Digital Test Stand

Remember 2001?
 I can recall bits and pieces.
 But through the magic of the Web I can recall my previous acts, many of them more notable for bravery than intelligence.  
Do it once, share it once, and be embarrassed many times.    

An example:  Remember the digital bathroom scale I bought to test the static compression strength of my propellant?  
Well I do, because I've been using it ever since.  

It looks like this:


Several descriptive words come to mind.  "Mundane" is perhaps the least insulting.

But it is an admirable thing,  It's cheap, sturdy, and highly functional.  It is still running on the battery I bought with it.

Even back then I wondered what was inside this $25. scale.  Obviously it had some kind of load-sensing device and electronics to read it and display the results.  
I wondered if it could be hooked to a computer.  

But I left it intact, suspecting that if I took it apart, it would never work again.  Many consumer appliances have suffered this fate in my eager hands.  
This scale was too young to die.  

The Digital Age

Recently Steve Ghioto built me a couple of nice test stands and the INA 125 amp that goes with them.  Studying his model and several other sources, I have learned to build my own little amps.  I have several of them now.
Thus I have become a load cell monger, ordering a couple from Aerocon Systems  and a couple more from Jeff Grey.  But my first test stand is still in the works.  

While engaged in a rare cleaning frenzy, I found this scale hiding under a cabinet.  I get a sudden urge to see if it can be disassembled and reassembled by human hands.  This promised to be much more interesting than cleaning.

It was.

Underneath, two small, strong springs hold the top to the bottom.  Bravely, I unhook them with a pair of pliers.

There are surprises under the lid:

Click Here  for a better view of these parts.

As expected, there is a cute little load cell connected to the electronics.  As unexpected, there is a clever system of levers that transfer a portion of the weight from the upper platform to the load cell.  All in a vertical space of less than 2 inches.  Somebody did some fine work on this frugal design.  

The levers distribute the pressure so that one can stand at any spot on the scale platform and get the same weight reading as any other spot.  
By reducing the pressure delivered to the load cell, this design allows the use of a much smaller cell.

Here is a link to the Health-O-Meter page on this scale.  They also offer a stealth model.

And they kindly provide a description of how different types of bathroom scales work.
This describes what I saw under the hood pretty well.

But digital scales may work in other ways - Richard Creamer told me that he had taken apart a different model and found that it used a mechanical scale with an opto-electric sensor.  As one stands on the scale, a perforated disk spirals down under pressure and an optical sensor reads its movement by counting the holes as they pass by.  Sounds like the way the ball-type computer mouse works, with its two slotted wheels.  Bottom line:  don't count on any given "digital scale" to work exactly like this one, you might want to peek under the hood before buying it.

Even the Health-O-Meter page lists different types of scales in their lineup.  One of their better ones is described as having four load cells.

Back to my project.  


The load cell has four wires.  I am guessing that the red and black are for excitation, the yellow and blue for signal.  That guess will prove to be correct.  Wish I had cleaned the dust off before taking the photo, but this thing had been sitting around the house for over three years - it's much cleaner than I should have expected.

Emboldened by the fact that I have just purchased an identical new scale from K-Mart for $17.99, I clip the load cell leads.  They are kinda short and very thin, so I solder on some extension wires and hook these to my load cell amp, Dataq unit, and computer.  I press on the dainty load cell and get a fingertip bump on the graph.  
It works!  

The load cell is very responsive.  One finger can exert 150 lbs and it doesn't even hurt.  Maybe this cell would be good for testing model rocket motors?  It's certainly not strong enough for HPR motors.  Oh well...  Nice try.

Then I remember that this scale is rated for 300 pounds with the lid on.  I have stood on it myself many times, and some heavier folks have used it too.  
Maybe whoever designed those beams and levers knew something.

So I replace the top lid, stand on the scale, and find that I get a nice, fat bump in the Windaq window.  
I pick up 60 lbs of barbell weights and step back on:  A bigger bump.  I get Teresa to stand on it with me (wheee!) and it still doesn't bottom out.

Another flash of activity - I remove the factory electronics and install one of my  INA125 amplifier boards in its place.  
I drill a holes in the side for a switch and an RCA jack, and pop-rivet in two battery clips.  This switch is almost too big.  There is a lot of horizontal space but very little vertical, and the edge of the weighing platform just barely misses the switch lever.  If I had put it 1/20th inch higher, it might have touched the platform, possibly interferring with those clever levers.  Next time I will use a smaller switch.

(Click for larger photo)

The load cell wires are:
Red:  Excitation +
Black:  Excitation -
Blue:  Signal +
Yellow:  Signal -
I drilled a hole for a power-on LED but didn't use it.  
There is already a red plastic window on the top of the scale, and the amp board is right under it.  
So I attach a clear LED directly to the terminal and bend it up and over so it can be seen through the window.  

Top back on, I put different weights on this scale and watch the computer screen to get a sense of how it might be calibrated.  The cell appears to have a no-load current flow of about 2.6 volts.  Fortunately, the Windaq software allows adjustment for this kind of thing, but it limits the voltage range that can be tested.  Good news is that the remaining 2.4 volts is enough.  (Any of you electronics geeks know a way around this?  I bet you do!)

After twiddling the gain on the amp to what seems like a happy medium, I calibrate the scale by stacking 10-lb barbell weights on it one at a time up to 100 total pounds, and reading the resulting voltages.  

The volt-per-pound ratio is surprisingly linear.  

First Test

No way I could wait until morning.  
Fortunately, I am out in the boonies and can make strange noises at night.  

The 38-360 casing is loaded with a single uninhibited grain of rcandy, 3.5 inches long.  
This should provide a vigorous test.  
The motor is strapped to a crude support which is set upon the scale.

Click to go to the Tests web page for more details.  

A bit glitchy, as you can see.  There are unexplained peaks at the beginning.  I often see these spikes using other test stands and believe they occur when the ignitor resists ejection.  I use large, crude ignitors.  But there is usually only one big spike.  The bumps that follow seem to occur at regular intervals, suggesting a resonance effect.  

This test also recorded a substantially higher total thrust than would be expected.  That makes me suspicious.  But it does crudely resemble an uninhibited grain thrust curve, the figures are within the realm of possibility, and I have never fired this exact configuration before.  More tests are needed.  Life is so hard.

Double Take

Funny that my best ideas come in my sleep.  Too bad I don't remember most of them.  Maybe that's why I need to sleep - the rational mind must defend itself against too many ideas.  

But this one occurred to me upon waking, when my defenses were down:  Piggy-back.   Place the smaller load-cell test stand Steve built on top of the bathroom scale.  The Dataq unit has four channels.  Use two of them.  Fire one motor and get simultaneous data from both stands.  See if they give similar readings.

It was after dark again, as this area is still on Daylight Wastings Time.  The motor is more conservative in design, burning two Bates grains in the 38-240 casing.  I stacked the test stands, calibrated them simultaneously with barbell weights, and mounted the motor.


Click to go to the tests page.  This is test 1-7-05B

They look kinda similar, don't they?  The left graph is from the load cell test stand A, using the 44lb cell from Aerocon Systems.  Note that this load cell handles much higher pressure than its nominal limits.  In my testing with known weights, its response is linear and accurate past 100 lbs.  The bathroom scale graph is not as smooth, and the readings vary slightly.  But it is pretty darned close!

The bath scale graph is much smoother than in my first test.  I get the idea that the weight of the load cell test stand may have something to do with that, holding the motor more firmly in place.  

OK, that was pretty good.  But this scale is rated to 300lbs.  Can it handle really handle that?

Double Take 2

I loaded the Loki 54mm motor with a single uninhibited grain 6-1/2 inches long - a little more than half the propellant length of this motor.  This should provide approximately the same thrust as a full load of Bates grains, but only about half the burn time.

It is placed in Steve's larger test stand, the one with a 500lb load cell, and this stand placed on the flabometer.  


See the sparks?  I have been testing ignitors using a pinch of Ti flakes instead of Magnesium turnings.  Mg ignitors are very reliable, I just want to know if Ti works as well.  Some of the flakes are getting flushed out of the motor as the thrust builds.

Click to visit this test on my Tests page

Again, the graphs show similar shape.  Again, the flabometer is a little rough, and this time it registers somewhat lower thrust than the 500lb load cell.  I wonder, though, if the weight, flex, and possible springiness of the load cell and its heavy base influence the reading of the bath scale.  After all, the flabomatic is getting secondhand thrust, from which some information has already been extracted, no?

Interim Summary

Results of these tests make me optimistic that the Flab-O-Graphter could be useful in testing small to medium sized rocket motors.  It certainly seems cost-effective, and could give an experimenter a good start toward a more sophisticated testing setup.  

I do not believe it will be as accurate or reliable as a well-built test stand using a certified load cell.  The long levers, while very clever, sseem likely to introduce artifacts.  It is possible that some of the roughness in these thrust curves is due to resonance artifacts in this system.

Damping that resonance is on my agenda.  Installing a foam pad under the levers might help.  I'm open to suggestions.

In all of these tests, the motor stand has been sitting loosely on top of the scale, held only by its own weight.  I will fasten it down on future tests.

As always, further tests are pending.  Good thing.  This is much more fun that whatever I ought to be doing.

Questions, comments, suggestions, and irreverent remarks are welcomed!

Postscript:  Two more tests have been done, both with the motor mount fastened to the weighing platform.  In the first of these, a delay grain failed, allowing propellant gasses to burn off some of the cover to this scale.  Fortunately, no serious damage was done.  I removed the scorched parts and tested again.  Fastening down the mount does seem to smooth out the thrust curve somewhat.

These are now on my Tests Page (scroll to the bottom)

Several folks have made suggestions, such as Mark LeBarre suggesting I brace the levers to reduce possible resonance effects.  I will try these as time permits!

Jimmy Yawn
Recrystallized Rocketry