Podcast: Play in new window | Download
Subscribe: Apple Podcasts | RSS
- Welcome Dr. Howard Johnson!
- He has published two reference books that are a must-have for the field of high speed signal propagation:
- Howard hails from Twisp, WA.
- Martin Graham, the co-author of his book, was also his longtime mentor at ROLM.
- Howard teaches a class at Oxford every summer. He also teaches classes throughout the US, both in public and private forums.
- He also has published articles regularly at EDN and other technical magazines.
- Other things mentioned during the show:
- Chris mentioned his article about Bell Labs and the Trickle Down Techonomy.
- Howard mentioned how the Voyager Space Craft actually experienced some cosmic ray data corruption, though it doesn’t happen often on earth.
- As connectors get scale smaller, signals get better. However, on boards as traces get longer, they also need to get wider.
- To continue increasing the speed of modern day comms, Howard believes we need to move to Multi-Level Communication (as we always do with every medium).
- The limits of channel capacity are governed by the Shannon-Hartley Law (referred to as Shannon’s Theory on the show).
- In “The Early History of Data Networks” by Gerard J. Holzmann and Bjorn Pehrson (there’s a “synopsis” here…), they talk about torches being used in single and multichannel modes.
- Howard helped define the Gigabit Ethernet Standard (with no help from “Ernie”!)
- Howard suggests “The Theory and Practice of Modem” Design, by John Bingham as a good starter text on the subject of encoding and data transfer.
- If you need a place to talk about signal issues, check out the SI-list, part of freelists.org.
- The IEEE EMC society also is a great place to meet other designers.
- At EMC meetings, they often watch related videos, such as the ones on Howard’s website!
- Howard responded to silliness relating to claims of “skin effect in audio cables”.
- The 90 degree question: Is it wrong to make right angles on your board layout?
- This rule was propagated by microwave designers who were designing with 120 mil line widths.
- Your board already has tons of 90 degree turns…in the vias on your board.
- It’s the added material in a right angle turn (beyond the normal width of a trace) that can add parasitic capacitance.
We had a wonderful time talking with Dr. Howard Johnson. It was great getting to know the kinds of work he does and the kinds of signal integrity problems he works with regularly. Please leave any questions you might have about the show or for Dr. Johnson in the comment section.
Adam Ward says
Thanks for including my question – fascinating to hear Dr Johnson’s insights into the industry and technologies, damn that guy’s clever.
What happened to “off the cuff” Chris? You with the pre-written introduction — getting all professional on us now. Hehehe. 🙂
Great episode fellas.
Chris Gammell says
That was actually Howard’s idea. I like it, think I’ll be doing it for future guests as well.
Carmen Parisi says
I thought that was a nice touch and wouldn’t mind hearing it going forwards. Sometimes I’m not 100% sure who you guys are talking to and it takes a few minutes into the episode before I figure it out.
Very interesting episode, thank you Howard, Chris and (unusually quiet) Dave 😉
I loved the story about evicting “Ernie”s 😀
Lots of very interesting references too.
Thank you again Howard for answering all our questions.
Brian J Hoskins says
*FANTASTIC* show this week guys, congratulations! I really enjoyed it. Jammed full of good technical discussion with a bunch of techy-humour thrown in. Can’t fault it!
Loved towards the end of the show, very interesting.
Wow guys what a great show this week. I really thought that it might be too technical for me, but that show was both informative and down right fun.
Phenomenal podcast, so much knowledge and humour. Loved it.
Alan Wolke W2AEW says
Great show – so much I can relate to! I participated in the FDDI PMD standard in the mid/late 80’s – and it was just as he described!
Excellent show guys. A bit more formal than usual, but that works well to break up the regular style which can get into a bit of a rut. Variety is certainly a good thing!
Very good episode, all 85 minutes! Thanks.
Carmen Parisi says
Great show this week guys. Definitely in the Hall of Fame in my opinion. I’m going to have to see if he’s going to do any lectures around Raleigh this year…
Not a high speed digital person but understood most of the discussions, Excellent show and a great guest who’s knowledge is manifest by his easy unflustered style. The Amphour is really starting to take off now and attracting the cream of the electronics world, well done Dave and Chris and a big thanks to Howard for his time.
Jelle Haandrikman says
Fanatastic show this week.
Keep ‘m coming.
I’m picturing a Schmidt trigger for the 16 level transceiver…
DM Stern says
Going back a few weeks, Dave was very critical about a Seleae logic analyzer knockoff. Does he realize that Itead (who he uses for PCB prototypes) make and sell them? (http://iteadstudio.com/store/index.php?main_page=product_info&cPath=29_31&products_id=221)
Golden. Best 1:26:32 in a long time 🙂
Rob Whitfield says
I think the main reason I hate people using right angle tracks is that they look far neater not being right angles.
Also no reason to have that extra bit of capacitance if you don’t have to!
Great show – shame I didn’t listen to it at lunch as I normally time my lunch break with it!
Thank you Dave and Chris and, of course, thank you Dr. Johnson for this great lecture.
He talked about capacitance and not inductance of right angle traces. I wonder if the inductance is significant. It is identical for different trace widths. It is probably small or he would have mentioned it, I guess.
A signal trace is a transmission line. It’s characteristic impedance is sqrt(L/C) where L and C are inductance and capacitance per metre. So both are equally important. L, caused by energy stored in the magnetic field (and hence current) may not change at a corner but C will due to the effect Howard described. So there is the impedance mismatch that causes reflections.
Usual rule of thumb is keep such geometric irregularities much smaller than a wavelength at the switching speads you need. So a millimetre irregularity corresponds to a few pico-seconds, becoming significant at a few hundred GHz. Here in laid back Oz, very few of my designs go that fast…
Great show! Great questions. It was real fun to listen.