Nick Knudson

Coming back for Round 2 on the Altium® OnTrack podcast is Ethan Pierce! At this year's PCB East conference, Ethan and Tomas Chester gave a detailed course on thermal design for PCBs, a topic which sometimes takes a back seat to other areas of PCB design. Give this episode a listen and let me know how you handle thermal issues in the comments! https://lnkd.in/g4hnS4da #pcbdesign #electronicsengineering #altium

35

6 Comments

Don't do noise tests near desktop computers! I was once testing a power supply our team had developed for a quantum sensing application. The power supply was designed to support 250W with very low noise (<20mVpp).   We had selected the power converters from TDK-Lambda (i6A4W series) that already guaranteed <20mVpp ripple. And additionally we added some filtering to attenuate the noise further. But during testing, I saw the noise-levels exceeding 50mV. If the power converters are certified to generate <20mVpp ripple, and we have additional noise attenuation, shouldn't the output be even more clean? It didn't make any sense. Upon investigation, it turned out that the oscilloscope was picking up radiated noise coming from a nearby desktop computer in the lab. 𝗪𝗵𝗲𝗿𝗲𝗲𝘃𝗲𝗿 𝘁𝗵𝗲𝗿𝗲 𝗶𝘀 𝘀𝘄𝗶𝘁𝗰𝗵𝗶𝗻𝗴 (𝗱𝗩/𝗱𝘁), 𝘁𝗵𝗲𝗿𝗲 𝗶𝘀 𝗻𝗼𝗶𝘀𝗲. The switching in desktop computer power supply could radiate noise. And these emissions could be absorbed by your device-under-test. Always shut down the switching electronics nearby when testing noise levels in your device. Have you encountered similar interference issues in your tests?

67

12 Comments

Couple of strategies going on with this prototype: 1️⃣ Partial prototype (don't worry about building the whole system, just the components that you're evaluating) 2️⃣ 3D printed flex circuit, with printed components. don't need to wait for your PCBA vendor to make parts that won't work. Print something as a stand in, and test the fit! 3️⃣ Use colors! Seems simple, but it helps tremendously when reviewing our discussing parts. "That gray part, no the other gray part, the one that goes up inside the big gray part..." This advice also helps in your 3D CAD model. Choose a few basic colors and try to stick with them from version to version. Keeping colors consistent makes comparison images much easier to compare. 4️⃣ Photo/video evidence. Take photos and videos of all your prototypes! Create Google Photos album for the whole project and keep stuffing the images and videos there. Scrolling thru this in the future is like a time capsule of design documentation. Good luck out there! #prototyping

82

4 Comments

Case study: Designing loose-leaf #RigidFlexPCB stack-up for enhanced flexibility When designing a flex circuit board, balancing structural reliability and signal integrity is paramount. In our recent project, we tackled this challenge with an innovative loose-leaf rigid-flex stack-up. Here are the highlights of the case study: 1. To achieve the desired flex thickness and bending radius, we transformed a flex design into a loose-leaf rigid-flex stack-up. 2. To prevent signal overlap and maintain signal integrity, we routed 253 impedance trace signals across two leaf structures with an air gap between them. -- Read the case study to learn more. -- Send this to flex PCB designers. #flexpcb #pcbdesign #electricalengineering

11

Saturn PCB Design, Inc. ok, so the link to the antenna matching video posted as a reply didn't get much visibility so I'll make it a separate post. Hope you like it :) https://lnkd.in/eWv2uJ4C #pcbdesign #pcbassembly #pcbmanufacturing #electronicsengineering #electronics #antennas #rfengineering

35

2 Comments

Explore All Your Iteration Options for C++ STL Containers! Curious about the different ways to traverse your favorite STL containers? My latest guide breaks down all the iteration options from integer indexing to iterators and range-based for loops. Perfect for any C++ developer looking to enhance their coding toolkit! Key Iteration Methods Covered: Sequence Containers: vector, deque, list, forward_list, array, string Container Adapters: stack, queue, priority_queue Associative Containers: set, multiset, map, multimap Unordered Associative Containers: unordered_set, unordered_multiset, unordered_map, unordered_multimap Summary: Integer Index: Available for sequence containers like vector, deque, array, and string. Iterator: Supported by most containers except certain adapters like stack, queue, and priority_queue. Range-based For Loop: Simplifies traversal for containers that support iteration. Read the full guide now https://lnkd.in/g34HmpWv #cplusplus #programming #softwaredevelopment #coding #developers #STL #C++ #TechTips #CodeEfficiency #CodingLife #LearnToCode #TechCommunity

17

4 Comments

Launching an electronic product requires compliance with EMI/EMC regulations. This is often costly 💲 and time-consuming ⏰, with a high failure rate ❌. Simulating EMI/EMC can reduce or even eliminate this pain. 𝗔𝗻𝘀𝘆𝘀 𝗦𝗶𝗪𝗮𝘃𝗲 is a tool you can use to analyze PCB layouts. The image below shows a simulation of resonant modes performed by our team at Oxeltech, helping us gauge the severity of EMI in our designs. It's wise to perform this simulation before using an anechoic chamber, which is expensive and often requires waiting for an appointment. The tool identifies potential EMI sources on the board, allowing for iterative corrections until the design is chamber-ready. How do you tackle challenges with EMI/EMC compliance in your projects? #PCBDesign #EMI #EMC #Electronics #Product

114

11 Comments

Do you know what it takes to design a differential pair that works past 30 GHz? If you are planning to take on PCIe 4, 5, or 6, USB 3.2, or any modern SERDES interfaces, I hope so! When we were building the JITX autorouter we wanted to be sure to address the issues plaguing modern designers. We found guidance in this DesignCon paper from Donald Telian, Kevin Rowett, and Ilya Teplitsky: https://hubs.ly/Q02wf7l70 Summarizing the key points for diff pairs here: 1. Above 8 GHz: have fear. Above 30GHz, everything is a stub. 2. Route as short as possible to minimize loss 3. Match P/N statically to within ~0.15ps 4. Match P/N dynamically to never accumulate 0.5ps of skew within a 100ps window 5. Size skew matching bumps below the relevant feature size so they become 'invisible' 6. Space out skew matching bumps irregularly to avoid impedance discontinuities. (the paper has a lot of other great lessons, so check it out) We make sure that JITX helps people design better boards, so when we asked ourselves "How should our autorouter draw differential pairs?" our answer is of course "Just do the Best Possible Thing." Put the computer to work making the detailed shapes so humans can focus on the more important stuff. Here's a video of the router matching skew to 1 femtosecond, enforcing dynamic phase matching, minimizing distance, minimizing impedance discontinuities, and avoiding obstacles. Feels good to offload all that to the computer! You can learn more on our website: https://hubs.ly/Q02wfdCL0 What do you think about this paper Robert Feranec, Benjamin Dannan, John Burkhert? Agree or disagree with the approach? #electricalengineering #signalintegrity #DesignCon

206

30 Comments

Checkout the US Patent https://lnkd.in/gYibwNKq because this shows how you can re-use the current that flows in the load of a Class-D driver, significantly increasing performance... The speaker load is doing double-duty: its making the audio output and at the same time making a buck regulator. Initially, it looks like we just have another voltage to switch to (there are other chips that do that) - that's an incremental win. But there is more to it than that. The charge on the capacitor that provides that additional voltage level can only have come from prior cycles of the audio output switching - thus it is re-using the same charge that generated prior analog outputs. Efficiency at full power is enhanced since the patent is working as the audio signal passes through zero, and at 1% of full power this patent configuration is giving at least double the efficiency. That makes a big difference to battery life. Check out some graphs at https://lnkd.in/g_D2Em9e I hope you find it interesting - message me if you want to know more.

24

14 Comments

Enhancing Reliability: Expert Tips for Designing Circuit Boards Dr. Christopher Jackson explains the value of designing products for their intended in-use application. View the entire episode (Introduction to HALT – Highly Accelerated Life Testing with Dr. Christopher Jackson here: https://lnkd.in/ey_7Rcbs #reliabilitymatters #halt #halttesting

5

One of my debugging war stories from a while ago: ⚡ I was porting my existing real-time audio application in C++ over to my new embedded audio DSP platform running linux. I had existing code that was running on Windows and Mac OSX and I was using a popular 3rd party C++ audio application framework. On my embedded linux platform my goal was to use the mainline kernel tree for cost of maintenance reasons. Unfortunately, I kept having problems where the audio output would get regularly interrupted resulting in a click. Eventually I decided to use a profiler to figure out what was going on and turned to the excellent Tracy profiler (open source and free!). With the help of Tracy of found out my audio callback was taking too long to return occasionally. What was interesting is that it always seemed to have something to do with a memory related function call. Eventually I started examining the memory usage behavior of my application and found out that the application framework I was using, had its own memory management classes which under the hood would call malloc to allocate 4 bytes (!). Due to the frequency with which these calls happened, I eventually understood that this resulted in memory fragmentation, which explained why some memory related function calls would take such a long time to return. Fragmentation can result in the linux kernel pre-empting your thread(s) while it performs its cleanup. The result is that my callbacks and other code took too long to return, and this resulted in missed real-time deadlines and thus audio clicks. To solve this problem I implemented my own memory allocator and re-implemented malloc and friends to use my allocator. I considered using off the shelf allocators, but got a bit paranoid due to the above experience and decided it was a good exercise anyway to implement one myself and be aware of its behavior under the hood. Additionally, I reviewed my code to pre-allocate as much as possible. Conclusion: if you are working on a real-time application, no matter what it is, whether it is an audio processing application, a low-latency proprietary trading application, or anything else that needs to have real-time or low-latency behavior, profile your application in various ways and be aware of the memory usage behavior of your application, and don't forget to examine the behavior of ALL 3rd party libraries you are using in your application (including profiling code or libraries). 🤓 #audio #realtime #linux #memory #lowlatency #dsp #hft #cpp

103

24 Comments

Dependable traffic sensors are made by #smartmicro.

23

When Does using VPX Make Sense? I was recently involved with a design using the VITA 78 specification for space-based applications. VPX is an open standard for creating very specific circuit cards for electronics systems. If you are unfamiliar, this will be a snoozer post so you should move on. The project involved a chassis design inside an orbiting telescope. The cards inside the chassis steer the imager and perform processing. Very interesting and exciting. Using a VPX platform makes perfect sense in this case. The technology is not changing, the system can perform all functions with adequate overhead capabilities, and the system will serve the platform throughout its life. Perfect fit! I am a big supporter of MOSA, SOSA, FACE, etc. when it makes sense. There are some applications that are less static, and the mission continues to evolve at a faster pace than the hardware providers can respond with solutions. This is when VPX does not make sense. Trends that tend to prevent VPX from being the right choice include: · Frequent increases in sensor resolution · Consistent change in software demands which impacts the necessary hardware capabilities · The amount and complexity (bit depth) of the data being processed ·  The latency requirements that drive network speeds (GbE, 100GbE, 200GbE, etc.) ·  AI and compute performance demand driving change into the architecture I realize I am calling someone’s baby ugly here which is a very risky thing to do, but hey, let’s call it a good discussion topic. VPX isn’t really commercial-off-the-shelf (COTS) and it takes several years to design, debug, build, test, and production-ize the modules. With the current technology cycle, the capabilities are nearly obsolete by the time the product is deployed in the first application. It is a timing issue. Again, great stuff and a very impressive engineering accomplishment but only relevant if the mission isn’t changing over time. These wheels were set in motion by William J. Perry, SECDEF in a 1994 memo to the DOD describing a “new way of doing business” for the defense industry. This changed the focus from designing leading edge technology for the military, then flowing that into commercial applications, to letting the commercial applications create the technology and adapting this to the military. Like it or not, that is where we are today. I don’t think it is a bad thing, just a wrinkle in making the DOD wildly successful. Challenge accepted! Be challenged, do important work, and make a difference. Jim

16

1 Comment

Flex PCB pre-layout considerations recommended by Vern Solberg 1. Determine the type of application 2. Design with your stack-up in mind 3. Work with your contract manufacturer 4. Stick to standard flex PCB design guidelines 5. Optimize your design for the specific application needs 6. Implement modeling and testing -- Read the article to learn more. #FlexPCB #ElectronicsEngineering #ElectricalEngineering

9

This is basically how Composition of Experts runs on Samba-1 for enterprise grade. For datacenter version, we are getting 1000 tokens per second with time to first word at 0.14s (correction 0.09s per latest release) doing same type of dynamic hot swap. Good to see world moving in this direction. Now let’s hot swap 1000 models and keep 1000 tokens per second!

73

4 Comments