Podcast: Play in new window | Download
Subscribe: Apple Podcasts | RSS
Our third (non-embedded) call in show!
- Callers
- Mahesh called in about a recommendation on soldering irons for his business in India.
- Sebastian was having his micro lock up when his wall wart was leaking current (assumed) onto the device and then it was shorted to chassis via a USB ground.
- Aaron called in asking about how to change careers
- Brian called in from Trinity College asking for advice on getting through an engineering program.
- Paul was also interested in changing careers, thinking about going from test engineering to design.
- Adam was wondering about resources for learning about high speed design.
- Dave and Chris both recommended Robert Feranac’s Fedevel Academy. He teaches high speed design using Altium.
- Rudy was wondering if we have tried OpenSCAD. Neither Chris nor Dave had.
- Rudy also has a bet going with his son about when self-driving cars will be implemented (5 or 10 years?).
- Andy is building a fridge that is meant to age steaks. He is building buttons for the control in KiCad and wanted to know about making hatched planes. This is not directly available in KiCad, but there are some workarounds.
Another great set of questions this show! We’ll be back for future shows where you can call in. Check out our post about how people could contact us this time and keep an eye out for future announcements.
Thank you to rfduck for the picture of the phone.
The bet is for for 1 dollar, just like Trading Places.
Anyone looking to do more PCB layout work should read…
High Speed Signal Propagation: Advanced Black Magic
…In my opinion it is just the bee’s knees.
Great show!
A thought for Sebastian. Check if any of your inputs are floating. Maybe noise is being coupled to one, and for one reason or another the firmware is going into another state because of it. Maybe it’s triggering an interrupt or incrementing a counter you hadn’t expected to change at that point of the program.
A tip for Brian. When the study load increases it’s important to make time to relax. It’s tempting to throw yourself into study and lock yourself away in the lab, but combined with the stress you’re already under, isolation is the last thing you need. Even if it’s half hour a day just to go for walk.
hey, I had that issue with coupling into high impedance inputs on one of my uC pins, so generally speaking your idea is a pretty good start to look at those “strange” phenomenons.
I was a bit unprepared for the interview, since this problem and with it my troubleshooting occured more than half a year ago.
I ran into that problem on completly different PCBs designs of mine, and I dont want to lie, but i think I also could test my fault conditions on an commercial board from olimex:
OLIMEXINO-5510.
So this is of course rather disturbing, since it is not only in my designs, but then again the code more or less 🙂 was written by my hands and the help of the internets.
Now, one important point to mention, that I did forget in the show:
The USB GND is always mains grounded, if one device is mains grounded, you can of course decouple the shield from GND via a 1k resistor to restrict the current once putting in the USB connector, but you can never do that to the GND line.
Now one could argue my wallplug devices are cheap quality, but as they are from a rather big manufacturer, XP power, or meanwell, I simply doubt that 🙂
And leakage currents and floating DC is of course an issue of switchin topology and design, I think in Germany leakage currents up to 0.5 mA are perfectly fine, at least when speaking of standarts.
So here the main points:
more than 3 different design, including one commercial, can be shut down or at least brought into trouble by attaching usb connector to it, as the AC of maybe peak 60V are shorted in a matter of µs to a real GND, this is enough to cause trouble.
Any ideas about that?
Another point just for your interests, and because i didnt make myself clear 🙂
0.5 mA are fine for the law, but actually if that current flows through your body, say wet fingertips or as I tried to explain, earplugs (ears are very sensitive I did find out ), you will definetly feel a bit of pain crawling up, and this is legal, those are perfectly designed PCBs, they dont have any issues whatever.
Thanks for any experience with your floating voltages.
First of all, I have a feeling that what you are doing is wrong (live-connecting together two circuits with different ground levels). It is much better to galvanicaly isolate them, but I understand that there is not much you can do about it now.
It is difficult to help you without being able to test things physically, but I would recommend you to try to bodge a few things without redesigning the board (by means of cutting traces and bodging wires): make sure that USB sonnector shield is brought directly to power supply terminals GND, so the GND of your board and USB shield meet each other in the single point near the terminals (kind of star-grounding), place some decoupling capacitance between that ground point and Vbus, then throw some inductance (ferrite beads etc.) in series between this point and your main board ground (same for vbus if needed), then place capacitive decoupling network next to inductance (probably you have one already near the power terminals).
Please, let me know how it turned out.
When you get stuck you need to ask the caller “What is color of their DVM?”
Definitely
For the callers looking at getting work, agreed a portfolio of work is very useful, as with the hobby items you build. Always take plenty of photos of your projects, and your notes…not that we like documentation or anything.
Personally myself, as a mature aged worker, I find I fall into the gap of either “over-qualified or no degree”. My best come back to the degree question was “What degree would you like”, the HR manager said “Any”, at which point I stood up and walked out.