Nut upsdev - Sep 2008 - tripplite smart2000rmxl2u product id 3014

First, apologies if this is the wrong mailing list to pose this question.  That
said, my question:

This is my first time using nut.  Just built nut-2.2.2.  The tripplite driver
does not support my version of the smart2200 which has a product id of 3014. 
Does anyone know about this ups?  There is a 3012.  I'm tempted to simply
add the 3014 to tripplite-hid.c to see if it will "just work".  Is it
safe to do?  Would it be useful for you who are developing this driver?  Any
sort of special commands you care to pass on, eg., to dump out information about
this version of the smart2200  that might be of use to you?  I noticed that all
the code does with 3012 is to set the battery scale to 1 which I'm guessing
is some scaling factor to calibrate the battery.

Thanks! Ken

On Sep 29, 2008, at 4:30 PM, Ken Teh wrote:
> First, apologies if this is the wrong mailing list to pose this  
> question.  That said, my question:
>
> This is my first time using nut.  Just built nut-2.2.2.  The  
> tripplite driver does not support my version of the smart2200 which  
> has a product id of 3014.  Does anyone know about this ups?  There  
> is a 3012.  I'm tempted to simply add the 3014 to tripplite-hid.c  
> to see if it will "just work".  Is it safe to do?
We generally recommend that you do not test things on a production  
system, especially if you are the first to do so.

Also, did you get a "possibly supported" message? You can add "-x
productid=3014" to the command line, and the driver should work.
> Would it be useful for you who are developing this driver?
Certainly - this product ID would be useful to add to the  
compatibility list, especially once we figure out what differences  
there are between that model and the others that we have seen.

You may be interested in the SVN "trunk" version of the driver - this
is slightly more recent than the 2.2.2 release, and it is easier to  
make patches against that. Let us know if you're looking for help on  
checking that out. We also have SVN snapshots available through our  
Buildbot installation: http://buildbot.ghz.cc/public/nut/
>   Any sort of special commands you care to pass on, eg., to dump  
> out information about this version of the smart2200  that might be  
> of use to you?  I noticed that all the code does with 3012 is to  
> set the battery scale to 1 which I'm guessing is some scaling  
> factor to calibrate the battery.
Most of the information should be available from 'upsc', once you  
start the driver with the aforementioned option, and then start upsd.

If you run into issues (e.g. basic information is missing, or if you  
want support for a specific variable that isn't in the upsc output),  
there are other diagnostics that we can run.

thanks,

-- 
Charles Lepple
clepple at gmail

On Mon, Oct 20, 2008 at 11:20 AM, Ken Teh <teh at anl.gov>
wrote:
> Charles Lepple wrote:
>>
>> On Fri, Oct 17, 2008 at 11:55 AM, Ken Teh <teh at anl.gov> wrote:
>>>
>>> I have it sort of working.  I was able to run the usbhid-ups driver
with
>>> the productid as you suggested but I have to run everything as
root.  Here's
>>> a quick summary of results:
[...]
>>> Today was the smoke test day.  I took the UPS off line power,
partially
>>> drained the battery with a electric heater, and monitored the
battery charge
>>> with upsc.  It kept dropping down to 1 and stayed there like
forever.
>>>  upsmon never went through executing the low battery notify event
or the
>>> shutdown.  I noticed that the UPS LEDs indicated adequate power (it
was
>>> still green) so this leads me to the next questions:
>>
>> What fraction of the UPS power capacity was the heater wattage?
Still interested in the answer to this one.

I don't think you mentioned how you were cutting line power, but Arjen
has pointed out that it is better to leave the ground connected (e.g.
use a power strip switch or circuit breaker to remove power during
testing). That could affect the accuracy of the charge meter.

Another thing is that you may need to fully drain the battery once to
calibrate the charge meter. The manual for the UPS may cover this
procedure.

[...]
> You
> mentioned debug logs during the discharge cycle.  Could you be more
> specific?  upsd logs?  upsmon logs?
I am referring to the log output from starting the usbhid-ups driver
with "-DDD" on the command line. This is basically the raw data from
the UPS. You can use something like "/lib/nut/usbhid-ups -DDD -a
name-of-ups | tee /tmp/tripplite.log", then cut power, and press
Ctrl-C after you have tried draining the battery for a while. If upsd
is running, you should also be able to retrieve status with upsc.

I assume you meant that you saw battery.charge go to 1 in the upsc
output. We probably will see that in the driver log output, so you
could just check that once to be sure the problem still exists.
>  Is there a specific piece of code I
> should try hacking?  tripplite_usb?  Or, something higher up?
The "tripplite_usb" driver predates your UPS. It is basically the old
Tripp Lite serial protocol wrapped in USB packets, as opposed to the
standardized HID Power Device Class (PDC) protocol that Tripp Lite,
APC and MGE use now.

Things specific to the Tripp Lite PDC UPSes are in tripplite-hid.c,
and there are some more generic parts in usbhid-ups.c, libhid.c/h and
libusb.c/h.

I am not yet sure what needs to be changed, though.

-- 
- Charles Lepple