Interfacing to a low value resistive sensor...
- Thread starter MartinM57
- Start date
So...Plan A - some Kelvin clips, a LM317T producing a constant current and a PICAXE differential READADC. I suppose the cheat's way to the latter is two READADC channels referenced to ground and get the differential in the code...
http://www.allaboutcircuits.com/vol_1/chpt_8/9.html
@Hippy - I was worried about the 3R end of the sensor causing problems with regard to the lead lengths - the sensor is about 6 feet away...
http://www.allaboutcircuits.com/vol_1/chpt_8/9.html
@Hippy - I was worried about the 3R end of the sensor causing problems with regard to the lead lengths - the sensor is about 6 feet away...
fernando_g
Senior Member
"So...Plan A - some Kelvin clips, a LM317T producing a constant current"
Good idea...both the Kelvin connection (true 4 wire) and the constant current to linearize the change.
Now, I know that the mere mention of analog components may cause cardiac arrest in some forum members but... and I have to say this carefully...don't want to hurt anyone's feelings, but.... gulp..... could you use an opamp as a true differential sense front-end before actually feeding the ADC input? It would also reject any common mode noise you may pick up on your wiring.
Good idea...both the Kelvin connection (true 4 wire) and the constant current to linearize the change.
Now, I know that the mere mention of analog components may cause cardiac arrest in some forum members but... and I have to say this carefully...don't want to hurt anyone's feelings, but.... gulp..... could you use an opamp as a true differential sense front-end before actually feeding the ADC input? It would also reject any common mode noise you may pick up on your wiring.
I thought I'd slug the voltage measuring lines with small and large capacitors so that the time gap in the PICAXE between two READADC commands was very small compared with the time constant that I ended up with, effectively dealing with common mode noise that way...and it reduces component count and stops me having to go over to the dark side and ask newbie questions about differential op amps...
Anything fundamentally wrong with that idea?
Anything fundamentally wrong with that idea?
fernando_g
Senior Member
It depends....
for high frequency noise, the answer is an unqualified yes.
For low frequency noise, it is another issue.
I ignore the environment you'll be using your project or the required accuracy, so the following discussion may not be applicable or even required.....but for everyone's benefit, let's discuss it.
To have some decent attenuation, the corner frrequency has to be at least a decade below your offending noise. Powerline noise, which is ubiquitous, is 50 Hz, so the filter would have to have a corner frequency of 5 hz.
With a minimum 3 ohm, this boils down to aprox 10,000 microfarads.
To reduce the capacitor size, you can add some resistance in series with the Kelvin leads. Since the Picaxe's ADC requires a maximum series resistance of 10 kohm, the capacitor would now be reduced to about 3 microfarads. This may appear OK, but to reject common mode noise, the two sides have to be balanced with tight tolerance components. This requirement eliminates the use of electrolytic, tantalum and X7R or X5R ceramics. Essentially this will require the use of poly film caps, which are quite bulky and expensive.
On the other hand, an opamp has excellent rejection down to DC.
And a diff amp is fairly simple, as shown in the attached image. If you want unity gain, make all resistors equal. The only two caveats with this circuit is that it requires 1% or better resistors, and usage of a rail to rail amp, like the TLC2272 to tie the -Vcc supply voltage to ground (and avoid a negative supply).
for high frequency noise, the answer is an unqualified yes.
For low frequency noise, it is another issue.
I ignore the environment you'll be using your project or the required accuracy, so the following discussion may not be applicable or even required.....but for everyone's benefit, let's discuss it.
To have some decent attenuation, the corner frrequency has to be at least a decade below your offending noise. Powerline noise, which is ubiquitous, is 50 Hz, so the filter would have to have a corner frequency of 5 hz.
With a minimum 3 ohm, this boils down to aprox 10,000 microfarads.
To reduce the capacitor size, you can add some resistance in series with the Kelvin leads. Since the Picaxe's ADC requires a maximum series resistance of 10 kohm, the capacitor would now be reduced to about 3 microfarads. This may appear OK, but to reject common mode noise, the two sides have to be balanced with tight tolerance components. This requirement eliminates the use of electrolytic, tantalum and X7R or X5R ceramics. Essentially this will require the use of poly film caps, which are quite bulky and expensive.
On the other hand, an opamp has excellent rejection down to DC.
And a diff amp is fairly simple, as shown in the attached image. If you want unity gain, make all resistors equal. The only two caveats with this circuit is that it requires 1% or better resistors, and usage of a rail to rail amp, like the TLC2272 to tie the -Vcc supply voltage to ground (and avoid a negative supply).
Attachments
Last edited:
Thanks Fernando - good info and a good heads-up on a suitable device.
http://www.allaboutcircuits.com/vol_3/chpt_8/9.html suggests using a four part op-amp (TLC2274 I guess) and buffering the two inputs with voltage followers. Any thoughts on that?
http://www.allaboutcircuits.com/vol_3/chpt_8/9.html suggests using a four part op-amp (TLC2274 I guess) and buffering the two inputs with voltage followers. Any thoughts on that?
BeanieBots
Moderator
Many op-amps come in packs of four, so, if you're not using the others for anything special, might as well use them as buffers. It won't cost you any more or use any more (just for Dippy) PCB real estate.
BeanieBots
Moderator
In amoungst 'noise' you need to include lead impedance so as Boriz says, the kelvin connections will take care of most issues.