rezabejd
October 28th, 2005, 07:48 AM
All,
Just wanted to pass on one of our recent experiences / findings.
Summary: DCS cabinet was built with no space between non-redundant DIN-rail mounted power conditoners. They've been getting hot for over 5 years now - constantly under power - and many (more than half) started putting a sinusoid of various amplitudes out on their associated segments.
We found this after experiencing a rash of communication erros on two unrelated segments in different parts of the plant. An oscilloscope revealed the imposed sinusoid (wavelngth about 100x FF communications).
While the problem caused mostly nuisance errors (controllers shedding mode) I think eventually, we'd have had a couple valves going to their fail positions. In my mind, an active physical layer diagnostic would have revealed this issue before > 50% of our power conditioners failed.
The good news is, that even with all the errors, the problems experienced didn't cause a major disruption of the process. Mostly just a nuisance for the operators, who had to go through a shift or two with persistent nuisance alarms, and having to run some loops in MAN or local auto.
Backround
When we built our system in 1999, there weren't any redundant FF power conditioners on the market. When we split apart the case of the (non-redundant) power conditioner recommended by our DCS supplier, we saw IC's and fuses, which was a concern. Our host manufacturer was able to come up with a redudnant power conditioner, or more correctly, one where the non-redundant components were all "simple" devices (no fuses & no IC's (integrated circuits)) the idea being that such a configuration had a MTBF comparable to 250 ohm resistor, the single point of failure in conventional systems.
We only did this for a small percentage of the segments (c. 15% or so) because the redundant solution (at the time) was large form factor and expensive. So we still had between 50 and 60 segments with the non-redundant power conditioners.
The current documentation for the power conditioners we use recommends 10mm spacing between adjacent conditioners, but if this spec existed in 1999, it was missed by both us and our systems integrator. Nowadays many suppliers' products are fixed to a backplane that imposes some spacing already.
As hosts and devices continue to get better, those of us using this technology will find it's the physical layer, things like a shield shorting out to ground somewhere, or water in a device, (or a failing power conditioner) that presents the greatest threat to system robustness.
Something more of us are experiencing now, is the risk presented by project execution in a running plant. When some project adds a valve or a measurement that get's commissioned while the plant is on line, another threat to the physical layer occurs . . .
I con't attribute any process interruptions in the past 5+ years to physical layer problems, but as any installation gets increasingly "long in the tooth", the risks presented by physical layer issues increases. As does complacency.
So, it's great to see a couple of the key suppliers of physical layer hardware coming out with products that have the potential to address this need.
HTH
John Rezabek
ISP Lima
Just wanted to pass on one of our recent experiences / findings.
Summary: DCS cabinet was built with no space between non-redundant DIN-rail mounted power conditoners. They've been getting hot for over 5 years now - constantly under power - and many (more than half) started putting a sinusoid of various amplitudes out on their associated segments.
We found this after experiencing a rash of communication erros on two unrelated segments in different parts of the plant. An oscilloscope revealed the imposed sinusoid (wavelngth about 100x FF communications).
While the problem caused mostly nuisance errors (controllers shedding mode) I think eventually, we'd have had a couple valves going to their fail positions. In my mind, an active physical layer diagnostic would have revealed this issue before > 50% of our power conditioners failed.
The good news is, that even with all the errors, the problems experienced didn't cause a major disruption of the process. Mostly just a nuisance for the operators, who had to go through a shift or two with persistent nuisance alarms, and having to run some loops in MAN or local auto.
Backround
When we built our system in 1999, there weren't any redundant FF power conditioners on the market. When we split apart the case of the (non-redundant) power conditioner recommended by our DCS supplier, we saw IC's and fuses, which was a concern. Our host manufacturer was able to come up with a redudnant power conditioner, or more correctly, one where the non-redundant components were all "simple" devices (no fuses & no IC's (integrated circuits)) the idea being that such a configuration had a MTBF comparable to 250 ohm resistor, the single point of failure in conventional systems.
We only did this for a small percentage of the segments (c. 15% or so) because the redundant solution (at the time) was large form factor and expensive. So we still had between 50 and 60 segments with the non-redundant power conditioners.
The current documentation for the power conditioners we use recommends 10mm spacing between adjacent conditioners, but if this spec existed in 1999, it was missed by both us and our systems integrator. Nowadays many suppliers' products are fixed to a backplane that imposes some spacing already.
As hosts and devices continue to get better, those of us using this technology will find it's the physical layer, things like a shield shorting out to ground somewhere, or water in a device, (or a failing power conditioner) that presents the greatest threat to system robustness.
Something more of us are experiencing now, is the risk presented by project execution in a running plant. When some project adds a valve or a measurement that get's commissioned while the plant is on line, another threat to the physical layer occurs . . .
I con't attribute any process interruptions in the past 5+ years to physical layer problems, but as any installation gets increasingly "long in the tooth", the risks presented by physical layer issues increases. As does complacency.
So, it's great to see a couple of the key suppliers of physical layer hardware coming out with products that have the potential to address this need.
HTH
John Rezabek
ISP Lima