Every Starlink I have installed has significant coil whine. I really wonder if they adhere to emissions limits, as I can't see how testing in an EMC chamber would capture real world emissions when the unit is actually communicating with satellites.
You can simulate it communicating with satellites in the chamber. Especially since if they're at X or Ku band, and coil whine is in the audible range, it really shouldn't interfere with the measurement?
It is my experience that products with coil whine tend to have unwanted RF emissions, as coil whine is an unwanted emission. The hardware engineers I know have enough pride in their product to fix such issues before production. I just don't have the equipment needed to properly evaluate what the Starlink is actually emitting as you have to look at upwards of 20+GHz of spectrum. Spectrum analyzers with that kind of range cost north of $100k. That the author is inferring what the Starlink is doing from one tiny slice of spectrum at a time is very impressive.
The other issue of concern to me is that devices like Starlink are largely software defined. Changes to the software will change the emissions profile -- that's just a fact of life when software controls everything and products continue to be developed beyond the initial launch date. If emissions testing is only done with beta firmware, who knows if what's actually shipping behaves in the same manner as the product actually tested and certified.
Every Starlink I have installed has significant coil whine. I really wonder if they adhere to emissions limits, as I can't see how testing in an EMC chamber would capture real world emissions when the unit is actually communicating with satellites.
You can simulate it communicating with satellites in the chamber. Especially since if they're at X or Ku band, and coil whine is in the audible range, it really shouldn't interfere with the measurement?
It is my experience that products with coil whine tend to have unwanted RF emissions, as coil whine is an unwanted emission. The hardware engineers I know have enough pride in their product to fix such issues before production. I just don't have the equipment needed to properly evaluate what the Starlink is actually emitting as you have to look at upwards of 20+GHz of spectrum. Spectrum analyzers with that kind of range cost north of $100k. That the author is inferring what the Starlink is doing from one tiny slice of spectrum at a time is very impressive.
The other issue of concern to me is that devices like Starlink are largely software defined. Changes to the software will change the emissions profile -- that's just a fact of life when software controls everything and products continue to be developed beyond the initial launch date. If emissions testing is only done with beta firmware, who knows if what's actually shipping behaves in the same manner as the product actually tested and certified.
That seems like cost cutting leading to coil whine. With that said, even some $1000+ GPUs have had coil whine issues.