Roll your own SPI interface to a molding machine

I am going to be interfacing a Fanuc robot to a Toshiba molding machine. Is it acceptable or good practice to use Ethernet between between the robot and PLC for the permissives (Permit mold motion, permit mold close, ect)or should I do this with discrete I\O? I believe you could configure the Ethernet I\O on the robot side to fail off if comms are lost. Additionally you wouldn't have to worry about the contacts sticking together. Also would it be acceptable to wire the molding machine inputs (Using the molding machine PS) directly to transistor inputs and use a relay output module for the molding machine permissives or should I run both through interposing relays? I know on the machine side they are going through interposing relays that are just regular ice cube relays but not sure if this is done only because SPI is an added option or for best practice. I know you can get a SPI from Fanuc but it is quite bulky and would prefer not to use it.

As always,many thanks
Bill

I understand there are two questions here
1. Can you use E/net between the robot and the Macnine for a safety interface.
- I ask what is the safety level of the interface?
- What is the risk analysis of the machine? - (loaded question)
99.99% of robots are a CAT.4 (Sil.4) safety risk
Not much diference with Molding machines either

2. to Use interposing relays or not.
Both machines use different power supplies, you do not want the risk
of shorting inputs or outputs
So use Interposing relays

Thanks for the reply. SPI is not actually a safety interface but rather a machine robot interface standard used in the plastics industry. I believe it stands for Special plastic Interface. Among other things it gives a input to the robot telling it not to go into the mold because the mold is closed and then the robot will give the machine a input not to close the mold when the robot is in the mold area. There is a output from the molding machine giving the status of the safety gates but that signal goes through a regular relay so it is not actually a safety output but for status only. My concern is maintaining communication or should I loose communication failing in a way that the robot won't crash into the machine or that the machine won't close the mold on the robot. I plan to have the PLC handling the interface (between the robot and the machine)so that I can program other features and I also need to know machine status in my program.

I did a lot of robot interfaces back in the 80s and 90s when discrete was the only option (EUROMAP 12). Everything was single-channel too, so sticking relays was an issue. I think the newer interfaces are all dual channel for the E-stop and guards, which is great. I don't know how you could easily maintain the same safety integrity over Ethernet.

Keep in mind that this is IS a human safety interface. The molding machines guard circuit gets tied into the robot safety, so if a guard on the IMM is open, the robot knows about it. Same with the robot access doors. Their status must be shared with the IMM, because the non-operator side of the IMM is usually open for the robot.

IMO, you can use Ethernet IN ADDITION to the hardwired interface, but I wouldn't use it INSTEAD.



-Eric

Agreed. I would wire the molding machine gate and E-stop SPI circuits into the robot E-stop circuits and visa versa. I plan to use Ethernet from the robot into PLC for the machine non safety interface signals and then use the PLC I\O to interface with the molding machine SPI (again only for the machine non safety interface signals). Since the other signals can only effect machine operation when E-stop and gate circuits are enabled they present no safety risk to the operator. The robot will enter the machine on the non-operator side so I will have to defeat the gate circuit on that side but it can only be accessed by entering the robot fence gate that will be wired into the both the robot and molding machine E-stop circuits. I will do as suggested and use interposing relays. Would opto-couplers be a better chioce considering I could crash the robot if one of the relays got stuck?

In that case, Ethernet is fine. Personally, I would hardwire the Mold Fully Open (Robot Entry Permit) and Robot Clear (Mold Close Permit) signals as a physical backup of the Ethernet signal.

Where were you planning on using the opto-couplers? I'm guessing for the non-safety signals? Yes, opto-couplers would probably be better than mechanical relays. The E-stop and gate relays will need an additional normally-closed contact for monitoring by the safety relays, so I think you'll need force-guided relays. I like the Omron G7SA series.

http://www.ia.omron.com/data_pdf/data_sheet/g7sa_ds_csm1269.pdf



-Eric

I was thinking I would use a light curtain on the non operator entry side with safety relays and wire the permit mold motion and permit mold close signals through them so that when the robot breaks the light curtain these permissive are taken away. That way it would be redundant. Yes, the opto-couplers would be for the non saftey signals and I would use safety relays for the E-stop and gate circuits.I think I can configure the Fanuc comms so that they go to 0 if I lose communications. That way I would lose the permissives and prevent the robots from crashing into the mold or the mold closing on the robot. Thanks for your input.

I meant to say that I would wire the permit mold open and permit mold motion robot permissives outputs in series with the light curtain safety relays

wdd0422 said:

I meant to say that I would wire the permit mold open and permit mold motion robot permissives outputs in series with the light curtain safety relays

Click to expand...

Redundancy is always a good thing!...

I am curious about why there is a light curtain between the robot and mold. I don't know your workcell layout, but isn't there a already a continuous guard encompassing the robot? Why the need for a light curtain in place of the non-operator gate?...



-Eric

Yes there is a fence. My thinking was that if someone was inside the fence and the gate was unintentionally closed for some reason then the motors could get turned on and the mold could move. Also if for some reason the robot was stopped in the mold and they put the machine in operate without robot they could shut the mold on the robot. Seems unlikely but I have been burnt before when I listened to people that said "oh that'll never happen" finally just as a redundant in case some how those permissives got stuck on somehow because it would be so catastrophic if that happened. The EOAT alone is 2 sided with 16 vacuum pickups on each side,vacuum sensors on each. It's fairly sturdy and also pricey but I doubt it would take a direct hit into the mold or if the mold closed on it.

Better use of light curtain

Actually this a 2 part system I use with robots on mold machines so I can ignore the SPI for all the reasons already listed.

First I use the light curtain to detect if the Robot is in the mold area. You can tie the output of the Light curtain controller to the mold machines safety circuit with a muting function. The robot will break the light curtain beam when it enters the mold area.

The second part is a temposonic rod that ties directly into the robot controller. This allows you to see the mold machines position without relying on relays. To make this work you attach your magnet to the moving platt while the rod mounts to the mold machine frame. The only thing you need then on the SPI interface is the command to close the mold.

Basically the way it works: As long as the temposonic rod sees the mold machine is open far enough the robot is allowed to enter. If the light curtain beam is broken and the mold is closing or closed to far then the mold machine safety is kicked in and the mold machine comes to a screeching halt.

I actually designed this for a mold machine that did not have a SPI interface. While I will not say it is "full proof" I have not heard of it failing. I did add code in the robot controller that also said the temposonic rod had to read within a certain range or a fault happens.