Actions
Forum Thread Details
  • State Suggested Answer
  • Replies 25 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 6493 views
  • Users 0 members are here
Related

Testing rear window defroster grid

Former Member

Hello,

I would like to develop a testing method to detect window defroster grid failures. Rear windows has 12 heating wires painted to the glass and cca. 2% are faulty pieces.

I tried to measure the resistance, current and voltage drop of good parts and compare results with those where was at least 1 broken wire, but the difference was smaller than the tolerance specified by the supplier. (Other problem is that the resistance is greatly depending on the temperature)

Now we are testing visually with a thermo foil but it takes more time and sometimes the operator mistakes or can forget to correctly check thermo foil. It is also very important to not hurt parts during the testing process, so best option would be contactless testing if possible... image

I'am thinking about to make a tool with 12 sensors, 1 sensor to every each heating wire to sense a flowing current in a wire. But I'am not sure that there is any sensor avaible in the market for such a purpose. Finally PLC have to check the outputs of every sensors, if all the 12 wires are good (current is flowing through them) gives visual signal or in case of failure sound alarm or red light flashing.

 

Here is the specification of rear window defroster:

Power: 167,37W ±10%
Current: 12,87A ±1A
Resistance: 1,0 Ohm -10%/+20%
Wire thickness: 0,48mm
In the attachment you can find technical drawing of the window.

Any ideas and suggestions are welcome.

Best regards,

Ate

Attachments:
image
Parents
  • 0

    Hi Lansm

    I have a heat sensor that I use in the lab that has a very low mass thermistor. I suspect that it could be moved across a wire to detect if any heat was being produced. None of our solutions so far will tell you where in the wire there is a break, just that there is a break somewhere. I suspect that some sort of a capacitive sensor dragged along a broken wire might be able to detect the break where there is a potential gradient across the break but I have no empirical experience with this.

    John

  • 0 in reply to jw0752

    A strip of thermal paper across the screen might work.

    We use thermal printers for Flight Strips and they are reasonbly sensitive.

     

    Mark

  • 0 in reply to mcb1

    I think Ate is saying that the operator forgets to visually check the strip (or there is a belief it could damage through contact - but I would have thought that would

    be a valid test), but it could all be automated, either with a visual camera examining the strip output, or FLIR has industrial thermal cams (e.g. just as an example, FLIR Axx series

    spotted at a trade show, but other models may be more suitable) and they have an API where you can select the region of interest for example.

    However, it costs (very high resolution might be needed to identify hot spots from a thinned out track, much like a CCTV camera has for it's field of view limited pixels) -

    the cost may be negligible for an industrial use-case though.

     

    There are cheaper ways for broken wire  - Panasonic has GRIDeye - twelve of the sensors, one aligned to each wire, would be sufficient to monitor the twelve wires are a

    cost of under $500, and low-cost microcontroller(s) would be sufficient to process the data (I2C).

    It would identify which of the 12 wires is not heating. There is also Melexis but I've had difficulty obtaining their products easily - although generally they work well from what I could tell.

    Also Honeywell and Omron Omron and GEGE (no longer available) have single-pixel or low count linear arrays of thermal imaging type sensors (thermopile).


    For a go/no-go indication for the entire window (e.g. if it is not for repair in a production line, but for complete replacement) where you may not care which of the 12 elements

    are faulty, I think any resonant circuit method could be worth a shot simply because it doesn't need a jig for thermal alignment, (and so is a quick and cheap method to

    eliminate quickly if it doesn't work) and possibly the easiest (with the LDC1000LDC1000) although this is a pure guess.

Reply
  • 0 in reply to mcb1

    I think Ate is saying that the operator forgets to visually check the strip (or there is a belief it could damage through contact - but I would have thought that would

    be a valid test), but it could all be automated, either with a visual camera examining the strip output, or FLIR has industrial thermal cams (e.g. just as an example, FLIR Axx series

    spotted at a trade show, but other models may be more suitable) and they have an API where you can select the region of interest for example.

    However, it costs (very high resolution might be needed to identify hot spots from a thinned out track, much like a CCTV camera has for it's field of view limited pixels) -

    the cost may be negligible for an industrial use-case though.

     

    There are cheaper ways for broken wire  - Panasonic has GRIDeye - twelve of the sensors, one aligned to each wire, would be sufficient to monitor the twelve wires are a

    cost of under $500, and low-cost microcontroller(s) would be sufficient to process the data (I2C).

    It would identify which of the 12 wires is not heating. There is also Melexis but I've had difficulty obtaining their products easily - although generally they work well from what I could tell.

    Also Honeywell and Omron Omron and GEGE (no longer available) have single-pixel or low count linear arrays of thermal imaging type sensors (thermopile).


    For a go/no-go indication for the entire window (e.g. if it is not for repair in a production line, but for complete replacement) where you may not care which of the 12 elements

    are faulty, I think any resonant circuit method could be worth a shot simply because it doesn't need a jig for thermal alignment, (and so is a quick and cheap method to

    eliminate quickly if it doesn't work) and possibly the easiest (with the LDC1000LDC1000) although this is a pure guess.

Children
No Data