posted by Bryon Moyer
Numerous systems tend to get used for verifying SoCs, and, with software now in the picture, the range is extended even further. We’ve talked before about the use of simulation, virtual prototypes, emulation, and prototyping as ways of getting both hardware and software to work, and to work together. Including their unification.
Synopsys recently took a move towards unification by bringing their Virtualizer virtual platform tool and their HAPS prototyping tool closer together. What this is means is that a design can be implemented with some parts in Virtualizer and some in HAPS and the two systems can talk to each other while running.
They actually run the SCE-MI 2 interface (traditionally found in the emulator-to-host connection), running over their UMRBus. This allows transactors to speed the interchange of data.
The architecture is very AMBA-centric; much of their DesignWare catalog relies on AMBA, and AMBA is popular, so this isn’t a big surprise. They’re open to other busses on an “ask us and we’ll consider it” basis.
The actual use of the tools isn’t so integrated. The two sides have separate programs that you run to manage them – there isn’t one unified interface that can talk to both sides. But this is partly due to the fact that they don’t traditionally see one person doing the whole thing. In the early stages, system integrators/architects would use the Virtualizer side and FPGA guys would implement the HAPS side; they would tag-team to get it up and running. Once that’s all done, then software programmers could use it (using computers more moderate than those required for the FPGA-building tools, for instance). So a single console might not have an associated use case.
The design partitioning process is also manual (although they could see the future possibility of tagging a design to automatically build the virtual and FPGA sides). Cross-triggering between the two sides is rudimentary.
This capability will be generally available in August. Why announce when they did? I’m guessing because they couldn’t talk the DAC guys into rescheduling the conference to August…
You can find more info in their release…
posted by Bryon Moyer
The next thing after MEMS is… OK, time’s up… NEMS. Of course. From micro to nano.
One of the materials that seems inextricably linked to NEMS is carbon, even if carbon isn’t a defining element of NEMS. More explicitly, at UC Irvine, they’re doing work on what Prof. Marc Madou calls Carbon-MEMS (or C-MEMS).
Manufacturing techniques at the nano level can be quite different from what we’re used in at the MEMS level. C-MEMS uses organic precursor chemicals to lay down a structure. When processed/heated, these structures lose 80% or more of their volume and reduce to almost pure glassy carbon. This presents an attractive alternative to trying to actually machine carbon, which is notoriously difficult.
At the recent MEMS Business Forum, Prof. Madou illustrated one type of structure his team has worked on called a “wash-line nanosensor.” These consist of a series of carbon posts; wires are then strung from post to post (like a wash-line) above whatever the substrate below is. This makes the wire accessible from all sides and distances it from any effects the substrate might have on it.
But how to make such a structure?
The posts can be made by layering an appropriate photoresist polymer down, patterning and exposing it to create the posts, and then reducing those polymer posts to glassy carbon. Step 1 done.
He showed that, given the right polymer goop, discharged through a syringe or nozzle in the presence of a voltage between the nozzle and substrate, you can “spin” a thin bead of the material which, on its own, just kind of mats up like spaghetti on a plate. They refer to this as “electro-spinning.”
Done in the presence of the posts, and spraying for 2-3 seconds, they found that the extruded polymer thread would naturally start at one post, swirl around a bit, then drift to the next post, swirl some more, etc. The spaghetti cap on each post made ohmic contact, but they wanted something not quite as messy.
So they moved the substrate with the posts a bit closer to the nozzle and then moved the stage on which it was held. Rather than the thread going where it wanted, they could direct the thread from post to post in a controlled manner, with each thread making a clean, simple connection on the top of each post.
There’s lots of magic here in the materials and viscosities and all of the other parameters involved. But, stepping back from all of that, it represents a dramatically different way of building an electromechanical structure.
posted by Dick Selwood
Almost every presentation that I go to talks about the fast moving development cycle and the rapid obsolescence of products. So it was a real change to get a release that talked about longevity.
Foster Transformers are boasting about product returns, not normally an issue that people are proud of. In this case the first example was a door bell transformer that was making strange noises. It was indeed making a humming noise, as the wax that protected the core from moisture and damped the intrinsic hum had broken down. Since the wax was put in place in the 1940s, it was perhaps not surprising.
The second problem was caused by a user shorting out a transformer. Until this happened the transformer had been working happily since 1967. Foster used this to plug that the company’s new model is designed to survive shorting.
And finally a real failure: a transformer installed in 1960 actually gave up the ghost.
Well done Foster. In a throw- away society you are a beacon of common sense.