STPM33/34 RMS Current Calculation Error

[bsahan]

Hi everyone,

I met a problem when I measured the current rms value with stpm33 ic.I'm using this IC with resistive loads and other load profiles with low crest factor. But I met a case with a high crest factor load like as SMPS type loads. 

I collected the momentary current datas from STPM33eval board over SPI lines. And than calculate the RMS current value using with point by point datas. The calculated value asme as real value. I compared the calculated value with true RMS values that are measured with other power analyzers as called as Vitrek PA900 and Schneider PA module.

I compared the current values with resistive and refrigerature type loads. They are same for all measurement device.

Is it possible to make any mistake when I config the STPM?

You can find my measured values and voltage-current graph at below.

Blue: Actual Voltage Orange: Actual Current

Load	STPM33 Current RMS(mA)	STPM33 Current Calculated RMS(mA)	Vitrek PA900 RMS(mA)	Scheneider PM2200 RMS(mA)
High Crest Factor SMPS Load	119,9	199	198	198
Low Crest Factor Load (Resistive Load)	1033	1033	1033	1033

 

[MasterT]

I see circuits has LPF on both current inputs, likely the problem. You can try to de-solder cap's C25 / C33.

 

[bsahan]

Thank you for your response. I have tried your suggestion before. I am afraid the problem is not hardware-related. Because I understand it from the actual values I collected using the same eval board. I think there is a structure where I can configure the method used by the STPM for RMS calculation or another register-level change. Otherwise, it would be strange if I were the only one encountering this problem.

[Andrew Neil]

---

@bsahan wrote:

 

Load	STPM33 Current RMS(mA)	STPM33 Current Calculated RMS(mA)	Vitrek PA900 RMS(mA)	Scheneider PM2200 RMS(mA)
High Crest Factor SMPS Load	119,9	199	198	198
Low Crest Factor Load (Resistive Load)	1033	1033	1033	1033

 

---

So you are just one or two LSBs out in a couple of hundred - so no more that 1% error ?

That doesn't sound bad to me - what were you expecting ?

[bsahan]

Hi Andrew,

Thank you for your reply. It seems that the focus is on the calculated value for the error ratio. However, I don't need the momentary value; I used it only for detecting the error in the RMS readout.

Let's focus on the table again, specifically the first and second columns for comparing the two RMS values. The first column and first row define the STPM33 RMS value read from the RMS register. The second column and first row define the calculated value from momentary points. There is a significant difference of about 40%.

Best regards

[Andrew Neil]

---

@bsahan wrote:

Let's focus on the table again

---

OK - I'll colour it:

---

The first column and first row define the STPM33 RMS value read from the RMS register.
The second column and first row define the calculated value from momentary

---

That's these 2 cells:

They only differ by 0.9 in 199 - that's under 0.5%

 

[bsahan]

This topic is still current and I’m still awaiting your support.

[bsahan]

Hi again,

I opened a ticket explaining my issue for this error. However, the STM team stated that it’s not possible to resolve this issue with the STPM33. They pointed out that the reason is due to the current signal having a high peak. Based on their recommendation, I should handle this with raw data. But this is not possible with my design, as I am using UART communication. I didn’t expect an IC that performs RMS readings to make such a fundamental error. If I have to perform RMS and all other calculations with raw data, I don't fully understand how the STPM differs from an ADC with a high input voltage.

[Gabriel_Paulino]

Hi bsahan,

I would like to know if you were able to solve this problem.

I am working on a meter that must meet class 0.2 measurements, and the main loads will be SPMS converters.

I inserted this chip into the project with the greatest confidence because it is from a manufacturer like ST, and I never expected that there could be such a grotesque error in the component.

Please, if you managed to find a solution to your problem and it is possible, share it with me.