There's a new construction zone drone in my town, but this one is unique in that it's situated at the top of a hill and 120-degree curve. As a result, almost no detector can see it until it's just a second or two away from the "kill zone" if it were a real cop.
I had my V1, X50, and XTR-690 available for testing against this K-band source. Since it was a light-traffic area, I made a few passes... the PSL was 45mph.
Each detector was tested separately. A passenger started a stopwatch when the detector first went off, and I (the driver) said "STOP" when I had line of sight with the radar source, which would be a certain lock for a real cop.
Also note: To level the playing ground, I decided to "punish" the V1. This is a city area where K-band noise was abundant. As such, I'd typically run my V1 in little-L mode on 4-bar automute, and take note if the V1 goes off but not take any action unless it breaks the mute barrier. For this reason, for my comparison I'll note when the V1 initially alerts, but I personally am comparing the V1 to the others by when the automute barrier was crossed. Whether or not this is fair/accurate is a different argument that I'll save for another day.
I set my cruise control at about 50mph, simulating the average driver rolling into a construction-zone speed trap.
V1: Started alerting at the base of the hill. A liquid-smooth linear ramp-up, about 3/4 of the way up the hill the automute barrier was crossed and the V1 was at full strength before reaching the top. 4.5 seconds from automute cross to the kill zone.
X50: Alerted later than the V1's automute cross point. The alert came in at 3 bars and then immediately pegged full strength. About 2 seconds from initial alert to the kill zone. Tested with POP on to see if it'd react earlier, no difference.
XTR-690: Under default FILTER mode, alerted about 4 seconds before the kill zone. As the driver, I note that it alerted after the point the V1 would cross the mute barrier but not much after. Ramp-up was a bit jagged but about 2 seconds later it jumped to oscillating between 8-9 signal strength. Under FILTER 2, the detector started alerting a bit later, but due to the speed and testing variances I couldn't get a very accurate (numerical) assessment of how much impact that was. I'd say the difference wasn't terribly big.
I controlled my speed at 10mph via first gear and gas pedal (cruise control doesn't reach this area). My 300C has a digital speedo in the console which allowed +/-1mph precision to this speed using a combination of engine noise pitch and careful watching of the speed indication. This was intended to simulate GOL like conditions with low-speed testing.
This test showed the GOL-type alert order: The V1 was beeping away RIDICULOUSLY early, at the base of the hill, the 8500 X50 starts its alert at the halfway point of the hill, where the V1's automute broke off, and the XTR-690 didn't start alerting until about 3/4 the way up the hill.
I think this test shows a very peculiar phenomenon: While the X50 is head and shoulders to the XTR-690 in K-band sensitivity and in testing absolutely obliterates the XTR-690 on this band, in the real world the difference margin was very minimal. Just a few precious seconds of "filtering" time can be the difference between a save or kill... Don't take controlled test environment results too seriously in the real world, because in the real world, NOTHING is controlled, and a ticket is a ticket even in these unpredictable cases.