Intel AP-419 and Frequency Tolerance w.r.t. IEEE 802.3?

[AJone171]

Hi,

According to AP-419, Revision 1.2, September 2004, Section 3.3 interprets the IEEE 802.3 Frequency Tolerance as +/-50ppm, "referenced to a standard temperature of 25° C".

Should I interpret this as +/-50ppm @25° C, with a wider tolerance for temperature, built into the IEEE 802.3?

While I've been able to correlate that the IEEE 802.3 Frequency Tolerance is given as +/-50ppm, for 100Base-TX, I've not been able to correlate whether it is referenced to a standard temperature of 25° C, or whether that represents the total ppm budget including Nominal Frequency Tolerance, Over Temperature and Ageing factors. The latter seems more appropriate for such a timing specification, because the accuracy will be required for whatever temperature range and lifetime the network components are required to operate in.

Clarification of what revision of 802.3 the AP-419 might align to might help me understand this. If anyone can assist, I'd appreciate it.

Thanks,

Andy Jones

[idata]

Hello AJ_1962,

Thank you for posting in Intel Wired Ethernet Communities. We will investigate the requested information in this post, and provide an update as soon as possible. Please do not hesitate to contact us with any questions.Best regards,Daniel DIntel Customer SupportAgent under contract to Intel

[idata]

Hello AJ_1962,

The frequency tolerance is the allowable deviation from the specified frequency at a specific temperature, usually 25° C as specified in the document. Frequency stability is the deviation from nominal frequency at 25° C over the rated temperature range. For example:Percentage vs PPM.01 % = 100 PPM.005% = 50 PPM.001 % = 10 PPMPlease let us know if you have any other questions, or this clarifies everything.Best regards,Daniel DIntel Customer SupportAgent under contract to Intel

[AJone171]

Hi Daniel,

Thanks for replying, but I think you have missed my concern with how AP-419 is worded.

I understand that manufacturers will define a frequency tolerance, usually specified at 25° C, with other factors such as drift over temperature and ageing factors to be added in where necessary.

I would expect an engineer to evaluate the sum of these factors to be assessed against the published figure in IEEE 802.3.

I would expect IEEE 802.3 to define a frequency tolerance that must be satisfied by all parts of a network, across it's entire network, requiring consideration of the right component selection to suit the lifespan of each node of the network and the temperature range that each node is expected to operate in. I understand that if Packet Length is constrained on a given network, then there may be scope for relaxation of network frequency tolerance, but that is another matter.

I would not expect IEEE 802.3 to specify a frequency tolerance requirement that was meant to represent a certain temperature, unless it went on to specify a temperature or ageing related factor, but it does not, in the revision I've been able to view (-2008).

However, AP-419 is quoting a frequency tolerance from IEE 802.3 and - as I read it - claiming that that quoted figure is the frequency tolerance for 25° C, and this would then imply some other budget in IEE 802.3, to allow for the temperature range, but as far as I know, there is not.

It is far from clear to me that IEEE 802.3 tolerances should be interpreted in this way, but it seems that is what is being stated in AP-419. Am I misinterpreting AP-419, or misreading IEEE 802.3, or looking at the wrong revision?

I cannot find anything other than AP-419 that suggests the IEEE 802.3 figure is a frequency tolerance that relates to a specific temperature that can then be exceeded when environmental temperature is considered.

On what basis is AP-419 suggesting that that the IEEE 802.3 tolerance for 100BASE-TX can be satisfied at 25° C, and then exceeded for a wider temperature range and as the product ages?

Other manufacturers application notes that I have looked at apply the same interpretation as myself, so for each node on the network, the sum of manufacturer oscillator frequency tolerances ( @25° C, + node temperature range, + ageing) has to be checked against the IEEE 802.3 frequency tolerance.

Thanks,

Andy Jones

[idata]

Hello AJ_1962,

Please allow us to look into this inquiry further.Best regards,Daniel DIntel Customer SupportAgent under contract to Intel

[idata]

Hello AJ_1962,

Thank you for your patience. Oscillators and crystals are referenced at a general lifetime value of X ppm. IEEE indicates 50ppm, Intel at 30ppm . 30ppm accounts for any temp or age factors drifting outside 30ppm while staying in the 50ppm limits. Newer parts want tighter ppm specs. Almost all of the timing devices sold today will still fall inside 50ppm spec range. Datasheets from the manufacturer of the part will often include this information.Best regards,Daniel DIntel Customer SupportAgent under contract to Intel

[AJone171]

Hello Daniel,

Thanks. It makes sense that it should be interpreted as ±50ppm lifetime, and I agree a ±30ppm part will do the trick so long as temperature and age factors keep it to within ±50ppm. Most references in AP-419 are consistent with that, but I'm still left with the impression that AP-419 could do with a tweak, in the areas of:

Section 3.0 Table 1 Crystal Parameters, where a Temperature Stability of "±100 ppm at 0° C to 70° C" seems unsuitable, unless the user was actually using a much narrower temperature range.

Within that same table, for any product with a decade or more life expectancy, I think I would expect to see a much smaller ageing value too, although I note that as we have discussed, a crystal may have a higher first year figure and lower ageing for subsequent years.

Section 3.3 Frequency Tolerance "The frequency tolerance for an Ethernet physical layer device is dictated by the IEEE 802.3 specification as ±50 parts per million (ppm). This measurement is referenced to a standard temperature of 25° C."

Section 3.4 Temperature Stability states " Manufacturers may also list temperature stability as 50 ppm in their data sheets". This doesn't seem to be a particularly helpful example figure to quote, as it would be all of the IEEE 802.3 budget, and the basic part and ageing elements would have to add up to zero.

Best Regards

Andy

[idata]

Hello AJ_1962,

Thank you for the information. Hope we were able to answer all of your questions. Please let us know if you come across anything else that needs clarification.Best regards,Daniel DIntel Customer SupportAgent under contract to Intel