Yeah, I'll dump the unipis from my project, this and (after updating the ide) "Your project has been created with (old version), please stick to that version forever to avoid errors" did it for me. Yes, S7 or Wago PFC are more expensive but concerning the hours I dumped into troubleshooting the are the better deal.
Posts made by _andi_
-
RE: Weird PID Behavior
-
Weird PID Behavior
So,
I try to do some heater controlling and no matter what I do the DFactor stays zero. The only thing which brings some action is to use a negative D value which brings dampening but on the ramp-up and not when reaching the target value. Long story short: Can't for the live of me damp the overshoot the PI Part is producing and I need an "overdampend" setup because the customer is sort of intolerant concerning overshoot. Also the underlying Formula is sort of weird.
Has anybody any advise or has implemented a working PID with anti windup for Mervis?
Thanks, Andi
-
RE: Deploy to muliple Neurons
Yeah, three one-wire temperature sensors. The upside is that we average the values, so we don't really care which one is which. But if I can't find a way to loop over the sensors without the bus mapping then this feature would become much more involved. Or some shell/sscp/??? magic to replace some one-wire address values after deploying...
-
RE: Deploy to muliple Neurons
Hi Martin,
thanks!
AUEPLC / Neuron. Mixed in the project abbreviation for the thing, sorry.Sure, command line would actually be appreciated, better for automation.
Yeah, I guess what I'm doing is sort of unusual. It's augmentation for an industrial plant, plenty of little heaters/fans to squeeze a bit more efficiency out of an old workhorse. Since operation is not critical we opted to try the neuron/mervis combo since the price point is really attractive compared to lets say siemens or wago units.
-
RE: Deploy to muliple Neurons
Hi, thanks.
It's also possible to save some time by copying a AUE and live with the crude naming style. Would be nice to be able to change the prefix.
But as soon as you change for example settings on a modbus server you have to change it for every single PLC. Annoying for the six Neurons we're using atm, a dance I'm not looking forward to for the full installation (40ish PLCs).
-
RE: Weird debugging / startup behavior
... so, the trigger for this behavior was a division by zero on the first call to the function block (getdeltatime4...). Combined with the "stop on exception" setting the plc does just that, which is fine. Annoying though is that neither IDE nor HMI bat an eye on that, you have to go to "system status" and then to the very bottom to see that your plc just stopped all operations. Is there any way to catch such errors? I'd love to get an email when something like this happens...
-
RE: Deploy to muliple Neurons
so, I guess there is no documentation...
I sort of managed to be able to deploy one program to multiple plcs by not using autogen at all and doing io mapping in a way that it works if I copy&paste the plc in mervis and attach another physical plc to that copy. Far from ideal, I am not looking forward to doing this with 30ish plcs...
-
Weird debugging / startup behavior
So,
I added a function block in mervis. When I deploy my solution, the plc starts up in some stopped state and I have to connect to it via debug and do right-click, resume normal tasks. WTF?
Also, the HMI is working just as normal, acception new values, no error messages whatsover.
How do I turn this off/is there some documentation for this feature somewhere out there?
Andi
-
Deploy to muliple Neurons
Hi,
so I've a bunch of Neurons, all should run the same software (control for a couple of similar heater boxes). From my point of view these only differ in ip-address and some one-wire ids. Is there a manual somewhere on how to do such deployments with the least amount of hassle?
Thanks, Andi
-
RE: ModbusTCP Server Real (datatype) Problem
Ah, thanks! I just clicked "add variable" and didn't realise I can / have to actually edit the offset afterwards. Might be worth a feature request to change the gui to automatically check the previous value and adjust the starting element accordingly.
-
RE: Safe DO/PWM Operation with Open Collector
In case somebody else stumbles over this: Another Problem is that the Digital Output is not 0V on 0 PWM / LOW due to the transistor inside the unipi. The voltage (around 0.6V) is just enough to kick on the fan on occasionally / keep it running from time to time (starts around 0.4V).
I mitigated all issues with a wago 859-791 optocouple for each PWM, a rather pricey solution.
-
RE: ModbusTCP Server Real (datatype) Problem
Hi,
thats what I'm, doing. Just having the real works fine, adding the int breaks the real.
https://drive.google.com/open?id=1W7ZJpO1MUiSXMdvRWszUj2BQtqj9uxx2
Thanks, Andi
-
ModbusTCP Server Real (datatype) Problem
Hi,
I'm using Mervis RT, setting up a modbustcp slave (also called server). When I add a variable of type real and choose "Identity" as conversion method I get one Int on the bus instead of two. Is this a known issue or am I missing some setup?
Thanks, Andi
-
RE: Safe DO/PWM Operation with Open Collector
Hi Martin,
thanks for the answer! I'm using the current mervis image on the unipi. Sadly I can't use the analogue out because it's already in use for a valve :-(
Thanks, Andi
-
Safe DO/PWM Operation with Open Collector
Hi,
I'm running a Mervis on Unipi Neuron s103 with an ebm-papst fan (W3G200 Series, 220V, can hurt fingers), setting RPMs with (low power) PWM. The papst PWM is either controlled by 0-10V or PWM using the same wire. So with an Open Collector I've a pullup resistor to 10V. The Transistor on the DO has to become conductive to pull the wire to ground. As long as the DO is low, the fan sees 10V/100% duty cycle and goes to full throttle. This is quite unsafe, the fan spins up as soon as I power up the unipi. I tried to set Inital Values for the DO and PWM but that does not seem to work as expected.
Now I'm wondering what the best practise for this kind of situation is? Should I run the 10V through a relais, sacrificing another digital out? Or can I run DGROUND and AGROUND to a relais? That would have the benefit of providing a sort of "arm" circuit for the implement on the analogue out as well. Or is there any other way I couldn't think of?
Thanks, Andi