Advise for Voltage buffer

[anishkgt]

Hi All,

is there are requirement for a voltage follower when using an NTC wouldn't it be possible to use the same 5v. Secondly would the attached be ok to be used as a voltage buffer ?

[Ian.M]

Well, that OPAMP's not going to do anything useful - Configuring a rail-to-rail OPAMP as a unity gain voltage follower then tying its input to the +5V positive supply just gets you a slightly crappier source of +5V.

What are you using that 'temp' voltage output for and what temperature range do you need?

[xani]

Your opamp noninverting input is connected directly to its own power supply rails. At the very best (rail-to-rail opamp) it's the same, at worst it will be worse than having power supply used directly.


If you want to have stable better-than-power-rail reference for temperature sensing at least use zener as source of voltage, not positive rail.

If you want to filter out any power supply noise, make a resistive divider giving say 4/5 of supply voltage and a cap and use that as a source for opamp.

If all you do is rough temperature sensing without much care for <5 C accuracy, just connect it directly to 5V

[anishkgt]

Well, that OPAMP's not going to do anything useful - Configuring a rail-to-rail OPAMP as a unity gain voltage follower then tying its input to the +5V positive supply just gets you a slightly crappier source of +5V.

What are you using that 'temp' voltage output for and what temperature range do you need?

I was hoping to use this (https://www.digikey.com/products/en/sensors-transducers/temperature-sensors-ntc-thermistors/508?k=ntc&k=&pkeyword=ntc&pv69=80&pv16=12442&FV=1c0002%2Cd880007%2Cd8c000c%2C1f140000%2Cffe001fc&mnonly=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=500) a 1⁰C accuracy with a max temp of 120 to 150⁰C. I would be using it measure the temperature of an electrode while spot welding.

[Ian.M]

That's only specced up to 125 deg C, and up near 125 deg C  you are only going to get a change of 0.9mV per deg C.   Not particularly useful without amplification, and with high pulsed currents for spotwelding in the close vicinity, its output voltage is likely to be swamped by the induced noise.

[anishkgt]

updated the schematic adding 10k to both inputs. Would that make it any good ?

I was simulating on LTSpice with LT1028 op-Amp and that gave about 3.03v with a 5v input. Adding a vRef IC would cost about 3$ so i was hoping to use this method instead.

I guess i should be better off with this (https://www.digikey.com/products/en/sensors-transducers/temperature-sensors-ntc-thermistors/508?k=&pkeyword=&pv252=86&FV=1c0002%2C1f140000%2Cffe001fc%2C1140050%2C1140363&mnonly=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=500)

[anishkgt]

Sorry for the dumb questions and attachments.

Your opamp noninverting input is connected directly to its own power supply rails. At the very best (rail-to-rail opamp) it's the same, at worst it will be worse than having power supply used directly.


If you want to have stable better-than-power-rail reference for temperature sensing at least use zener as source of voltage, not positive rail.

If you want to filter out any power supply noise, make a resistive divider giving say 4/5 of supply voltage and a cap and use that as a source for opamp.

If all you do is rough temperature sensing without much care for <5 C accuracy, just connect it directly to 5V

I am just not sure which op-amp to start with for this application as a voltage buffer. I have the LMV321M5 in the bom so i was thinking of using that instead of adding another component.

If you want to have stable better-than-power-rail reference for temperature sensing at least use zener as source of voltage, not positive rail.

like zener with 4.7break down in series with the non-inverting input ?

[anishkgt]

 i think i\ve managed to get vRef of 3v3v while searching for a voltage buffer.

Could somebody please advise.

[xani]

Change a voltage to SINE (7 1 1000) (1khz sine at 7V dc and 1v amplitude) and see what happens when supply voltage changes

Then move diode to be between ground and noninverting input and see what happens.

Then remove R3 (so it is just one diode and one resistor in series with middle connection connected to opamp) and see what happens.

Basically zener "stabilizes" (.... more or less, depending on temperature and current) voltage between its 2 leads. If you want ground-referenced voltage you need to put zener between ground and resistor (or better, current source). Resistor divider only makes sense after zener, if for example you need voltage that is stable but smaller than zener voltage.

But if all you want to do is to measure temperature, get a thermocouple (common K ones go > 600) and a converter like one of those:

https://www.sparkfun.com/products/13266
https://www.sparkfun.com/products/13314

Thermocouples operate on very low voltages so DIYing it is more work and rarely worth it

[anishkgt]

Thanks i get what you meant, the zener seems to keep it stable 3.2v though V_breakdown is 3.3v. Does it vary that much in practical too would the LMV321M5 be ok here ?

For reference i've added both results for those who would need to know the difference between using a voltage divder vs zerner for refernce voltage.

[xani]

Shouldn't matter much for just buffering DC, as long as amplifier can output desired voltage (some opamps are rail to rail, some are not).

If you want a bit more stability you can just swap TL431 into it (which have 3rd pin that you can use to adjust its voltage, maybe even add a pot to fine tune it).

[anishkgt]

The TL431 is an excellent one but that i would increase the BOM list with a unique component moreover i would not need that much accuracy. I guess i would be ok to go with this.

[mikerj]

Is you are sampling the output with an ADC then simply use the same reference voltage for both the ADC and your thermistor/divider chain.  This gives a ratiometric measurement, so you could use a reference derived directly from the 5v rail.

[anishkgt]

Actually the MUC is an arduino running at 5v i was thinking of connecting it to the 11v supply rail that is before the 5v regulator and after rectification because the rated zener break down is voltage is 3.3v so thought I'd add a separate voltage divider. bad idea ?

But how much voltage can this zener take while reverse biased ? from the datasheet(https://www.diodes.com/assets/Datasheets/BZX84Bxx(x).pdf) i can see Zener Voltages from 2.7V - 39V. Does that apply when reverse biased as well ?

[Rerouter]

why not just run the NTC off the micros VCC itself, the ADC reading when using VCC for aref is ratiometric, meaning fluctuations in supply voltage are not an error source,

as for the buffer, i am going to guess you were told to use one due to the higher resistance of the NTC, the arduino generally wants a source resistance of only 10K, however there is a workaround, the sample and hold capacitor is only 14pF, in order to get a good ADC reading, you want at least 4 time constants of the sampling cap to ensure the voltage in it is very close to the voltage your measuring,

Well normal 328p arduinos only sample on 1 clock cycle at 125Khz, or 8 us (micro seconds), and the RC time constant of 10K source resistance and 14pF is 0.14us, meaning you have far more than 4 time constants, assuming you don't add a crazy amount of capacitance to the wiring.

In other words, you would be unlikely to see a loss of accuracy by measuring your ntc directly, even though its not recommended by the spec sheet,

[anishkgt]

i am going to guess you were told to use one due to the higher resistance of the NTC,

actually yes. using a 3V3 would give more accurate reading than using a 5v ref. I actually did not think of sampling the signal just thought of reading it straight from the NTC.

In other words, you would be unlikely to see a loss of accuracy by measuring your ntc directly, even though its not recommended by the spec sheet

I guess there has to something to it if it is said in the datasheet. I suspect too, using just the 5v as a ref would be ok, will have to get the board fab'd to see how things work out.