Initial budget discussion

Added by J. Simmons over 12 years ago

Wanted to get the ball rolling on research for the budget. Post items we are considering using, and links to sites where we can buy them and the price.


Replies (15)

RE: Initial budget discussion - Added by Greg Moran over 12 years ago

From the book: "50 Model Rocket Projects for the Evil Genius"

Project 46: Measuring Exhaust temperature (Thermocouple)
Paralax thermocouple kit includes 2 thermocouples that are suitable for rocket motor exhaust temperature readings: the Type K (chrome/Alumel) and the Type J (iron/constantan) thermocouples. The third sensor, Type T (copper/constantan) is only rated to 400 deg C.

Link to the google shopping page $34.95 + S&H

RE: Initial budget discussion - Added by Ben Barnett over 12 years ago

I believe that type K thermocouples are preferable to "J". The "K" is more resistant to oxidizing environments and more resistant to corrosion in most environments. I don't know the chemistry of rocket exhaust, but I expect it to be somewhat corrosive.

I must study the I/O capabilities of the Arduino, and the signal-conditioning board in the proposed kit. This points directly to the Off-The-Shelf vs custom design question. I probably have some "K" wire in stock, and would have no problem designing analog signal-conditioning. Note that the wire in the kit appears to have PVC insulation which is NOT appropriate for our application. Thermocouple wire with refractory insulation is readily available from Omega Engineering (www.Omega.com). Omega is my personal favorite source for measurement supplies, including O-T-S solutions for measurement requirements.

Ben

RE: Initial budget discussion - Added by Aaron Harper about 11 years ago

Ben, you are correct that rocket exhaust is corrosive, but it is also fouling, particularly in the outer portions of the flame. This makes life interesting for the motion components of the test table, whether it be linear bearings or a drawer slide. This is less of an issue when measuring temperature though, since any coating would be sloughed off in the discharge before it could build up to a degree where it would dampen or otherwise affect the readings.

As far as doing some signal conditioning and amplification of the thermocouple signal, you need an amp, and you cannot solder the connections or you will destroy the effect by adding another metal to the mix. Foran amp a reasonable price, check out http://microcontrollershop.com/product_info.php?products_id=4643. This came from a discussion on thermocouples at SparkFun forums at https://forum.sparkfun.com/viewtopic.php?f=32&t=31407 While SparkFun does sell a suitable thermocouple, they do not sell the amp breakout, though oddly enough they stock the parts for one. Adafruit has one at a higher price at http://adafruit.com/products/269?gclid=CJmswJGi47QCFYw-Mgod8GYAjw

Another issue is that the insulation on most thermocouples, including the one at SparkFun is only rated to 900F. This means everything that enters the stream except the junction itself should be shrouded with something to protect the wire. While this is traditionally done with a ceramic bead, I was initially looking at Mylar-Aluminized-Kapton (MAK) which I have on hand. The trouble is that MAK is not flame retardant at all. The stuff I have will withstand 2000 degrees for 15 sec with some degradation, but will light on fire with a lighter. It may be possible to use a paint-on boron-nitride coating such as those used in the casting industry such as http://www.zypcoatings.com/Datasheets/BnHardcoat/BnHardcoat.htm to protect the thermocouple wire through it's usable life. While this product is WAAAAY outside the budget, it may be possible to get a sample or donation from the company if they know what we're doing.

RE: Initial budget discussion - Added by Ben Barnett about 11 years ago

Thermocouples are something that (unfortunately) I do understand. First, use thermocouple wire that is stiff enough to be self-supporting in the exhaust stream, then don't worry about the glass-fiber insulation. It might turn to fuzz, or even melt, but the thermocouple wire needs some space to avoid stem-effect cooling. Omega doesn't stock nearly the variety of TC wire that they did even 5 years ago when my job left for China. I think that my former employer has a 1000 ft. spool of Omega p/n GG-K-20 wire, and I may be able to run off with about 50 ft. You might need smaller wire if you want really fast time response. I don't really know the characteristics of your rocket exhaust, but the mass-flow must be pretty high because mass X velocity is thrust. If you really need faster response, #26 wire might still be mechanically strong enough to not be blown away. Find it here:
http://www.omega.com/ppt/pptsc.asp?ref=XC_K_TC_WIRE&Nav=temh06
...and scroll down to GG-K-26.
The braided glass insulation loses most of its mechanical strength above about 500 C, but you don't need mechanical strength near the junction anyway. #20 wire is most likely strong enough to hold its position in the exhaust stream, so it would need no mechanical help. What you DON'T want is alumina protection tube anywhere closer than about 10mm to the outer part of the exhaust stream. Alumina has very high thermal mass, and will act like a major heat-sink if it is too close to the junction.

Another suggestion is to use Omega's “miniature” thermocouple connectors at a safe distance from the exhaust stream so that the thermocouple wire doesn't get any repeated flexing while changing the rocket motor. Type K wire work-hardens quickly, then breaks easily. The metal clamp on the plastic connector should probably be at least 50mm away from the exhaust. I generally put a piece of polyolefin shrink-tube over the thermocouple wire where it goes through the clamp to keep from fuzzing the glass-fiber insulation. A simple “X” pattern resistance weld will work for the thermocouple junction, but if you don't have a spot-welder or something similar, a molten-salt weld will work. Molten-salt is fairly easy with a bit of borax, a carbon rod, and a transformer that puts out about 3 volts at about 25 amps.

Your $11 thermocouple amp is cheaper than anything that I could do in small quantity. I see the amp pc-board is laid out with enough copper around the input so that the chip is the same temperature as the thermocouple connector so it can measure input connector temperature and provide cold-junction compensation. My designs used a YSI 44016 thermistor under the input connector and did the cold-junction compensation in analog, but that makes it very expensive compared to your COTS amp for $11. However, the analog amp is much faster than 5 samples-sec. I never tried, but I think that I could get the bandwidth up around 1KHz if you want real speed.

If you want cheap refractory cement, look here:
http://www.menards.com/main/heating-cooling/fireplaces/fireplace-accessories/furnace-cement-fireplace-mortar-gray/p-1705842-c-6845.htm
...but don't buy it yet because I have plenty left from a recent chimney mod. If your thermocouple seems to have a problem with flow-induced vibration, not unlikely with finer wire, add a few strategically placed beads of furnace-cement to damp the vibration. You could even run a piece of steel wire (like a heavy-duty paper-clip) from the connector clamp to about 10mm from the outside of the exhaust stream and add a bead of furnace-cement at the end of the steel wire.

RE: Initial budget discussion - Added by Aaron Harper about 11 years ago

I see what you mean about the conversion time. .17-.22 sec is just too slow in a 6 second burn, we'd only get 30 to 35 readings, certainly not enough to see a rise or "chugging" as the result of a poor burn, nor would it give us data we could reconcile to a real time event. Even with something like this http://www.ebay.com/itm/Type-K-Thermocouple-Ceramic-Kiln-2372F-High-Temperature-Probe-Sensor-/110990651658?pt=LH_DefaultDomain_0&hash=item19d78ef10a , the readings would be slow, and from what I can tell from some basic research, this is a function of k-type thermocouples. Is this correct? Perhaps what we are looking for in the next test stand is an optical sensor or pyrometer?

I found this http://www.melexis.com/Infrared-Thermometer-Sensors/Infrared-Thermometer-Sensors/MLX90616-779.aspx but the response time is still .15 to .22 Sec.
http://www.futureelectronics.com/en/Technologies/Product.aspx?ProductID=MLX90616ESFHCAMELEXIS5018703&IM=0

RE: Initial budget discussion - Added by Aaron Harper about 11 years ago

I had a thought... If what we or the industry needs isn't available for a reasonable price, we make it and open source the design, right? What if for a later version of shepherd we made our own open hardware pyrometer as a level 1 kite?

Open Hardware Pyrometer

Purpose
To produce a method of measuring and displaying the discharge exhaust temperature of a rocket engine for future versions of the Shepard Test Stand.

Construction Process
  1. Take a webcam and remove the IR blocking filter.
  2. Add an optical low pass filter to allow the IR to pass, but none of the visible light.
  3. Optically split the image with a prism
  4. On the bottom half of the image let it image a regulated heat source where the temp is known, possibly a small light bulb with a thermocouple and regulator.
  5. Feed the upper half of the image with the filtered optical image of the rocket engine's plume.

Principle of Operation
The brightness of the upper image, relative to the intensity of the lower image will give you a ratio to multiply the known temp of the regulated source by to derive the temp of the plume at a scan rate of 30-60 readings per sec. Note that as the brightness changes, the camera will electronically iris down, but will do so on both halves of the image, which preserves the ratio. This data could be displayed as a value in real time with the rest of the readings from the control system. If the image were automatically cropped to include only the top image and normalized to register the peak temperature as the maximum scale in a post production process, this would provide a 30-60 fps movie of the thermal energy during the engine run. This image analysis and storage is something a Beaglebone or other ARM processor running Linux can handle. For something similar, but without the reference or normalization, have a look at http://youtu.be/qeuJKUIw220.

RE: Initial budget discussion - Added by Jeremy Wright about 11 years ago

The pyrometer is an interesting idea. We ended up deciding not to measure the exhaust temp in the 1.x versions of Shepard, but we'll need to measure it at some point. We are measuring the casing temperature in one spot though.

As far as cheap thermocouple amplifier chips with higher sample rates go, this one was suggested to us at the Open Hardware Summit. The biggest problem for me using it on Shepard 1.1 will be that it doesn't come in a DIP package. Not a show stopper, but not as convenient.

RE: Initial budget discussion - Added by Aaron Harper about 11 years ago

I like the fast rate on this, and I can fix the DIP issue: https://www.sparkfun.com/products/495

The trouble I am finding is that if I have read correctly, the K-type thermocouples have a set rate at which they can respond to a change in temp, which coincidentally is a lag of .17 to .22 seconds. Now, I don't remember this from school, so either I was absent that day or I am wrong. Ben, would you care to weigh in here?

RE: Initial budget discussion - Added by Jeremy Wright about 11 years ago

That's a good find. The Arduino ProtoShield we're using has SOIC pads, but I worry that my soldering skills and equipment aren't up to the task. Any suggestions?

RE: Initial budget discussion - Added by Aaron Harper about 11 years ago

It's not that hard to do. Have a look: http://youtu.be/5uiroWBkdFY

  1. Clean, flux, and tin the pads.
  2. Tack down the opposite corners.
  3. Put a good amount of solder on the iron's tip and reflow it over the pads (don't worry about bridges)
  4. When the solder is on both sets of legs, use a clean dry iron to heat the pads and legs.
  5. This will reflow the solder and should clear up bridges. If not use a bit of solder wick.

Worst case, stick it in a static bag, put that in an envelope, and mail it to me. I'll be back in the lab on Tuesday the 22nd, and my workbench has not been packed up yet.

RE: Initial budget discussion - Added by Ben Barnett about 11 years ago

Re: Thermocouple response time - is entirely dependent upon the thermocouple and the medium being measured. Fine-wire thermocouples can respond in a couple milliseconds when welded to a lamp filament for example. On the other end is the ceramic-kiln pyrometer thermocouple that was linked here. The wire is heavy-gauge, and the insulators add even more thermal mass. In the ceramic-kiln application, this is fine because fat wire has a much longer life at high temperature, and the kiln temperature responds to electrical power variations with a time-constant on the order of many minutes. Rocket exhaust has high mass-flow, which will certainly favor faster response, but #20 wire is still pretty massive, which will slow the response. If I can get my act together (kinda unlikely) I will get some #20, type K wire and test the response in a propane torch flame. The setup will be simple using a torch with push-button ignition. Just clamp the thermocouple wire to the torch tank with a large hose-clamp so that the junction is at the tip of the inner flame. Turn the valve off and watch the response. This will be the “still-air” response, which I expect to have a T-C of several seconds. Turn the valve on, push the button, and record the response. This will be the response with the mass-flow of the torch flame. This will still be slower than what you are looking for, but the torch flame mass-flow is probably a couple orders of magnitude below that of the rocket exhaust. If the response is too slow, go to finer wire. Just a thought here: Using an X pattern weld with the excess wire clipped a bit past the weld should give more surface-area exposed to the flame, while adding only a little mass, which should also speed response compared to the ball produced by a molten-salt weld.

Re: Optical pyrometery – is dependent upon the temperature and also the IR emission profile of the item being measured. As an example, an oxidized Inconel fixture with an alumina plate cover moving through a tunnel-furnace has the Inconel and alumina at the same temperature. But the Inconel looks “brighter” than the alumina. However they both look about the same color, so I would guess that a differential measurement at two or more wavelengths would be much more accurate than a simple intensity measurement. I think that the video at 30-60 FPS is a great idea. I suspect that, given the high temperature of rocket exhaust, measuring the ratio of green to red might give a fast and accurate temperature measurement. The only problem I see with this concept is the opacity of the outer part of the “flame”, which would be much cooler than the inner “flame”. I just dunno, but video analysis would be a very easy experiment to perform. Does anyone have video of any of your test firings where the exhaust “flame” doesn't saturate the sensor? Astronomers measure the temperature of distant stars by analysis of visible light, so maybe rocket exhaust is hot enough to measure without an IR channel.

RE: Initial budget discussion - Added by Ben Barnett about 11 years ago

Interesting indeed!

But I think not applicable to the current discussion. Note the top of the temperature range at 300 C, far below the temperature of rocket exhaust. This type of imaging uses wavelengths about 100X longer than visible light, 70um vs 700nm. From the article:
"...Fully assembled and tested IR-Blue. You are ready to start seeing the world in far infrared..."
Note the term "far infrared".
I have never seen an Estes rocket motor fired on a static test stand. Can you see the "flame" within the cloud of smoke? What color does it look like? Red? Orange? Yellow? White? These would all be measurable in visible or NEAR infrared. Does anyone have video? If so, please send a few representative frames to:

Small files size is more important than high resolution. 160 X 120 is plenty of resolution for color analysis.

Another question: How did this discussion get from "Future Design" to "Initial budget"?

RE: Initial budget discussion - Added by Aaron Harper about 11 years ago

This indicates there are more folks than us working on the issue. The thermopile sensors he uses still have that .2sec lag. He did a nice job on integration though.

Whatever this sensor becomes, will not be used in a version 1 Shepard, even 1.1 due to the limited budget. The soonest we could see this becoming useful is version 2.0, but more likely 3.0 or beyond. As such, I doubt we'll be using Estes engines on these for anything other than a test run. We should probably focus on performing measurements on some of the real high power motors to make sure we meet that need about the same time Shepard is ready to run them. Consider these: http://youtu.be/0srcCA3OUyY and http://www.ddeville.com/derek/CSXT.htm

I have a feeling that the visible flame color has more to do with the chemical composition (metal ions) of the fuel than it does with the temp. Potassium hydroxide for example will make the flame appear a striking purple without making a flame hot enough to do so naturally. Consider this: http://youtu.be/jJvS4uc4TbU. This is my reason to block visible light and focus on the infared, where both frequency and intensity are tied directly to the energy being released.

How'd we get here in a budget discussion? The road was long and winding. It isn't the first time an aerospace project had spin off projects... In fact, I would say that is fairly traditional, and I'd be worried if it didn't happen. Now that you and I are asking that next question, perhaps we should create a project to evaluate this specific project and tasks. Worst case, it goes nowhere and we just delete the entry or leave it as a learning example, complete with post-mortem discussions and project closure documentation. I'll have the project up in an hour or so, and you'll be listed/informed.

(1-15/15)