Next up was 10W. It took 2m 16s to reach 100 C and would have kept rising had I not turned on the fan. With the fan on, the temp decreased to 75 C and stayed there. I left it on for a couple of minutes but no Majik Smoke appeared, so I shut it down.
I reversed the connections to the attenuator and noted no difference in readings.
So, it can take 10W for minutes at a time without a fan, and certainly it could have withstood 10W for longer than that with a fan. If I was testing a transmitter I wouldn't key it up for minutes at a time even adjusting the power per the manual.
So it looks like it can be used in a power meter made from the AD8310 to extend the range to something like -73 to +40 dBm which is a pretty darn good range. I could get fancy by sensing the temp and turning on the fan automatically, maybe at 60C. But The simpler approach would be to just have the system turn on the fan anytime you are selecting to measure power over +13 dBm (switching in the 40 dB att.) which was my max power level in the power meters I made to date.
I only chose the small footprint 10W 40 dB attenuator so the case wouldn't be too big. But there is no reason a person could not use a bigger case and fan and make a power meter to handle -73 to +50 dBm 100W (100W) ... or more.
I'd have a manual switch to select the higher range, which would flip the coaxial relays and turn on the fan at the same time, so the user has to realize to switch in the attenuator. But, even with the expensive commercial power meters, if you the user aren't paying attention and apply too much power with bad settings or setup, you are going to pay for it. Not paying attention is the reason why some people can't have nice things. Can't get past human error in the end.
To paraphrase some lines from 2001: A Space Odyssey
"Well HAL how do you explain this power meter burning up when it had a clearly labeled switch for the user to select a built-in attenuator?
"Well, I don't think there is any question about it. Things like this have cropped up before, and it has always been due, to human error ... "