Back to Sensing Family





    Support Documentation for the SFP Family of Precision Current, Voltage and Temperature Measurement ICs and Modules

Application notes

Calibration technique for SFP10x family of measurement ICs
External shunt connection notes
Interpreting the data for the SFP family of products (understanding Two's Complement)

Measuring multiple voltages with the SFP100 and SFP101

Research reports

Effect of current measurement gain error in battery monitoring
Impact of current measurement offset error in SOC monitoring of HEVs

Current sensing signal chain encompassed by SFP family

Current sensing signal chain

Technical FAQ

I have some technical questions specific to my application.  Is there anyone there that I can talk to?
We offer comprehensive customer support. Please email your questions and basic application information to and we will answer you promptly.

Does Sendyne provide customization for the SFP family?
Yes.  Please contact us for more information (

Can the SFP work on an AC system?
Providing a DC current sensor functionality, the SFP was not built to work on an AC system.

Where can I find a shunt that will work with the SFP IC?
The SFP will work with any shunt, made from any material.

Does Sendyne offer the SFP IC in SMT reels?
Tape and reel is available in quantities over 2,000.  Please contact us for more information.

Can Sendyne provide the PCB layout and BOM for the modules?
Yes, upon request.

Could the SFP101 be used to monitor high side and low side current at the same time?
The module is capable of measuring both high side and low side currents.  If you want to measure both simultaneously, you will need two sensors.

I understand the standard SFP evaluation module comes with a 100 micro-Ohm shunt.  Can I use a different shunt?
Yes, we can provide you with a module to which you can attach your own shunt.  Please see application note which explains the proper method for connecting a remote shunt. 

I purchased an SFP101 kit without a shunt, in order to attach one of my choosing.  How do I properly set up the SFP control software?
Customers that have purchased the SFP evaluation module without the standard 100 micro-Ohm shunt should begin by reading the Shunt Selection Guide, found in the Help menu at the top of the main screen of the SFP101SFT software application.
Values for both the Full Scale Current and the Full Scale Voltage parameters will need to be entered on the Shunt Panel of the software. 

Does the SFP module support various battery chemistries available in the market, e.g. lead-acid, gel batteries, Li-Ion, etc. 
The SFP can work with any battery, or any application that requires precise measurement of current, accumulated charge, voltage and temperature. 

If a current spike occurs in the time between reported data samples, will the SFP lose the current spike sample?
High speed current spikes will not affect the accuracy of the averaged current measurements or the Coulomb count measurements. Even when the SFP is being asked for data at a very low rate, such as 1 Hz or 0.1 Hz, high speed events will be captured and integrated into the measurement.

The SFP module utilizes heavy EMI/RFI/antialiasing filters in front of the two amplification channels used for current sensing. All high-frequency components of the current signal are properly integrated by these filters, up to several GHz. 

Does the SFP have flash memory for storing the Coulomb count so that it can be retrieved after a power down?
If the IC is powered down, all data will be lost.  

Can this IC also support bandwidths up to 100 kHz with sufficient
sample rate?

The maximum report rate of the IC is 200 Hz.  However, the IC reports the true averaged value for the DC currents that are not bandwidth limited, to several hundred MHz.  The antialiasing/RFI filters in front of the current-sense inputs perform this function.
The IC reports an averaged value between the present and the last data read request.  That value is accurate even if the signal has high harmonic contents, to hundreds of MHz.  For this reason, it is not necessary or recommended to read data values at a high rate, because 10Hz, 1Hz or even slower data rates are sufficient.

How accurate does the LDO that provides the power supply need to be?
The LDO U1 in the schematic is not a very precise voltage regulator.  We use the model with ±0.4% typical and ±3.0% maximum tolerance. 
The LDO must only provide a stable supply voltage that is defined in the Electrical Specifications of the datasheet.  This value must be between 2.375 V and 2.625 V, with 2.5 V nominal (i.e. the IC needs a 2.5 V ±5% supply). 

Does the SFP allow for customizable input voltages of the shunt’s measurements?
Yes, please see the discussion in the SFP101 datasheet, page 5.

Are there other current ranges available or can we customize it?
The current range is defined by the shunt you use and the setting of the full-scale input voltage for the current measurements’ channel.  The SFP will work with any shunt of any material.  For example, the standard arrangement on the SFP101EVB uses a 100 micro-Ohm shunt (73 micro-Ohms of which is used for sensing), it can take 100 A continuous and 513 A peak using ±37.5 mV full-scale input voltage; a 25 micro-Ohm shunt can take 300 A continuous and 1500 A peak that is still digitized correctly using ±37.5 mV full-scale input voltage.

Please explain the details of current offset error and drift vs. temperature.
For the SFP, offset error is not rated for drift.  Rather, the maximum deviations within the extremes of the operating temperatures are specified.  The guaranteed specification is less than ±150 nV divided by the resistance of the shunt, over the entire operating temperature range (-40 °C to +125 °C).
Please see the performance chart in figure 6 on page 17 of the SFP101EVB datasheet.

Please explain the details of gain drift vs. temperature.
The current value’s magnitude error and drift depend on the changes of the resistance of the shunt and the gain of the amplifier. Figure 7 on page 18 of the SFP101EVB datasheet shows the maximum deviations, due to both of these errors combined and including the error from the built-in voltage reference. This combined error is specified at -0.4 % to +0.3 % over the whole operating temperature range of -40 °C to +125 °C.
Please note that this performance is with the standard, 100 micro-Ohm shunt on the SFP101EVB; most of the -0.4 % to +0.3 % error is coming from the shunt itself. 
This error can be greatly reduced if the shunt’s thermal compensation functionality of the SFP101 is utilized. Figure 8 on page 18 of the SFP101EVB datasheet shows the chart of magnitude errors when thermal compensation is employed.

What is the expected voltage value for the ISOFHV net indicated on the schematics in SFP101EVB datasheet?
The Isolated-Filtered-High-Voltage net ISOFHV is typically at 4.3 – 4.7 V.  If a different circuit for the isolating DC/DC converter is employed, the remainder of the circuit would be able to run normally if the voltage supplied between ISOVSS and the top of C2 is from 4.3 V to 10 V.  Most of the circuit is supplied with 2.5 V from U1, and only the gate drivers for the input switches are supplied from this ISOFHV voltage. 

How does Sendyne perform calibration of the SFP modules?
The SFP modules are calibrated at room temperature for the current measurements.  In order to check the deviations of the measured value vs. temperature, and assuring the specified maximum deviation of -0.4% to+0.3%, we routinely scan the boards after the calibration over the full operating temperature range of
-40 °C to +125 °C.  Please note: this is simply the temperature stability performance of the shunt itself as this specification is for the thermally uncompensated shunt.
In our lab we have built an instrument, SCS200, based on a closed-loop lab-grade Hall-effect sensor with a purpose-built, thermally-stable load resistor.  The SCS200 is able to provide currents up to 200 A DC and measure them with an estimated 50 ppm uncertainty.  That current is passed through the shunt connected to the SFP.  Readings from the SFP and SCS200 calibrator are recorded and processed by custom software; then the calibration values are written into the SFP.

In turn, we calibrate the SCS200 by an Ohm-Labs precision shunt, with estimated <10 ppm uncertainty.

Currently, the best specifications we guarantee for tightly specified and individually-calibrated units, is 0.05% (500 ppm). Therefore, our calibrations are performed with a comfortable TAR/TUR (Test Accuracy Ratio / Test Uncertainty Ratio) of 10x.
Our conservative opinion is that an unused, properly stored SFP evaluation module will experience less than 0.05% drift in one (1) year.  For an SFP module in use, other factors such as humidity, the current magnitudes, and the time spent while heated will also affect performance.  That said, the SFP uses a shunt which is typically specified to drift no more than 1%.

Is there a mechanism to calibrate voltage measurements?
The SFP100 IC does not have voltage calibration functionality. If desired, it could be implemented within the host MCU or PC/PLC that is controlling the SFP100.

The SFP101 IC has the voltage calibration functionality. The calibration value is stored on-board in the non-volatile memory; it is applied automatically if this functionality is enabled (however, the application of the calibration value must be disabled for performing the voltage calibration itself).

The voltage calibration follows the same logic as the current calibration, in that the calibration constant is permanently stored on the IC, and affects the measured result by multiplying the raw measured data by a gain adjustment; this adjustment is calculated as [1 + ½ {calibration_value}], while calibration value itself is a number between -1 and +1, with 16-bit resolution; thus the calibration adjustment can be from x0.5 to x1.5, and the calibration function is able to accommodate initial errors of the voltage divider from +100 % to -33 %. Resolution (granularity) of the voltage calibration is 15 ppm at the calibration adjustment value of 1 (i.e. around a few percent of the nearly-correct divider ratio).

Can the SFP and shunt be calibrated?  
The SFP100 IC (legacy) only provides a non-volatile register to store the shunt’s calibration value. It is up-to the controlling host MCU or PC/PLC to implement the adjustments due to the calibration. Also, the method and logic of the calibration is controlled entirely by the customers.However, one such method is described in the SFP100EVB datasheet on pages 23 and 24; this method is used by the SFP100SFT software application.

On the other hand, the SFP101 IC provides both the non-volatile register for the shunt’s calibration value and the ability to apply that value automatically. The method and logic of the calibration is the same as described for the SFP100; it is discussed in the SFP101EVB datasheet on pages 27, 29, and 30. The functionality of the SFP101, with respect to the shunt calibration, is especially useful since the automatic application of the calibration value is effectively employed not only to the current values, but also to the accumulated charge counters.

To apply the calibration value, the raw measured value is multiplied by a gain adjustment; this adjustment is calculated as [1 + ½ {calibration_value}], while calibration value itself is a number between -1 and +1, with 16-bit resolution.  Thus the calibration gain adjustment can be from x0.5 to x1.5, and the calibration function is able to accommodate initial errors of the shunt’s resistance from +100 % to -33 %. The resolution (granularity) of the current calibration is 15 ppm at the calibration adjustment value of 1 (i.e. around a few percent of the nearly-correct shunt’s resistance).

How can I use the SFP for high-voltage sensing?  
The voltage measurement input on the SFP10xEVB modules is rated only for the maximum continuous voltage of 150 V.

The underlying reason is that due to the small overall dimensions, the board can only achieve 150 V continuous DC input (this in turn is due to the maximum continuous rating for the 1206 resistor component used for the upper resistor in the voltage divider that scales the input voltage to the nominal +/- 1 V input on the SFP). In order for SFP10xEVB to be useful for the higher input voltages, it is necessary to utilize the external resistor or resistors. For example, with three (3) 1 MOhm resistors connected in series to each other and to the voltage sensing lead (and each of these is capable of at least 150 V continuous rating), the unit would be able to measure 600 V continuous.
With a special request from the customer, we can supply not only the SFP101EVB, but also a set of resistors pre-assembled with a lead wire, for the customer’s required continuous input voltage range. These resistors are used when calibrating the SFP101EVB for voltage, and the calibration result is permanently stored and automatically applied, if such functionality is desired (in other words -- the digital data provided by SFP101 is automatically adjusted by the calibration constant, if this function is enabled).  The value of the external resistors must be entered into the SFP101SFT software application in order to provide the correct read-out of the voltage measurements.

Does Sendyne have a method for multiplexing voltage inputs? 
For an application note for the possibilities on how to multiplex several voltage inputs to the SFP, please click here.

Is there more than a single ADC on the IC?
Yes, there are two (2) independent 24-bit A/Ds. running independently, sync'd to each other, but they are both serving at least a couple of inputs, and are used in a multiplexed mode.

What is the conversion rate of each ADC channel?
The conversion rate for the two (2) internal A/Ds is fixed at 800 Hz. The two parallel channels for the current sensing are thus sampled at 400 Hz each, using the first A/D. The voltage input, processed by the second A/D, is also sampled at 400 Hz. The rest of the measurement channels (4 thermistors and 1 reference resistor for the thermistors' temperature measurements) use the remaining throughput of the second A/D.

What is the highest continuous sampling rate one SFP IC can achieve?
There is only a single effective sampling rate of 400 Hz, however, we only allow read-out at 200 Hz rate, and recommend read-out at 100 Hz maximum for any practical application.

The actual "effective" sampling frequency is controlled by the frequency of requests from the Host controller, the IC does not stream the data on its own.

The SFP was designed for DC measurement applications, for both current and voltage measurements. However, even with relatively low "effective" sampling rate, the IC properly integrates all high-frequency components all the way into GHz.  When the current being measured has high frequency components, they will be properly integrated and incorporated into the reported DC value. Operations at relatively slow sampling rate allows us to control the offset voltage of the current-measurement channel to better than 120 nV over the full operating temperature range of -40 to +125 °C.

Is there internal averaging or filter functions?
There is very flexible averaging functionality on the SFP101. The data reported from the SFP101 is automatically averaged, depending on the frequency of the data requests from the Host controller. Each data report will have an average of exactly the number of (400 Hz) samples that happened since the last request for data. There are no rounding-off errors in this averaged data due to high-precision math being done in the averaging algorithm.

The IC also accumulates a long-term average for the current measurements, i.e. the Coulomb Count, for monitoring of the total charge amount that has passed through the shunt.

Does the SFP ADC input have a software filtering function?
The SFP has a built-in digital filtering function.
Any measurement reported in response to the Host's request is an averaged value for the time interval from the last report; this is effectively an integrating filter. However, this filter's bandwidth automatically adjusts to the frequency of requests for data from the Host. If the Host asks for data at a slow rate, the noise in the reported measurement values will be attenuated. All calculations are done with such a high resolution (using 32-bit or 64-bit calculations, as needed), so that the results do not have any truncation or rounding errors, typical of some digital filters.) To filter data more, one simply has to ask for data at a lesser rate from SFP. For example, it is possible to ask for data only once per hour; the value will have an exact average of current during the 3600 seconds of that hour.

Can you explain the continuous internal calibration? How does this relate to the AUTO_CAL and FORCE_CAL bits?
Continuous internal calibration is Sendyne's proprietary method for reduction of the offset error. No user intervention or control is required or desired.
AUTO-CAL and FORCE_CAL control bits allow disabling and starting the internal calibrations with accurate timing; this is used during comprehensive factory tests only.   AUTO-CAL and FORCE_CAL control bits should be left undisturbed for normal operations.

How could a compensation table be acquired for a custom solution with a board-mounted shunt?
When a question of the shunt's thermal compensation arises, it is important to begin with the specifications:
1.) What are the desired accuracy and precision of the measurements? (i.e. measurement uncertainties?)
2.) What is the desired operating temperature range?
3.) What are the maximum current, and maximum current transients that the shunt should handle?
4.) What is the desired typical and maximum heat dissipation?
5.) What is the tolerable cost and total quantities for production?
Once these questions are answered, then a solution can be determined.
If thermal compensation is needed, then it could be based on:
a) the published manufacturer's specifications for the shunt, or
b) measurements of a limited number of the shunts, or
c) individual measurements of each shunt.

In the case of a) and b), it could be expected to half the maximum error resulting from the deviation of the shunt's resistance over the operating temperature.
In the case of c), very high performance and reduction of errors to the tune of 10-100 times can be expected at a cost of performing complicated and time-consuming procedures.

Conceptually, in order to make the compensation table:
First, the Resistance vs. Temperature characteristics of the shunt are captured by running the assemblies in the thermal chamber over the desired temperature range. Second, the above characteristics are filtered, and fitted to a polynomial of sufficiently high degree (this is to reduce the "granularity" of the temperature measurements). Third, from the high-resolution R vs T data, a compensation table is created which is a reciprocal of the R vs. T characteristics.  In other words, when an entry from the table is multiplied by the "raw" current reading, a correct value results.

Depending on the total quantity of the parts, different approaches are possible:
1) If quantities are small, and the equipment for the tests outlined above is not available, then Sendyne can offer a service for creation of the compensation tables, on negotiated contract basis.

2) If the required equipment is available to the customer, then Sendyne will assist with the set-up for the creation of the compensation tables.












© 2010-2016 Sendyne Corp company| technology | products| media | contact