MicroVoltage fuse (Now "Sensing the Flame-Front of a Plasma")

[jim hardy]

I found a manual at http://www.ni.com/pdf/manuals/371303n.pdf

By GND i trust you mean one of these GND pins on the analog side, and NOT the earthing pin of a wall receptacle..

I've been unable to find how one tells that front end software whether it is to expect your inputs to be wired single ended or differential.

 

[davidwinth]

I found a manual at http://www.ni.com/pdf/manuals/371303n.pdf

I've been unable to find how one tells that front end software whether it is to expect your inputs to be wired single ended or differential.

View attachment 222883

Yeah, I have never understood that part of the device either, Jim Hardy.

Tom G. Looking at the results from yesterday's tests with clear eyes this morning, it seems the resistor did make the slope on the backside drop off faster. The image is below. As a reminder, this is with the 2nd and 3rd line disconnected from the battery, so that only the first line is powered (gaps 1,4,7). And a 100k Ohm resister is run from line 1 channel input to DAQ ground. I only have one of those resistors, or I would have tried it with all three lines.

I want to go get several resistors to try out this morning because the signal is too low with 100k Ohm. Would a larger or smaller resistor make the signal bigger? If I try with a 50K Ohm resistor, will that make the signal bigger or smaller than the 100k Ohm? Also, would the shielded cables help this part too? I would like to order them but they are expensive and I will hold off until I get someone saying it should help. Unfortunately, my boss will not let me post pictures of the experimental setup. I apologize for that. The gaps are made from Aluminum wires, 20 gauge.

 

[Tom.G]

@davidwinth If the material of the flame bed is conductive, connect it to the DAQ Ground. Also what is the material?

I've been unable to find how one tells that front end software whether it is to expect your inputs to be wired single ended or differential.

That is in the "Interactive Control Panel" software that comes with the hardware. See: http://www.ni.com/example/54386/en/

http://www.ni.com/cms/images/devzone/tut/Interactive_Control_Panel_2.JPG

 

[davidwinth]

Tom G., so it is on the software side that the signal is interpreted as individual or differential mode?

 

[Tom.G]

Tom G., so it is on the software side that the signal is interpreted as individual or differential mode?

Not necessarily (and probably not). The software likely reconfigures the hardware for single-ended or differential operation.

I want to go get several resistors to try out this morning because the signal is too low with 100k Ohm. Would a larger or smaller resistor make the signal bigger? If I try with a 50K Ohm resistor, will that make the signal bigger or smaller than the 100k Ohm?

In this usage, the smaller the resistor the smaller the signal, but the faster the trailing edge. Another possibility is to add an amplifying stage between the sensor gap and the DAQ input. That way you can adjust the resistor value at the amplifier input for best (compromised) results and make up the signal loss with the amplifier. For testing you could even use an Audio amplifier with a resistor across the input, then feed the DAQ with the amplifier output. The amplifier must be able to handle an input signal up to a few volts, so an amp with a "Line" input would be best. A starting value for the input resistor could be 15k Ohm, or get a 50k or 100k pot (potentiometer, variable resistor). If you can't beg/borrow/steal an amp with a Line input, an input resistor around 4.7k would be a value to start with.

Also, would the shielded cables help this part too? I would like to order them but they are expensive and I will hold off until I get someone saying it should help.

Probably not worth it at this point. If they would not be exposed to much heat, say less than 140F, they are pretty cheap, around $15 per 100Ft. You can find it as Microphone cable, Guitar cable, or just Shielded Cable with a Google search. If you are in the USA you could even use coax cable as used for cable TV. Home Depot has it for around the above price, it is pretty hard to work with though. Again, some of the Home Depot stores also carry shielded cable.

Bummer on the photos but I understand. How about the material of the flame bed? And the fuel could be helpful too.

Technical stuff for those lurking here:
The overall project is trying to characterize flame-front movement using spark gap style sensors feeding a DAQ. No spark occurs, the gaps are powered by 48VDC and the flame plasma provides a current path to the DAQ. The plasma impedance is in the megohm range and the DAQ input resistance is 144k with the capacitance unspecified. This yields a current source of 7uA to 70uA with flame presence and just the DAQ input circuit otherwise. The stated problems are long fall time of the signal and crosstalk between signals. Many details of the setup are unavailable to us so we are playing 20 questions and 'try this and report.'
End Technical stuff

 

[jim hardy]

Also, would the shielded cables help this part too?

One experiment is worth a thousand guesses...

I would like to order them but they are expensive

Surely you can find some TV co-ax laying around. Shield to 'GND' and center wire for signal .

 

[davidwinth]

Thanks again. I looked at electronics parts amplifiers and I would have no idea how to hook one up so I am just going to get double the batteries and use smaller resistors instead to get sharp quick signals. Hopefully I can find a combination that works! Also, I do have 200 feet of cable coax lying around which is plenty.

Thanks for your time.

 

[jim hardy]

The software likely reconfigures the hardware for single-ended or differential operation.

That might be by switching in a differential amplifier for true differential measurement, hardware solution
or maybe digitizing voltage at both pins and subtracting them, software approximation of true differential.

this image at the NI tutorial isn't live. @davidwinth See if you can make the real one select single-ended ?

 

[Tom.G]

See if you can make the real one select single-ended ?

That screen shot already shows single ended. The Red circled one selects differential.
The tab labeled "RSE" is defined in the documentation as Referenced Single Ended.

 

[Tom.G]

I looked at electronics parts amplifiers and I would have no idea how to hook one up

Someone in your company may have a HiFi amplifier laying around you could borrow. If they do, they probably know how to hook it up; One of your sensors or group of sensors to an input and the speaker output to one of the DAQ channels.

Your idea of higher input voltage and lower resistor values is certainly worth a try... and rather easy too.

 

[jim hardy]

The tab labeled "RSE" is defined in the documentation as Referenced Single Ended.

i missed that. I assume he's got his set for RSE?

 

[jim hardy]

Okay I'm confused by the behavior of that multiplexer when it is presented with an open circuit at its input. His charts show that he's measuring zero , i assume that's zero volts ? When connected to just a bare wire awaiting a flame front ?.

Is that zero volts or zero change from something ?

Reason i ask is
here's the input circuit according to the manual at http://www.ni.com/pdf/manuals/371303n.pdf
I labeled the three resistors and two voltages; why will be apparent in a minute.

Let us examine what is the voltage (Vout) handed to MUX over the range of Vin which is -10 to +10 volts...

Sum currents at node Vout, positive current enters the node.
(Vin - Vout) / (R1) + (Vref - Vout) / R2 - Vout / R3 = 0

After a whole page of algebra i come up with

Vout = (Vin + Vref X R1/R2 ) / ( 1 + R1/R2 + R1/R3 )
which, inserting values for resistors and Vref
resolves to
Vout = Vin / 8.350 + 1.230 volts (rounded to four significant figures)

When Vin = -10 Vout = + 0.03239 volts
When Vin = 0 Vout = + 1.230 volts
When Vin = 1.397 Vout = 1.397 volts
When Vout = +10 Vout = 2.427 volts

(Suggesting a single ended PGA and ADC running on 3 or 5 volts ?.)

Aha ! The plot thickens... take a look ===>> if Vin = 1.397 so is Vout,
meaning zero current through R1 at that and only at that value of input voltage.
Conversely, an open circuited input which disallows current through R1 will deliver exactly the same voltage to the MUX as would an input of 1.397 volts.
That's why the manual says this on page 17.

So - i do not understand from where came those readings of zero in his charts.

If he's set the device for Single Ended he should have got 1.397 volts prior to conduction through his gaps, not zero.

If he's wired his inputs for Single Ended and set the device for differential , it explains the zero readings
and maybe the crosstalk
because channel A0 is referenced to pin 1 for Single Ended, but to pin 3 for differential.
A gap wired to pin 3 and read differential will show up on A0 not on A4.

I'd suggest to @davidwinth he copy the image above from tech manual into Paint and show us where his inputs are wired, and where battery negative landed, and if he's reading single ended or differential, and what gain setting he's using.

That'll get me out of my present conundrum , if only into a deeper one.

old jim

 

[jim hardy]

(Suggesting a single ended PGA and ADC running on 3 or 5 volts ?.)

48 volts at Vin would hand the MUX 6.98 volts.
If it's a 5 volt or 3.3 volt chip i don't know what it would do. I hope your gap wires haven't touched.
The 127K resistor would probably protect it from damage but it cannot work with that kind of overvoltage. The voltage will bleed into other channels of your MUX. See that Analog link i referenced.
I'm accustomed to +/- 15 volt MUX IC's. Never used this new low voltage analog stuff.
old jim

 

[Rive]

I don't like this problem. It is a perfect example for endlessly poking an under-specified setup in hope that some miracle will make it work at the end. It is just - not engineering.

The first thing I would do is to put some 5.1 or 3.2V zeeners on the low leg of the resistor dividers.
The second, is to physically remove every 'sensor' from the setup, except two. Then with that two remaining sensor repeat the experiment and check whether what the sensors seems to sense is real (in this case it'll be sensitive to the arrangement of sensors).
Third, is to get a real (maybe: analog?) oscilloscope and check the output signal of the 'sensor'. Then design a real sensor based on those measurements, with a power supply and a schmitt-trigger built in.

 

[jim hardy]

It is a perfect example for endlessly poking an under-specified setup in hope that some miracle will make it work at the end. It is just - not engineering.

A question well stated is half answered. This one is neither.

It is difficult to troubleshoot a secret gizmo. I've done what i can.

Flame detectors in furnaces conduct just microamps and in only one direction. They apply 120 VAC and look for DC current.

 

[Tom.G]

@davidwinth

Flame detectors in furnaces conduct just microamps and in only one direction. They apply 120 VAC and look for DC current.

Well that's a clue to something I've been wondering about! How about trying it with the battery polarity reversed.

 

[Tom.G]

@davidwinth Try going back to one sensor per DAQ channel to see how that affects the signal interference. From you image in post 22 (https://www.physicsforums.com/threa...-front-of-a-plasma.942881/page-2#post-5966324), the DAQ sampling rate seems high enough that you could estimate the peak locations fairly accurately.

The second, is to physically remove every 'sensor' from the setup, except two. Then with that two remaining sensor repeat the experiment and check whether what the sensors seems to sense is real (in this case it'll be sensitive to the arrangement of sensors).

This was done per posts 21 thru 30-something, and none of us properly followed thru.

More like some resonant or reflecting behavior associated with the flame front.

Re-reading some earlier posts, combined with more recent report results, there is a lot to support that. Especially if the fuel is distributed over the length of the sensing area, whether or not it is done explicitly or as flamefront in an expanding gas flow. For instance, assume the experiment is done in an enclosed space such as a gun barrel. I can see how pressure waves could increase conductivity at earlier sensors, and also how gas flow could induce signals at later sensors just before the flame-front arrives. See the image in post #18.

The other supporting info is that the experiment lasts about 15ms with the speed of sound around 1100 ft/s and the OP stateing that the sensor wiring is about 8 feet long.

All these seem to point to a highly flammable fuel, probably a gas, that is not a high explosive (i.e. flame front is not supersonic), in an enclosed space. It could also be the unstated reason the OP is resisting the test of a fan blowing across the experiment to gather further data.

If the above conjectures are indeed facts, then the signals are very likely to be real in the sense that what is being measured is the plasma density at the gaps.

 

[Rive]

If the above conjectures are indeed facts, then the signals are very likely to be real in the sense that what is being measured is the plasma density at the gaps.

Some signals are too close to each other, so some kind of coupling on the side of the electronics is likely to be present - but it is just not consistent.

Ps.: just noticed that blue also part of the same forward-and-backward pattern...

 

[sophiecentaur]

Fuses don't operate on voltages so your question makes no sense.

A fuse works by melting. That involves energy dissipation which is VI. There has to be a finite PD across it for this. It's not normally an issue for mains or 12V but with low voltage circuits and, perhaps, highish internal resistance power sources, a fuse may just sit there getting for ever, just getting warm. Tin is selected as it has a low melting point, of course but . . .

 

[jim hardy]

Troubleshooting 101
spit the system in half to localize the trouble, ,
First question : is the unexpected behavior in the flames or in the measurement system ?

Replace 48 volt battery with a 6 volt one so as to not overdrive electronics
superglue a piece of tinfoil to a fingertip and drag it along the line of sensors

Crosstalk gone or still present ?

 

[Averagesupernova]

No disrespect intended but the fact that the OP was expecting to blow fuses through a flame leads me to believe ANY of this is above their head. Basic electronic troubleshooting skills seem to be absent. It is very difficult to help someone with something like this with the level of experience that is apparently lacking. To expect folks on an internet forum to be able to fix this under those circumstances is expecting a lot and to do it with no pictures is about impossible. Folks with the same general understanding of electricity that are on opposite sides of a keyboard/computer screen can accomplish something but this seems rather pointless. Again, no disrespect intended. I wish I knew more about what the OP is doing because it interests me. I have had little experience with flame sensing besides thermocouples and cad cells, and the fact that I know microvolt/microamp flame probes work since they are in many furnaces.

 

[sophiecentaur]

I have had little experience with flame sensing

Me neither but I would have thought that an IR imager would be the way to do it. A video camera with a filter over the lens which just passes IR would give a time resolution of better than 20ms and nothing special would be required in the way of experimental equipment,
Of course, there is nothing wrong with negative experimental results so the difficulties constitute a learning experience.

 

[Asymptotic]

No disrespect intended but the fact that the OP was expecting to blow fuses through a flame leads me to believe ANY of this is above their head. Basic electronic troubleshooting skills seem to be absent.

A fair point, although I've known freshly-minted techs straight out of school who had yet to experience their first "real world" design and diagnosis cycle, and found themselves in similar straits.

Troubleshooting heuristics aren't necessarily taught in school, and I've seen otherwise capable folk overwhelmed by their first foray into the field. One of my favorite books on the subject is "Debugging" by Dave Agans (although there's a soft spot in my heart for Polya's "How to Solve It"). This poster from Agan's website sums up the rules quite nicely.

 

[Averagesupernova]

A fair point, although I've known freshly-minted techs straight out of school who had yet to experience their first "real world" design and diagnosis cycle, and found themselves in similar straits.

Troubleshooting heuristics aren't necessarily taught in school, and I've seen otherwise capable folk overwhelmed by their first foray into the field.

You are quite correct. I was once in that position. That does not change the basic scenario. When I was in that situation being a greenhorn I could not have imagined trying to solve it in this manner. I needed some side by side guidance.

 

[jim hardy]

I'm really suspicious of software . It was not my field and i know just enough to be dangerous.

Absolute value of signals wired single ended and read differential ?
Indexing mistake reading that FIFO buffer into an array to plot ?

When you can't see what's wrong you have to make everything prove itself right.
Test the data acquisition with something that's too simple to fool you. Like the microswitch from a microwave oven door - there's plenty of them in trashpiles. When you can see the contact bounce for about a millisecond you have achieved good data capture.

Then characterize the flame detector signal by measuring one gap by itself at high speed with no averaging or smoothing..
I searched for high speed photography of a flamefront but the sites all wanted seventy-nine dollars to view it.

old jim

 

[jim hardy]

You are quite correct. I was once in that position. That does not change the basic scenario. When I was in that situation being a greenhorn I could not have imagined trying to solve it in this manner. I needed some side by side guidance.

Yes, we all started as greenhorns. It's too easy to forget how mysterious it seemed at first.

 

[Tom.G]

A video camera with a filter over the lens which just passes IR would give a time resolution of better than 20ms and nothing special would be required in the way of experimental equipment,

His experimental run lasts 15ms.

 

[Rive]

His experimental run lasts 15ms.

Yeah. The equipment actually fits with the task, it's just not known what it does and how should it be used...