Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?

To: John Walliker who wrote (65378)	2/5/2001 4:14:57 PM
From: Ali Chen	Read Replies (2) \| Respond to of 93625

No need to demonstrate your viscous pedantry. You better appply it to many less obvious areas of RIMM design. First, we both perfectly know we were talking about wide-bandwidth square-wave-like signals. Then why did you bring out the academic example about impedance matching at a single frequency in the first place? To mislead less sophisticated and semi-literate individuals here, and to make a fool of your opponent? Therefore, your second answer must be correct: "This is not appropriate to the much wider bandwidths used in Rambus." Now you are trying to squirm with "limited" bandwidth. Wrong again. Every signal in the real world has a limited bandwidth. So what? More: When you make a section of a trace "narrower", what do you think happens to the trace impedance? You are trying to sweep the problem under the rug. No matter what you do, the impedance mismatch problem is not going away, it just slightly changes it's shape. Of course, it is possible to slightly optimize the most ugly reflections, but the trick of trimming traces just disperses the waves to some extent. But in the whole transmission domain the phases of the reflected waves still may combine in some less predictable way, with the same final effect: the loss of bit of information. In addition, the use of instrumentation with inadequate bandwidth only aggravates the problem. You just cannot identify the problem if you do not see it in a heavy-filtered oscilloscope. Here comes the difference you so deliberately trying to hide: yes, the DDR load is a tree, but the RIMM load is space-time distributed, so the wave propagation effects play critical role. For less sophisticated reades, please take a look at the following picture ednmag.com in order to understand the background of the problem. Also, as I explained in many other posts, there is no such thing as "constant current driver", especially if the Vcc rails are only 2.5V. So your implication that the driver has an infinite impedance and therefore does not affect the matching, is one of many official Rambus lies. You mislead people by creating a false impression that RIMM design is perfect. It is not. With no respect, - Ali