[PSUBS-MAILIST] Pressure Transducer

Wed Jul 22 16:28:00 EDT 2020

Sorry. The bipolar range is probably why you're using 15 bits instead of 16. Forget I said anything.

Sean

Sent from ProtonMail mobile

-------- Original Message --------
On Jul. 22, 2020, 13:46, Sean T. Stevenson via Personal_Submersibles wrote:

> 0.5% full scale on a 500 psia transducer is 2.5 psi, which corresponds to a little over 5.75 ffw. I respectfully suggest that this may not be a sufficiently accurate transducer for the application, depending on whether that error is predominately scale error, or offset.
>
> Also, check the error plots on the ADC datasheet. Total error will be cumulative: transducer error plus (or multiplied by) ADC error. Unless I'm reading the datasheet incorrectly, it also looks like full scale is +/- 6.144 V (bipolar) on that ADC. Step resolution in that case is [(6.144 - (-)6.144)/(2^15 -1)].
>
> Sean
>
> Sent from ProtonMail mobile
>
> -------- Original Message --------
> On Jul. 22, 2020, 13:04, Jon Wallace via Personal_Submersibles wrote:
>
>> Sean,
>>
>> Might be a false alarm. As I was writing a detailed report for you I realized the discrepancy is looking like only 1.97 psi which is within the specs of the .5% full scale accuracy.
>>
>> The transducer is 500 psia, 1-5v analog output with 0 psi at 1v. Three wires (power, ground, output), length of 3 feet. Data sheet accuracy is published as .5% full scale.
>>
>> The ADC is ADS1115 (sourced through Adafruit) and configured for a range of 0-6.144v.
>>
>> 6.144/32768 = .0001875 v per step over the full range of the ADC. Mathematically that puts 1v around step 5333 and 5v around step 26666. This then provides a range of 21,333 steps for the PSI range, or .0234378662166596 psi per step assuming linear consistency.
>>
>> I'm simply testing atmospheric pressure, obviously at the low end of the transducers scale. Confirmed pressure of 14.7 via multiple weather sources. 14.7 psi should produce an ADC output of around 5960 calculated as 14.7 / .0234378662166596 = 627 steps, and then added to the base 5333 representing 0 psi. In reality I am seeing the ADC produce 5878 which is 82 steps or 1.947 psi less than expected. So this is an almost 4.5 foot depth discrepancy that will be more noticeable at shallower depths than deeper ones. I thought it was a lot more than that which had me frustrated, but after reviewing the numbers for the 498th time it looks like the transducer is reporting within spec. If you feel like doing the math I'd appreciate confirmation.
>>
>> Through empirical testing I have been able to adjust software to "calibrate" the water depth transducer to 14.7 psi. It matches almost exactly to two decimal places my analog 0-60 psi Honeywell cabin pressure device. I've also tested my software calibration with an air tank pressurized up to 50psi and have found that to be nearly perfect as well, at least within two decimal places which exceeds my requirements. So the software calibration may give me the accuracy I need, I'm just not sure it will translate accurately all the way up to 500 psi. I'll have to build a smaller, simpler, display unit and find someone with a pressure tank that can take the transducer up to it's limits.
>>
>> Jon
>>
>> On Wednesday, July 22, 2020, 12:57:32 PM EDT, Sean T. Stevenson via Personal_Submersibles <personal_submersibles at psubs.org> wrote:
>>
>> Jon - What are the specs on your SSI transducer? Are you using voltage, current or unamplified? 15 bits on 500 psia should give you 0.01526 psi resolution, provided the input range of your signal conditioner matches the output range of the transducer. Check that. There could be a range mismatch, and possibly a unipolar / bipolar mismatch which would double the achievable step size (e.g. if that 15 bits corresponds to -10 V to 10 V, but the transducer signal is unipolar etc.).
>>
>> You will likely see some temperature drift if the transducer is not temperature compensated within the range of service conditions, and I would also check for grounding of the transducer body, and shielding of the signal cable if it is not 4-20 mA.
>>
>> If it is a current transducer, is it loop-powered or independently powered? If it is amplifed (voltage or current signal), check the stability of your power supply. If it's a bridge, that's even more critical. On the signal processing side, check how you are obtaining the reading. Acquisition frequency of the ADC vs display update frequency, etc. It could be that a lowpass filter on the measurement will make it behave better.
>>
>> Sean
>>
>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>> On Wednesday, July 22, 2020 7:11 AM, Jon Wallace via Personal_Submersibles <personal_submersibles at psubs.org> wrote:
>>
>>> Sean,
>>>
>>> I think you've talked about sourcing pressure transducers in the past, do you have any recommendations for ones that you have found reliable? I'm getting quite frustrated trying to interface the SSI Technologies P51 series with a 16-bit ADC (using 15 bits) and unable to get results that are anywhere close to my calculated/expected results even considering worst case accuracy limits. I don't mind paying more for a sensor that is going to perform close to it's datasheet specs. Specifically, looking for 500 psia with .5% or better accuracy. I'd prefer .25% or better accuracy but can live with .5%.
>>>
>>> Jon
>>
>> _______________________________________________
>> Personal_Submersibles mailing list
>> Personal_Submersibles at psubs.org
>> http://www.psubs.org/mailman/listinfo.cgi/personal_submersibles
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.whoweb.com/pipermail/personal_submersibles/attachments/20200722/079e42a6/attachment.html>