SoftLogix?

[Stupidav 1]

I am getting ready to do a new project where I am going to be having between 4 to 6 Servos and some where around 48 feedbacks from encoders, analog devices, and only a few I/O. The system is going to be moving around quite fast, simulating the suspension on an automobile, while monitoring the stresses on the frame at many different points. I plan on having some type of datalogging happening as well. My main concern is getting the highest update rate into the datalogging system as possible. I have only played with SoftLogix, and so I am not sure if it would be a better solution rather than several ControlLogix processors (I say several because, I will have one doing the motion, and the other specifically for feed back). Either way I go I plan on having a sperate PC for the data collection. If it runs on a PC would I be able to get an increased scan time, thus increased RPI capabilities? Initailly I was going to use a "Motion Controller", but after some additional research, I found that I am not going to be any better off, so I am sticking with what I have had the most experience with (AB). I have seen sytems that can do simular testings, but they are all proprietary and this project will be testing a variety of suspensions and frames each of which having a life span of only a couple years before the frames and suspensions are changed. Any feedback, ideas, or critizisms are greatly appreciated. Edited 20 Apr 2006 by Stupidav

[TimWilborne 108]

Personally I would go with Controllogix I'm not for PCs in industrial environments but that is an argument for another thread. My big concern with Softlogix is although it is similar to the Controllogix it is not very popular. The first thing I would ask myself is why? I would probably do this with a single processor. I have done dual processor applications in the past and found that there was not enough performance gains to justify the headache. We eventually converted these to single processor applications and found we actually increased performance and response time by doing this on one of them. 6 axis servo and 48 feedback devices doesn't sound like to much load to me. But would need specifics to run numbers RPI capabilities as I understand them are really not scan time dependent. They have more to do with the amount of information between devices and network (including Controlbus) load.

[Gerry 13]

Since this is a test-bed application as opposed to a production application, I can't see any reason to exclude SoftLogix from consideration. Running on a 3.x GHz P4 it would out-pace a L6x - which will help the motion planner. I've worked with SoftLogix 5 but never with SoftLogix 5800. I would guess that there are many similarities. For instance, it seems to run much more reliably on the server editions of Windows; you'll likely need to tweak some time-slice settings (SL vs Win) to get optimum performance. I believe there is a 32-axis limit on a CLX processor, so I'm curious how you plan to handle the 48 encoder feedbacks.

[Stupidav 1]

I know what you mean TW, I have been too nervous to use a PC for any machinery control, but in this case it is just a testing station / non production. I have some what wondered why it isn't as popular myself. Initially I really wanted to use it on a project, untill I saw the price, and I never had a project untill now that the cost justification could be there, but I am just not sure, and don't feel like reading thier 1 of thier many 462 page manuals. Also like I first mentioned, way back then I had a hard enough time getting a PC to come off from a screen saver with out the Blue Screen of Death, let alone running a machine. Thats why I haven't done anything with it. I can only guess that these are the main reasons that it hasen't gone anywhere as well. But with newer, more relieable, PCs and windows becoming more stable, is partially why I am contenplating going this route. The main reason why I would use two processors, is to some what simplify the data aquisition. I would use the one for the test stand control, operation, and monitoring. The second would be just for all the feed back devices, conversion of raw data, and data transfer to the data aquisition system, no control. Thanks for your opinon TWControls! Eventhough I could just use a HSC or somthing for many of the encoders rather that having actual Axes for each. Thanks for your opinon as well Gerry!

[TimWilborne 108]

I don't understand this statement. What are you needing to gain from hooking these to a servo card instead of a HSC? You could create Virtual Axis off of feedback coming from HSC Edited 21 Apr 2006 by TWControls

[Gerry 13]

As I understood it, there are 4-6 servo axes plus 48 separate encoder inputs plus analogues. Maybe the 48 feedbacks are a mix of encoders and analogues - I don't know. What's not explained is how often the positions (velocities? accelerations?) of the separate encoders are to be sampled or how precisely the samples need to correlate in time to the 4-6 servo positions / velocities / accelerations. I don't think a virtual axis driven (how? - move? jog?) from sampled HSC data will provide a useful mirror of the encoder's actual motion.

[BobLfoot 301]

Just a wild hare of a thought, but the Micrologix 1100 with it's $500 list price, ethernet and analog capability might be just the unit to create a DCS system with.

[Stupidav 1]

I haven't put too much thought in to using a something like a Micro or somthing else for the DCS, so I am not going to rule it out, as long as it can meet my rigerous needs. I haven't done the exact calculations yet, but for current thoughts, imagine the wheels on your car going down the road. Hitting pot holes and speed bumps and other bumps at speeds of 70-80Mph. Well 4 Axes are basically going to the up and down travel simulating all the bumps. The 5th Axes will be basically simulating a trailer attached to the frame. I am not sure about the 6th yet, just leaving the possibilites open. The feed backs are going to be various devices, many along the lines of Linear position transducers, optical alignment sensing, load cells, accelerometers, and strain gages. There total amounts of each have not been determined as of yet. Basically this system will monitor different all the different stresses to all the different parts of the frame in as close as possible to a real scenario, without going to the test track, or even having a fully assembled vehical to see what fails first. For the most part I plan to do somthing along the lines of create a motion profile on a PC and download. When the system is started, the I hope to record the what the servos are actually doing and all of the sensing with the fastest sample rates as possible. Eventhough 100ms would probably be ok, i will be satisfied if I can get less than 10ms. I didn't necessarily mean that all the encoders were going to be hooked up to a (or several) servo card, I just haven't had an opertunity to see what the best way to get the encoder signals back into the system. I have done it both ways, but never compared on to the other. Thanks again for the feedback, its helping me focus on this project and all the possibilities and learning that can come out of it.

[Mark Buskell 0]

I had a dynometer on a 503. The 503 was linked to a computer. The data was coming in to fast for the computer to capture it real time. This was ten years ago. What I did was capture the data in the SLC and then after the test, transfer the captured data to the desktop computer to display trending and such.

[Stupidav 1]

That will have to be another aspect that I better give some thought to, bringing in all the data at once could potentially use some drastic resources. Where as like you said, retaining some infocould be stored in the Logix system an "exported" at a later time for evaluation, could prove to be critical. Although I kinda wonder what the differences would be in this situation between SoftLogix versus Control Logix. Thanks for your info.

[OkiePC 14]

The last time I looked at Softlogix5, I noted that there was no way to get high-speed local I/O. That was quite a while back though, and there may be something available now. Anyway, if all the I/O is updated as remote I/O that may impact your ability to make high speed control decisions. Your logic engine will be at the mercy of the communicaiton efficiency. Now, with controlnet, or even ethernet IP that may not be a problem, but I would go with the C-Logix CPUs and controlnet for remote I/O if necessary. Edited 11 May 2006 by OkiePC