Hi,
We have a customer that shared his observation as below. Appreciate your help
I am attempting to use the INA250 current sense IC in the feedback path of a "current mode" thermoelectric cooler driver. While working on the design, I observed that the frequency response curve obtained from the TI-provided SPICE model does not match the datasheet. This prompted me to purchase the INA250EVM evaluation module and obtain a frequency curve through experiment.
These are my observations (see attached graph):
* The SPICE model has its -3dB point at 90 kHz, with a -20dB/decade roll-off afterwards, a gain margin of +22dB, and a phase margin of around 75 degrees.
* The datasheet has its -3dB point at 60 kHz (30 kHz lower than the SPICE model), and has a roll-off of around -60dB/decade.
* From experiment, I found the -3dB point to be at around 40 kHz (another 10 kHz lower than the datasheet), the roll-off to be about -20 dB/decade, and the phase shift to be about 180 deg/decade.
The most concerning result is that I found (from experiment) that at the point where there is 180 degrees of phase shift, there is still 25 dB (!) of gain, meaning that the amplifier is not stable.
I obtained the frequency response by removing R101 and R102 (disconnecting the VIN+/VIN- terminals from the shunt resistor), connecting TP106 (REF) to earth ground, the GND pins (via T101.2) to power supply negative (at 17V below earth ground), and VS (via T101.1) to power supply positive (at 17V above earth ground). A function generator with an earth ground referenced 50 ohm output was connected across VIN+ and VIN-. A high impedance 10:1 scope probe with a low inductance spring clip ground lead was connected to the output (TP107). Thus, the INA250A1 was operated with a 34V supply (within the 36V limit) and VREF was kept below the datasheet-mandated 18V max w.r.t. device ground. The generator output was set to ~27 mV (the lowest it would go) to avoid clipping or slew-rate limiting, although the device appeared to show signs of slew rate limiting starting at 80 kHz.
My questions are:
(1) What would explain the large discrepancy between the datasheet, SPICE model, and observed frequency & phase response? Why am I finding that the INA250A1 is unstable?
(2) Am I doing something wrong in characterizing the device?
(3) Is there some way to externally compensate the INA250 to ensure that it is stable?
(4) The datasheet shows only device gain curves. Are TI-measured phase curves available?
Because I am using this device with a switch-mode converter operating at 500 kHz, the INA250 will be exposed to some amount of noise past the point of 180 degree phase shift.
ina250a1_freq_response.pdfina250a1_freq_response_redone_msox3104t.pdf