A silly (?) question to you chemists out there

All Recipes and Chemistry!

Moderators: richardh08, Boophoenix, Lloyd

User avatar
Pyro-Gear
Site Admin
Posts: 3034
Joined: Wed Jul 10, 2013 8:24 am

Re: A silly (?) question to you chemists out there

Post by Pyro-Gear »

Just a thought guys what about trying Red Glyptal Insulating Paint.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

That might be worth a try. But at around £50 per quart (or more) it's a bit pricey.

Other brands are available. I see that there are temperature-resistant paints that claim to withstand 800ºC. As far as I can tell, they should also provide a sufficient degree (no pun intended) of electrical insulation.
Even when I'm wrong, I'm convincing.
User avatar
Pyro-Gear
Site Admin
Posts: 3034
Joined: Wed Jul 10, 2013 8:24 am

Re: A silly (?) question to you chemists out there

Post by Pyro-Gear »

http://www.ebay.co.uk/itm/RED-AIR-DRYIN ... 2548.l4275 yes there other varnish’s out there but that was the first thing I thought of, could be worth a shot Richard.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

"I see that there are temperature-resistant paints that claim to withstand 800ºC"
--------------
A thin coat of "exhaust manifold" spray paint might suit.

I've use the stuff, an it maintains its film AND its color at the hottest part of the manifold.

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
Pyro-Gear
Site Admin
Posts: 3034
Joined: Wed Jul 10, 2013 8:24 am

Re: A silly (?) question to you chemists out there

Post by Pyro-Gear »

I have used this before on an engine I rebuilt some time back it is very good http://www.ebay.co.uk/itm/EXT-manifold- ... Sw9N1V3HXf
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Definitely worth a try then. Thanks for the tip, guys.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

As I rather suspected, my first attempt at adding insulation didn't really work. Not only did it make very little difference, but it also made the grains a lot harder to ignite.

My can of heat-resistant paint arrived in the mail yesterday, so I'll give that a go. But first I'll have to make up another batch of thermocouple assemblies - sigh!
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

I still seem to be at the stage where I’m learning more about thermocouples than what I’m trying to use them to measure. And much of what I find out, I don’t understand.

As part of the assembly process, I attach copper leads to the wires by twisting each wire around a hook in the lead and then clamping it tightly with pliers. The twisted pairs themselves are generally made a day or two before I use them.

A few weeks ago I made a batch of assemblies and, immediately afterwards, I measured their resistance. All of them had a resistance of 3 to 4 ohms except one, whose resistance was much higher. Despite the fact that it doesn’t wet either of the metals I use, I applied solder to the joints between them and the leads and the problem seemed to go away. Except that when I re-measured their resistance a few days later, they were all at about 1.2 ohms. At the time I was a bit surprised, but didn’t think much more about it. When I used them, they all seemed to work as expected, with very little variation in output.

Yesterday I made another half dozen (I’m getting faster with practice) and, again, one had a much higher resistance than the others – over 200 ohms. Again, I soldered the junction, which helped a lot but only reduced the resistance to about 35 ohms. This morning – about 12 hours later – I measured the resistance again and it had fallen to 3 ohms.

It’s beginning to look as though, for some reason or other, the occasional assembly shows a much higher resistance than the others, but that they all show a drop in resistance over time, and I’ve no idea why.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

Man, Richard! That just isn't right.

Perhaps the wire needs to be 'de-oxidized' before twisting. That's the purpose of various corrosive fluxes.

I can't believe industrially-made thermocouples would 'age into spec'. I'm also pretty comfortable saying you won't get ideal performance from them until you solve this particular puzzle. <shrug>

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Lloyd, I'm sure I ought to be more worried about this than I am. But...

I think I've convinced myself that this is a problem with the connections to the lead wires rather than with the twisted pairs themselves. It happens with both newly made twisted pairs and ones that were made a week or more before final assembly. With more examples, I'm seeing that it doesn't happen in all cases. No doubt it would be better if I could find an improved connection method, thereby (hopefully) proving the problem was definitely due to that cause but, given the resources I have, I'm more or less stuck with how I do it now.

In any case, it seems to settle down within a day or two of assembly. I test each thermocouple before use and, so far, I've seen no significant difference in response, regardless of whether the resistance has changed in this way or not. The performance doesn't need to be ideal, as long as it is consistent - which is what I seem to be seeing.

I guess I mean that I intend to take the pragmatic approach of 'if it works don't fix it' - even if I don't understand everything that's going on.

On another tack, I've been experimenting with the heat-resistant paint. I tried applying it with a fine paintbrush, but that didn't work too well. I ended up having to dip them, which means that the twisted pair is also coated - which I hoped to avoid. Also, I see that, to be fully effective, the paint should be hardened for an hour at 160°C. It will be interesting to see how the rest of the assembly deals with that! All in all, I'm not too hopeful that this approach will fix anything.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

"... which means that the twisted pair is also coated - which I hoped to avoid."
--------------------
I believe that's exactly what you will need to do. The film thickness will be important, so as not to limit thermal transfer too much.

My suspicion is that the twisted junction is what's being affected when it's hot. If nothing else, the UNsoldered wires will increase in length as they get hot, and loosen the twists until they cool some. The paint might help mitigate that some. Solder would obviate the effect entirely, up to the temperature at which the solder flows.

As for your lead connections... have you considered crimped sleeves, instead of just twisting and 'mashing' the union?

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Well, Lloyd, you've been right about a lot of things, so let's hope this is another one. The good news is that all the assemblies survived being baked and now all have resistances around 1 to 2 ohms. Apart from a couple of cases where a tiny 'dew drop' formed at the very end of the twisted pair, dipping seems to have added a really thin, uniform layer over the wires, so I'm hopeful it won't affect the temperature measurements too much.

I hadn't thought about crimped sleeves, although Ken did make that suggestion - for the twisted pairs themselves - as a possible solution to the dropping signal problem, at the time when I suspected the twisted pair was going open circuit. Given the smallness of the wires, it wouldn't be easy to do, but I'll bear it in mind if the problems persist.

I've been looking into the contact issue and am now suspecting that the main culprit is the connection with the tungsten wire. Connecting via crocodile clips to a few centimeters of the molybdenum wire shows a resistance of an ohm or two, but doing the same thing with tungsten gives a wildly fluctuating value of anything from a few ohms to over 250 ohms!

Ken has suggested that the difference in expansion coefficients between the wires and the copper leads might give rise to stresses that, over time, improve the electrical contact. The fact that everything seems to have settled down nicely after the assemblies were baked at 160°C lends some credence to this idea.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

"I've been looking into the contact issue and am now suspecting that the main culprit is the connection with the tungsten wire. Connecting via crocodile clips to a few centimeters of the molybdenum wire shows a resistance of an ohm or two, but doing the same thing with tungsten gives a wildly fluctuating value of anything from a few ohms to over 250 ohms!"
-------------------
You've probably thought of this, but: I know the wires are tiny, but you might try burnishing the grip points with a VERY fine (say 1000mesh) carborundum paper before trying the clip thing. If that rectifies the problem (even for a few minutes) it would be a clue that there's some sort of oxide or nitride layer growing on the wire, until it finally passivates it against further attack.

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Well no, I hadn't thought of doing that, but it would definitely be worth a try.

However, after all the alarms and excursions, I've just run a simple calibration test on a set of 10 assemblies. Four of them were made two or three weeks ago and the rest were made over the last two days. They have all been paint-coated and baked, and include a couple of examples that initially showed an anomalous value for their resistance. The test was simply to take measurements (every 10ms) while holding the thermocouple in the steam emerging from a boiling kettle. In each case the readings fluctuate a little, but averaging over 10 second's worth of data gives a value with a standard deviation of about ±2%. Averaging over all 10 samples shows that they differ from each other by around ±3%. Also, there is no statistically significant difference between the results from the old and new assemblies, nor between the 'problem' samples and the rest.
Even when I'm wrong, I'm convincing.
User avatar
biffo
Posts: 278
Joined: Mon Oct 13, 2014 9:25 pm

Re: A silly (?) question to you chemists out there

Post by biffo »

Well Richard you are experiencing some strange results,for who knows what reasons .One of the things I learned about thermocouples is that you can make what is called a cold junction by using another dissimilar metal for connections like you might be doing with the copper leads.there are you must know compensating cables to be used with the different junctions if you need to be a long way from the temp controller. Sorry if you know all this Richard regards B
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

The cold, or reference, junction is at most only 2cm away from the hot one, and the leads are copper from that point to the amplifier. I'm relying on the fact that the wires themselves are thin and not particularly good conductors of heat, that the hot junction is only hot for a few seconds and that the copper leads are - relatively speaking - effective heat sinks. Given those factors, I'm assuming that the cold junction is at or close to the ambient temperature. The few results I've got so far seem to suggest that's a reasonable assumption.

Apart from the peculiarities that I've observed, the main issue is that the assemblies take so long to prepare. Given that both in their assembly and in building up the crackle core and the prime coating by slurry dipping there are long periods of waiting for stuff to dry, it can take up to a week to make a batch.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Over the last few days I’ve managed to get some more data, and done the basic reduction and analysis.

The first thing to note is that my attempt to coat the thermocouples in heat-resistant paint did nothing to prevent the dip in output at the moment of explosion. Either the coating didn’t provide an effective insulation, or the dip is caused by something else. I’ll have a bit more to say about that later.

I didn’t get good results from every trial: of 13 runs, one looks as though the thermocouple malfunctioned and another resulted in an empty data file (?!). I’ll clearly have to take another look as my software; perhaps I’m not being as careful as possible in detecting all possible error conditions. Also, there seems to be a recurrence of the spiky interference that I thought I had eliminated when I redesigned the amplifier. It appears in 6 of the 11 successful runs – and in one of them it stops halfway through the data. It can be ignored, so it doesn’t invalidate the results, but it is annoying and it would be good to isolate the cause and get rid of it.

The samples I tested were made from a variety of compositions, some of which were designed not to explode, and they all behaved broadly as expected – although two of the ones I thought might produce a report, didn’t. Interestingly, of the six exploding samples, two showed weak signals on the open circuit detector, coinciding with the sudden drop in the main output, and a third showed a modest signal, but the other three gave no indication. A zero signal on all outputs (shown by 3 out of 6 samples) indicates a short circuit, whereas zero on the main outputs and a significant O/C signal (shown, weakly, by the other 3) indicates an open circuit condition. Perhaps, on balance, the most likely cause is a short circuit, but it still isn’t entirely clear. As things stand, I don’t know how to eliminate the effect, so it looks as though I will have to live with it for now.

Apart from these issues, the outstanding problem is that of calibration. Although my rough measurements indicate there is no great difference in performance from sample to sample, I don’t have a simple, accurate and reliable way to create a true temperature scale. The best I can do still leaves a degree of uncertainty – and a lot of extrapolation beyond the reference points – in getting to the temperatures I need to measure.

Added later:
Looking more closely at the 'O/C' signal, I see that its magnitude is roughly related to the rate at which the main signal falls - the steeper the fall, the greater and the longer lasting the signal. I wonder if it is just caused by the amplifier overshooting. If that is the case, then it seems most likely that I'm seeing the thermocouple being short circuited.
Even when I'm wrong, I'm convincing.
sambo
Posts: 957
Joined: Sat May 16, 2015 6:04 pm

Re: A silly (?) question to you chemists out there

Post by sambo »

Just out of interest, what kind of temperature range do you need for calibration, can it be narrow or is there sufficient hysteresis that it needs to be at the end of range ?
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Well Sam, Mo melts at 2896K so that puts an absolute limit on the range I can measure. But what I've done so far suggests that I don't need to go that high. Only rarely do I see the temperature going over the limit set by the gain of my amplifier, which is currently at about 1400K - the bulk of the data up to now suggests that the temperature range of real interest is in the range of, say, 600 to 900K. Obviously it momentarily goes higher than that in the explosive phase, possibly up to the boiling point of copper at 2835K, but the 'short circuit' issue means that I can't measure that anyway.

I use the W/Mo combination because it's the only one I've come across that will survive the conditions I subject it to, and is available at a reasonable price in wires that are thin enough for my purposes. But it isn't a standard thermocouple, so I don't have any published data to go on. In the absence of information to the contrary, I'm forced to assume that the response is (at least approximately) linear with temperature.

At the moment I don't see many options for calibration, given that I'm stuck with no more than 'kitchen chemistry' techniques. Apart from the ambient temperature, I only have two reference points; the boiling point of water and the melting point of KNO3 (607K). There are a few other salts whose melting points I could use, but they are all around the 600K value. There are substances with higher melting points, but the practical difficulties start to proliferate as the temperature increases beyond, say, the softening point of my pyrex glass apparatus. Even at around 600K the results are harder to get than I originally expected.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Over the weekend I took another look at the calibration issue.

I used a more careful setup to measure the response when a thermocouple is immersed in molten KNO3 and got some significantly better data. Using several thermocouples seems to confirm my earlier suspicion that different examples give similar results, within a few percent of each other.

Overall, given that there is also some uncertainty in the calibration itself, I believe that I can measure temperatures to an accuracy of, at worst, around ±5%. That isn't particularly brilliant, but it's enormously better than the estimates I got from photographs.

I could possibly do better if I further refined the calibration process - or if I calibrated each thermocouple individually, but I don't want to have to do that if I can possibly avoid it.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

I've just had one of those moments, but I'm not sure if it's an 'Aha!' or a 'Doh!'.

I thought, "What if the open/short circuit signal is representing a genuine negative output?". I therefore changed its sign and tacked it onto the gap where the main signal dropped to zero. The result wasn't very convincing.

Then it suddenly struck me that the 'open circuit' detector is connected to stage 1 of my amplifier, and therefore gives a signal about 32 times smaller than my standard +ve output. So I multiplied it by 32 and tried again. Here's the result:
NegativeThcplOutput.png
Apart from a bit of a glitch around 4.6 sec, where the signal changes sign, the two signals merge more or less perfectly. And even the glitch is understandable, given the change in scale: there's a 5-bit (x32) change in scale, so '0' from the ADC could be any value from 0 to 31. I obviously need to tweak my circuitry so that there is a smaller discrepancy between the two signals.

It looks to me as though the signal really does go negative while the explosion occurs, and that that the cause is neither a simple open- or short-circuit. So what's going on?
You do not have the required permissions to view the files attached to this post.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

"It looks to me as though the signal really does go negative while the explosion occurs, and that that the cause is neither a simple open- or short-circuit. So what's going on?"
---------------------
Adiabatic cooling, when the gasses (and solid combustion products) expand with such velocity and mass that there results a partial vacuum at the original position of the grain?

I'm just guessing, but it actually sounds reasonable as I repeat it to myself.

Deliberately cooling one of your thermocouples might give a clue as to how much of a temperature drop it requires to create that signal...

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

At first sight, that seems a plausible suggestion, Lloyd. But if that is the explanation, then - whatever the relationship at low temperatures - it gets very cold. Assuming the relationship remains linear, then the predicted lowest temperature is -124K, which is physically impossible since absolute zero, or 0K, is the lowest you can go. (If you could, that would be nearly -400°C, or -690°F, and I've now seen a couple of cases where the voltage goes even more negative).

I could take your suggestion and look at the output when I cool the junction, but I think I can argue, on physical grounds, that adiabatic cooling can't be the answer. Clearly, since everything grinds to a halt at absolute zero, the generated emf will have ceased to change with temperature by then, so the slope of the graph of voltage against temperature must be zero at that point. That means that the average slope, between room temperature and absolute zero, must be less than the measured value of the slope at or around room temperature. In consequence, the drop in temperature corresponding to the negative voltage I measure must be even greater than the one I just calculated - which is even more physically impossible.

Or, putting it another way, what you are suggesting is that if I stuck my finger in the cloud of vapour surrounding an exploding crackle grain, it would get frozen rather than burned. I don't intend to try it.

The signal appears to be real. Looking at the positive aspect, that means it ought to be possible to deduce something about what is happening during the report phase. The only problem is that I can't think of any physical process that would produce that kind of signal. Without such an explanation, no conclusions can be drawn. Right now, I'm absolutely stumped.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

Richard,
If I had any money, I'd 'bet' that the low-temp output of that thermocouple isn't linear with temperature.

If my money were on it, I'd be chilling one of those to see what the effects were.

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Lloyd,

I know there's a flaw in my previous argument, in that the response curve could be S-shaped, but I don't think that is usual. Here's how a few examples behave at low temperatures (from zero Celsius, where they intersect, down to about 3K). As you can see, most flatten out as I pointed out in my previous post.
NegativeResponse.png
I'll definitely try chilling one, but there's a limit to how low a temperature I can use. At a push, I think I can get my freezer down to -23°C, but that's about it. I don't have access to anything colder.

However, I still can't believe crackle could get anything like as cold as that, but we'll see.

One thing that really bothers me is, if the gas/vapour gets colder than ambient temperature, how come it is still able to emit so much light? I realise that you are saying that it might only apply to the very core of the explosion and therefore only affect what is left behind, but all my physical instincts are telling me that's fairly unlikely. On the other hand, I've been wrong before - and I have absolutely no idea what else could cause the emf to go significantly negative.
You do not have the required permissions to view the files attached to this post.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

"On the other hand, I've been wrong before - and I have absolutely no idea what else could cause the emf to go significantly negative."
0000000000000

Heh! As have we all, sir! I also don't 'have a clue', just a theory. That's why we first speculate, then try to confirm!

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

I have to say it's not looking good for the adiabatic cooling hypothesis.

I had to think about how best to make a measurement at low temperature. The twisted pair is so quick to respond that I couldn't think how I could keep it at the right temperature while making the measurement. In the end I went for a 'backwards' solution. I added thermal mass to the reference junction by wrapping it in many turns of insulated wire and then covered it in a thick layer of foam insulation. I then put the whole thing in the freezer (at about -21°C) and left it there for a couple of hours. I then connected it as quickly as possible to the amplifier and made the measurement with the twisted pair immersed in water at 20°C. The result, added to my previous high temperature measurements, is shown here:
ColdCalibration.png
It seems to confirm that the slope of the graph decreases as the temperature falls towards absolute zero, which means that the negative output values I've measured can't be caused by a drop in temperature of any magnitude.

As a further check, I made the modifications to my circuit to improve the response to negative voltages and ran a quick test, with the reference junction at 21°C and the twisted pair dipped in a mix of water and ice. The response was more or less compatible with the above results, but a bit off. I have since realised that the discrepancy was caused by the fact that my changes have disturbed the zero offsets on a couple of stages of the amplifier. I'll have to try again after I've readjusted all the offsets. I'll only report the results if they contradict my current findings (but I don't expect them to).

The cause is now a total mystery. I think I'll have to have a little chat with my solid-state physics professor/friend.
You do not have the required permissions to view the files attached to this post.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

OK,
That seems like pretty good evidence that it is not adiabatic cooling. How about this theory:

For a short time, the entire thermocouple is surrounded by a 'cloud' of very conductive combustion products. (not plasma, but metal and metal oxide vapours)

It's not a 'dead short', but diminishes the voltage output to near-short values.

If I'm correct, you never see that phenomenon unless the particle explodes.

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Knowing my amplifier, I can definitely say that it is set up so that, for a short-circuited input, all outputs are zero (plus or minus one or two mV). The 3 'normal' outputs (at gains that differ from each other by a factor of 2) give non-zero ADC readings only if the input voltage is positive, and the O/C detector (I think I'm going to have to change its name!) only gives a non-zero output if the input goes negative - or open circuit, when its output saturates at the positive rail voltage. If the input is positive and partially or fully short circuited, there can never be a non-zero output on the O/C channel.

As far as I can see, the results so far eliminate as possible causes a) a short circuit (partial or otherwise), b) an open circuit and c) low temperatures. Any intermediate range output on the O/C channel must be due to a small but negative voltage across the input, however it comes about.

What if the two wires, surrounded by a conducting medium, turn into a voltaic cell? Probably unlikely and, even if that were the case, I don't know that it would help, as I have no idea how to predict its sign, even less the output value, or how it would vary with conditions.

I suspect we are both scraping the bottom of the barrel for ideas - I know I am.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

Richard,
Your idea that the thermocouple is becoming a 'cell' is not so wild. All sorts of dissimilar metals will act as a voltaic cell when in contact with an electrolyte.

You can test that one!

Immerse a working thermocouple in various electrolytes (like concentrated salt solutions, acids, bases, etc.), and see what it does!

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

The thing that worries me about that explanation is that, as a cell, it will be fairly effectively short circuited by the twisted pair. But you never know, and it won't take much effort to give it a try. I might also try making a cell with just the two wires, not twisted together. At least that could tell me if the emf was of the right sign.

Apart from that, I think I'll put trying to find an explanation on the back burner for now. I'll continue to measure it, as the signal clearly has to have some physical meaning - if only I knew what it was. Who knows, maybe more measurements will end up providing some new clues, or perhaps somebody might come up with a more convincing idea.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

I've just done a really crude trial, attaching one of my thermocouples to a multimeter and then dipping it in conc. hydrochloric acid. Not too surprisingly, there was no measurable voltage, but there were indications that it might have been triggering the sign indicator on the display. I then dipped a broken example into the acid and got a definite signal. It was highly variable, up to around 200mV and, perhaps more significantly, of the right sign.

In a way, I'm hoping that voltaic action isn't the cause, as I don't see how I could get any useful information from the measurements, but it certainly isn't yet ruled out.

I guess I'll have to try a more controlled experiment tomorrow.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Today I did a more careful trial, with the thermocouple connected to my amplifier. I took the opportunity to run a slightly wider set of tests so that I could also check out the recent changes I made to the circuitry.

I measured the outputs with the thermocouple immersed in a) concentrated hydrochloric acid at room temperature and – as controls – b) ice water, c) water at room temperature and d) condensing steam. Since my circuitry can now detect both positive and negative voltages, I ran each test twice, once with the thermocouple connected normally and once with the leads reversed. That allowed me also to check out the gain of the O/C channel relative to the other outputs, which turned out to be close to what I expected.

Dipping two dissimilar wires in a conductive medium ought to generate an emf, and that is what I observed yesterday. However, if the wires are in contact, and if they are perfect conductors, the measured emf would be zero. Since tungsten and molybdenum are not particularly good conductors, I suspected that a twisted pair, when dipped into a conducting solution, might still be capable of generating a small but measurable emf.

When I made the measurement, I found that there was a detectable negative voltage – which is what would be needed to explain the crackle data – but that it was very small indeed; typically about one hundredth of the value I see in the crackle data.

Now it could be argued that an experiment performed at room temperature with a liquid conductor isn’t likely to have much to say about a process taking place at 800K or hotter in a conductive gas – and, at least, the experiment does give a voltage of the right sign. But, overall, I have to accept that the experimental evidence I have so far obtained doesn’t settle the argument either way.

As I said in my previous post, I’m going to give this topic a rest for now, and go back to making more measurements on crackle. I’ll just hope that, sooner or later, some fresh inspiration will turn up – either from me or from someone else.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

My solid state expert appears to be on holiday, but I had a chat with another of my physicist friends, who suggested I see what happens with different types of thermocouple. At the moment I only have one other wire, nichrome, that might survive, so I tried making a twisted pair out of it and molybdenum. A quick test reveals that the combination is pretty useless as a thermocouple, with an output of around 1% of the W/Mo pairing, but it will be interesting to see what happens in the explosive phase. Unfortunately, it's been raining more or less continuously for the last 3 days, so the trials will have to wait until the weather improves.

Depending on the results, I might look into other combinations, but there aren't many options for selecting wires that are thin enough and don't either melt or oxidise under the conditions I subject them to.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Well, the nichrome wire proved to be fairly useless. Not only is the output feeble, but the twisted pairs turn out to be unlikely to survive. However, the fragmentary data I got from them does suggest that they also show a drop in output as the grain explodes - but there's not enough meaningful information to perform a sensible comparison.

I've got a couple more wires on order and I'll try those out when they arrive.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

While waiting for more wire samples to arrive, it occurred to me that, as another control, it might be worth also looking at different types of explosion. I embedded thermocouples in a small charge of BP and an even smaller charge of 70:30. Both showed a sudden rise in temperature, followed by a more or less exponential decrease back to room temperature, without the sudden negative signal that I’ve been seeing with crackle compositions. So, there’s definitely something different about crackle; I just wish I knew what the difference is.

Incidentally, all the tests I’ve been doing are showing that the W/Mo twisted pairs are incredibly rugged. Virtually all of them survive intact – and have turned out to be reusable. I can’t detect any difference in performance between new and used samples, and there doesn’t seem to be any evidence of the aging that thermocouples are said to show. Perhaps that’s because I’m using pure metallic elements rather than alloys, and/or because they are only subjected to high temperatures for very brief periods of time.

In one of the 70:30 tests, the binding I use to hold the wires in place was completely ripped off and the twisted pair itself ended up fully embedded in the end grain of its wooden support. But it still worked fine!

Oh, and I got to talk to my solid state guy this morning. Unsurprisingly, he didn't come up with an immediate answer, but I've got him thinking about it.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Having noticed that the dip in output was less pronounced for crackle mixes containing only small amounts of CuO, I thought it might be related to the amount of copper in the explosive products. There is no dip when I make measurements on either BP or 70:30 KClO4:Al but there is a noticeable (but significantly less than I see with crackle) dip in output in both cases if I mix in some copper powder.

To provide a further check, I quickly developed a couple of mixes containing either Pb3O4 or Bi2O3 and no CuO. Neither composition is particularly good, so I only got one decent measurement for each of them. However, both show a very strong fall in output, so the presence of copper can't be the cause.

Maybe it's due to the presence of any kind of metal vapour? But even if it is, I don't have any idea about how that could cause the output to swing as far negative as I'm seeing.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

I've just made an amazing discovery!

I'm not entirely happy with my calibration of the W-Mo thermocouples, particularly as I haven't been able to use any reference temperatures greater than about 600K. So I went on an internet hunt for any information I can find on the topic. I found an obscure, short paper that mentions that it has been known since at least 1964 that they show a reversal of their output at temperatures exceeding about 1300°C !!!!!!!!! That might explain a lot.

So far, I haven't been able to locate any more specific calibration data, but it must exist somewhere.
Even when I'm wrong, I'm convincing.
dave321
Posts: 388
Joined: Sat Apr 25, 2015 11:53 pm

Re: A silly (?) question to you chemists out there

Post by dave321 »

well done, it sounds like that may explain a lot Richard.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

Richard! That's a marvelous find!

Perhaps someone has even done some plots of what the voltages are at various temperatures above that 'reversal knee'. (or perhaps, another selection of metals will be necessary... irk!)

Thanks for that update.

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

'irk' indeed, Lloyd.

That paper has a few numbers for temperatures of 1300 to 2000°C, apparently taken from a Russian (GOST) standard. No units are given, but an intelligent guess that they are in mV seems to work out. Unfortunately, as yet I have no data in the 1000 degree gap between that data and my own calibrations.
ErgardtPlusTwo.png
As shown in the graph, applying a simple-minded quadratic fit to the combined data appears to work very well, but predicts a maximum value for the output that is less than I have measured. Maybe I can fix that with a bit more playing around with the numbers, but if I tweak the data too much I'm not sure I'll end up with a physically meaningful result. If I do, then it shouldn't be too hard to create a single temperature-time graph by stitching together the data from the two halves of the relationship.

But even if that works, there is a region around the maximum (say, from 800 to 1000°C) where the output is insensitive to changes of temperature. Of course, that is exactly the range where interesting things appear to be happening. Sod's Law strikes again!

Maybe I'll end up having to switch to a different pair of metals. But finding ones with sufficiently high melting points at a price I'm prepared to pay might be a problem. I've found a supplier, but their prices are astronomical - even for copper wire, they are typically charging around £100 per metre (and don't even ask about more exotic materials, such as alloys of Ti and Rh!)

P.S. Lloyd, since you were on-line early today (our time) I assume you have reached your place of safety. I hope so, and that all goes well for you.
You do not have the required permissions to view the files attached to this post.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

"P.S. Lloyd, since you were on-line early today (our time) I assume you have reached your place of safety. I hope so, and that all goes well for you."
---
Indeed. And it appears the storm is worst on the west coast of Florida (and I am near the East coast). But the storm is SO large that we are expected to get winds in excess of that required to damage our shop/barn. OTOH, unless a tree falls on our home or a tornado is spawned over it, it should withstand the winds OK.

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Glad to hear you're all OK, Lloyd. Let's hope everything survives back at home.

I've spent most of today doing sums. I revisited the calibration data, but this time I included all my own measurements, including the ones at low temperature, where the calibration curve starts to flatten out. To account for the extra twists and turns, I used a higher power curve fit. The best result seems to be a fifth order polynomial, giving the result shown below.
FifthOrderFit.png
The good news is that the peak agrees pretty closely with the maximum reading I get from the thermocouples.

After a lot of fiddling with the data, I came up with the following stitched-together temperature/time plot.
WMoUnravelled.png
Before anyone gets too excited, I should point out that I had to tweak the data quite a lot to get the various segments to join up reasonably smoothly. I had to scale the data by about 8% before a time of 4.7s (a little after the peak) and by about 13% from there on. My only justification is that this particular thermocouple might be slightly less sensitive than the average, and that there could be a slight, temporary reduction in sensitivity following the intense heating. Anyway, it works - sort of. There is still a slight mismatch at the switch-over points, at about 950K.

Because of the tweaks to the data, I can't be sure the whole curve is accurate, but I'm fairly confident that the measured peak, at around 2000K, is more or less right.

Well, it took me most of today to get to this point with only one of my sets of data, and there is still that 1000 degree gap in the calibration data to worry about. I might have to bite the bullet and fork out for some more suitable wires.
You do not have the required permissions to view the files attached to this post.
Even when I'm wrong, I'm convincing.
User avatar
Lloyd
Posts: 1902
Joined: Sun Aug 09, 2015 1:43 am

Re: A silly (?) question to you chemists out there

Post by Lloyd »

I won't get "excited", but that sure looks a lot more like the sort of curve I'd expect!

Lloyd
"Pyro for Fun and Profit for More Than Sixty Years"
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Here's another 'fixed' temperature-time graph.
G1P1Fixed.png
As with the previous one, I had to tweak the data a bit - scaling the data on the cooling side differently from the rest - to get the different branches to join up smoothly. This example and the previous one are the two easiest ones to fix in this way; most of my other data would need too much adjustment for me to really trust the result.

I'm beginning to think that the heat cycle really might temporarily fatigue the junction - or perhaps there is a hysteresis effect, with the response being different depending on whether the junction is heating up or cooling down. Given the unusual response of this pair of wires, neither of these two options would surprise me. It's a pity that the greatest uncertainty is precisely in the region where all the interesting stuff is happening.

I'm definitely going to have to switch to a different pairing, probably a type G junction (W and a 74:26 W/Re alloy). I've got calibration data for that one up to about 2500K, which at least gives me a staring point, and I already have one of the wires which will cut down a bit on the expense.

I'm also going to take the opportunity to modify my wire-twisting jig to make it a bit easier to use.
You do not have the required permissions to view the files attached to this post.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

My W/Re alloy wire has just arrived!

I'm going to have to be a lot more careful with how I make the thermocouples, as they will now cost around £4 each instead of the few pence-worth of materials in the old ones. The redesigned wire-twisting jig should help. Also, with any luck, like the old W/Mo ones, the majority of them ought to reusable. But I'll obviously have to check whether their characteristics change after each heating/cooling cycle.

The new thermocouple's output will be much larger, which is a good thing, but it does mean that I'll have to make another change to my circuitry to keep the signal within the range of the A/D converter. All in all, I reckon it will be a couple of weeks before I'm ready to get new data.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

I've made 10 of the new thermocouples.

I've just checked them out at 100°C (condensing steam from my kettle!) and they all give the same output, via my amplifier, to better than 2%. Converted to a voltage and a cold junction temperature of 0°C, the output at 100°C is 0.307mV, compared with the 'standard' value of 0.29mV.

Given the crudeness of my measurements, that seems good enough agreement for now. I'm going to take it to mean I can at least start making measurements by just relying on the 'standard' calibration data. Of course, that may need to change later, depending on the results I get.

Now I have to work out by how much I need to reduce the gain, and make the appropriate changes to the amplifier.
Even when I'm wrong, I'm convincing.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

I've finally got to make some measurements. I used three different crackle compositions, with two measurements - plus a single test of just the prime - for each. Out of the nine trials only one thermocouple was destroyed, and the remaining 8 appear to be fully reusable.

The results look much more reasonable - no mysterious dips in the output! But it will take me a day or two to reduce and analyse the data.
Even when I'm wrong, I'm convincing.
User avatar
Pyro-Gear
Site Admin
Posts: 3034
Joined: Wed Jul 10, 2013 8:24 am

Re: A silly (?) question to you chemists out there

Post by Pyro-Gear »

All this sounds very positive Richard, it is good news that the thermocouple is holding up and can be reused because they're not exactly cheap.

Look forward to the data results.
User avatar
richardh08
Site Admin
Posts: 2226
Joined: Wed Aug 07, 2013 1:54 pm
Location: Bedfordshire

Re: A silly (?) question to you chemists out there

Post by richardh08 »

Some initial results are starting to come in. The new thermocouple definitely works a lot better than the old one, but it is clear that I'll have to make a lot more measurements before the full picture emerges.

The graph shows a recording of a grain made to Lloyd's recipe, ignited with a BP prime. The red curve is the raw R+G+B value from a video, indicating how the surface brightness varies. The video recording didn't catch the point of explosion at all!
LloydNo2.png
One problem is that everything is happening a bit too fast. For example, it is certain that the real peak temperature is a lot higher than 1600K, but the twisted wire pair just can't respond fast enough. Also, the report I heard was a double one, indicating that some degree of spalling was taking place, which may well have distorted the data. I think I'll need to slow the mixture down a little, probably by using a coarser MgAl, before I try again.

However, the results so far seem to confirm what I suspected from the old measurements; there is a definite pause in the rise in temperature just before the explosion takes off. Again, as I suspected, it seems to be in the precise temperature region where the old thermocouples had virtually zero sensitivity!
You do not have the required permissions to view the files attached to this post.
Even when I'm wrong, I'm convincing.
Post Reply