[Nut-upsuser] Liebert PSP

[Charles Lepple]

On Sep 21, 2010, at 1:16 PM, Will Cooke wrote:

> On 19 September 2010 22:16, Charles Lepple <clepple at gmail.com> wrote:
>
>> Remember when I said you were really close? Apparently, now that  
>> we've
>> figured out that it isn't really "0.000000" for both cases, I think  
>> you just
>> need to multiply the value by 1e7 in that function, sort of the  
>> same way
>> that some Tripplite voltages are multiplied by 10:
>>
>> http://trac.networkupstools.org/projects/nut/browser/trunk/drivers/tripplite-hid.c#L155
>>
>> This function multiplies by the variable "battery_scale" because some
>> Tripp-Lite product IDs get this right, and others don't - so the  
>> scale
>> factor is set at initialization time. You might want to try a similar
>> approach, just in case a new model comes out.
>>
>> The input and output voltages look like they need a similar  
>> adjustment, and
>> the frequency probably needs to be divided by 10. Actually, this was
>> mentioned in another email thread in July:
>>
>> http://lists.alioth.debian.org/pipermail/nut-upsuser/2010-July/006104.html
>>
>> ... and that seems to indicate that BatteryPresent and maybe other  
>> booleans
>> need a scale factor, too.
>>
>
> Hi again,
>
> I've made some progress with the sub-driver!
>
> In order to work support for Vendor ID 10af probably needs to be
> removed from the Belkin HID driver which tries to support the
> LiebertPSP at the moment.
>
> I've attached my efforts thus far.

The problem is that there is a Liebert PowerSure PSA which uses the  
same vendor and product ID:

http://thread.gmane.org/gmane.comp.monitoring.nut.devel/4232

> Please remember, I'm not a programmer - so please feel free to point
> out my glaring mistakes.  A lot of the new functions have been
> borrowed wholesale from other functions and random web pages without
> really understanding how they work.
>
> What it does now that it didn't do before:
>
>  Report OL instead of OB when on AC
>  Report corrected values for voltages & frequency

Cool! Can you please send the output of "upsc your- 
ups at localhost" (adjusting the second argument for your configuration)?

> I expect that the values for things like "NeedReplacement" will also
> need converting, but I haven't checked yet.

Also, some of the names (first column in the tables) are not quite the  
standard NUT names.

> What are the next steps?  Obviously I will check those other values
> (NeedReplacement etc) and pass them through the conversion if
> required, but what about other metrics?  Are ALL the features of the
> UPS automatically added to the stub driver?  If not, how would I go
> about discovering them.  For example, let's imagine the UPS could
> report an estimated "time remaining", how would one discover this
> feature, or is it impossible without a data sheet?

Which file did you adapt for this? belkin-hid.c?

Basically, the "feature discovery" happens when you run the usbhid-ups  
driver in "explore" mode, and you might not have the same set of HID  
Usages.

Details on how to run this against your UPS are here: http://buildbot.networkupstools.org/~buildbot/docs/latest/website/developer-guide.html#hid-subdrivers 
  (under "4.21.3. Writing a subdriver").

You can then compare the created files with the ones you just posted,  
and see what pops up.

Actually, if you wouldn't mind, can you post the output of running the  
driver in explore mode with "-DD"? (gzipped again, please)