Sensatronics Model F (serial temperature monitor)

[Frank Shute]

On Wed, May 14, 2008 at 12:51:53PM +0100, Dominic Marks wrote:
>
> Frank Shute wrote:
> >On Tue, May 13, 2008 at 03:53:58PM +0100, Dominic Marks wrote:
> >  
> >>Hi list,
> >>
> >>Does anyone have experience with these devices?
> >>
> >>http://www.openxtra.co.uk/products/environment-monitors/temperature/model-f/prodmodelf.html
> >>
> >>    
> >
> >I programmed (in C) a similar thing to this on a university course. It
> >was very easy to program. Although, I see the sensatronics site have
> >some sort of plugin for nagios.
> >
> >The only possible problems I see with it are that it's a serial port.
> >How long are these going to hang around?
> >
> >The price of the probe is outrageous! It's just a thermocouple. You
> >could probably get one from Maplins for a tenth of the price unless it
> >has some sort of funky coupling. Maybe check the spec of the probe on
> >the sensatronics site and see if you can bag something similar on the
> >cheap.
> >  
> Thanks, I was surprised at the price as well. I looked at IPMI, SMBus on 
> the board itself and it just doesn't work at all. I've also got five 
> other working sensors but I'm a little wary to trust them. There are the 
> SMART temperature sensors on the discs (4 of them) and a sensor in the 
> UPS which I can read from the server. I've often seen SMART temperature 
> sensors reporting improbably high values, say +150c so I don't have much 
> faith. The ones in this particular server are reporting ~45c and the UPS 
> is reporting an internal temperature of ~30c. These seem fairly 
> reasonable. The other slight concern is that these values are not 
> exactly the information I'm after. I should check the switch in there 
> too actually, there is a procurve which might have some temperature 
> sensors on it...

I agree with you, these temperature monitors aren't going to give you
a satisfactory answer to what the temperature in the server room is.
They're all going to vary dependent on the load on your server.

> >Beware that when placing the probe you don't want it touching
> >anything. Maybe you can make it dangle inside your servers case
> >without actually touching it.
> >  
> Ok. I'm really interested in the temperature of the comms room as a 
> whole as it has barely any ventilation. It's surrounded by two metre 
> thick stone walls so I'm hoping it will be quite well sheilded from 
> external heat. It's been fine during the winter when I first looked at it.

Then I think that Sensatronics gadget would be the answer. You can
just dangle the probe down from the ceiling somewhere in the centre of
the room.

The cheapskates answer would be to use a max/min thermometer but it
doesn't warn you if the temperature becomes unacceptably high.

I had look at the manual for the gadget on the Sensatronics site and
they've even got a sample perl script in there, so there's going to be
no problem getting it to work with FreeBSD.

Should be easy enough to adapt so that it fires off an email to you if
the temperature is > x°

I tried to find some information on the probes ie. how their
resistivity varies with temperature but I couldn't find it.

I also had a look at Maplins. They've got a couple of thermocouples
but they don't tell you either!

So it looks like you'll have to buy the outrageously priced probe. The
nylon one should do the job. Hopefully, you don't have to cover the
cost yourself.

> >Test your probe by placing it next to a mercury thermometer and
> >comparing readings.
> >
> >Do keep us all informed if you decide to get one.
> >  
> Will do.

Excellent.

> 
> Thanks for the info Frank,

No worries.

> 
> (p.s. Thunderbird tells me it will wrap my lines at 72 chars, I suspect 
> it lies. Apologies in advance.)

Seems OK to me.

> Dominic

Regards,

-- 

 Frank 

	
 Contact info: http://www.shute.org.uk/misc/contact.html