Re: AW: Readback of output variables in adsDriver

[Torsten Bögershausen via Tech-talk <tech-talk at aps.anl.gov>]

Hej again,

I haven't manage to find the time to set up a test system yet -
hope that you have more luck.

Here are the good news.
After chatting with out expert:
epics-twincat-ads does support read back of output records !

One precondition is the '?' at the end of the OUT field:

<https://urldefense.us/v3/__https://github.com/EuropeanSpallationSource/epics-twincat-ads/blob/045f40f2e80057025befe015b89b80288157c479/adsExApp/Db/adsTestAsyn.db*L12__;Iw!!G_uCfscf7eWS!a7O2n5uR7qlxzkr_aV81uRdqnKyuvf9l5qwggdWT010Ihel6kTj_imLAZsLVjZxb-MOwsMHmnOk00TXTcII$ >

The "SetFAmplitudeRB" record should do what you ask for:
Writing to the EPICS PV changes the value in the PLC.
Once the value in the PLC is changed, that change should be
seen in the output record.

HTH - and please let us know, how it goes.

On 2024-04-15 18:21, Torsten Bögershausen via Tech-talk wrote:
> Hej all,
>
> I did have a short look into the driver -
> adsTestAsyn.db does have a READBACK, but I couldn't find anything
> in the code, to switch it on or off.
> But I know for sure, that people from SLAC have contributed a bulk read,
> so far only 1Hz.
> Similar to Jure's driver, so far nobody had asked seriously enough
> for READBACK so no time had been invested here.
>
> Having said that, why do you not get any data ?
>
> Wrong "ads port" ?
> I sometimes run wireshark to find out, what is going on.
>
> /Torsten
>
>
>
> On 2024-04-15 16:54, Meyer, Markus via Tech-talk wrote:
> > Hello Jure, Mark and Torsten,
> >
> > Thank you all for your responses and explanations of the 2 cases.
> >
> > > I kindly ask you to open a pair of issues on the Github repo. I can't 
> > > say when we we'll be able to take a closer look (we're stretched 
> > > pretty thin right now), but having open issues helps raise the priority.
> > I will do that and maybe I can help with some code after I have 
> > finished my evaluation of the other available drivers.
> > > Case 2, at least, is probably not that hard to implement. Case 1, on 
> > > the other hand, is a bit thornier because of the design of the driver.
> > It may be OK to do the readback after iocInit only in a cyclic interval.
> >
> > > We here at ESS use a different driver for ADS, and if that one can do 
> > > the readback ? I haven't checked - but I highly suspect that it doesn't.
> > I have compiled the latest source of the driver from ESS [1] and can 
> > connect to my plc but unfortunately I get no values from the plc at 
> > all. I am still investigating what I am doing wrong here. I expect a 
> > readback because it's explicitly mentioned in the adsTestAsyn.db of 
> > adsExApp.
> > Thanks,
> > Markus
> >
> > [1] 
> > https://urldefense.us/v3/__https://github.com/EuropeanSpallationSource/epics-twincat-ads__;!!G_uCfscf7eWS!cWY-oNbq-4U-gg1N5Q8SvI_E4DLNAd9tVgk5xrioIeV8ejfmVohmnmNpsp_rKDUJGlJGz2RJXqJ_poyPHDdx8g$
> >
> > -----Ursprüngliche Nachricht-----
> > Von: Tech-talk <tech-talk-bounces at aps.anl.gov> Im Auftrag von Jure 
> > Varlec via Tech-talk
> > Gesendet: Montag, 15. April 2024 12:49
> > An: tech-talk at aps.anl.gov
> > Betreff: Re: Readback of output variables in adsDriver
> >
> > Hello Markus,
> >
> > Mark made a nice breakdown of the two cases. Right now, the adsDriver 
> > module supports neither. We are aware of this, but honestly, nobody 
> > has needed it badly enough yet :) . I kindly ask you to open a pair of 
> > issues on the Github repo. I can't say when we we'll be able to take a 
> > closer look (we're stretched pretty thin right now), but having open 
> > issues helps raise the priority.
> > Case 2, at least, is probably not that hard to implement. Case 1, on 
> > the other hand, is a bit thornier because of the design of the driver. 
> > It is focused on doing bulk reads, which are needed to get sufficient 
> > performance with lots of variables. But such reads can only be done 
> > after all the records are initialized and ADS variables made known to 
> > the driver. So, initial readbacks were disabled for simplicity. They 
> > can be added for sure, but it won't be a simple bolt-on.
> > Cheers,
> > Jure
> >
> > On 4/12/24 18:11, Mark Rivers via Tech-talk wrote:
> > > Caution: This email originated from outside of Cosylab.
> > >
> > >
> > > Hi Markus,
> > >
> > > There are 2 separate cases here:
> > >
> > > 1) The PV value getting set to the PLC value during iocInit.
> > > 2) The PV value getting set to the PLC value when the PLC values 
> > > changes after iocInit.
> > > Case 1 is handled as follows:
> > > - During iocInit the ao record device support (devAsynFloat64.c) does 
> > > a read from the driver to get the current PLC value.  If that read 
> > > operation returns asynSuccess then the PV should be set to the PLC 
> > > value, otherwise it is not changed.  This definitely works fine with 
> > > the Modbus driver, for example.  I am not sure about the adsDriver, 
> > > since I have not used it.  You can add debugging to the driver to 
> > > track that down.
> > > Case 2 is handled by adding the line you did to database:
> > >       info(asyn:READBACK, "1")
> > >
> > > That should make the ao record value update to match the PLC value 
> > > when the PLC value is changed outside of EPICS.  Note that this line 
> > > is NOT required for getting the behavior described in Case 1 above.
> > > Mark
> > >
> > >
> > > -----Original Message-----
> > > From: Tech-talk <tech-talk-bounces at aps.anl.gov> On Behalf Of Meyer,
> > > Markus via Tech-talk
> > > Sent: Friday, April 12, 2024 10:25 AM
> > > To: tech-talk at aps.anl.gov
> > > Subject: Readback of output variables in adsDriver
> > >
> > > Dear all,
> > >
> > > I am evaluating possibilities to communicate with a plc and other
> > > hardware from beckhoff. I have successfully build the adsDriver [1]
> > > device support and it seems to work very well for input variables. I
> > > can also write output variables, but if the value gets changed on the
> > > plc itself (e.g. by the interlock logic or by another connected
> > > protocol) it is not synced back to the pv. It also seems that the
> > > current value is not initially read on startup of the IOC.
> > >
> > > For example, this is the pv definition:
> > >
> > > record(ao, "$(P):IC:CB01-SSet") {
> > >       field(DTYP, "asynFloat64")
> > >       field(PREC, "2")
> > >       field(EGU,  "%")
> > >       field(DRVH, "100")
> > >       field(DRVL, "-100")
> > >       field(OUT,  "@asyn($(PORT) 0 0) REAL W P=PLC_TC3 
> > > V=HMI.IC.CB01.SSet")
> > >       info(asyn:READBACK, "1")
> > > }
> > >
> > > Before I start the IOC, the plc value is set to 2. If I start the IOC,
> > > the value of the pv is set to 0. The Timestamp is updated. The value
> > > in the plc gets not changed. I expected, that the pv value is also set
> > > to 2. Writing a value to the pv is working as expected.
> > >
> > > Did I miss something here? What can I do to subscribe for changes of
> > > output variables?
> > >
> > > Thanks,
> > > Markus
> > >
> > > [1]
> > > https://urldefense.us/v3/__https://github.com/Cosylab/adsDriver__;!!G_
> > > uCfscf7eWS!eovU-aYXFfsZIvbTx6Y3tTGma4zwbWLi1xy-LUyW3RghomFH88t9jP45toK
> > > nZq8oo1pt5gT5tQhGnPd9QXmCruzGkJg$