LM5056A: Power reading lower than Vin X Iin

Hi,

Can you provide more details? Such as how lower are we talking about? The device is not ideal, so rarely you get exactly what you would otherwise on paper.

Regards, Guang

Hi Guang Zhou

The problem is not the precision it's the power reported from MFR_READ_AVG_PIN and MFR_READ_PIN do not correspond to Vin X Iin up to 21%. But the accuracy of Vin and Iin are within spec. Here is some measurements:

Thanks

Rick

Hi Rick,

Thank you for the data table.

It appears power measurement accuracy improves as power level increases. I don’t know the root cause yet, but this is a common phenomenon in measurement applications. At lower signal levels, nonidealities, especially offsets dominate. How were the Vin and Iin value obtained, readings from the device or external measurements? The value of shunt resistor play an important role in this too.

Regards, Guang

The Vin and Iin are the LM5056A reading and do not correspond to Pin reading from chip. As I said the Vin and Iin are accurate as spec when compare to multimeter reading. The table show simultaneous reading of Vin, Iin and Pin. And I just calculate Vin X Iin to compare self error of LM5056A reporting from software with coefficient define in table 39 of spec.

Can you replicate reading of Vin, Iin, Pin from eval board at different tension and load to replicate my issue please.

Here is an example of sensors reading: in that case Vin x In =173W and it report 164.69W why??? I want to keep Highest, avg and max (without glitch idealy max 4.87kW is impossible).

lm5056-i2c-1-58
Adapter: SMBus I801 adapter at f000
Vin:          23.97 V  (min =  +0.12 V, max = +88.91 V)
                       (avg = +24.00 V)
Vout:         24.05 V  (min =  -0.04 V, avg = +23.97 V)
Temp:         +35.6°C  (high = +125.0°C, crit = +150.0°C)
Power In:    164.69 W  (highest = 212.37 W, avg = 164.69 W)
                       (max =   4.87 kW)

Rick

Hi Rick,

Average power does not equal to average of Vin times average of Iin. LM5056 averages on all instantaneous power readings.

Regards, Guang

Sorry it miss the current reading in previous example.

Again Live reading Vin 23.69v X Iin 1.79A = 42.4W and live Power In is 38.34W 10% error why??

And the average Vin X In = 43.3W compare to average Power In read 38.34w 11% error.

If the average is good for Vin Iin and Pin it should be really close, as we calculate the error is live 10% and average 11%.

Is it possible to replicate the reading from LM5056A on eval board please.

lm5056-i2c-0-58
Adapter: SMBus I801 adapter at f000
Vin:          23.69 V  (min =  +0.12 V, max = +88.91 V)
                       (avg = +23.67 V)
Vout:         23.62 V  (min =  -0.04 V, avg = +23.62 V)
Temp:         +26.7°C  (high = +125.0°C, crit = +150.0°C)
Power In:     38.34 W  (highest =  61.00 W, avg =  38.34 W)
                       (max =   6.14 kW)

Current In:    1.79 A  (max = +54.55 A, avg =  +1.83 A)

I look in eval board doc https://www.ti.com/lit/ug/snvu201/snvu201.pdf?ts=1616775992778&ref_url=https%253A%252F%252Fwww.ti.com%252Ftool%252FLM5056EVM

and I find and example iin 0.006A x 48.02v 0.288W and PIN read 0.245 this is 17.6% error on internal LM5056 calculation of power??

Hi Rick,

Instantaneous power accuracy will probably degrade at low power levels. I’m afraid what you saw is real, just like the example in the User Guide. You may use the same datasheet condition to verify accuracy:

Regards, Guang

Hi Guang

The question is not accuracy, I do not compare to multimeter measurement.

The question is why instantaneous POWER reading do not correspond to Vin  MULTIPLY by Iin? It should be done in that way in LM5056. So is it the linux sensors driver that modify the register reading with not appropriate coefficient? If the LM5056 is build that way to get over 20% error power conversion calculation internally compare to is Vin X Iin the power register is unusable. Do you understand?

Rick

Hi Rick,

Datasheet only specifies accuracy under certain conditions as I mentioned previously. It has the following statements regarding conditions other than specified:

It also provides how-to on finding custom coefficients as another option:

......

Regards, Guang

Thanks Guang

Forget it.

We will do a software patch and read Vin , Iin and multiply to get good power input sensors.

Close issue.

Rick

Hi Rick,

Agreed, that would probably the optimal solution for the current situation.

One last note, we just introduced a family of high voltage, high accuracy power monitors - INA228/INA229/INA237/INA238/INA239. These are I2C and SPI compatible. I would migrate toward these new devices for future projects unless PMBus is a must.

Regards, Guang