Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: [802.3af] IEE802.3af and IEC60950 - safety issues

I would suggest that you find a consultant familiar with EN60950/IEC60950. He will be able to assist you far better than this reflector.
Again, let me stress that the actual clearances required will be a function of your particular application, device characteristics, and operating environment. These may be determined from an exhaustive reading of EN60950. It isn't fun or easy, but it's the only way. Let me give you some examples:
a) your PD is an IP phone - its environment is a standard office (no salt water, conductive dust, etc.) and there is no source of mains voltage in your PD. Your device is SELV and will have the most relaxed requirements for spacings/clearance/creepage.
b) your PD is a motor controller - the control logic runs from 802.3af power, but the motors operate on 440 VAC. The controller is in a steel mill. In this case, your environment is completely different from the benign office cubicle. EN60950 would not even be the appropriate safety standard; in the US UL508 for industrial controls would be the correct one, and in Europe the industrial section of the CE Low Voltage Directive. Spacings, creepage and clearances would be much more severe due to the environment and the high mains voltages.
The term "adequate separation" is undefined precisely because the requirements may vary per the above examples.
Finally, Clause defines several methods of measuring isolation that use testing methodology defined in EN60950:1991. It  does not indicate conformance with EN60950, but simply uses the test methods of EN60950. Please note that three different methods are available for doing the isolation testing and the implementer may chose whichever one they want.
 Prior to P802.3af, isolation in 802.3 was intended to prevent earth loops and to eliminate common-mode voltage problems with the data communications interface. The 802.3 standard is not a safety standard. The amendment of P802.3af is also not a safety standard, but provides pointers to EN60950:2001 to formally recognize the concept of the Limited Power Source and to align isolation testing with the 2001 edition. Again, these pointers are used to define testing techniques for determining isolation for the data communications interface. P802.3af will open a maintenance request to determine of the references to EN60950:1991 may be updated in all cases to point to EN60950:2001.
Steven B. Carlson
Co-Chair, ESTA Control Protocols Working Group
Chair, ESTA ACN Task Group
Chair, IEEE 802.3af DTE Power via MDI Task Force
Secretary, IEEE 802.3 CSMA/CD Working Group
High Speed Design, Inc.
11929 NW Old Quarry Road
Portland, OR 97229
FAX 503.626.4206
-----Original Message-----
From: owner-stds-802-3-pwrviamdi@xxxxxxxxxxxxxxxxxx [mailto:owner-stds-802-3-pwrviamdi@xxxxxxxxxxxxxxxxxx]On Behalf Of Charles Palmer
Sent: Friday, March 28, 2003 11:56 AM
To: stds-802-3-pwrviamdi@xxxxxxxx
Subject: [802.3af] IEE802.3af and IEC60950 - safety issues

Dear Bob, Steve and others   
I've now got the 802.3af D4.2 spec, as well as EN60950-1:2001 and I've done my best to understand them. (Note I have EN60950, not IEC60950), as far as they apply to the design of a PD. Can you check that I've got it right?
1    802.3af clause 33.4.1 says separation in in accordance with IEC60950 clause 6.2, which in turn calls for "adequate separation", without defining this. Compliance is tested by performing tests specified in 6.2.2 and EXPLICITLY STATES that the "... requirements of 2.10 regarding the dimensions and construction of CLEARANCE, CREEPAGE DISTANCE and solid insulation do not apply...". Is this really saying that all I have to do is pass the electric strength tests, and not comply with any creepage and clearance specs?
2    If that is the case, there still must be some useful guidelines I should apply to creepage, clearance, insulation etc. Any pointers?
3    I note that 802.3af clause 33.4.1 calls for either 1500Vrms steady-state, or 1500V pulses. The EN60950 spec calls for higher voltages for the impulse tests that the steady-state tests, and this seems to be the policy of IEEE802.3 (in, for example). Is it really the intention of the 802.3af spec that the steady-state and pulsed voltages should be the same? (Just checking!)
Many thanks.
Regards -
Charles Palmer
Technical Director, DSP Design Ltd
email: charles@xxxxxxxxxxxxx
7 Tapton Park Innovation Centre, Brimington Rd, Chesterfield S41 0TZ, UK
ph: +44 (0) 1246 545 918