Ultrasonic circuit design

Projects / Project Swiftlet  "That old Spice magic..."

I tried building a discrete driver in LTSpice along with the transducer model from Microchip. The driver design is a modified version of the pseudo push-pull from The Art of Electronics coincidentally for the 2000V ultrasonic transceiver. The modification allows me to tristate the output just like a bus driver.

I set up the model to drive 8 cycles at 40kHz, then the output low for 500us before going tristate. So what happens is that there is still a lot of kinetic energy stored inside the transducer even though I shorted its output. There is only so long you can short the transducer as it affects how close the range finder would work.

Note: using transceiver model I got

LTSpice simulation at transceiver

I even tried 8 additional cycles that are 180 degrees out of phase, the amplitude decreased a little and then built up again. I think I have gone down a dead end with this approach.

There are others that managed to get a single transducer to work. It would require a special transducer! That explains it. 1.2ms is about 0.4m. I'll stick with the separate Tx and Rx transducer approach.

I have built a LTSpice model of the receiver based on the reverse engineered schematic. The circuit would work at 3.3V. (I used the LMV324 Spice model as I couldn't find the SGM324 one.)

Here is the frequency response plot with the 2 resistor values change that Emil suggested to move the frequency to 40kHz. Not sure if that's already been incorporated into the new modules.

LTSpice - Ultrasonics receiver circuit Gain, Phase vs Frequency plot

Because of the virtual ground rail (trace in blue) capacitance, it'll take at least 2ms before it settle down. The output would follow the DC bias. Since the comparator circuit detects DC level, so it will be affected by this. Looks like we could use a better virtual ground that settles more quickly - I can probably use that spare opamp which used to be the comparator. The other approach is simply to power up the circuit ahead of time.

So the big question are there any improvements in the current rev. of the module?
A couple of the resistor values was change around the bandpass filter of the current version of the modules, but the filter center frequency is still about 1/2 of it should have been.

The new mid rail reference impedance is too high and resulted in an oscillation. That does not agree with real life results. Looks like I'll need to do some more work.

I have decided to mount the temperature sensor SE97BTP onto the ultrasonic transducer daughter trying to thermally isolate it from the body heat. The thermometer is used for temperature compensate for the change in speed of sound. The chip is designed for monitoring DDR temperature and comes with 256 bytes of EEPROM (for SPD). The thermal pad of the package is soldered to the ground plane of the daughter card. I am going to play around with the placement to get better thermo isolation.