BOOSTXL-TUSS4470: BOOSTXL-TUSS4470 and MSP-EXP430F5529LP - Getting ADC values using energia at 1MHz

Itamar,

The TUSS44x0 devices are analog front end only devices that return an analog output at the VOUT pin. The MCU's integrated ADC is used to capture and digitize this echo envelope, so you will need to configure the MCU's ADC settings to modify the resolution and sample rate.

The TUSS44x0 can only output the demodulated log amp output signal, so this is your only echo envelope output option.

Hi John,

Energia is currently the limiting factor preventing 1MHz operation because it forces a much slower clock in background files that the user is not recommended to access. I believe the DCO frequency used by the timer function I have created for the TUSS44x0 library is forced to reference ~1.045MHz instead of something more ideal like 4-16MHz to allow for a proper clock divisor to reach 1MHz.

The MSP430F5529 is capable of a master clock of 25MHz, so if you were to create your own project in Code Composer Studio (CCS), you will have complete control over the clocks used throughout the MCU. It is possible to enable a 1MHz bit bang or timer based toggle of the IO1/2 pins on the TUSS44x0 using the MSP430F5529, just not with Energia.

The problem with Energia is that many features reference the same clock from the MCU (SMCLK). So if you were to change the clock settings in "C:\ti\energia-1.6.10E18\hardware\energia\msp430\cores\msp430\wiring.c" file, you could unintentionally break a different feature used by Energia, such as the serial communication to the PC. I've tried editing the Energia init_clock function in the past, but created more problems, so it is a known limitation of the IDE. Eventually, our team will need to create a CCS example to enable 1MHz. In the meanwhile, I hope the Energia project can serve as a good example for the structure required by your custom CCS project.

FYI, the MSP430F5529 in Energia uses a 16MHz master clock by default. You can change the master to 25MHz so that the timer based burst toggle increases from a 260kHz max to a 400kHz max in Energia. To change the master clock in Energia, update the following line in the boards.txt file found in the source directory of the Energia IDE:

Form:

MSP-EXP430F5529LP.build.f_cpu=16000000L

To:

MSP-EXP430F5529LP.build.f_cpu=25000000L

I know this doesn't solve the 1MHz problem, but enables you to use the 400kHz transducer alternative you mentioned. I don't see why a 400kHz transducer can't be used for level detection with millimetre accuracy. The main disadvantage of the 400kHz option will be a longer ring decay, meaning the minimum detectable distance may be worse than the 1MHz, while the 1MHz may not be able reach as long of a maximum range as the 400kHz.

Hi John,

You can likely get some additional support from the MSP430 MCU E2E forum if you have questions about the internal ADC capture. Just create a new post targeting the MSP430F5529 as your device in question.

The entire ADC capture sequence is abstracted to Energia's analogRead function, which I know runs relatively slow compared to what is possible. The C-code details of Energia's analogRead function can be found at "C:\ti\energia-1.6.10E18\hardware\energia\msp430\cores\msp430\wiring_analog.c" for reference. At line 349 of this file, I found this note regarding analogRead:

// The below ADC configuration applies this rule to all channels right now.
// ADC10CLK = 5MHz / 5 = 1Mhz
// Tsample = S&H / ADC10CLK = 64 / 1 MHz = 64 us
// Tconvert = 13 / ADC10CLK = 13 / 1 MHz = 13 us
// Total time per sample = Tconvert + Tsample = 64 + 13 = 67 us = ~15k samples / sec

The ADC clock and sample+hold times can be much faster when you create you custom CCS project. There may be some working high speed and resolution ADC examples at dev.ti.com in the Resource Explorer.

Hi John,

Yes, the GUI is able to capture the result of transducers up to 1MHz with ~6us step resolution for a 12ms record length. In the case of a 1MHz (1us step) transducer, the waveform will definitely be under-sampled. However, even when bursting a 1MHz transducer with a single or few pulses, the amplitude of the ring decay is large enough to still create a wide-enough echo signature for the MSP430's ADC to capture. The bigger problem for a 1MHz transducer when using the EVM-GUI is that there is a delay between bursting and the ADC turning on to start sampling on the MCU. The MCU code is currently not able to both burst and sample the ADC simultaneously, so there is a bit of a blind zone created by the GUI. For 1MHz evaluation, I recommend that you first start with an external oscilloscope to monitor the VOUT pin so you can still use the GUI to configure the device, but eventually move to a high-speed ADC to capture the 1MHz results more efficiently on your own hardware solution.