User Satisfaction Survey Follow-up Action Plan

As we said previously regarding our customer satisfaction survey, we wanted to share with you our plans for improving key areas that you highlighted in the survey. These key areas were software quality improvements, a more robust model library and advanced training.

Software Quality 

Test Hardware:

What hinders our ability to fully test our software products in all of our customers’ environments is the absence of a full complement of complex test hardware. Obviously, we cannot design and build the routers, base stations, computers, aviation controls, missile control boards, etc. that you, our customers, build. Nonetheless, as part of our QA process for every release, we use several very complex boards that customers have given us and we have found this to be very beneficial. We would like to test our software on other hardware. To encourage your assistance in this regard, we would present you with an iPOD or an XM satellite radio with subscription card (your choice) for helping us. We are looking for boards that you no longer need (prototypes, for example). We must be able to power up the board and run it. It’s okay if the board has faults (actually, that is a good thing). We’d like to dramatically increase our library of hardware test platforms. We believe this will have a major impact on improving the quality of our software to the benefit of both of us. What do say? Will you help?

Note: For this to be successful we need the following:

• Boards that work
• A powered chassis or test fixture with a power supply. Failing this, we need good information on how we will be able to power up the board with an off-the-shelf power supply.
• Your management’s approval, so no one gets in trouble. (We can sign a specific NDA, if needed.)
• The ScanWorks project for the board with all the input data so we can rebuild the project and re-use it to test each ScanWorks release.

Code Inspections:

We have done code inspections on key segments of code and have found that this is the best way to identify errors early in the development process. This year we are committing more of our developers’ time to this activity. We are at least doubling the time developers spend on code inspections and we expect this will improve software quality.

QA Resource:

Some of you know that Eric Sweeney has been an application engineer for many years at ASSET. Last month we moved Eric from the support team to Quality Assurance in development. We believe his application expertise will help us improve our QA testing. That is his job.

Change Communication:

We have identified one area where we should communicate better with you. In some cases, we have been lax in highlighting a data input violation and this has caused problems later on. For example, in the next release of ScanWorks we might tighten up ScanWorks to flag this issue. Unfortunately, people then complain that they had been using ScanWorks in a certain way and now the new release doesn’t allow the old methodology anymore. To address this issue we will have a new section in our Problems.txt file to educate everyone on the issue and our reasons for “tightening” things up. This will be implemented as part of ScanWorks 3.8 and carried forward in future releases.

Beta Testing:

It is not your job to test our product, but we cannot test ScanWorks or any of our products in every way an engineer might use them on new designs with new devices. I am sure you understand the challenge as you deal with the same issues in doing your job. Our plan, as always, is to continually add major new functionality to ScanWorks over time. As we do so, new functionalities will be subjected to beta testing much as we did with our recently soon to be released DFT Analyzer product. This beta test proved to be very successful and we expect to have the same kind of success with our beta testing program in the future.

Longer-Term Plan:

In our plans for the next generation of ScanWorks we have included an extensive beta field test so we can be certain it will be our highest quality product ever and that you will be confident upgrading to it. This field test will require months rather than weeks. In addition, we believe the new system will address some lingering legacy problem “work-arounds” in the most effective manner. Also, we are implementing a software DFT methodology for improving quality that’s similar to the hardware DFT methodology we champion with you. We will share more information on these items in 4Q06.

We will be implementing other improvements in our internal processes, but we feel the changes outlined here will have the most significant effects on software quality. We want all of our users, not just some of them, to be “very satisfied” that ScanWorks is a high-quality product. If you have other suggestions with regards to software quality, please let us know.

Models 

Your survey comments about device models focused mostly on these issues:

• "We want more models in the library."
• "We want validated models."

More, More, More:

We believe we have the largest model library in the industry and we add models every month. But we agree with you that the growth of the library must accelerate. ASSET will be focusing more engineering effort on model-creation in 2006. We also have plans to replace many of the cluster models that are “disable-only” models with multi-part models.

However, we all must share responsibility for the models in the library. Of the 7,681 models in the library, 99.35% have been submitted by ASSET personnel. The library encourages the sharing of models among ScanWorks users, but few are doing this.

Validation of Models:

Your comments were mostly about memories, and, yes, we admit this is a challenging issue. Customers deploy new memory designs very quickly; often faster than we can acquire samples and develop models. When customers request delivery of a model faster than we can acquire the sample, those customers may be the first to validate the model for that device. But ASSET must do a better job at following up to ensure that the model was validated and to include this information in the model's information page. This will give other users higher confidence in the model. And if we learn that the model is not accurate, we must correct it and re-validate it when we receive samples.

We have formed a team that will create an internal process which will make ASSET more proactive with new memory architectures and designs. We are asking device manufacturers for their cooperation so we can get this information earlier. This is another area where you can help. If you learn that your design team is going to use a new memory design, an early warning from you about the new memory will help us order samples sooner. Then, we have a better chance of having the model validated before you need it for a test project.

Training 

In the area of training, the two most common suggestions were:

• Classes on more advanced topics.
• A more advanced design/project in the introductory class that exposes users to more real-world issues.

Advanced Training:

In 2006 ASSET will deliver courses addressing advanced topics. We have already created a new one-day course on 1149.6. Another new course on scripting is being developed. Moreover, our main development course will be expanded to better cover topics such as CPLD programming, flash programming efficiency and debug techniques.

We could use more input from you to ensure we work on the topics that are most important to you. Please respond to this email to give us more feedback.

Real-World Training Board and Project:

Yes, the ScanLite training PCB design must be improved upon. We will introduce a new training board early next year. It will have more devices, multiple chains, 1149.6 and include device types you are more likely to see in today's designs. Since this new board will also serve as an additional test case in our software QA process, this will also drive the design specification for the board.

Again, let us take this opportunity to thank you for your participation in our User Satisfaction Survey. When you see this survey again in the future, we hope you will continue to share your thoughts with us. In the end, your participation benefits all ScanWorks users.

Sincerely,
Glenn Woppman, President and CEO
and
Alan Sguigna, Vice President of Sales and Marketing