pid control configuration

Ed Tanous ed.tanous at intel.com
Tue May 14 03:32:02 AEST 2019


On 5/13/19 9:32 AM, Patrick Venture wrote:
> On Fri, May 10, 2019 at 2:21 PM Vijay Khemka <vijaykhemka at fb.com> wrote:
>>
>>
>>
>> On 5/6/19, 9:20 AM, "Ed Tanous" <ed.tanous at intel.com> wrote:
>>
>>     On 5/2/19 4:24 PM, Vijay Khemka wrote:
>>     >  My understanding here is pid data are only defined for temp sensors which will generate pwm values for fan.
>>     >
>>
>>     The above is not how phosphor pid control operates.  The temperature
>>     controllers "request" a given fan RPM, then the fan PID controller
>>     (which is scanned much more quickly than the temperature controllers)
>>     attempts to drive the fan to that speed.  This ends up being a two stage
>>     feedback loop that can perform better than a single PID loop.
>>
>> Ed, we are not using cascade control here in facebook. We only use outer loop which gives data for fan to drive.
>> Is it possible to still run Phosphor pid controller with single loop?
> 
> Your cascade PID could be a pass-through -- so that it receives the
> set-point from the sensor PID and then just sets it directly within
> that loop..  It may require a new PID controller that is just a
> pass-through to explicitly exist -- i haven't dove into
> phosphor-pid-control in a while, but you probably need to write a
> basically empty PID controller object type.
> 

+1  That is an option for you.

>>
>> Regards
>> -Vijay
>>


More information about the openbmc mailing list