Electromagnetic Compliance: Anechoic Chamber Not Required

Yesterday, I reported on Paul Cunningham's announcement of a new product, System VIP, in my post System VIP: Logistics for Cache-Coherent Systems. A few minutes later Paul announced a second product, the Clarity 3D Transient Solver.  This is the next product in Cadence’s system analysis initiative, breakthrough EM simulation technology, and delivers up to 10X performance with virtually unlimited capacity and test measurement accuracy.

Anechoic Chambers

You can't sell an electronic product without getting CE or FCC (or equivalent) certification. In fact, for medical devices and aerospace, the requirements are even stricter. This doesn't just apply to obvious electronic products like laptops or automotive ECUs (electronic control units). It even applies to household goods like washers and dryers. Furthermore, this isn't just some sort of nice-to-have certification—you can't sell the product if it fails.

There are two main aspects that need to be measured, known as susceptibility and emissions. Emissions mean that the device does not emit electromagnetic radiation above the appropriate threshold. Susceptibility means that the device performs correctly even in the presence of a given level of electromagnetic noise. For example, an ECU in a car must not radiate above some threshold. But also, it must not fail if someone turns a laptop on in the car, or several people all use their cellphones.

The usual way to test this is in an anechoic chamber. I only knew the word "anechoic" in the context of audio where it is a room surrounded with sound-absorbent baffles. In fact, twenty years ago when I last worked for Cadence, I was in an anechoic booth to record the voiceover for a DVD we made of one of my keynotes. It is a weird experience since, without headphones, I didn't sound at all normal. But the same approach is used to absorb electromagnetic radiation. So to test a device, it is put in a chamber. A standard test is to have an antenna to make measurements 3m (~10ft) away. For example, the image above shows a drone being tested.

Physical testing is an expensive proposition, mostly because a prototype needs to be built. It is especially expensive if the test fails, since then the prototype needs to be re-engineered, a new one built, and the test repeated. For example, creating and testing an automotive ECU apparently takes 10-12 months. It's not just cost, it's also time-to-market. This requires that physical prototype testing be replaced with simulation. But legacy tools with old architectures scale poorly and require multiple machines with terabytes of memory, in effect requiring expensive, dedicated hardware.

Emissions and susceptibility are the most important parameters, but there are others. For example, cellphones must not radiate in a way that too much radiation penetrates your head. On the other hand, heart pacemakers are programmed wirelessly, and they need to work correctly when they are inside a person's chest. Antenna placement can affect how effectively a device like a smartphone will operate. Remember "antennagate", when the iPhone 4 would not work properly if you held the phone the wrong way.

Clarity 3D Transient Solver

At CadenceLIVE Europe yesterday, during his keynote, Paul Cunningham announced the Clarity 3D Transient Solver. This is a product that does all the things that I just described in the context of an anechoic chamber, except in software. Eventually, the anechoic test will probably be needed for final compliance, but using Clarity 3D Transient Solver means that the earlier tests, especially ones that require any redesign, can be done entirely on servers. Typically, this is done in the cloud, and Cadence CloudBurst is ideal for this since it requires no customer IT setup and is 100% web-based.

The Clarity 3D Transient Solver is massively parallel, multi-threaded, and distributed. The architecture allows the solver to be run on hundreds of CPUs optimized for both cloud and on-premises distributed computing. It has virtually unlimited capacity with near-linear scaling as compute resources are added. There is no loss of accuracy compared to physical test measurement. Note that this is measuring an entire system: an ECU, a mobile phone, a drone, even a whole car, or a heart pacemaker inside a body. The tool reads design data from all standard chip, IC package, and PCB platforms and offers unique integration with Cadence implementation platforms.

A typical use is to simulate the 3m test that I mentioned above. Clarity 3D Transient Solver is used to model the anechoic chamber, the system being tested, and the antenna 3m away. It then uses finite-difference time-domain (FTDT) methods to work out all the required values needed for compliance. Since it is a time-domain method, FDTD solutions can cover a wide frequency range in a single simulation run.

Clarity 3D Transient Solver will be available in Q1 2021.

Experience Developing Automotive ECUs

I said earlier that developing an automotive ECU takes 10-12 months. In Japan, one of the major developers of automotive ECUs subcontracted by tier-1s is UTI (Ultimate Technologies Incorporated). They have been using Clariity 3D Transient Solver and this allows them to take three months (30%) off their design cycle. Satoshi Utsumi, UTI's CEO, related their experience:

As a premier engineering service provider, Ultimate Technologies focuses on quick, efficient, and first-time-right designs. The Clarity 3D Transient Solver from Cadence allows us to simulate with test measurement accuracy so we can predict what will be measured during EMI testing, thereby ensuring our customers pass EMI compliance checking on the first pass while dramatically reducing the number of prototype designs. This allows us to shave up to three months off automotive ECU design cycles, reducing design cycle time by as much as 30 percent. With Clarity 3D technology, we can quickly iterate and improve design quality while meeting customer schedule demands.

UPDATE: Ben Gu presents a video on Clarity 3D Transient Solver (4.5 minutes):

Sign up for Sunday Brunch, the weekly Breakfast Bytes email