After reselling/gifting all my old Tempest and Air+Sky gear in summer 2020, I recently picked up the new Tempest’26 to see how it has evolved from the original Tempest and even the old Air+Sky from its earliest days on Indiegogo.
Here’s a long review.
I’m comparing it to my other gear, including a early 2009 Davis VP2, some Ecowitt T+H sensors, a CoCoRAHS manual rain gauge, as well as some previous testing with Ecowitt’s tipping rain gauge.
Quick assessment
-
no real changes from the earliest days. Looks like a WF station, acts like a WF station.
-
if you don’t need actually accurate rain, it’s fine, but you can get there for under half the cost with the competition (Ecowitt).
Longer review follows…
The good story:
-
the station was ‘much’ easier to get online than the original first days of Air/Sky and of Tempest. It can’t get much easier. Very very easy to do now.
-
liked the little level included in the pole mount. I know it’s only a few cents cost but it’s a nice touch.
-
build quality seems fine at least initially. The unit sure seems solidly built. It seems much heavier than I remember, but my memory might be hazy there.
-
like the new hub a lot especially the form factor and antenna
-
the new temp/hum sensors seem right on the money, within a tiny fraction of a deg-F from the VP2 and Ecowitt sensors and some reference breadboard rigs I built, but of course temperature+humidity are easy pickings.
The possibly not as good story:
-
wind to me seems to read low but I can’t quantify due to differences in where the Tempest is sited versus the VP2. We had multiple windy periods and I know the windows shaking in gusts are certainly more than the 14 mph the Tempest was reading. I similarly couldn’t previously quantify whether Ecowitt’s ultrasonic wind was good either due to siting limitations. Wind is so sensitive to siting that it’s not too easy to quantify one way or the other.
-
I had to cover the retina-burning super-bright glowing LEDs that were positively blinding with electrical tape. Please give us a switch to turn the LEDs off. Please.
-
as I’ve mentioned in other threads, it is ridiculously complicated to turn NC on/off via either the mobile app or the website. Ridiculously complicated. There’s no reason why we have to jump through these hoops for simple toggles.
Please pay some kid to tweak your website and app as a Christmas break project. There has to be some kid with a usability interest who can put themselves in the user’s shoes and make it more usable. At a minimum let us display raw+corrected rain on the same screen.
The ‘I dunno’ story:
- we almost never get lightning here, so can’t say more there other than the history lesson. Previously the sensor was basically useless even with the crowd-sourced (ugh) lightning readings. I’ve had it go wild with no rain nearby. I’ve had it read nothing with the house shaking from rare booms directly overhead. Guess we’ll see when our rare lightning happens. C’mon thunder-snow !
The we’ll see in a few months story:
-
this is the gloomy ‘big dark’ season here near Seattle, so I can’t test whether the Tempest temperature readings have the same flaw that the original one had where it read low as the sun came up, taking a while for the temperature to catch up to reality (as the case heats), and then read high during the day and after sundown (as the case holds that heat and cools to ambient very slowly).
The daylight-only fan-aspirated Davis VP2 ‘barely’ has this issue, and of course Ecowitt T+H sensors and the old Air do not have any issues since you mount those in the shade.
I recall there used to be a temperature compensation based on solar and wind, if I remember correctly, but WF has never given us any way to tell how much that compensating is tweaking things. I know many folks have asked for that kind of info for years to no avail.
-
solar - I can’t measure whether the solar/uv values are reasonable until we get back to the sunny time of year in June-September next year. Previous Tempest hardware read absurdly high well beyond the maximum theoretical limits for illuminance and uv for this latitude/longitude. So we’ll see next summer.
The ‘bad’ story:
-
documentation - would it really be ‘that’ tough to publish expected accuracy ranges for the Tempest’26 like every other vendor on the planet publishes ? How about providing an actual teardown howto for the reportedly replaceable battery ?
Isn’t there a High School kid you could hire over Christmas break to document this information and link it into your web ? This isn’t hard to do. You just have to want to do so.
-
bigger solar panels - there is zero documentation on how much bigger/better they are, how much energy they can capture vs. the original Tempest, whatever. It all looks like market-speak to me in the absence of any quantifiable data. I don’t understand why this seems to be so secret.
-
the new battery - again, no info on how it differs in specs from the original Tempest
The disappointing really bad story:
-
Undoubtedly rain is still the weak spot in the Tempest. The haptic rain technology is not good in my experience from day one dating back to the Indiegogo days. Nearcast is still no help, trying to make a weak sensor implementation less week with magic math. That simply is not helpful and does not work. More data below…
(I might note that Ecowitt’s seemingly-stolen-technology haptic rain sensor is 'far’ worse still, but those are $79 sensor suites without any magic math that is WF’s secret sauce.)
Here’s my rain data for comparison, with data from a Davis VP2 with its latest tipping gauge versus Tempest both without and with NearCast corrections enabled. I omitted the CoCoRAHS manual gauge for brevity since it always ‘exactly’ matches the VP2 readings. Exactly.
I might add that I did some previous tests with an Ecowitt tipping gauge that was also pretty much right on the money versus CoCoRAHS and VP2, within a percent or two daily. I’m not understanding why a $40 Ecowitt tipping gauge measures rain more accurately than a $399 list Tempest.
The Tempest is way off. NearCast is way off with seemingly random higher/lower corrections.
(example - we’ve received 4" rain in the last 4 days as the ‘atmospheric rivers’ slam the Pacific Northwest. Some sites 5 miles east have almost 7". I don’t understand how NC can possibly deal with that kind of localized rain (or no rain), but WF is so secretive about how they cook up corrected readings that it’s impossible to know at all).
Uncertain yet
-
solar charging - I have the Tempest running in weewx using the excellent @vreihen driver and have been measuring how the Tempest battery level lines up with how dark it is here this time of year. I have the data so far, just need to tweak my weewx setup to do some graphs of battery voltage vs. accumulated lux for the day. Currently holding around 2.47 V.
Here’s voltage vs. solar so far - note that the daily low might be ‘after’ the daily high depending on when the sun was shining that day, but it should give an idea of the voltages I’m seeing. I might get around to graphing voltages over time as the gloom continues here.
It sure would be nice for WF to publish what voltages the Tempest’26 need to stay out vs. how many accumulated lux-hours are needed to stay charged. Again - too many secrets regarding specs.
(another example - Garmin publishes how many lux-hours you need to keep their solar watches charged up. It’s accurate data. This is another place where it isn’t hard to provide quality info if you want to)
