why do plc's have so little memory still?

In my experience, for you to get away with installing a system based on a Linux SBC you'd have to be selling something that no one else had... and that would be hard as well if you weren't in charge of the manufacturing of said boards. No one would be willing to pay reengineering time when a controller went down because you couldn't source parts.

I've seen stuff like that in the field... but the manufacturer makes their own controllers and IO's and there is perhaps one manufacturer that offers something remotely similar to their product so there's no real choice there. One thing that they also have is a fairly open system where you see exactly what the program is doing, despite not being able to modify it... I can only imagine the amount of hours taken to develop something like that.

Valid points my friend, but not going to use an SBC, it'll probably be Advantech's UNO-1172AH. I need C1D2.

This is one of my dilemmas -- I'm seeing a lot of people taking hobby boards and adding interfacing hardware (convert 24VDC logic to 3.3VDC logic, etc.), but no one bothering to do approvals.
 
Arduino makes many different models, but colloquially this refers to their Uno model, which is a lame 16 MHz 8-bit CPU with no FPU, basically representing the early days / dark ages of hobby boards. They did this on purpose so they could get something cheap, available, and simple out for students and hobbyists to play with. Those CPUs cost cents.

30 euro from the maker of the board is not what I call cheap.
Plus, up until a few years ago, if you asked an electronics engineering student to produce a development board based on the ATmega8 family, you'd get an Arduino. There is no magic on that board at all... it's taking the basic design from ATMEL, swap out expensive components by others such as the crystal and add a USB - TTL bridge and you're good to go.
The original bootloader was basically the translation of an application note into code.
The real strength was making functionality that is usually not so simple to achieve as it involves programming protocols or dealing with registers into something an art student could use.

Until Art students are required to program industrial facilities, the Arduino is going to stay where it is.


Same thing with Raspberry Pi, the first one really sucks, it's an outdated ARM chip that doesn't even have out of the box support for most Linux distros (at least it didn't until people went nuts about them).

Getting to the point, the hobby/maker market is now FLOODED with new and better stuff. Most Linux boards have multi-core modern ARM chips that you would find in your phone, with specs to match. The non-Linux ones are also using modern ARM chips, but of the microcontroller flavor (Cortex-M, not Cortex-A).

What applications would you use a raspberry pi instead of a PLC? And which customer would accept it?


Relevant question, has anyone else seen OPTO 22's digital I/O system for the Raspberry Pi?

No, but I've seen this.
https://www.rugged-circuits.com/24v-industrial/24v-industrial-shield
Sorry, couldn't resist it. I'm sure I've seen one website with something like that available, but it may have been a kickstarter page.
 
@cardo: I don't understand what argument you're trying to make, I said Arduino is ****ty and you're saying it's ****ty too, so we agree. I was saying there are much better boards out there now.

I don't understand the 'art student' bit either, the primary users of these boards are from technical fields. People are doing very advanced things with these (not necessarily Arduinos, hobby boards in general).

Regarding RPi, you would use it or a PC instead of a PLC for things that a PLC can't do, like FFTing analog signals, or running a database, or higher precision math. FYI, RPis are on plant floors right now.

Siemens, one of the largest and most popular industrial automation companies on the face of the planet, came out with this: http://w3.siemens.com/mcms/pc-based-automation/en/industrial-iot/pages/default.aspx a Linux IoT gateway compatible with the Arduino's pinout. What does that tell you?
 
Last edited:
$75.00. Where's Modbus for my HMI? Easily expanded? DIN rail mount? Etc...
Recent article for Arduino vs PLC. PLC is the latest from Automation Direct.

I wouldn't buy that if you paid me to buy it. I brought up the OPTO 22 and Siemens stuff to make the point that major companies see the value in hobby boards, and the potential they hold.

To clarify, I'm arguing on the software and computational power fronts, PLCs win hands down for real industrial digital and analog I/O, for now.

Modbus for your HMI? I'll write you a TCP server in a few hours tops, if not less. I had an internal project I was working on where I used an Arduino Due to poll values from a Sierra vortex meter and a Micro Motion, the program took less than an hour from start to finish. Raw comms are easy on the Arduino.
 
Do you realize that the DirectLogic PLC line is over 20 years old? It's like comparing a ControlLogix to a PLC 5.

And the ControlLogix platform is close to 20 years old. . .
 
jraef
I still have a 3.1 machine that I keep alive. I also have 2 backups that I bought used. All this for one customer, a 35 year old machine that just keeps cranking on and making money.
I am the only one who can repair it or make changes because I kept that old software. The quotes they got for upgrades would take the machine out of production for 6 weeks.
 
It is total BS. I don't think the PLC manufacturers are colluding. I do think they are trying to 'milk' their old designs for as long as they can.

Most microcontrollers that have REALS also have the ability to support DREAL. The problem I see is where would you need DREALs? A REAL provides more precision that can be measured or manufactured. We are changing some of our internal code to use DREALs. We have found that when doing complicated calculations for motion that DREALs still can only provide about 10 digits of accuracy. REALs provide about half that.
I bet most of you never need that kind of accuracy unless you want to use a DREAL to count integers past 16,777,???.

The problem I see with using DREALs and LINTs is with communications. Ethernet packets are only so big so there is a limit to how many bytes of data can be sent as 'consistent' or atomic data. Sending commands and reading the status over multiple packets makes it hard to keep things synchronous.

Our latest motion controller has 512 MB of ram. We are not using it all. In fact less than half. After a while old ram sizes are hard to get and actually cost more than larger ram sizes. We use 5 ram chip to get 512 MB. Why 5? ECC. We have been bit too many times by getting bad lots of rams.

Our latest processor is a P2020. It is a dual core processor that generates 7 watts of heat. That is 6 times more than our previous design. We had to design an aluminum heat sink to dissipate the heat. Heat is a problem. Our design has been tested in thermal chambers. It can be done. Still the cabinet needs to dissipate that heat.

In the PLC manufacturer's defense. They really are selling software or firmware in hardware units. What you don't realize is that when you buy a PLC or one of our motion controllers, 99% of the programming is already done for you.
 
I remember AB PLC3 days....128MB memory module was over $10,000.

A 16K memory card for the Series Six went for $4,000.

I remember needing to upgrade from the 12K board (4K of programming space, 8K of registers) to the 16K board (8K program, 8K registers) and complaining to the GE guy about not only having to spend $4,000, but toss out my $2,000 board as well.

He said "Yeah, the cost is excessive, especially considering there really isn't much difference between the boards".

That got me curious, so I compared two boards, and sure enough, the 12K board had jumper J7 (Zero Ohm resistor) clipped out of it. GE had actually made the board as a 16K board and then crippled it by clipping the jumper out.

I put the jumper back in, and voila, I had a 16K board. :)

So what could you do with around ~50K of memory on a Series Six?

We had one PLC that had 48K that ran three conveyor systems and had ~2000 I/O. With diagnostics for the conveyor system that took up 60% - 70% of the program. And it all ran in under 80 milliseconds.
 
@cardo: I don't understand what argument you're trying to make, I said Arduino is ****ty and you're saying it's ****ty too, so we agree. I was saying there are much better boards out there now.

Sorry, I replied without having seen your other reply. I agree that Arduino is nice for home projects and prototyping in certain situations (NASA uses them for prototyping and proof of concept, for example) but never for industrial applications. And the same goes for a raspberry pi sort of board too unless you take the time and expense of proving it.

I don't understand the 'art student' bit either, the primary users of these boards are from technical fields. People are doing very advanced things with these (not necessarily Arduinos, hobby boards in general).

From Arduino's website:
Arduino was born at the Ivrea Interaction Design Institute as an easy tool for fast prototyping, aimed at students without a background in electronics and programming.

You can look about the motivations behind it and projects and you'll see the creators give quite a lot of enphasis on art students being able to create things based in electronics. That's where the art student comes from.

And yes, even with Arduinos are doing advanced things... I have one connected to the internet querying Transport for London about when is the next bus arriving near my door. Sure, most of the smarts is in a server somewhere, but it's still pretty amazing what an 8 bit processor is doing.
But I would never even consider having one of those in an industrial facility controlling a process.

Regarding RPi, you would use it or a PC instead of a PLC for things that a PLC can't do, like FFTing analog signals, or running a database, or higher precision math. FYI, RPis are on plant floors right now.

Siemens, one of the largest and most popular industrial automation companies on the face of the planet, came out with this: http://w3.siemens.com/mcms/pc-based-automation/en/industrial-iot/pages/default.aspx a Linux IoT gateway compatible with the Arduino's pinout. What does that tell you?

The RPi's on the plant floor aren't replacing the PLC... they are replacing a computer running windows or linux. The problem with that sort of thing is that someone will have to support it and that may be where companies won't get behind it.

Intel, the largest processor manufacturer in the planet launched development boards aimed at the developer community... they killed them off one or two weeks ago, what does that tell you?

Siemens makes anything through which electrons go through... they are taking a punt with this. It may pay off, or it may not... and for sure they have that into account.
Also, there are a lot of people jumping on the IoT bandwagon... but I'm not so sure that it's not more than a fad. Time will tell, I guess.
 
The problem I see is where would you need DREALs?

My programs use some temperature correction equations that involve adding and multiplying numbers that are magnitudes different, like 1 + 1e-5 + 1e-9 + 1e-13, and so on. I'm tired of calcs not matching Excel, because Excel is decent enough to use the de facto standard (64-bit precision).

P.S. You should be getting around 15 accurate digits with an IEEE 754 compliant doubles
 
Epy said:
P.S. You should be getting around 15 accurate digits with an IEEE 754 compliant doubles
We do if only 1 calculation is involved. I need to test motion profile generators. Most of the time the resolution is close to 12 to 13 digits. However, I must test for worst case.
I have run a test that tries 200 trillion motion profiles. The worst case had errors where the precision was only 10 or 11 digits.

1xE-13 is pretty small but if you really need it then I agree, REALs won't do.
Even with DREALs you would lose lots of digits adding 1 and 1xE-13
 
@cardo: Intended audience != real audience, technical people use hobby boards

I would bet my left nut Intel did that because their Quark processor that they developed for this was a massive failure in every way. Quark is an x86-compatible processor but has the same instruction set as a friggin 486, so even the crappiest ARM chips run circles around it. In addition, in certain cases a LOCK instruction (basic building block for implementing shared memory) causes the CPU to crash. The Galileo board they killed off used this POS Quark.

Have you looked inside a ControlLogix card before? Most of them have ATMEL chips, you know, the same people who make the main chip on the Arduino.

Hobby boards are not (all) unreliable cheap ****, most are made using the same damn chip manufacturers used everywhere else: ATMEL, STMicro, TI, NXP, etc.

Like Peter mentioned, the real difference is the software/firmware.
 
We do if only 1 calculation is involved. I need to test motion profile generators. Most of the time the resolution is close to 12 to 13 digits. However, I must test for worst case.
I have run a test that tries 200 trillion motion profiles. The worst case had errors where the precision was only 10 or 11 digits.

1xE-13 is pretty small but if you really need it then I agree, REALs won't do.
Even with DREALs you would lose lots of digits adding 1 and 1xE-13

Makes sense. If you need more precision, you can go quad precision :)

As far as what you said about 64-bit and communications goes, I plan to use 64-bit reals to do everything, but will provide 32-bit downcasted reals of the final calculations, similar to some Krohne products.
 

Similar Topics

I am working on backing up some Little Star PLC's for a customer. I am able to download to them but can't upload. Has anyone had any dealings with...
Replies
9
Views
2,144
Today, Orbital ATK fired the latest of their five-segment solid rocket boosters (SRB's), at the test facility in Promontory, UTAH. The center of...
Replies
2
Views
1,673
please help me to buy a set of s7 400 and s7 300 for my laboratory. we want to have a pcs7 hardware configuration for making a little and low cost...
Replies
7
Views
4,780
I would like a little help ID'ing a HMI - PLC combo unit. I have looked this thing over really well, and I have yet to find a brand name on it. I...
Replies
5
Views
2,177
Hello everyone, I'm really interested in playing with PLCs, i'm thinking of picking up a TECO GENIE II SG2, its relatively inexpensive. As for the...
Replies
14
Views
3,427
Back
Top Bottom