Hi Robert, The HDMI spec (section 4.2.4) details the electrical characteristics of the TMDS lines, including rise and fall times. I think your interpretation of the spec isn't strictly correct in that "rise time / fall time" actually means "rise time OR fall time". If it were the case that rise and fall times were meant to be different then they would be on different lines of the table. Also, the risetime must be greater than 75ps, not less than.
However the spec is different between HDMI 1.3 and HDMI 1.4 HDMI 1.3 75ps < Tr/Tf < 0.4 Tbit HDMI 1.4 75ps < Tr/Tf Presumably they've removed the 0.4 Tbit because if your rise time starts to infringe on this parameter then you'll be in danger of closing the eye and infringing the central mask of the eye diagram (instant fail, no conditional pass here). Also, it might not have been compatible with the higher clock frequencies and data rates called up in HDMI 1.4 (speculation). You are right in that it doesn't specify matching of rise and fall between the pairs. However all the measurements I've made on HDMI sources show good matching between the rise and fall times for each pair. The worst deviation I've measured was about ±5ps on our products. RM >> "Wouldn't steering current through either conductor have more potential for EMI generation from physical reality limitations, than if the same cabling were to be driven using LVDS with matched impedances?" You aren't strictly "steering the current" as each line in a TMDS pair has it's own termination pullup to the 3V3 termination supply. From what I've seen >from my measurements on some sources, HDMI drivers can take a small amount of current (1 to 2mA) in the "high" state, presumably to control the rise time and any ringing. In my experience with HDMI and EMC, you will have significantly more issues with cheap cables than from any other issue. The area from transition of cable to connector is often poorly shielded with long "pigtail" ground connections. Measurements with near field probes clearly show this to be a major problem area for emissions. Also, because the data frequency is 5 x the clock frequency you get the harmonics ganging up on you! For a 74MHz TMDS clock you'll likely see 370MHz and 740MHz harmonics. FYI the convention for HDMI is to refer to the "source" (transmitter) and the "sink" (the load e.g. a TV) I hope this helps, Best regards, James James Pawson Leading Hardware Engineer EchoStar Europe T: +44 (0)1535 659000 e: [email protected] From: Robert Macy [mailto:[email protected]] Sent: 18 November 2009 23:30 To: [email protected] Subject: [PSES] HDMI Transmitter: EMI Tutorial EMC Group Members: The following questions came to mind after looking at the HDMI electrical interface specifications. Especially, questions regarding the potential of EMI generation for the technique that is specified in HDMI 1.2, assuming 1.3 is similar. The spec looked like it could cause problems, because instead of matching rise and fall times, they have tr and tf specified VERY unbalanced, at least potentially, ie, risetime < 75ps and fall time < 0.4Tbit [related to bit time] Doesn't that yield a huge spike of common mode at each transition? It was my understanding from reading the spec that the transmitter consists of a pair of current switches to shield ground, where either is on, much like driving the cable with ECL Logic, except in the HDMI transmitter the impedance was kept as high as possible, approaching infinite. I realize using current switches may give you 6 dB larger signal over LVDS for the same power supply, but...here's the big question: Wouldn't steering current through either conductor have more potential for EMI generation from physical reality limitations, than if the same cabling were to be driven using LVDS with matched impedances? And one can only drive over shorter distances using current switches? Plus, there has to be ringing, after all, you are only load terminating the line and never source terminating. Shouldn't long lines should be be terminated at both ends? Also, consider a standard 100 ohm differential shielded cable: It is my understanding that most 100 ohm 'differential' cables barely make it to 75 ohm and 75 ohm to shield and 300 ohms between the two conductors. I believe some of the Belden models for balanced twisted pair show 61 ohm, 61 ohm and only 560 ohm between. Is it possible to even get better 'forgiveness' to the balance and have 100-100 and 200 between them? Now consider an unbalanced cable caused from manufacturing tolerances, any 'unbalance' in the cable will exacerbate EMI generation, because the shield current is subtracting large numbers to get zero. In other words, driving a slightly off balance pair of conductors, CAUSES EMI when driven with LVDS. But, that effect must be much less when driving with switched currents, only the voltage levels become unbalanced? Any papers covering this in detail out there? What happens when you go through a connector where the structure usually becomes 50 ohm to shield, 50 ohm to shield, and zero between during this transition? Has anyone compared the two driving techniques, current switching to LVDS? Has anyone compared the maximum distance obtainable for the two techniques? Regards, Robert - This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. To post a message to the list, send your e-mail to <[email protected]> All emc-pstc postings are archived and searchable on the web at: http://www.ieeecommunities.org/emc-pstc Graphics (in well-used formats), large files, etc. can be posted to that URL. Website: http://www.ieee-pses.org/ Instructions: http://listserv.ieee.org/request/user-guide.html List rules: http://www.ieee-pses.org/listrules.html For help, send mail to the list administrators: Scott Douglas <[email protected]> Mike Cantwell <[email protected]> For policy questions, send mail to: Jim Bacher: <[email protected]> David Heald: <[email protected]> - This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. To post a message to the list, send your e-mail to <[email protected]> All emc-pstc postings are archived and searchable on the web at: http://www.ieeecommunities.org/emc-pstc Graphics (in well-used formats), large files, etc. can be posted to that URL. Website: http://www.ieee-pses.org/ Instructions: http://listserv.ieee.org/request/user-guide.html List rules: http://www.ieee-pses.org/listrules.html For help, send mail to the list administrators: Scott Douglas <[email protected]> Mike Cantwell <[email protected]> For policy questions, send mail to: Jim Bacher: <[email protected]> David Heald: <[email protected]>
