Slow comunications through a Hirshmann Spider II 8tx unmanaged switch

[nmcf]

I am working on an installation of a new production line that has about 100 of these Hirshmann Spider II 8tx Ethernet unmanaged switches. We have had no problems with about 85 of these Hirshmann Spider II 8tx Ethernet unmanaged switches.
All of these switches are intended to operate at 100-Full Duplex
The LAN connection is a Cisco set for Auto-negotiate
All downstream devices are set to auto-negotiate.
Connected devices include Siemens S7-3xx PLCs, Siemens WinCC HMIs, Fanuc Robot controllers and Cognex cameras – all using Profinet communications.
The 15 Hirshmann Spider II Ethernet 8tx unmanaged switch that have a problem all have the Cognex cameras attached. The switches that do not have cameras do not have problems.
SYMPTOMS of the failure : After a power down reset of the Hirshmann Spider II Ethernet 8tx unmanaged switch (power down of the switch is the only way we have found to restore normal operation), the switch will operate normally for between a day and a week – then will the switch will start being a data bottleneck. This slow comm Results in failed connection between programming PC and cameras, Camera to PLC comm. failures, HMI to PLC comm. Failures and Auto backups that take a long time or fail to upload.
Troubleshooting so far.
The Hirshmann Spider II 8tx Ethernet unmanaged switch has no diagnostics and there are no DIP switches or settings of any kind.
A test was created to determine the failure. A project (5MB, 540 files, 70 directories) is download from a server through the switch. A download from the Server to a PC will normally take 1 second for the data transfer and 4 seconds for the unzip. When the Hirshmann Spider II 8tx Ethernet unmanaged switch is in failure mode, the project download can take between 10 seconds and 8 minutes – the longer the time since the Power down reset, the worse the symptoms. Amazingly, the data always comes in correct, if a timeout does not occur on the app.
The Hirshman rep supplied us with 2 Hirshmann RS-20 8tx unmanaged switches to try. We have installed these in place of the Hirshmann Spider II 8tx Ethernet unmanaged switches in several locations. The Hirshmann RS-20 8tx unmanaged switches work flawlessly in all locations.
Next the Hirshman rep supplied us with a Hirshmann RSB-20 8tx Managed switch to try. This managed switch worked flawlessly in 2 locations and reports no errors after several weeks. It also reports all ports using auto-negotiate and all port running at 100-full.
Identical production line are being installed at 2 other sites. These sites are reporting similar problems.
Does anyone have any suggestions ot troubleshooting tips?

 

[VinceWhirlwind]

As the switch is unmanaged, you will have to infer what it might be doing by checking your PC logs and stats.

Use perfmon, eventvwr to find events and interface stats/errors.

 

[nmcf]

The plant LAN drop to this Hirshmann Switch is from a Cisco Catalyst 3750 Series PoE-48.
The I.T. group cannot find any problems from the Cisco stats for this LAN Drop.
This machine has a Siemens Simatic S7-315 PLC, a Siemens Simatic Multi-Panel MP277 (running Windows embedded). 2 Cognex Insight 5100 Cameras. This particular machine does not use the Ethernet for the MP277 HMI, although the HMI is connected to the Ethernet – there is a ProfiBus connection for the HMI to PLC comms.
The cameras do not store any comm. Stats, but the PLC has a few stats:
Dropped received packets – no resources 0
Bad received packets 0
Received Octets 2,732,093,433
Dropped sent packets – no resources 0
Bad sent packets 0
Sent Octets 20,234,253
Interesting that we do not get errors when the Hirshmann is powered down to restore normal operation.
Do you think that temporality adding a PC with PerfMon will give us any info? I am not familiar with perfmon – am not sure of it’s capabilities. There are spare ports on the Hirshmann to connect.
Have also thought about connecting a PC with wireshark, but have been told that another site with a similar problem with the Hirshmann Spider II tried wire shark with no new clues.

 

[VinceWhirlwind]

As the "going slow" affects a normal data transfer between PC & Server, I would be looking at Perfmon on that PC to see if I can see anything odd, as well as eventvwr on that PC, plus wireshark to see what the data transfer is actually doing at a TCP level.