UPDATED August 24, 2018 at the bottom of the post with parametric data at different voltages and gains
I have a project in mind that will require op amps and have been testing the Texas Instruments MSP430FR23xx microcontrollers with built in op amps. Yesterday I did initial tests with the MSP430FR2355 LaunchPad and looked at how gain can be set from inside the microcontroller. The input voltage of 32.7 mV was very close to the lower rail (the rails are GND and 3V3) with gains from unity to x33. The error as a percent was somewhat high at lower gains, but Vin was also closer to the rail than the values specified in the datasheet (40 to 100 mV). In this blog I will look at the effect of Vin on the accuracy of the op amp in follower (buffer) mode.
Thanks jc2048 for suggesting this test - you nailed it.
The code used is similar to yesterday but the gain was set using the following snippet.
// Select Buffer Mode
SAC_PGA_setMode(
SAC0_BASE,
SAC_PGA_MODE_BUFFER);
The test setup consists of a MSP-EXP430FR2355 LaunchPad, a bench power supply, two multimeters, a 10k 1% resistor, breadboard and jumpers. My bench power supply is not ideal for this as it is fiddly to set and low voltage levels cannot be set with great accuracy. The power supply is connected directly to the noninverting input of the op amp on the FR2355. The gain is set internally to unity. The output is connected through the 10k resistor to ground. Testing showed that the 10k resistor had minimal effect on the gain. The input voltage was set at 3.2V and gradually lowered to 0 V by adjusting the bench power supply while recording input and output voltages.
Here is a photo taken near the start of the testing with the input voltage set to 3V.
I will go off on a slight tangent now and point out the new Tenma 71-1020 multimeter which I selected and received after a element14 contest. It is a 40,000 count multimeter with SCPI. I may have got the last one as I don't see it on the Newark site anymore. Thank you element14! It looks like the display on the little Aneng multimeter is more readable than the Tenma but in person the Tenma has better contrast and is easier to read. Back to testing...
In the photo the bench multimeter on the left is reading the noninverting input and the meter on the right is showing the output from the op amp. Pretty good - less than 3 mV difference and 0.1% error. I'll take it.
Below is the compete test data in tabular form.
I started at 3.2V and worked my way down. The data was so good that I stopped after a few readings to make sure I wasn't measuring the same point with both meters. But towards the lower end the error began to increase and the results around 32 mV agree closely with what was measured yesterday.
Below is the percent error plotted as a function of the input voltage.
The voltage output swing from the supply rails is given to be 40 to 100 mV in Table 5-25 of the datasheet. The datasheet also gives the Gain to be in the range of 0.99 to 1.01 with inverting mode, follower mode. The observed value just meets the datasheet specification at 100 mV in this test. This marginal showing may be an artifact of my test setup as things were not so clean at the lower end and the bench power supply and use of a breadboard is less than ideal. All other data at higher Vin is well within what the datasheet specifies.
UPDATE August 24, 2018:
In this update the tests have updated with the following parameters changed from that shown above:
- Op amp module 1 was used instead of op amp module 0
- The op amp was in noninverting mode rather than buffering mode (gains of 3 and 9 were tested)
- Vin was limited to the lower nominal range of 40 mV to 200 mV as this is the range of interest for my application
Here are the results in tabular form:
This is the second of four op amps on the FR2355 and this one gets under one percent accuracy well within the datasheet specifications.
These tests are a bit tedious with my current bench power supply (fiddly and no automation) and I have enough data now to make me believe the FR2355 can be used for what I have in mind. So, the next step is to design a proof of concept prototype.
Comments, suggestions, and corrections always welcome.
Links
MSP-EXP430FR2355 LaunchPad Initial Look and Operational Amplifier Test
Top Comments