Last week was definitely not a good one for semiconductor startups. Every other day an announcement appeared proclaiming the end for one startup or another. There was Evident Technologies, a company that specialized in Quantum Dot Products and Light Emitting Diodes (LEDs), which filed for Chapter 11.Then there was CSwitch, a startup specializing in configurable switch array chips for next generation networks, which seized operations according to sources. And finally, there was MetaRAM, a company that worked on quadrupling the DRAM capacity of existing systems using existing DIMMs.

We profiled MetaRAM in March of last year, shortly after the company emerged from stealth mode. It was backed by several prominent venture capital firms including: Kleiner Perkins Caulfield & Byers, Khosla Ventures, Storm Ventures, and Intel Capital. This just shows you that having prominent VC backing is not a guaranteed indicator of success. Already back then we had a couple of concerns regarding the MetaRAM technology: First, with increasing DRAM frequency, how long would MetaRAM be able to hide the latency of their chipset via clever buffering of reads and writes? Second, it was inevitable that memory controllers would enable support for ever larger amounts of memory, possibly making MetaRAM technology irrelevant? Whether any of these was the actually reason for the company ceasing operations we might never know. The company’s website seems to be down, and as far as I’m aware nobody has been able to reach any of the company representatives for an official comment.

The issues that lead to troubles for the other two semiconductor startups are somewhat clearer. CSwitch was simply unable to raise another $10 million in funding in order to bring its chips into mass production. The company is currently negotiating to sell its assets, and with some luck we might see the technology hit the market yet in the same way Ambric’s technology did earlier this year. Evident’s issue was a bit different, in that the company announced this it was filing for bankruptcy protection due to exorbitant costs associated with a patent infringement case that was brought against the company by a large California-based life science company. The good news here is that while in Chapter 11, Evident intends to continue their research and development, as well as commercialization of their technology.

Over the last few months we have seen several innovative semiconductor startups raise some serious money, nevertheless it seem that as a whole, Venture Capitalists (VCs) seem to be a pessimistic bunch at the moment - at least the three guys Rick Merritt interviewed for his "Silicon Stratups get the Squeeze" article. Among the VCs interviewed were Andy Rappaport from August Capital, Mark Stevens from Sequoia Capital, and Lip-Bu Tan from Walden International. The main reason for the pessimism is well know to anyone in the chip industry, namely the prohibitively high cost of bringing a new fabless semiconductor startup to market.

I won't recap Rick's whole article for you can read it on your own, but there were a few interesting takeaways. According to Andy, if companies doing big SoC designs adopt open interfaces, there will be a chance for startups to at least play a part in supplying designs for specific functions to be integrated onto the SoCs. Further, he points out that system designers will have to do more in software in the future as chip choices will become limited. Mark very much agrees with this point of view, adding that with IPO exits largely non-existent these days and with established companies paying less and less for acquisitions, the return on investment (ROI) for funding semiconductor startups is simply no longer there. Lip-Bu is even more pessimistic, arguing that a substantial percentage of the approximately 2,200 companies funded over the last decade will have to fold and that the R&D mantle will pass once more to the big chip makers. He further points out that plenty of talented engineers, who will not be interested in joining these conglomerates, will choose instead to pursue careers in the clean technology and solar areas. Of the three guys interviewed, Lip-Bu seems the one most committed to semiconductors, but you better approach him only if the majority of your company is going to be located in China, India, or Taiwan. He firmly believes that the semiconductor center is moving west rapidly. If green technologies are not your thing and you want to stick with semiconductors, you better brush up on your mixed-signal and analog skills since both Andy and Lip-Bu prefer these areas at the moment.

I'm somewhat surprised that not one of the interviewed VCs discussed opportunities for low-power semiconductor startups in the life-sciences area. It seems to me that this field in particular could benefit from novel analog and mixed-signal designs and should be well suited for startups. I'm also somewhat skeptical about the large sum of money, on the order of $100 to $200 million, that the VCs claim are needed to get a fabless semiconductor startups funded. Additionally, a budget of $2 million dollars for a verification team per month as claimed by Andy seems unreasonable for a startup - even large chip design houses would find this to be expensive. Most companies we have written about on this site fall somewhere in the $20 - $60 million range when it comes to total funding, which while not cheap compared to a software startup is a far cry from the numbers that the VCs are suggesting. It almost seems that the exaggerated price tag that VCs are attaching to semiconductor startups is their indirect way of saying that they have found cheaper investments that have a quicker ROI elsewhere.

We first covered Powervation and the company’s Auto-Control DC/DC technology at the beginning of this year, while mentioning several other startups that were developing digital solutions for typically analog problems. A few months have passed and Powervation recently notified us that the company’s inaugural product, namely the PV3002 has become available. The PV3002 is an Auto-Control Dual Phase Digital Power IC and is targeted at the Computing, NetComms, and Storage markets, although many additional applications will surely take advantage of it. The PV3002 consists of a Digital Signal Processor (DSP), a RISC processor, and several analog blocks to make it a complete mixed-signal System-On-a-Chip (SOC). It can provide 1 or 2 phase operation and several of these ICs can be utilized in parallel to enable load sharing. As mentioned in our previous post, the key feature of this IC is its ability to monitor the output voltage in order to compensate for variations in line, load, capacitance, and inductance.

Performance and stability are provided through the use of a feedback loop for which the compensation level is controlled dynamically through a single parameter referred to as “MOJO.” The exact implementation of MOJO is obviously a trade secret, but one can obtain somewhat of an idea about what it entails from several publications. The company’s own website has a nice little primer titled DC-DC conversion with Auto-Control, which examines the typical digital DC-DC converter design, the inherent limitations brought on by the digital feedback loop, and how adaptive control can relax some of the performance constrains while still keeping the system stable. If you want to delve a little deeper and have IEEE access, you can find the following two recently published papers: Current Share in Multiphase SMPCs by Digital Filtering and Current Share in Multiphase DC-DC Converters Using Digital Filtering Techniques. Anthony Kelly, Powervation’s VP of Digital Control, must be one busy guy for all of these publications are authored by him.

The PV3002 comes in a 5x5 mm package and can deliver up to 80A to the load. The key parameters for the IC are specified in the table above. Tests at beta sites have show a 10% efficiency improvement at light load over existing solutions and a gain of up to 30% in system energy savings. Since the power-converter is completely self-contained, it can be plugged into any board and operate directly without any user intervention – this is referred to by Powervation as Plug-and-Power technology. Should some configuration be necessary, a Digital Power Center Interface GUI is also provided to make the configuration a breeze. In quantities of 1000, the PV3002 is currently available at $2.75 a pop.