PID Autotune Removed from Studio 5000 v33

Why??? My guess is its because that's what your most familiar with. Because it surely cannot be for performance, price, scalability, flexibility. I know it's not for those reasons, can't be.

My customers choose what benefits them and its been mostly RA, Siemens, and Omron. At one time Modicon was on that list. I have no say on what corporate choices they make. So familiarity comes with use and for 40 years RA, Siemens for PLCs and Fanuc or ABB for robots has been what I needed to know to do the work. If one of my customers choose Beckhoff I would get familiar with that too.
 
It is now an add on license option for $ 713

After a week of nobody @ Rockwell
being able to tell me why my Auto tune was grayed out...

I finally received an answer in the from of a brand new tech note

Sales did not know it
Tech Help did not know it
Engineering did not know it

So if you need Autotune & don't have $ 713
Avoid the upgrade to Firmware v33

Or do as we did & downgrade to v32 until we get the paperwork done

Do you mean the PIDE instruction?
 
Last edited:
9324M-RLDRT131 or 9324M-RLDRT131M seem to be the part numbers. Helpfully, the product name does not contain the words Autotune nor Studio, instead "Tuner For Logix5000 PIDE Function Block"
 
Re Beckhoff vs. 'classic' PLCs.

I find Beckhoff very interesting. But to me it seems to be aimed very much on machines, not so much plants or processes.
I say this because Beckhoff is not so strong in features for PLCs that must run 24/7 and must be able to be programmed, and maintained without shutting down the PLC.
Programming while running, hardware option handling at runtime, RIUP, redundant hardware and redundant networking.
I think that Beckhoff have some features in these area, but the coverage is not so complete and not so strong as some of the classic PLC vendors.
I think that you can say the same about CoDeSys based platforms as well.
This is just from my casual glance and from following topics regarding Beckhoff over the years. So I might be wrong or some things may have changed.
 
Re Beckhoff vs. 'classic' PLCs.

I say this because Beckhoff is not so strong in features for PLCs that must run 24/7 and must be able to be programmed, and maintained without shutting down the PLC.

This applies to Rockwell. Particularly for process with PlantPAX.

Speaking about process, PIDs and Rockwell, their PID implementation is really poor for process ( the Velocity algorithm) since it works on Delta Error rather than error.
 
Speaking about process, PIDs and Rockwell, their PID implementation is really poor for process ( the Velocity algorithm) since it works on Delta Error rather than error.


I am curious why you say that, since the maths for Position and Velocity equations are equivalent; the main difference is the location of the bookkeeping for Integral term's offset.

Does the issue you have with that related to floating-point roundoff and/or precision?
 
Why??? My guess is its because that's what your most familiar with. Because it surely cannot be for performance, price, scalability, flexibility. I know it's not for those reasons, can't be.



20 years ago.....
I was rolling with a new flip phone that had this cool new feature where they integrated a camera into the phone. Imagine that, a camera in the phone! It was great because you could take pictures with your phone and display them on the little screen on the phone!

Apple had this really cool product too, called an 'Ipod'. It was actually a portable HDD that could store up to 5000 songs on it!!! Bye-bye Walkman portable CD player! You had be careful carrying it around though because it was not a solid state drive, so a hard bump would make the song skip. It was a bit big and clunky too, but I still managed to get the thing in my front coat pocket.

Sarcasm aside, the point is a lot has changed in the past 20 years. Don't tell me about 20 years ago. The automation industry landscape is changing.

BMW

Toyota

It's price... and I don't mean purchase price but rather total cost of (lifetime) ownership.


I was a Fanuc and Rockwell (generally combined) integrator for about a half decade... I had a hard time understanding why customers wanted to spec everything about what we did in controls. I felt I was being constrained by legacy gear in some situations, and software architecture in others.

And then I went back to (no travel!!) work at a local (huge) mill. This mill has equipment that has been running, in some areas, since the late 80s / early 90s. There are about 100-110 PLCs here. The entire electrical department is 10-11 guys currently. 7-8 of them have been here for >20 years. You may not like to hear this (you said you didn't) but it's real life reality in many places. They can support the PLC5 and SLC based equipment very well. It becomes tenuous as we get into Logix5000 (AOIs; etc) and impossible when they have to jump over to the 2-3 PLCs that are Mitsubishi, for example, or even Logix5000 that is done in FB and/or ST.

Given the high cost of down time and the ability of the electricians here, if machines were still sold with PLC5 hardware, they would honestly consider it.

We really don't care about what makes the integrator's life easy because 1. he's getting paid and can bill for our requirements (as I did when I was an integrator) and 2. He will not be in this plant to support his baby for the next 30 years after it is born. It takes very few hours of production downtime to justify getting the hardware / software that the local group is already familiar with, up front. It's basic business. This mill is in a remote facility and it is very hard to attract new (and ABLE!) talent. 45 minutes away, you can work at several Amazons and dozens of other huge name manufacturers, so trained and able help will often leave for greener grass.

A good example of this... I did 4 palletizing 410 robots with a sorting infeed and a full pallet wrapper that went to a GE facility that produced lightbulbs. Part of their spec included using GE PLCs. It was their name and they had a plant full of them. I advised them that (PLC) development time would reach 3x that of Rockwell and I got the answer that I myself gave above; "the cost up front matters little when you are honest about the total cost of ownership of the machine". They had a crew that knew GE PLCs inside and out (must have been in-house trained haha). They also specified ladder with EVERY rung commented in a way that explained what it did. They understand the work and that it would cost more like that (take longer), but they recognized the savings in support time. We built what they wanted, even having never touched a GE PLC previously (nor since lol).

I'm fighting tooth and nail to keep a Beckoff machine out of the plant now, because it will be the only one here. Out of the 500+ people that work here, there is no one with even exposure to Pascal / ST besides myself, and the new machines are all ST. Our guys can very easily use ladder for machine troubleshooting. None of them will be able to say the same about Beckoff.

So yeah.... your ipod is a great step-up from a Walkman/Discman for sure. But what if you had no computer (or ability) to use with it. If you had 500 CDs but were MP3 'tarded (and with no computer) then you really needed a new Discman to continue listening.

I think a lot of integrators have lost sight of this, or never had it to begin with. It's great that you love ST and can see the power of using it. It's unhelpful for a great many users that have supported ladder for decades though, and this is very often the reality.

That's my 2 cents. I'm not anti-any-brand but for ease of support, I will pick what the base here is already able to properly support.
 
Last edited:
Spot ON Frost999, exactly what I've experienced.

As an aside, I did a palletizing Fanuc at the GE Light facility in Bucyrus OH a few years back and they were following the corporate standard of GE PLCs also. Those guys knew them in and out. Wouldn't be the same facility, would it? (Funny, to configure the DeviceNet network I had to use RSNetworx.)
 
Spot ON Frost999, exactly what I've experienced.

As an aside, I did a palletizing Fanuc at the GE Light facility in Bucyrus OH a few years back and they were following the corporate standard of GE PLCs also. Those guys knew them in and out. Wouldn't be the same facility, would it? (Funny, to configure the DeviceNet network I had to use RSNetworx.)

No I think they do flouros there at Bucyrus don't they? This project was done back in 2006(ish) at their northern VA facility that manufactured incandescent bulbs. I'm pretty sure they shuttered that plant about a decade ago with the larger consumer push for CF/LED.
 
Last edited:
No I think they do flouros there at Bucyrus don't they? This project was done back in 2006(ish) at their northern VA facility that manufactured incandescent bulbs. I'm pretty sure they shuttered that plant about a decade ago with the larger consumer push for CF/LED.

Yes, and I know that had more than one plant but thought I'd ask. I thought I heard GE sold their lighting division so who knows what will be allowed if it isn't closed.
 
Given the high cost of down time and the ability of the electricians here, if machines were still sold with PLC5 hardware, they would honestly consider it.
Loss of production is always the most expensive bit of each project. Also worth mentioning, from the plant's point of view:

- Existing licensing and support contracts. Not a problem with Beckhoff, but throw other suppliers in the mix and it gets silly fast.

- spare parts. If I'm in control of what gets installed, I keep bare minimum of stock in hand. Processor cards and such add up really quickly too.

- Corporate agreements... It may not make sense to an integrator because he hasn't seen our pricing agreements. I barely ask quotes other than a certain process company because their discounts are, so far, unbeatable by other brands. This will not make sense to an integrator until he sees our pricing list.

- Certification. Particularly with Intrinsically Safe Instruments, I want to revert to the same IS barrier family and instrument because I have a circuit certificate created for that combo and is less verification work that I need to do.

I am curious why you say that, since the maths for Position and Velocity equations are equivalent; the main difference is the location of the bookkeeping for Integral term's offset.

Does the issue you have with that related to floating-point roundoff and/or precision?

The issue I find is that the calculation of error is not PV-SP. It's instead
E0 = PV[k] - SP[k]
E1 = PV[k-1] - SP[k-1]
E = E1-E0

So you have to pay an awful lot more attention to the time required between PID cycles or otherwise it will barely move despite having a massive difference to error.
 
The issue I find is that the calculation of error is not PV-SP. It's instead
E0 = PV[k] - SP[k]
E1 = PV[k-1] - SP[k-1]
E = E1-E0

So you have to pay an awful lot more attention to the time required between PID cycles or otherwise it will barely move despite having a massive difference to error.

Ah. I am fairly certain that those statements mean @cardosocea does not fully grok how the Velocity Form of the PID Algorithm works.

First let's re-write those lines with slightly different notation, also correcting

  1. the minor typo ("E = E1-E0" should be "E = E0-E1"),
  2. and the major misunderstanding about what (E0-E1) actually is i.e. it is not the "error:"

E = E[k] = E0 = PV[k] - SP[k]
E[k-1] = E1 = PV[k-1] - SP[k-1]
ΔE[k] = E[k]-E[k-1] = E0-E1


The main points are that

  1. the first equation is the calculation for the error, E, of the Error for the current update (update "k"), and
  2. the final equation is for ΔE[k], the change in Error, not the Error (E) itself; this is the key misunderstanding here.
    • Simply put, since the Pterm = Kp E (Positional Form),
      • then ΔPterm = Kp ΔE (Velocity Form)
TL;DR


With those details sor'ed, let's look at the two forms, Positional and Velocity, of the PID algorithm (cf. this link):
xxx.png
So the for the current (k) and previous (k-1) updates, the positional forms are
Code:
                            k
CV[k]   = Kp E[k]   +[SIZE=2] Δ[/SIZE]t Ki Σ (E[j]) + Kd/[SIZE=2]Δ[/SIZE]t [COLOR=navy]Δ[/COLOR]E[k]
                           j=1

                           k-1
CV[k-1] = Kp E[k-1] + [SIZE=2]Δ[/SIZE]t Ki Σ (E[j]) + Kd[COLOR=navy]/[/COLOR][COLOR=navy][SIZE=2]Δ[/SIZE]t Δ[/COLOR]E[k-1] 
                           j=1
Noting that
[Sum (Σ) of E[j] for j=1 to k]
is equal to
[Sum (Σ) of E[j] for j=1 to k-1] + E[k]
we can re-write those two Positional forms as follows:
Code:
                           [I][COLOR=red][B]k-1[/B][/COLOR][/I]
CV[k]   = Kp E[k]   +[SIZE=2] Δ[/SIZE]t Ki[COLOR=Purple][B] Σ (E[j])[/B][/COLOR] +[SIZE=2] Δ[/SIZE]t Ki [COLOR=blue][B]E[[I][COLOR=red]k[/COLOR][/I]][/B][/COLOR] + Kd[COLOR=navy]/[/COLOR][COLOR=navy][SIZE=2]Δ[/SIZE]t[/COLOR] [COLOR=navy]Δ[/COLOR]E[k]
                           [COLOR=purple][B]j=1[/B][/COLOR]

                          [COLOR=purple][B] [COLOR=red][I]k-1[/I][/COLOR][/B][/COLOR]
CV[k-1] = Kp E[k-1] + [SIZE=2]Δ[/SIZE]t Ki [COLOR=purple][B]Σ (E[j])[/B][/COLOR]              + Kd[COLOR=navy]/[/COLOR][COLOR=navy][SIZE=2]Δ[/SIZE]t[/COLOR]Kd[COLOR=navy]/[/COLOR][COLOR=navy][SIZE=2]Δ[/SIZE]t[/COLOR] [COLOR=navy]Δ[/COLOR]E[k]
                           [B][COLOR=purple]j=1[/COLOR][/B]
Subtracting the latter from the former, the second terms (Δt Ki Σ(E[j]) for j=1 to k-1) cancel, and we are left with:
Code:
CV[k] - CV[k-1] = Kp (E[k]-E[k-1]) + [SIZE=2]Δ[/SIZE]t Ki E[[I]k[/I]] + Kd[COLOR=navy]/[/COLOR][COLOR=navy][SIZE=2]Δ[/SIZE]t[/COLOR] ([COLOR=navy]Δ[/COLOR]E[k]-[COLOR=navy]Δ[/COLOR]E[k-1])
Finally, we

  • Add CV[k-1] to both sides,
  • Substitute ΔE[k] for (E[k]-E[k-1]) in the first (P) term on the right side,
  • And expand E[k]-ΔE[k-1]) in the third (D) term on the right side,
    • ΔE[k] = E[k] - E[k-1]
    • ΔE[k-1] = E[k-1] - E[k-2]
    • So
      • ΔE[k]-ΔE[k-1] = (E[k] - E[k-1]) - (E[k-1] - E[k-2]) = E[k] - 2E[k-1] + E[k-2]
which yields this:
Code:
CV[k] = CV[k-1] + Kp [COLOR=navy]Δ[/COLOR]E[k] + [SIZE=2]Δ[/SIZE]t Ki E[[I]k[/I]] + Kd[COLOR=navy]/[/COLOR][COLOR=navy][SIZE=2]Δ[/SIZE]t[/COLOR] (E[k] - 2E[k-1] + E[k-2])
Which is essentially identical to the Velocity Form of the PID algorithm that A-B claims to use. QED.

It's all just bookkeeping ;).
 
Last edited:

Similar Topics

Hi I'm working with twdlmda20drt twido PLC. I can work with the pid but i can't work with autotune, i read the whole instruction sheet but it...
Replies
2
Views
2,486
When I tried to use the autotune feature, my CompactLogix complained that the input was not stable enough prior to the step change. I had to tune...
Replies
0
Views
1,936
I have a process temperature control unit that has 5 pid loops. The problem is when the setpoint is changed, all the pid settings need to be...
Replies
1
Views
2,175
Hello all. Anyone used the AutoTune Function on an M340 w/ Unity Pro XL??? Got an injection mold machine retro. and I would like to try it out. I...
Replies
2
Views
6,352
My first job had something to do with PLCs but then I switched my trade and went on to become a "Design engineer". Recently, I've designed a PID...
Replies
4
Views
7,926
Unregistered
U
Back
Top Bottom