Maker Pro
Maker Pro

Mains powered variable resistance

R

Raveninghorde

I was given an interesting spec by a customer.

He wants a wall mount RH sensor with a resistive output which emulates
a 3k NTC thermsistor curve. So 0% RH is about 10k, 25% 3k, 50% 1k etc.

However the only power available is 230V ac and there is no space for
a transformer. And the design is cost sensitive as the quantities are
good. The first batch of 5000 are needed in 8 weeks.

So The design needs a variable resistance output which meets SELV
requirements.

My first thoughts are a capacitive dropper running a PIC which will
provide a pwm output to an opto isolator which will control the
resistance of a FET. Then all I need is a way of correcting the FET
resistance but that is not easy as it's on the isolated side of the
opto.
 
R

Raveninghorde

Depending on what hooks up to it, you may do a lot better with a heavily
filtered switched capacitor (or switched resistor) for the resistive
element on the output end -- trying to accurately control the resistance
of a FET is going to be problematic.

As long as the thing connected to the outside end can stand a
capacitively loaded "thermistor" then a switched capacitor should be
accurate enough to control open-loop, with duty cycle or whatever.

I've asked for a sample of the kit that this unit has to drive so I
can check how it responds to different inputs.
Irregardless, if you need to work over a large temperature range you're
going to be challenged: thermistors can present an enormous range of
resistances.

Maybe I wasn't clear.

I am simulating the response of an NTC thermistor for a humidity
sensor. So 0% RH gives the same resistance as 0C, 25% RH has the same
resistance as 25C, etc.

Temperature itself isn't a big issue.
 
I've asked for a sample of the kit that this unit has to drive so I
can check how it responds to different inputs.




Maybe I wasn't clear.

I am simulating the response of an NTC thermistor for a humidity
sensor.  So 0% RH gives the same resistance as 0C, 25% RH has the same
resistance as 25C, etc.

Temperature itself isn't a big issue.

use an NTCs control temperature with another NTC + FET as heater ?

might want a few NTCs to get above ambient

-Lasse
 
C

Cydrome Leader

Tim Wescott said:
How about a pair of "power" wires going into a box and being ignored,
while you connect a 3K NTC thermistor to the "output" side?


Hahahaha
 
L

linnix

Except I missed the part about wanting humidity to read as temperature.

So the joke's on me :/

I don't really get the joke. You need enough power to heat up the NTC and read it's resistance by something else. Of course, it's overkill or overspec. They are probably too lazy to redo the device reading the NTC.

Alternatively, mount a 12V auto battery behind the wall to power it. Perhaps, they will get the joke of how ridiculous the spec is.
 
F

Fred Bartoli

[email protected] a écrit :
use an NTCs control temperature with another NTC + FET as heater ?

might want a few NTCs to get above ambient

Not a bad idea at all.
Have two identical NTCs, one on each side of the PCB (for isolation).
Say top = primary side, bottom = secondary side.

Have a heating transistor on the top/primary side along with the sense
NTC, while your "simulated" resistor lies on the back side.

Have a small U or Omega shaped cut into the PCB so as to limit heat
conduction to other regions of the PCB, and get as identical as possible
temperatures on both sides.

Except you'll have either to figure how to cool the PCB to 0°C or you'll
need to use a higher resistance NTC so that at your max ambient, the NTC
resistance is above your targeted 0°C resistance. Of course, if you need
to reach the 100°C resistance at reasonable temperatures, you'll want a
higher beta NTC.
 
S

Spehro Pefhany

I was given an interesting spec by a customer.

He wants a wall mount RH sensor with a resistive output which emulates
a 3k NTC thermsistor curve. So 0% RH is about 10k, 25% 3k, 50% 1k etc.

However the only power available is 230V ac and there is no space for
a transformer. And the design is cost sensitive as the quantities are
good. The first batch of 5000 are needed in 8 weeks.

So The design needs a variable resistance output which meets SELV
requirements.

My first thoughts are a capacitive dropper running a PIC which will
provide a pwm output to an opto isolator which will control the
resistance of a FET. Then all I need is a way of correcting the FET
resistance but that is not easy as it's on the isolated side of the
opto.

How much do you know about what is connected to the resistor?
 
S

Spehro Pefhany

I'm just a fan of dummy parts, and stuff like that.

microchip makes weirdo parts like the OP wants

http://ww1.microchip.com/downloads/en/DeviceDoc/22017a.pdf

That would be an easy way- the "wiper" resistance will limit the high
RH end and the resolution will also be crude at the high end (but RH
measurements are pretty cruddy usually anyway) Probably want 1024
steps .. eg. AD5161. The voltage of both of the rheostat ends has to
lie within the power supply voltage limits of the digital pot. It
would be easy enough to opto-isolate the SPI signals, but it might be
easier to just use a tiny SMPS to generate a single isolated supply,
since you'll need power on the "thermistor" side of the barrier
anyway, and presumably the isolation from the mains will have to
comply with safety standards.
 
J

Jamie

Raveninghorde said:
I was given an interesting spec by a customer.

He wants a wall mount RH sensor with a resistive output which emulates
a 3k NTC thermsistor curve. So 0% RH is about 10k, 25% 3k, 50% 1k etc.

However the only power available is 230V ac and there is no space for
a transformer. And the design is cost sensitive as the quantities are
good. The first batch of 5000 are needed in 8 weeks.

So The design needs a variable resistance output which meets SELV
requirements.

My first thoughts are a capacitive dropper running a PIC which will
provide a pwm output to an opto isolator which will control the
resistance of a FET. Then all I need is a way of correcting the FET
resistance but that is not easy as it's on the isolated side of the
opto.
The only way I can see that happening in low part count is a photo
resistor opto-isolator being PWM with a SIN to give you
the log output you're looking for. THis would offer the isolated R you
need.

As for the cap supply, there was a thread here not to long ago on that
and I came up with a couple of ideas that seemed to me, to be a viable
application. If you're interested I can post that spice on that one.


Jamie
 
R

Raveninghorde

What RH sensor type are you required to use?

...Jim Thompson

I've not chosen the RH sensor yet. I already use a Sensirion SHT11 on
another product. Easy to use and doesn't need calibrating.
 
J

josephkk

Depending on what hooks up to it, you may do a lot better with a heavily
filtered switched capacitor (or switched resistor) for the resistive
element on the output end -- trying to accurately control the resistance
of a FET is going to be problematic.

As long as the thing connected to the outside end can stand a
capacitively loaded "thermistor" then a switched capacitor should be
accurate enough to control open-loop, with duty cycle or whatever.

Irregardless, if you need to work over a large temperature range you're
going to be challenged: thermistors can present an enormous range of
resistances.

Hmmm. Make the RH measurement and translate with a half-DAC switching a
resistor tree (or maybe a r/2r ivy).

?-)
 
S

Spehro Pefhany

Use a FET in series with resistor. Control the FET on/off by PWM so FET +
resistor will appear as variable resistance (at average).
No need for analog correction.

Often thermistors for this kind of application are measured by
alternately inserting them into an RC oscillator circuit with a single
precision reference resistor- and a bit of math is applied which
cancels out the supply voltage and capacitor value and linearizes the
reading. The oscillator frequency would typically be in the kHz to
handle the large dynamic range of the thermistor with reasonable
resolution at high temperatures, so not so easy to apply this kind of
solution.

Your suggestion might work if it was a simple ratiometric ADC reading.


Best regards,
Spehro Pefhany
 
J

Jasen Betts

I was given an interesting spec by a customer.

He wants a wall mount RH sensor with a resistive output which emulates
a 3k NTC thermsistor curve. So 0% RH is about 10k, 25% 3k, 50% 1k etc.

how about a carbon pile resistor tied up with rawhide, catgut, or
other hygroscopic material

To meet the resistance target it'll probably need a special
composition for the blocks, perhaps 90% cement, 10% carbon
 
Top