Flame ionization detector

[hamdi.tn]

Hi everyone 

well am trying to design an FID circuit, so as the name say it should detect if there is a flame or not ,by measuring the current through the sensor.
a FID is powered by a high AC voltage (around 100Hz could be less could more) and when there is a flame it act like a diode, so the circuit will see an ac current with dc component, it's the dc part we have to measure (uA current).
So basically what the circuit have to do is to extract the dc part of the signal, and amplify it to get the current value.
What am thinking of is a simple LPF to get the dc then a simple high gain amp.
In my mind this should work, in real word am sure there is infinite reason for it not to work.
Any suggestion, recommendation are welcome.

[Kleinstein]

I think using an AC voltage makes things only more difficult: The ion current is usually small, in the pA to maybe nA range. Having a 100 Hz, kV AC voltage will cause significant capacitive currents, you have to separate it from. Even 1 pF of capacity would cause a µA size AC current. Getting rid of a AC current that is something like 100000 times larger is not simple. Compare that to measuring mV DC voltage with 230 V AC superimposed.

I see no real problem using a DC voltage - this is likely the easy way to get rid of the AC current.

[SeanB]

What is wrong with the simple thermocouple sensor? Simple, easy to interface ( opamp and a gain setting resistor to get the mV signal high enough) and rugged and reliable.

Ionisation detector what about using the same chips as used in smoke detectors, just without the Americium pellet, and there you get the benefits of a good gain stage, built in power driver and voltage regulator and even a low battery detector if needed, along with micropower operation. MC14468 is a candidate.

[rolycat]

I did a teardown of an old flame ionization detector here on the forum last year.

I don't know much about them, but it was a lovely bit of engineering, with very careful attention paid to shielding and an exotic Analog Devices electrometer op-amp module at its heart. Voltage ratings on the input resistors were up to 28 kV.

Alas, the amplifier itself has been dismembered, but I still have the circuit boards.

[hamdi.tn]

I think using an AC voltage makes things only more difficult: The ion current is usually small, in the pA to maybe nA range. Having a 100 Hz, kV AC voltage will cause significant capacitive currents, you have to separate it from. Even 1 pF of capacity would cause a µA size AC current. Getting rid of a AC current that is something like 100000 times larger is not simple. Compare that to measuring mV DC voltage with 230 V AC superimposed.

I see no real problem using a DC voltage - this is likely the easy way to get rid of the AC current.

DC still a candidate for the final design, just squeezing my mind and making an effort with AC

What is wrong with the simple thermocouple sensor? Simple, easy to interface ( opamp and a gain setting resistor to get the mV signal high enough) and rugged and reliable.

Ionisation detector what about using the same chips as used in smoke detectors, just without the Americium pellet, and there you get the benefits of a good gain stage, built in power driver and voltage regulator and even a low battery detector if needed, along with micropower operation. MC14468 is a candidate.

didn't think of that ... thanks will take a look at the mc14468

I did a teardown of an old flame ionization detector here on the forum last year.

I don't know much about them, but it was a lovely bit of engineering, with very careful attention paid to shielding and an exotic Analog Devices electrometer op-amp module at its heart. Voltage ratings on the input resistors were up to 28 kV.

Alas, the amplifier itself has been dismembered, but I still have the circuit boards.

  i saw a nice document for the analog device model 210 ... awesome old school hardware is awesome 

[KerryW]

You will need"
Power supply
DVM
A piece of wire
A candle
 
Set DVM to read  Volts.  Connect red lead to + of power supply.  Connect wire to 0 of power supply.  Place the other end of the wire in the candle flame.  Place the black lead of the DVM in the candle flame, but don't let it touch the wire.

Note the reading on the DVM.  Blow out the candle and note the new reading.

It really is that simple.

[hamdi.tn]

You will need"
Power supply
DVM
A piece of wire
A candle
 
Set DVM to read  Volts.  Connect red lead to + of power supply.  Connect wire to 0 of power supply.  Place the other end of the wire in the candle flame.  Place the black lead of the DVM in the candle flame, but don't let it touch the wire.

Note the reading on the DVM.  Blow out the candle and note the new reading.

It really is that simple.

yap and how this can be connected to a uC 
i will just put my finger in the flame to check if it's there it really is that simple 

[KerryW]

I used a voltage comparator and a pot so the customer could set the threshold, but you could use a comparator input or A/D input if you have one.

[LaserSteve]

If you need an instrumentation grade unit find a copy of

"A solid state amplifier for the flame ionization detector used in gas chromatography."
Journal of Physics E: Scientific Instruments  Volume 4   Number 1 
D F Meigh and E H Oetzmann 1971 J. Phys. E: Sci. Instrum. 4 66 doi:10.1088/0022-3735/4/1/018

Uses discrete transistor pairs and a dual  mosfet  input stage  to make an differential amp and delivers a voltage proportional to the flame's conductivity. Its sensitive enough to see when tiny trace amounts of ions are added to the flame. I work at a university so I can't post the PDF here, when I view it.


If your just sensing if a flame is lit, there are simple little deep UV, gas filled,  sensor tubes that ignore room light and act like a neon lamp or really high gain phototube when the flame is burning. Hamamatsu makes them. Google "UV Flame Sensor".