3.5 volt reference + or - .020 volts

[rfranzk]

Hello All,

I am working on a project at work where an instrument we operate has a sensor that utilizes an auto id function. The id volts for this particular sensor is 3.5 volts dc + or - .020 volts. These particular sensors are prone to the id function going out of spec and that stops the instrument from functioning as it thinks there is no sensor attached. A new sensor is $1350.00 and when these fail the other sensor functions are just fine. I have disassembled and inspected one of these and it appears to be a simple voltage regulator with a voltage divider providing the id voltage. I just haven't had time to troubleshoot the real source of the problem and attempt repair as the components are all surface mount and difficult to work with.

I am looking for a work around until I can come up with a permanent solution. The instrument has a +15 volt supply that I can tap into for a source that measures 15.8 volts dc. I measured current draw on the id circuit of 0.00ua so it must be high impedance.

I was thinking just a 3 wire add on that utilizes the available + 15 vdc and provides the needed 3.5 volts by either a 5 volt regulator or reference with a voltage divider or a zener diode. The temperature environment has swings from 5 degrees C to 40 degrees C so it needs to be stable in that range.

I can assemble just about any kind of circuit but don't have the knowledge to design what is most effective.

Any assistance would be appreciated.

Thanks,

rfranzk

 

[alec_t]

I'd use a regulator based on a TL431.

 

[unclejed613]

i agree, a TL431 would probably be your best bet here. since the tolerance is about 0.6%, you want the "B" grade part (TL431BC) which has a tolerance of 0.5% as shown on page 12 of the [datasheet].

 

[AnalogKid]

You might get the voltage you want with two fixed 0.1% resistors. 2.00K and 4.99K look like good starting numbers. If not and you need a pot to tune it in, make sure the temperature coefficient of the pot and the fixed resistor(s) track.

Here is an online 431 calculator: https://solderer.tv/tl431-calculator/

TI has a 431 calculator spreadsheet: https://www.ti.com/tool/tl431calc

ak

 

[rfranzk]

Thanks for the input. I will review the datasheet more carefully for more questions concerning resistor values for the voltage regulator drawing and design parameters on page 25
Thanks

rfranzk

 

[Pommie]

Replace the existing voltage divider with a multiturn pot. What is the voltage at the top of the existing divider?

Using the normal power rail will probably vary too much.

Mike.

 

[kubeek]

+/-20mV at 3.5V is just a bit less than 0.5% accuracy, which the ordinary tl431 cannot provide. I suggest finding a variant that can do that, or some other precision reference.

 

[AnalogKid]

The LM4040 comes in a 0.1% version, but it doesn't multiply up like the 431. You would have to start with the 4.096 V part and divide it down. This should work as long as the calculated value of the two divider resistors in parallel is 0.1% or less of the input impedance of whatever this thing drives.

ak

 

[Pommie]

Could it be that the accuracy of the sensor is dependant on the voltage reference? Can you measure a good and a bad sensor both at the top of the divider and at the junction?

Mike.

 

[ChrisP58]

The LM4041 comes in an adjustable 0.2% version. It's basically a TL431 with a 1.23V internal reference. https://www.ti.com/lit/ds/symlink/lm4041c.pdf

Keep in mind though, if you're using a pot to set the voltage, the real accuracy will depend on that of the voltmeter you're using when making the final adjustment.

 

[rfranzk]

Could it be that the accuracy of the sensor is dependant on the voltage reference? Can you measure a good and a bad sensor both at the top of the divider and at the junction?

Mike.

The sensor in question is a temperature/baro press sensor and I have not had a new one here long enough to tear down and measure for comparison. The temp and bp sensors do however function linearly and as expected even when the id function doesn't.

+/-20mV at 3.5V is just a bit less than 0.5% accuracy, which the ordinary tl431 cannot provide. I suggest finding a variant that can do that, or some other precision reference.

The +/- 20mV is the stated specification by the manufacturer. In real life application I have just used a 10 turn precision pot to adjust a voltage with success at around +/- 40 mV. I think the TL431bc variant at .5% will suit my needs. I would at least like to start with that as it is pretty simple.

Here is an online 431 calculator

Thanks for the calculator. That will help. Just a quick question on the one from Texas Instruments. The bottom yellow input cell asks for IDiv/IRef. I don't see exactly what that input should be from the datasheet. I haven't had a chance to review it closely and probably won't till later tonight.

Thanks for everyones help and input!!

 

[rfranzk]

Hello All,

I used the TL431b and things work just as I needed. A little more testing to confirm things stay stable with temperature swings and we are set. Thanks again for everyones input.

rfranzk.

 

[crutschow]

The data sheets shows <10mV change in the 2.5V reference voltage over your temperature range, which should be well within your 20mV requirement for the 3.5V output.