Modern Device Forum

On the information page of this sensor is written that the maximum permitted supply voltage is 12 Vdc, and that exceeding this voltage can damage the sensor. However, according the schematic this sensor is powered from the internal 9 Vdc linear voltage regulator (Microchip LP2951D). This chip can handle a maximum input voltage of 30 Vdc though.

The input voltage only power the voltage regulator and the opamps. Why can't the supply voltage exceed 12 Vdc? Or might the schematic not up to date anymore?

Furthermore no information about the opamp can be found in the schemartic. Is the output voltage rall to rail? Meaning it can output (almost) to its Vcc?

The output of Opamp C is lowered by a factor of 3.5 by the two resistors R88 and R77. What is the tolerance of these resistors? Is this 0.1% as well?

To increase accuracy of the board I'm replacing the temperature sensor for the TMP236A2 from Texas Instruments as this drop-in replacement is more accurte then the exsisting MCP9701A (typical 0.5 degrees, max 1.2 degrees tolerance vs typical 1 max 2 degrees) This will affect the temperature output curve, but that's just replacing a few lines of code. Maybe a tip for others as well to increase the accuracy of the entire board as this chip can be sampeled freely.

Theo_NL,

You appear to be correct, checking the schematic that the board should handle 15 volts without an issue. Everything else is running off the regulator output. One possible issue is that the regulator will be generating more heat, since it's just a linear regulator. It's far enough away from the sensing system that I don't think it will be an issue in most cases.

The output of Opamp C is lowered by a factor of 3.5 by the two resistors R88 and R77. What is the tolerance of these resistors? Is this 0.1% as well?

Correct - for accuracy. Most accuracy issues are coming from the sensor components.

The op-amp is a rail to rail type but doesn't really get near the rails anyway.

To increase accuracy of the board I'm replacing the temperature sensor for the TMP236A2 from Texas Instruments as this drop-in replacement is more accurte then the exsisting MCP9701A (typical 0.5 degrees, max 1.2 degrees tolerance vs typical 1 max 2 degrees) This will affect the temperature output curve, but that's just replacing a few lines of code. Maybe a tip for others as well to increase the accuracy of the entire board as this chip can be sampeled freely.

That will increase the accuracy of the temp, but it's not likely to do much for accuracy of the wind speed. We'll look into the price for the TI part if it's a drop in replacement. Thanks,

Paul

Thanks Paul for the confirmation. I will use two of these sensors in a heat recover air ventilattion system, whereby it is important that two different air flows have exact the same capacity. Meaning if I can lower the tolerance of each board I will do so

One other thing that might be interesting to implement in the future is to improve the effective output voltage of the circuitry. Currently the output span from 0 to 90 Mph is approx 1600 mV (approx 1400 mV at 0 Mph and 3000 mV at 90 Mph)
A 3.3 volt ADC with 12 bit accuracy has a typical tolerance of 1 bit, plus approx max 0.5 % due to tolerances on the ADC used reference voltage. The 1 bit (equals 0.8 mV) creates a read out tolerance of ± 0.8% on the calculated wind speed. The ADC reference voltage tolerance creates an uncertainty of another ±2.5% in these calculations.

Meaning the calculated wind speed has a tolerance of almost ±3.5% which solely based on the ADC conversion. (on top of all other tolerances caused by other components)

The 'B' Opamp can be changed slightly to not use it as a buffer but to use it to have an effective output voltage range of almost 3.3 volt. Then the complete ADC span can be used at the user its micro controller/Arduino, and therefore also lowering the tolerances caused by the ADC conversions.

This can be achieved by adding a resistor with value 0.333X (X can be anything from 1KΩ to 47 KΩ with a 0.1% tolerance) between its output and negative input, add a resistor with value X between the negative input and GND and use a 3.3 V reference voltage (use one with a tolerance less than 0.5%) which is connected through a resistor with value 0.5X to the negative opamp input as well.

When using the R77/R88 not as a 3.5 voltage divider but as a factor 6 voltage divider then the output span goes from 0.25 Vdc at 0 Mph to 3.1 V at 90 Mhp, having an effective voltage span of 2850 mV, and therefore lowering the ADC tolerance with a factor 2.

The 3.3V reference voltage also can be used to power the temperature sensor (few µA, should not be a problem), eliminating the 3.3V supply voltage chip as compensation for the added reference voltage chip.

Effectively this change is just adding a few resistors, and modify one chip.

When playing a bit with the resistor values only two or three different 0.1% tolerance values can be used on the board to save on manufacturing cost. (placing two in series/parallel can be cheaper than using on different resistor)

Maybe this might be useful for you (and yes, designing electronics is my daily job )

paul wrote: Wed Sep 26, 2018 4:52 pm

The op-amp is a rail to rail type but doesn't really get near the rails anyway.


Paul

At 55 Mph and 25 °C the output of Opamp C is 9.9 Vdc. The board can be powered with 10 Vdc according specification, meaning in this situation a rail to rail opamp is required though.

At 90 Mph and 25°C the Opamp-C output voltage is 11 Vdc. Meaning a supply voltage of at least 11 Vdc is required.

Can the supply voltage properties be checked and updated? It seems that the given 10-12 Vdc range is incorrect as 10 Vdc is too low in certain cases, and beyond 12 Vdc seems no problem. I don't know the Opamp its Vcc limits, but it might be that these will limit the supply voltage.

Can the supply voltage properties be checked and updated?

There is a linear regulator after the input. So a higher input voltage won't achieve what you want. There is a voltage divider that could be changed to allow a higher working voltage. We could do this on request for an order of a few pieces.

We prototyping a new model with boost system that should solve this problem in a more elegant fashion.

It seems that the given 10-12 Vdc range is incorrect as 10 Vdc is too low in certain cases, and beyond 12 Vdc seems no problem. I don't know the Opamp its Vcc limits, but it might be that these will limit the supply voltage.

Current model is TLV2434AIPWR , absolute max is 12 volts and recommended is 10V, probably the reason we chose 9V as working voltage. We didn't anticipate people using these with hurricane strength wind speeds, but they seem to work.

I'd say a high speed model is in order for this application with a different op amp and - as you say - slightly different voltage divider specs, and some high speed testing.

paul wrote: Mon Oct 08, 2018 9:44 am
Current model is TLV2434AIPWR , absolute max is 12 volts and recommended is 10V, probably the reason we chose 9V as working voltage. We didn't anticipate people using these with hurricane strength wind speeds, but they seem to work.

This is a remarkable Opamp. According its specifications its recommended supply voltage is 10 Vdc maximum. Although 12 Vdc is the absolute max, this value is not recommended.

The linear power supply does require at least 9.3 volt at its input to create its 9 Vdc output. Meaning the safe operating voltage range is just 9.3-10 Vdc. Or 9.65 Vdc ± 3.5%, which is very tight and puts a serious constraint on any power supply used.

I would recommend people therefore to reconnect pin 4 of the Opamp (its Vcc) to the 9V bus, and not to the input voltage of the pcb as it currently is. By doing so the safe operating voltage is anything between 9.3 and 20 volt. Above this voltage the linear power supply will heat up significantly..

Furthermore one generic advice for people to not draw any current from the pcb V_OUT pin; even any pull up or pull down resistors must be avoided as these resistors can influence the V-OUT voltage.

We prototyping a new model with boost system that should solve this problem in a more elegant fashion.

If you need any assistance or review input just let me know; I design aviation electronics as a profession for almost 10 years now and might provide some input if desired

I would recommend people therefore to reconnect pin 4 of the Opamp (its Vcc) to the 9V bus,

I believed that is was connected after the regulator - I'll check on this.

You appear to be correct, checking the schematic that the board should handle 15 volts without an issue. Everything else is running off the regulator output. One possible issue is that the regulator will be generating more heat since it's just a linear regulator. It's far enough away from the sensing system that I don't think it will be an issue in most cases.

Is the goal here (15V operation) to be able to capture much higher wind speeds with more accuracy?
If so then tweaking the regulator resistors would provide some of those benefits while still powering at 12 volts.
Currently the regulator is set just above 9 volts.

To change the heat differential on the active sensing element (and thus change the curve - favoring higher wind speeds) it would be necessary to change the voltage divider values that govern the balance between the active and passive sides of the wheatstone bridge. If you do this it should be done fairly slowly, because it is possible to fairly easily set the active element to temperatures that will stress the active sensing element.