a typical specification, in theory this would means that we can expect everything (e.g. also +/-20%). But I think that's not the case, because then the sensor is unusable. So my question is how this specified 2% relates to the real accuracy when ordering e.g. 10k of sensors and testing them all. And are the sensors factory tested on this spec?

0 Victor Salomon over 1 year ago in reply to Eelco Simonse

TI__Expert 3910 points

Hello Eelco,

Typical specs, such as the ±2 RH% accuracy spec, are based on statistical analysis of characterization data from one or more wafer lots, in the case of HDC1080, the accuracy (±2 RH%) value represents the mean or average value of the distribution within ±1 Sigma.

Regarding whether we test these values in production, here are a couple of snippets from the HDC1080 datasheet.

(3) Limits are ensured by testing, design, or statistical analysis at 30°C. Limits over the operating temperature range are ensured through
correlations using statistical quality control (SQC) method.

(4) Typical values represent the most likely parametric norm as determined at the time of characterization. Actual typical values may vary
over time and will also depend on the application and configuration. The typical values are not tested and are not guaranteed on
shipped production material.

If you are still in the process of selecting a part, I might recommend looking at newer generation devices, such as HDC2080 or HDC3020 (see comparison), both of these devices have a "Maximum" spec in addition to the "Typical" spec, meaning we do test devices in production.

We also tend to have more characterization data in newer devices like our HDC3020 (from our 3rd generation of humidity sensors), including testing over a wider range of conditions, even beyond what the datasheet shows. Here is a document that includes some of the additional data we have for this generation: HDC302x Silicon Users Guide (Rev. C).

Let us know whether you are choosing a new part, or looking at an existing design, either way we'd be glad to answer any other questions you may have.

Best Regards,
Victor Salomon

0 Eelco Simonse over 1 year ago in reply to Victor Salomon

Prodigy 125 points

Hello Victor,

Thank you very much for your clear answer. We have an existing design with HDC1080 (for 5 years) but now we face some extreme deviations (+7%). We did not design the product ourselves, so our development partner is investigating the issue. Because I was a bit curious about the sensor specs, I posted my question. If our partner wants to have more information, they will get in contact with TI themselves. Maybe replacement by HDC3020 will solve a lot of 'specification discussions' for us. Thank you for the good advice.

Best regards,

Eelco Simonse

Sensors

Sensors forum

HDC1080: Typical Accuracy