Collision avoidance radar devices generate bursts of RF energy called ‘Chirps’. Which are transmitted (TX) to be reflected and ultimately received (RX). A Chirp is a sinusoid that increases its frequency linearly. The linearly changing frequency of each chirp can be used to extract vector speed information about the target. EBand Radar devices produce chirps over a bandwidth of 1GHz and, as in this example, near 76GHz.
blog posts related to semiconductor test
Semiconductor front end automation took place decades ago. The sensitivity of the front end process was requiring it and the always equal form factor of the wafers was supporting it nicely with a standardized transport media, the FOUP (front opening universal pod). It provided a standardized interface for all material handling and processing equipment and the wafer handling is, in general, very reliable and jam free. Automation companies were therefore able to address all technical requirements of the flexible automated material handling between standalone equipment in the fab.
On the other hand the semiconductor back end has had to deal with all kinds of different package types and form factors, resulting in different transport media, such as bulk, tube, metal magazine, tray and reel. Even though SEMI has started to standardize this media, there are still many different form factors to consider. This has made it impossible to come up with a global material loading and unloading type for all equipment. It was therefore less costly to move backend into low labor countries to be able to manage and reduce the handling cost.
In addition the packages often have very loose tolerances making it very difficult to handle the devices reliably and jam free through the backend equipment once they were singulated. This leads to a test floor operation with many human machine interactions for clearing jams. But a test floor with frequent human and robotic interaction can lead to a less safe factory operation and may infringe on existing safety standards.
However, over the last 30 years two automation approaches were attempted in the backend. Only one was successful. Fifteen years ago strip test, and its integration in back end production lines seemed to be the Holy Grail solution to automate the test floor. While strip test is still around, the lines disappeared pretty quick again from the test floors with some exceptions. The integration of trim & form assembly equipment with test handlers and final packaging equipment was not successful because of the typically different speeds of the integrated pieces of equipment, causing the OEE to drop significantly compared to a flexible standalone operation.
The only successful backend automation ‘line’ that survived and found its justification is the turret handler. These handlers combine test and final packaging. These handlers are a fast and reliable solution for simple devices, and applications, when long setup times and limited parallelism and temperature test capability are acceptable. Turret handlers are the dominant solution for the discrete jelly bean test market.
Why is the backend getting serious in the move into automation? What boundary conditions changed? Who are the drivers?
With the new generation of collaborating robots there is the ability to get AGVs (autonomous guided vehicles) with robots onto the test floor in parallel with human operators and technicians without any safety concerns. This was not possible before and is now opening options that used to be prohibited due to safety concerns. The automated material handling of standalone gravity, pick & place, strip test and turret test handlers is becoming feasible and manageable.
Historically the transition of test floors into low labor countries used to be one of the answers to manage and reduce the backend cost. However, these low labor countries are now getting into significant cost pressure themselves. And European semiconductor companies, that want to keep backend operations in Europe, are forced to get even more efficient and go for the next evolutionary automation step to reduce the cost of the equipment idle times during material exchange. Because of this effect, Europe has been at the forefront of factory automation efforts with its so called ‘Industry 4.0’.
Singapore has a different reason for embracing factory automation. One of the first low cost manufacturing bases in Asia is now quite expensive with regards to operational costs. In addition it has a people problem as it cannot source enough production workers. Singapore has limited living space, and with the country’s economic and industrial evolution, more high skilled jobs are required putting a squeezing the ability of production workers being able to afford to live in the country. The Singapore government therefore is funding development projects for factory automation and is subsidizing companies’ automation projects.
With these changed boundary conditions, and the mentioned driving factors, the first automation implementation projects began in Europe, Singapore and Korea. Once they get effective in HVM, it is expected, that the cost pressure will force its implementation in the lower cost countries. The automation race is on.
Please contact us if you are interested, how we prepared our handling equipment for test floor automation.
Today the semiconductor test market is very competitive. This is especially true in the consumable contactor market.
Low operating costs and low average selling prices create low barriers to entry. Micro-organizations plants themselves next to their sole customer and provides fast turn times at competitive prices and onsite support. Although this is acceptable for some it is a risky business model. Furthermore the depth of knowledge of the product and therefore the value add from these micro-organizations is limited.
Outsourcing components can diminish the value add of the end product and lead to finger pointing and delivery delays. These factors push organization toward more stable and established vendors that can not only provide fast turn times and good support but they can focus on R&D and new product development. The ability of these organizations to fund R&D has resulted in revolutionary “flat probe” technologies that combine both electrical and mechanical performance at a significantly lower cost point than traditional radial spring probe technologies.
Larger spring diameters allow more force with less spring length allowing shorter and narrower probes than possible with radial technology. Furthermore the external plunger surfaces allow superior plating than in the internal surfaces of a barrel. Hard base materials offer longer life with lower contact resistance. Finally, with proper attention to the “guts” of the probe design, flat probe technologies can be used for high frequency semiconductor test applications. This presentation will introduce various flat probe technologies and compare and contrast their designs against other flat probe technologies as well as against radial probe technologies.
Download the presentation of Jason Mroczkowski, Director RF Product Development and Marketing IPG, which was awarded “Best Tutorial” at BiTS 2017:Download the full presentation
Andreas Bursian, Director InStrip & InMEMS Products, authored an article for Chip Scale Review Magazine, in which he elaborates on the question of what the test requirements for MEMS sensor devices will be in the future. Before he goes into detail, he describes in general what our world will look like in the future shaped by IoT and Industry 4.0., and how this will drive MEMS and sensor technology. Industry 4.0 and IoT are small components of a rapid global change that experts tend to call the 4th Industrial Revolution. This revolution will change all aspects of today’s living, such as cash flow, data handling, job structure, and the political and social structures of society and the industrial production of goods.Download the full article published in Chip Scale Review March 2017
We recently attended the MWC in Barcelona. It is clear 5G is going to happen; it is just not clear on when it will arrive, and what it will actually consist of when it does.
In terms of when, some industry players stated 5G will be earlier than 2020, while others stated later than 2020. Those that ‘want’ it earlier, justified it as ‘the technology that is being used is well understood’; i.e Defense/Mil/Aero have used these technologies, in most cases having either invented or made the technology(ies) viable to manufacture and deploy – albeit in smaller quantities and much higher price points and footprint than what the consumer market for Wireless can support.
Those that want it to happen are the equipment and chip companies. The operators on the other hand are split. In the US, Korea and Japan, they are sufficiently capitalized or able to go to the debt or equity markets to raise the $250B+ it would take for the 5G build out. Europe is a different story, with over 120 operators, too much fragmentation, will likely holdback capital for investment. When looking at the investment levels, it is interesting to look at the 4G status. There is still significant runway for further LTE deployment. Not only will there be further deployment to increase the current penetration of LTE worldwide (Main 4 economies in Europe are only around 50%, US is better at around 80%. Asia has the highest rate of penetration with some countries in the mid or high nineties), but there is still significant improvement can be gained by the implementation of Rel-13 and Rel-14. Rel-12 which includes Carrier Aggregation is still not that widely used and this was released over 2 years ago. According to Liberty Global, Releases 13 & 14 will push the data rates beyond 500Mbps.
Even with the headroom available in LTE, most of the industry heavy hitters were exhibiting their 5G technology. Companies like Huawei, Ericsson and Nokia on the infrastructure side and Qualcomm, MediaTek and Intel on the chip side of the business. Exhibitions and demonstrations showed massive MIMO in both sub- 6GHz and the 28GHz bands with beam forming. The focus around 28GHz had the major operators talking about their “Test Benches” with their 5G Pre-Standard deployment.
The Automotive space, IoV or V2X; Vehicle to Vehicle, Vehicle to Pedestrian, Vehicle to Internet/Infrastructure, etc was pervasive in most all talks and in almost all booths. This is driven by the belief that the vehicle platform can bear the cost of IoV investment, while adding huge value. Example cited was: In 2013, 1.25M fatalities with 20M-50M injuries and an economic impact of $518B. The Autonomous Vehicle are seen a key to reduce these ‘accidents’ and requires big emphasis on Radar/LIDAR/Camera and the ability make sense out of these “sensors.
At the regulatory level, FCC policy in the USA is to use a ‘light touch’ and will strive to ensure unparalleled choice and competition, therefore FCC will set rules to maximize investment in 5G. The investments in the core and not just handsets and base stations.
No 5G discussion would be complete without some commentary on IoT . IoT is so much bigger than just Smart City or Smart Vehicle. The terminology used is that the 4th wave is taking place:
- Wave 1: Mainframe Computing
- Wave 2: PC Computing
- Wave 3: Cloud Computing
- Wave 4: Ambient & Cognitive Computing
IoT is Wave 4 and will encompass Drones and Robotics. Edge nodes will be on these devices to take physical info (sensors), digitize and analyze to allow for decision making; Robots on farms, Drones checking out pipelines/basestations/etc.
As an example of real world results in “ambient and cognitive”; measure the environment, process the measurements in the context of desired outcomes and then take actions steps to reach those desired outcomes based on the ‘ambient’ conditions measured (Plan/Do/Check/Act). Using this approach, IoT/Robotics on farms can cut water and fertilizer and weed killer usage by up to 90%.
So, how is this all playing out at the chip manufacturers? Below is a sampling of what some of the larger players were exhibiting:
Booth had a24GHz Radar targeted at exterior lighting. The use model is when a car approaches, the light comes on. When a person approaches a light comes on. If the expert system running the lighting/radar system detects that a vehicle and a person are approaching each other, the light will begin to flash to ‘warn’ of impeding accident.
Infineon also had a 60GHz radar implementation, which is used for Gesture Recognition. The 60GHz has fine resolution and is meant to allow for fine control of an interface vis-à-vis gesture recognition (GR). Finally our smart phones, or touch screens for that matter, can now become fingerprint free with GR.
Were demonstrating some of their designs being used in some of these trials; 28Ghz and 3GHz RF and MPU/FPGA. The 28GHz contains TxRx die, ABP die, PA die in MCM. All in a CMOS process. These ICs are 2×8, 4×4 and 8×8 MIMO. Exhibitors;
SK Telecom: 5G Fixed Wireless 28GHz with Qualcomm, Intel and Samsung IC. 800MHz IF, 192 antenna with 4×4 MIMO.
Showed fixed wireless at 28GHz using Qorvo PA/ABF. Showed a toy car moving back and forth, getting a beam-formed signal to/from a basestation.
One showcase in this booth was the Snapdragon 835 and dynamic anttenna matching; branded TruSignal. They had a 4G 5MHz wide dual carrier aggregation signal that was delivering 11Mbps for a streaming video signal. Then they turned on the “TruSingal Software” (dynamic antenna matching) which then boosted the data rate to 33Mbps. Note that this is a Rel-13 implementation; 1/10 of the of the CA using 100MHz IF capacity. Remember what we said about LTE runway.
What does this mean for test and measurement
From a test and measurement perspective, the focus was very much on the engineering and development labs. The main players we saw were Keysight, National Instruments and Rohde and Schwarz. All 3 demosntrated some form of Over the Air (OTA) testing. The major bands shown where at 28GHz and 70GHz. Included in these demos were how to calibrate OTA transmit attennas and manage much higher IF bandwidths.
National Instruments (NI) also conducted as session led by Leif Johansson, Principle Engineer Market Development RF & Comms for NI. They started the session with a video with key opinion leaders: Amitava Ghosh. Nokia Fellow and Head of Small Cell Research, Ted Rappaport of NYU Wireless Founding Director, Professor Mark A Beach of Bristol University (key research university in 5G Radio/MIMO/Antenna). The video also featured their chief architect (Ian Wong) and RF/SDR Director of Product Marketing, James Kimery. Although not really relevant when looking at the challenges of HVM testing of 5G IC’s, the video and presentation was informative nevertheless. NI talked about the four 5G Vectors:
Massive MIMO | mmWave | Multi RAT (GFDM/OFDM) | Adv Wireless Network; SRD, NFV, CRAN
The four days added context to all the information we at Xcerra are being exposed to on 4G and 5G and WLAN, IoT/IoV. LTE has significant runway with Rel-13 being implemented and Rel-14 getting ready to be released: These two releases will unleash >> 100Mbps sustained data rates, even into the Gbps range, along with LTE-LAA for IoT/IoV.
5G will happen, and like some of the other companies serving a different part of the 5G food chain we at Xcerra have started “test beds” and programs to help our customers qualify their 5G content, positioning Xcerra as the leader in 5G ATE space . When this will result in meaningful spend for 5G ATE tooling? Well, it depends on who you listen to. It could be 2020 or it could be as far out as 2023 before any meaningful 5G ramp.
We encourage you to join the discussion on 5G. Here are some initial questions. Please feel free to comment on any of them.
- Are you or your company involved in 5G technology development?
- What do you see in terms of deployment?
- When do you expect meaningful volumes of your 5G devices?
- What do you include in your 5G product/device portfolio?
- What are the test challenges you see with the introduction of the new 5G technologies?
Contact your sales rep to arrange for meetings on your 4G and 5G test needs. http://ltxc.com/contact-support
OSATs and ATE vendors are making progress in determining what works and what doesn’t in 2.5D packaging, expanding their knowledge base as this evolves into a mainstream technology. Most experts believe that full 3D packaging is at least five years away from mainstream deployment. 2.5D, in contrast, already has made inroads in markets where price sensitivity is low and demand for throughput to memory is extremely high, such as networking, server and graphics applications. Andy Nagy, Senior Director Marketing HG & TCI Operations at Xcerra, recently highlighted the challenges in test handling in an interview with Jeff Dorsch, Semiconductor Engineering. The article compiles thoughts of the key players in this market.
Read the full article here: http://semiengineering.com/2-5d-adds-test-challenges/
Semiconductor giant AMS has set some aggressive growth plans for the coming years. Achieving them will rely on a workforce with a will to win and out-of-the-box thinking.
Sensors are already all around us; they’re used in everything from smartphones to smart homes, to industrial automation and all devices that comprise the Internet of Things (IoT). With new applications constantly being developed, they’re only going to become more pervasive. According to Alexander Everke, Chief Executive at AMS. “Sensor technologies will be increasingly important in the future,” he predicts, adding that they’re replacing the human senses.
Read the full article in the CEO Magazine here: http://www.theceomagazine.com/business/alexander-everke/
Mobile broadband technology is beginning to crawl from commonly known 4th Generation Wireless (4G) transmission standards to fifth generation wireless IMT2020 standardization, also known as 5G. This 5G network technology will influence semiconductor test in two directions, an evolutionary track and a revolutionary paradigm shift. The revolutionary aspect of 5G targets massive amounts of bandwidth not previously thought of as accessible. Many technological challenges have blocked the reasonable implementation of 5G cellular technology. Consumer demand for rapidly growing amounts of bandwidth, has created the need to solve these challenges.
Recent millimeter wave (mmWave) band spectrum studies have put the solution for 5G in reach with large amounts of spectrum available in mmWave. Frequency bands under consideration include 28, 37, 39, 64-71, 71-76, and 81-86GHz, which are far removed from the less than 6GHz technology offered today. Shifting from 6GHz to 28GHz and beyond creates challenges up and down the value chain. This fundamental shift is why many data compression techniques currently in development will become the next evolutionary step towards 5G, also referred to as 4.5G. The techniques for 4.5G focus on better access within currently defined licensed spectrum.
R&D activities in studies beyond 6GHz frequency bands have been restrained because of the mmWave characteristics of limited transmission and wavelength. Studies on standards beyond 6GHz have lived in academia and the military but had limited consumer application. With limited consumer application there was not a great wave of interest in 5G as an acceptable standard. So, what has changed? Consumer demand for more bandwidth which is faster, smarter and less power hungry at a low cost!
Research from NYU Wireless, Ericsson, and many others has driven proof-of-concepts for mmWave applications to acceptable levels of reasonable realization. Areas of research interest include enhancements to the physical layer, interference mitigation, multiple-input-multiple-output (MIMO) antennas, network security, network management and many others. Bridging the gap and creating a learning environment is automotive radar and wireless LAN 802.11ad (also known as WiGig) specification standards. Both standards exist with promising levels of acceptance which drive the need to test semiconductors in the K-Band, and the V & W bands of mmWave cellular channels. Is there one key research breathrough which will put realization in hyperspace?
Since we are discussing a standard that is not yet ratified and the test requirements are not fully known, we must extrapolate to hypotheses based on Xcerra’s years of experience in RF test. This future technology has driven all of the semiconductor test equipment suppliers to rethink the world of RF test. Using our proven experiences in automotive radar and 60GHz WiFi we are applying what we have learned in this area to bring solutions to market. But creating the technology to test these devices is only half of the battle. These products have to fit within a model of production quality test while answering the requirements of mass production and aligning the costs associated with consumer expectations. How will the 5G revolution affect your business and testing operations?
Moving RF test from standard cable and pogo technology, which has been well understood for many years, to waveguides and Over The Air (OTA) connectivity is challenging. Waveguides create a new complexity where mechanical considerations are equal to RF signaling and measurement parameters. We have suddenly had to become plumbers to create and execute multisite test capabilities. For OTA testing, we are now required to speak the language of horn antennas and linear arrays rather than screwing down a Sub-Miniature version A (SMA) connector to a device under test (DUT) board. The benefit of OTA testing is that it allows us to move quickly to multisite test capability without the complexities of creating expensive multisite waveguide connection schemes. Where this becomes a challenge is the mathematical models required to beam-form a signal to the appropriate site for measurement or capture. As we move forward, test requirements will have to allow for the inconsistencies of sending and measuring signals through the air or accept the costs associated with mmWave. We all know where this is going to go! What role do you see OTA testing playing?
Over The Air test has changed transmission properties versus well known cable models. With the use of cables we have had the benefit of working with known properties regarding RF transmission, and when needed, can adjust for any inconsistencies along the signal path. One obstacle which is commonly taken for granted is the calibration of the Test system. Many systems today calibrate all inconsistencies without the user having to pay great lengths of attention. System calibration, user calibration, and de-embedding are commonly used today to ensure reliable, accurate measurements. Calibration and system to system stability at mmWave frequencies has all of us engineering new solutions for high-volume semiconductor production.
Lastly, we are all devoting time on the subject of Multi-GHz bandwidth requirements for the IF and RF for mmWave technologies. For starters, a 2.16 GHz bandwidth requirement for 802.11ad is driving semiconductor manufacturers, and back-end test suppliers towards a new horizon. Test instrument vendors, whether benchtop or production test, are making strides with up/down conversions to the 60 GHz mmWave band. Creating the 802.11ad modulation in IF is pushing sample rates beyond 1 Gbps, which is 4 times today’s 802.11ac standard. The issue with multi-gigabit sample rates is the heat created by the multichannel high speed ADC’s and DAC’s to achieve the 802.11ad modulation requirements.
In summary, paradigm shifts in our industry has created an opportunity that we are capitalizing on with our deep understanding of RF and mmWave. Our research in automotive radar test has given us the tools to move quickly towards broadband cellular 5G testing in the future. Expectations of production test in this arena will be realized in the near future. Therefore agreements to acceptable standards must be approached with a collaborative mindset between end customer, semiconductor supplier and test vendor, to achieve best practices for mmWave production test.
Andy Nagy, Senior Director of Marketing for the Handler Group at Xcerra has recently published on article about final test of WLCSPs in Chip Scale Review. He describes the short-comings of the established test process for WLCSPs particularly with the wider adoption of these package types to critical applications. The article elaborates on a new process, which supports true final test the packages – right before shipment
The semiconductor test equipment market has matured. You would expect an industry that came of age in 1980s would have matured over the last 35 years and in fact it has. The semiconductor test equipment industry has experienced consolidation, a sure sign of maturation. Two factors are important to consider with regard to it becoming a mature industry. First, market share shifts of 10 points or more are extremely difficult and slow to occur. The value of incumbency is very high and to achieve growth faster than competitors means uncovering new market opportunities through innovation of your existing products, applying existing technologies into adjacent or new markets, or through merger or acquisition.
A variety of technology and business trends impact the test equipment business cycles. In the full paper, which we invite you to download [hyperlink], we analyzed the impact of the following drivers n detail:
- IC unit volume expansion
- Equipment lead times
- Degree of customization of SOCs
- Chinese market
- Replacement of home-grown ATE by commercial solutions
- OSAT business model