Categories
Uncategorized

The Immaculate Conception of PXI  

By Carsten Thomsen

“If you started with a blank sheet of paper, what would the replacement of VXI look like?” 

Ron Wolfe, Owen Golden, Kevin Schultz, and Keith Odom and I were having a beer at the 1997 NI winter sales conference when I popped them that question.  

It was cold outside, and we were stranded at the 11th street Marriott (now a Sheraton) in Austin by an ice storm that paralyzed much of the South.  The afternoon “tradeshow” was just finished, where engineering groups displayed makeshift demos with homemade signs heralding their latest “birds nests” of prototypes.

The answers to the simple question gushed out:   

“we’d build on the CompactPCI standard”. 

“Yes it’s small, attractive compared to the prevailing C-size VXI”

“And it’s rugged” 

“Yeah, and it has lots of pins to spare on the backplane…future expandability”

“So we can even implement the Star Trigger from the D-size VXI boards!”

“And we’ll put the VXI plug-and-play framework on top”, and then Kevin Schultz committed: “I’ll make sure the we have ten modules ready for NI Week in August”, and Ron Wolfe followed up: “and I’ll get it on the cover of 25 trade magazines by NI Week”. I then summarized with “Let’s call it PXI”.  Which everyone did not agree with, but it still stuck.

“Yes, and let’s make it an open standard”. 

A few minutes later Dr. T wandered by, and Ron Wolfe gave the classic 30-second elevator pitch.  To which Dr. T said, “Sounds good to me”.  And Monday morning 30 engineers were working on the project.

But where was the business plan? The customer focus groups? The return of investment calculations? How did it align with corporate strategy?  

NI was a company that lived and breathed technology and customer engagement.  PCIbus products had successfully made a large impact in the market and VXI had demonstrated that there was a high-end market for multi-module virtual instrumentation.  The bulkiness of VXI put its entry level price too high for the broader market, while the 3U height of CompactPCI was less expensive, it visually sent a stronger message of being a “game changer” at a lower cost.  The many unused pins on the backplane ensured evolution to new communications standards (such as the multi-lane PCI Express) and still have room for multiple clock and trigger lines. 

So the answers to the above questions were obvious.  PXI was perfectly aligned to the NI culture and markets. It was a clear “go”. And the by-line soon became “it just makes sense”.  In addition to agreement on the core architecture, there was buy-in from sales, engineering, marketing and management, with a clear 8-month time schedule.  And Ron Wolfe believed we had just created a billion dollar market. A classic case of creative group think of employees clearly aligned to corporate goals. All the boxes were ticked, the technology was well-known, and it was simple to implement.  

And therein lay the biggest problem.  Back at corporate headquarters, where resources from multiple engineering groups were re-deployed to work on PXI there was a huge motivational problem. Because it was such a boring project.  Simply porting existing PCI boards to PXI was just a re-layout of existing boards, and some mechanical design.  There were no world-class, breakthrough innovations to challenge the engineering mind. And when I talked with Ed Loewstein about it, he complained that the 3U size, which marketing favored, was just too small to squeeze advanced RF products into.  Ed’s opposition settled the question of module size in my mind.  It had to be 3U.  Doing 6U would visually not make a statement in the customers mind…just another rehash of VXI.  But more important, images of Seymour Cray’s circular computer 

non-Copyright image from Wikipedia

reminded me of the laws of physics: if you want it fast, make it small. It gives shorter propagation times, hence better timing.  And then Moore’s law popped into my mind.  Semiconductor advancements would shrink the size and power requirements making it more and more feasible with time. And it would also challenge engineering.

Finally, I thought about Hewlett-Packard, National Instrument’s much admired competitor.  Their engineers would likely react the same way…much too small form factor to house anything meaningful. That would give NI plenty of time to build market share without any competition from them. 

As NI Week in August 1997 approached, it was time to set pricing.  The Power Point proposal put up on the screen used the standard 75% gross margin, and based on the cost-of-goods calculation, the PXI mainframe alone would cost over $3000.   Many in the conference room next to Dr. T’s office in Bridgepoint gasped at the price, fearing it could kill the project. But common sense prevailed and when the average system cost was calculated, the gross margin goals could be maintained.

Three weeks before NI Week97 a crew from NI flew up to visit HP in Loveland, Colorado.  It was pre-planned as a friendly get together, where we would start  with a short presentation and HP would then give us a tour of the facilities. First the PXI architecture was introduced and HP was invited, as the first, to join the PXIbus Consortium.  Several PXImodules were passed around the tables, including a tightly packed Slot 1 Controller.  The HP engineers said virtually nothing, except Paul Worrell who commented “it’s much too small to put anything useful in”. The atmosphere in the room turned chilly.

The meeting quickly ended and our planned tour of the factory was replaced by a trip straight out the front door to our rental car. It was first 12 years later that HP (then Agilent) joined the PXIbus Consortium, and Larry Desjardin was brought back from semi-retirement to rapidly drive Agilent’s first PXI products to market, while inventing the LXI bus along the way.   

As time went on,  National Instruments succeeded in squeezing more and more high frequency performance into the small 3U modules, and ultimately opened a facility in Santa Rosa, California, to which many Agilent high-frequency specialists were recruited. This march toward higher performance high-frequency modules, assisted by the precision backplane synchronization.

Thus, the comment by Ron Wolfe, when PXI first was conceived, “This is going to be a billion dollar business” became true.

Post-Script:

As this is written 25+ years after the fact, I would appreciate any corrections to make the article as historically accurate as possible. You can put them in the comments field or send them to me at ct@pxi.dk.



One reply on “The Immaculate Conception of PXI  ”

Extremely interesting. Thanks for playing such a key role in the creation of PXI, and for taking the time to record these valuable pieces of history! As someone who works with NI products for a living, I have tremendous admiration for the PXI platform. As a relatively young engineer, in my early 30’s, I can’t remember an engineering world without PXI. In fact, only after several years of using PXI I was exposed to VXI and realised that PXI had evolved from the older VXI – that’s how prevalent PXI had become compared to VXI. “And Monday morning 30 engineers were working on the project.” – it’s uplifting to see how quickly technical progress can be made when the culture is aligned and strong. It also raises the bar because most organisations move at a much slower pace. Thanks, Petru (Bristol, UK)

Leave a Reply

Your email address will not be published. Required fields are marked *

The maximum upload file size: 256 MB. You can upload: image, audio, video, document, spreadsheet, interactive, text, other. Links to YouTube, Facebook, Twitter and other services inserted in the comment text will be automatically embedded. Drop files here