Volume 2   Issue 3               Fall 2001

First on the Scene

Putting Sparklike's Revolutionary Gas-Fill Detection Device to the Test
by Randi Ernst

Windows produced today are marvels of technology, distant cousins of those produced just a few short years ago when what was not made of wood was made of clear glass and putty.

First came sealed insulating glass, then low-E coatings, inert gas fills and insulating spacers. We now have spectrally selective coatings and innovations like self-cleaning glass, while second-generation chromatic coatings are waiting in the wings. 

For window and glass manufacturers things have become complicated. It has been apparent for some time that test equipment and revised test procedures were needed to help sort out what we are producing and make reasonable assurances that products were being fabricated properly.

One of the holy grails has been the ability to ascertain the gas fill without destroying the unit. Dr. Hakim Elmahdy of the National Research Council Canada demonstrated a working prototype of such a device at the summer 2000 Insulating Glass Manufacturers Association of Canada (IGMAC) meeting in Toronto. (The Sealed Insulating Glass Manufacturers Association and IGMAC have now merged to form the Insulating Glass Manufacturers Alliance. See related story, Summer 2001 Door & Window Maker, page 50.)

Also last year, the U.S. Department of Energy awarded a contract to Aerodyne Research in Massachusetts to develop a unique gas sensor. Aerodyne has developed a sensor that can detect the amount of air in a sealed unit by passing a unique light beam through the glass and bouncing it back to a receiver. It is truly non-contact and non-destructive. It also may not be affected to the same degree by the various coatings as long as its light beam can shine through the window. The disadvantage is the reading of air instead of argon and therefore an indirect determination of the argon content.

Sparklike’s GasGlass

While still a prototype unit, the first gas-fill detection device to be offered for sale is Sparklike Ltd.’s GasGlass. Recently, I had the opportunity to work with one.

Does it work? Yes, it gave accurate readings. 

There may be difficulties with certain configurations, but based on the sample units we tested it yielded satisfactory results. We placed a mix of certified gas, 90-percent argon and 10-percent air, in test units with clear glass and three different low-E coatings. The GasGlass gave consistent readings showing a 90-percent fill (90.0 percent to 90.7 percent).

Problem Configurations

Which configurations did not read well?

It appears the Sparklike device uses a high-voltage tesla coil to generate the spark that creates the plasma. To obtain a reading it was necessary for the spark to jump across to the opposite glass lite. If the spark would or could not jump across, it created a starburst appearance on the glass. This became an instant way of knowing if you would get a reading or not: If the spark jumps across it will read; if it starbursts it will not.

If you apply the wand from the low-E side it will starburst and not jump across because of the metallic low-E coating shorting out the spark. If the gap was wider than 1/2 inch it would not jump across. (We tried 5/8-inch airspace and were unable to get a reading.) It could be that certain tints will have a metallic base and may not allow the spark to jump across.

With just air in a 1/2-inch unit it would not read. The literature states a minimum fill of 50 percent is needed to obtain a reading.

“We took readings in room light, against a backdrop and suspended in air. We also held a unit up to an outside window and took a reading 
successfully. More tests should be
done along these lines but initial opinion is that ambient light is not a factor.”

It did not appear to be dependent on ambient light. We took readings in room light, against a backdrop and suspended in air. We also held a unit up to an outside window and took a reading successfully. More tests should be done along these lines but initial opinion is that ambient light is not a factor.

Another area of concern may be the spark. The tesla coil idea has been around a long time. When studied years ago one concern was that the spark would burn the coating. The Sparklike has a 1-second-or-so spark at a lower setting than the tesla coil we have here at FDR. In retrospect, I should have made repeated readings in one spot with the Sparklike device and then inspected the glass but did not think of it at the time.

I did use a tesla coil here at FDR on a glass sample and after a few seconds you could see fly specs where the coating was burned by the spark. Additional testing should be done to see if the Sparklike damages the coating.

This would not be a big deal if the Sparklike were used just for process testing and certification, but it could pose a problem if it were used on production or field units.

Computer Interface

The interface is novel but has a couple of shortcomings that may be worked out for the production version. It uses a Compaq Ipaq palm computer to display the results. While clever, it had a couple of shortcomings. Our unit had been unplugged for a couple of days and had dumped its memory. We then proceeded to fumble around with unclear instructions until we were able to reload the software.

There is a histogram of sorts on the display but there is not a date/time stamp or other means of connecting the data to a specific test or unit. There is also (at present) no way of printing out or saving the data. I also wonder how well the Ipaq will hold up on the factory floor. While fine in the office or for lab-type tests, I would be concerned about the durability of the Ipaq and its touch screen when it encounters the sealant-covered paws of the typical glass-plant operator.

“In our tests we took at least three random location readings and it did not matter where on the unit we placed the wand as long as we got the spark to jump straight across.”

To test a unit you hold the flashlight-shaped wand on the non-coated side. The instructions say to be within a couple of inches from the spacer. I think this is so the spark does not short out over to the spacer bar.

In our tests we took at least three random location readings and it did not matter where on the unit we placed the wand as long as we got the spark to jump straight across.

You press a button on the wand and about two seconds later you hear and see the spark. It sparks for a second then stops. Then the Ipaq thinks for a few more seconds and displays the results. The total test time is less than ten seconds.

The manual is a few short pages and includes several sentences about the danger of the spark. It is a little lacking but that is normal for a new product. In talking to our customer who purchased the device, despite the seven hour time difference, he said Sparklike had been responsive via e-mail and phone calls.

The unit was provided with a 110-VAC plug and plugged into a wall outlet.

I received an e-mail from Sparklike’s head of marketing, Niklas Törnkvist, in which he stated, “all of the instruments that are on the market are prototypes, or as we call them, evaluation version instruments, not even trying to be the final thing.”

Törnkvist also stated, “Sparklike is working on an electronic sparker that is a lot better. This sparker will ignite the plasma instantly and hence eliminate all of the problems with the tesla used previously. All existing instruments will be upgraded into the new version once it is finally ready. With the new sparker that we are building there are no such windows that it would not jump. It, however, depends on the gas concentration.”

Regarding the interface he said, “We know all of the problems with the Ipaq and the previous version of it, the Aero. We assure anyone that we will not keep the Ipaq a minute longer than is necessary.”

Randi Ernst is president of FDR Design Inc. in Buffalo, Minn.

© Copyright 2001 Key Communications Inc. All rights reserved. No reproduction of any type without expressed written permission.