Kaze Master 5.25 serious thermometer bug

[BusError]

Well, I wonder how this one passed QA really.

Put a temp sensor under a fan, turn the fan on gently and read the temperature display.

The scale is not linear, it is completely bonkers in fact. This test is made under controlled conditions. no drafts etc.

Hint it goes like :
34.0
34.1
33.2
33.3
33.4
32.4
32.5
32.6
32.7
31.7
31.8
31.9
31.0
30.0
30.1
30.2
30.3
29.3
29.4
29.5
28.5
28.6

etc

Notice anything wrong here ? Well yes, someone reading the ADC got the conversion to decimal all wrong, and is not even using base 10 for the decimal, and it's in reverse too.

In other words
1) don't believe the thermometer
2) I'm returning my Kaze

BusError, from forums.overclockers.co.uk

[sumo]

Halo BusError,

i hope i understand you right, you put a thermo sensor under a fan. Than you turn on the fan and similar the temperature of the sensor falls. Right?

These little up-steps you saw in your values have many reasons: position change, airflow change, 2 s messurement time and measuring tolerance.

So we analyse your measuring and you can perfecly see the measurment rhythm of the Kaze Master Fan Controller. Every controller works with that rhythm.

I hope this was helpful to you.

Kind regards
Christoper - Scythe EU GmbH

[BusError]

Hmm I did the measurement in entirely controlled conditions, no drafts.

I can understand these kind of spikes on a system that has a response curve with a feedback loop, but here, I am just measuring a thermometer, not a fan controller. The fan is at a constant position and at a constant speed.

My own guess is that you are reading a 10 bits ADC, and fail to do the conversion to base 10 properly.
My guess is a strong one, I write embedded software for a living

[Shinigami]

Hi BusError,

How did you do the exact measurement for each second exactly?

The point is, that you might as well consider that there is something like a tolerance which we state and everything has such a tolerance. If you are looking for a very precise device, you might want to buy a tool from the scientific measurement segment. These devices should be accurate and might give you the opportunity to read the data with your pc. Just an idea

[BusError]

.1 Celcius is not rocket science, and /you/ made the display with 3 digits, I didn't make it. Now you are claiming that it's just to display a random number ?

I didn't do measurement to an exact timing.. Just the time it takes to scribble down the reading, which gives a pretty smooth ~2s interval as it turns out.

It doesn't seem to be a very complicated test to reproduce, and instead of denigrating my motive or methods, you could sit down with one of your unit and test it, then go and kick the programmer who doesn't know how to do a division, and the QA people who are not doing their job.