Earlier this week I was introduced to Paul Simon, Dutch co-founder of Qualtera, based in Montpellier, France. Having formerly worked at various Semiconductor giants like Philips, NXP & PDF in testing & yield management, Paul co-founded Qualtera in 2010 to make semiconductor testing easier & more efficient. Last week, the company launched Silicondash, a cloud-based system that allows semiconductor manufactures to take the large amount of parameter testing data and turn it into an analytic report in a matter of minutes, instead of days.
The manufacturing process is very complicated and expensive. The yield [for each wafer] is not 100%, the quality is not super.”
The standard process for testing a silicon wafer tends to take over a week. Designs are sent off, and wafers come back to be tested. Each wafer contains ~2,000 products with ~3,000 parameters each, meaning 6 million variables to check, generating terabytes of data per day. Most of the testing process is spent manipulated Excel sheets, trying to analyze the data and determine what the next wafer should include/exclude. Companies like NVidia, Apple & Qualcomm have teams of employees who do this every week. The process takes very long, and the results are not very consistent, due to each employee having different levels of competences in terms of testing.
What we saw in our previous jobs is that doing this yield process… is very complicated and takes a long time. It’s not very efficient”
With each wafer costing $1,000, and with big companies producing 3,000-4,000 wafers per day, the cost & risk are very high, and the value-added of shrinking the amount of time in each feedback loop, and improving the accuracy in the analysis, is invaluable. This is where Silicondash comes in. Paying per use, this automated Decision Support System (DSS) spits out charts and textual analysis quickly and accurately.
According to Electronics Weekly, the product is currently being beta-tested by major European IDMs. Simon boasts that their main competitors are over 10 years old, and lack the efficiency and user experience that modern technology product require: “People should find whatever information they need in 4 clicks,” says Simon.