EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: rrinker on September 08, 2016, 02:10:47 am
-
So today I got 5 more Arduino Nanos, and 5 RS485 adapters, to further my model railroad control project. This set didn't come with the pin headers soldered on, so I set about doing that, and then went to test each one. These came with the basic Blink program already loaded, they each merrily blinked away when connected to the USB cable. I then uploaded a modified Blink program and indeed they all took the program and executed it. Odd thing, they revert to the basic Blink when the USB cable is reconnected. So I decided to test with my bench supply after uploading the code. Hooked the first one up, the LED blinked some odd pattern, not the basic Blink and not my modified version. Strange. Tried #2 - same thing. Hooked the USB cable up, reuploaded my program, it works. Disconnect the USB cable - it dies. Then I happened to glance up at the power supply...
:palm: :palm: :palm: :palm:
Last time I was using it, I was testing the low current limits of the circuit I was working with, and I set the constant current limit about as low as it goes, sub 1ma (resolution of the readout is only 2 decimal places - HP 3610), and that was obviously too low to drive the Nano, as it had switched to constant current mode and the voltage being applied to Vin was barely over 3V. No wonder it was going wonky. Reset the current limit - each one tested fine, it retained my program in memory. But they still revert to basic blink if I connect the USB cable with no other power applied - with power externally sourced, it retains the last loaded program as it should. But a big self :palm: for having the power supply set for a ridiculously low current setting and not seeing that.
Bonus, got to try out my little fume extractor, and use my soldering station for the first time since setting my bench up.
-
I have certainly had the current limit on a supply set too low, and not noticed, a couple of times.
-
As I was reading, I feared worse. Of the possibilities, having set the current limit too low is the preferred error. :-+
-
While writing my masters thesis many many years ago now: I was frantically finishing some simulation work in the last week before the deadline. I spent a full day debugging a script I'd written to move some magnets and stuff around then run a finite element analysis on the magnetic field patterns. No matter what I tweaked I was still getting the same wrong answers. Panic was building steadily throughout the day until after about 12 hours work I realised that I was editing a script file called something like denzilsprojectv4_currentversion.whatever but every time I ran the script I was actually executing a file called denzilsprojectv4-currentversion.whatever
RAGE!
(Learning a proper version control system has been on my to-do list for the last 12 years as a result!) They all seem so complicated though!
-
When you experement with a large transformer driver and you accidently forgot to put a resistor on your 10,000uF 380VDC capacitor bus. :palm: (ended up getting 2 screwdrivers and a lightbulb to drop the voltage down enough to short it without damage)
-
Testing a switchmode battery charger with a bench power supply.
Puzzler was that the whole thing would start up, and work for a while into the flat battery but played silly buggers after a few hours.
Turns out that into a completely flat battery the charger was in CC mode and not drawing much power because the output voltage was small, so 5A * small voltage = small power = not a lot of input current.
4 hours later the battery voltage was up, and the input power demand was now sufficient to hit the bench supply current limit, at which point the interaction of the bench supply foldback current limit and the constant power nature of the switcher caused things to go very wrong.
Annoying to find because of the time constants involved.
Actually, one thing I learned from that project was that if something has a battery charging function, get that working early, it means you have the rest of the dev time to find bugs in the battery management.....
Regards, Dan.
-
Charged a lead-acid battery with my lab PSU and hooked it up to the battery connector of a quite expensive prototype board the next day. Some minutes later I noted that a logic IC got damn hot, but the board was just working fine otherwise. Holy shit. Turned out that I forgot to change the voltage and applied 14V to the board... :palm:
To my surprise, all other components on the same rail worked just fine during and after the incident. Would've never found that out deliberately.
Oh, I've got another one: Ever tried to debug an atmega using debugWire with a lowpass filter on the reset line? Yeah, took me a few hours to figure that out.
-
Big board full of 4000 series cmos logic, plugs into a card cage in an industrial system.
Kept getting calls from the plant that the thing was falling over once in a blue moon, completely unable to reproduce on my bench, hate those.
Transpired that the card cage at the plant had a faulty power supply module, but as long as enough of the external inputs were being driven to logic 1 the thing would power up sufficiently via the input protection diodes to sort of work, when too many inputs went low the flopflops that held the state tended to randomise.
Card got a power on LED, and undervoltage logic to hold it in reset.
73 Dan.
-
I realised that I was editing a script file called something like denzilsprojectv4_currentversion.whatever but every time I ran the script I was actually executing a file called denzilsprojectv4-currentversion.whatever
You're not the only one. On several occasions I've resorted to adding a rude word somewhere in the code I'm editing, just to force a compilation error. If it still compiles and runs, the file I'm working on wasn't the one being compiled at all.
-
Have you ever... (personal facepalm moments)
Sure, all the time. Just recently I found out (too late) that ultrasonic cleaning and hermetically sealed metal cans do not go together - apparently the bondwires just gave up. :palm:
-
worst thing I ever did was ages ago. I was a student and owned one of the first professional 21" CRT monitors (got that for free from a publishing company - 50kg weight, 300W consumption, entirely made from single transistors, each and every one socketed, with full schematics). I was working on a PCB layout and wanted to adjust the zoom factor to 1:1, in order to see how big my board would become. What did I do? I took that expensive EPROM which would go into that design (the ones with glass window on top), and held it onto the screen. After hearing a *zing* sound I realized that this was not a good idea. That 30kV static charge on the screen surface wasn't very healthy for the chip. In fact, one of the bond wires were missing after that incident :o
-
Failed my final term project using some motorola ADC, which had msb/lsb back to front. For example pin1 was MSG, ping 8 was lsb. When ever i adjusted my input signal, the output would swing wildly 128 pts. Being in-experienced, had no idea why I was getting what was effectively a square wave to an analog signal. The whole thing was wirewrapped so not easy to trace out etc. Later ended up Re-reading the DATABOOK to realise what was wrong. The lab techs were rushed off their feet by every other student racing to finish their projects, so no help there.
-
Was almost 2000 km away from my home in another country. Repairing a few devices. One of them (costs EUR 30k) kinda worked, but I was told it switches off. Tested it a bit, saw a fault. Then instead of disassembling, decided to see the fault "one more time" and it really was. Bzz.. sound, mosfet in DC/DC converter in the PSU board got shorted, 13V instead of 5 got on the power rail, and every board except one got fried :palm:.
-
Years ago in CRT TV time i had a very nice Sony big screen CRT TV , great image quality and everything. One day it popped and went blank. I disassembled, it had a single side PCB, so troubleshooting was easy and found the flyback power transistor dead and the current limiting low Ohm resistor and fuse blown on the flyback supply rail. The transistor was of some exotic part number, no big online parts suppliers stocked it, but found a small supplier online who actually was selling a kit that included the transistor and current limiting resistor, so i ordered.
Replaced the failed parts, switched the TV on, worked great.... For 5 min or so, then went kaboom again... Same fault, the transistor, resistor and fuse. This time i resoldered the PCB copper around the flyback transformer and all power tracks, and bought another replacement kit (actually two) from the supplier guy. Repaired again, switched the TV on, worked great... For another 5 min, then blew up again... WTF...
I still had a spare kit, so this tome i carefully inspected the PCB tracks for cracks around the flyback unit and the power supply, and resoldered the flyback transistor base driving circuit. Put new parts in again, switched the TV on and in a few min it went dead again with the same fault. That was enough for me and as much as i loved the TV, it had to go. The board had some nice power parts that i wanted to save, so i took the TV apart, cut the wires, unsoldered the wanted parts , and while holding the PCB in my hands ready to be tossed to the junk bin, i noticed a penny coin stuck under the flyback transformer, barely touching the transformer legs.... :palm: The coin was effectively shorting the supply rail to the collector of the flyback transistor. Apparently it fell through the vents from top of the TV casing and found its place under the transformer. It was too late to put the TV back together :rant:
-
I recently replaced an old metal can capacitor and forgot to ground the negative lead of the new cap to the chassis. This caused another smaller electrolytic to become reverse biased. Luckily I felt it get hot before it blew up. I thought I had That smaller cap in backwards. So I flipped it... Then I found the lack of ground issue, and ruined the second smaller cap when it ended up getting reverse biased once the ground on the if cap was correct. *sigh*. Two $2 HV caps wasted. Oh, well, didn't damaged the scope I was working on.
-
As I was reading, I feared worse. Of the possibilities, having set the current limit too low is the preferred error. :-+
I agree. Yesterday I wanted to see how much current a thin wire can hold without getting extremely hot. Unfortunately my high power PSU got a programmed default to 0V and 80A with the output enabled. I had to connect a new wire after entering the voltage because the first wire went up in smoke. Needless to say I reprogrammed the startup defaults.
-
Just after I started at HP Melbourne I picked up a storage (Analog) oscilloscope (HP 1727A if I remember correctly) and just plugged it in to start diagnosing. It seemed to me that the CRT came up very fast indeed. Turning the intensity down didn't save me - it burned the storage mesh pretty well. It was about then that I checked the mains setting - it was set to 110V and I'd plugged it into 240V. And it was then that I realised that we were a regional repair centre and received items from south east Asia for repair.
So I had to 'fess up to the boss. He wasn't happy having to swallow the cost of a new CRT and I was left in no doubt I wouldn't be forgiven a second time.
And now, 35 years later and living in a 110V country I *still* always check the mains settings!
-
Many years ago, while an avionics tech in the USAF, I was working on an instrument in a B-52G that had 28VDC power for the lighting. As I was working on it I leaned in and my upper lip came in contact with the steering yoke and the 28VDC went through my lip into the yoke and felt like a knitting needle had been pushed through my lip. VERY painful...
Brian
-
I put the ground clip of my scope to the 5v rail of a circuit without thinking. Normally that would probably not be THAT bad, however that 5v was fed by an ATX PSU, which can deliver an incredible amount of current on that rail. It did not explode, it did not make a noise, it just got really hot and smoke immediately started pouring out of the actual wire. I immediately realized my mistake and shut off the PSU. The wire itself was too hot to touch.
I'm just glad my scope is ok. That probe however, does not work anymore. :P
-
I put the ground clip of my scope to the 5v rail of a circuit without thinking. Normally that would probably not be THAT bad, however that 5v was fed by an ATX PSU, which can deliver an incredible amount of current on that rail. It did not explode, it did not make a noise, it just got really hot and smoke immediately started pouring out of the actual wire. I immediately realized my mistake and shut off the PSU. The wire itself was too hot to touch.
I'm just glad my scope is ok. That probe however, does not work anymore. :P
did a similar thing, but mine was a 200v 5kva transformer with one output earthed. Which I didnt realize, otherwise I wouldnt have my scopes gnd clip to the other side. Killed a PC, some USB gear, and most of my nerves. The 15000$ scope did luckily survive.
-
I've shorted my scope to a power socket several times. Never blew up the probe though, usually just blows up what it's connected to.
-
had just successfully repaired ATX psu, and prepared for a first powerup open frame at my desk. Switched on AC, lights coming out with a loud bang, home fuse tripped, entire house dark, family in panic (they know me). After disconnecting it with a flashlight and getting back electricity I had to realize that I forgot to put the loose earth cable aside. Found it right underneath the PCB. :palm:
-
This just in......if you are testing your latest vintage test equipment score read the manual or study the faceplate before assuming something is not working correctly!
I was trying out a Fluke 8010A meter at lunch time. It is a 1985 vintage non-autoranging DMM. There is also an 8012A version of this meter which shares the same basic faceplate and the legend for the switches. I could not figure out why the higher ohm ranges worked but the 2ohm and 200ohm did not, I would get an -1 open circuit reading, or maybe '000' shorted reading (non of that decimal point Tom foolery here). I looked at the schematic and tested the input resistors, a transistor, MOVs and PTC. I then looked at the switches again and noticed the two lower ranges said 20ohm/2mS. What the? So I read the front of the manual, the 8010A only has kOhm and higher ranges, the 2mS nonsense was for conductance testing (you press in two range switches at a time to select 2mS). The 8010A also has a 10A range. The 8012A has a low ohm board in it but does not have the 10A range.
After learning how to use the meter it now seems to work fine. Embarrassingly I used this same meter for 4-5 years in the late 1980's/early 1990's. Doh! :-DD
-
Quite some time ago, was doing some layout work. Wasn't quite that experienced with Altium yet. Started from a 6-layer template but I only needed four. Instead of getting rid of the layers, I modified the output settings to not generate Gerbers for the unused layers. Worked nicely. The project was pretty much done and I left on Hollidays.
Off course, there was "one more change" to make. I had a laptop with me, so I set up the project again, did the change, regenerated the Gerbers and got on with drinking Ice-Tea and doing nothing.
Unfortunately, I had to use a different Output Job file. So the 2 unused layers got to be the midlayers. The board went in production and pretty much everything was fused together (every single via through those planes was shorted together). The customer bailed out (had another bloke push the button to generate the gerbers and went in a "first time right" production), he shafted me with the production cost and didn't pay a single bill.
To this day, I can't believe nobody at the board fab house cought this.
-
So today I got 5 more Arduino Nanos, and 5 RS485 adapters, to further my model railroad control project. This set didn't come with the pin headers soldered on, so I set about doing that, and then went to test each one. These came with the basic Blink program already loaded, they each merrily blinked away when connected to the USB cable. I then uploaded a modified Blink program and indeed they all took the program and executed it. Odd thing, they revert to the basic Blink when the USB cable is reconnected. So I decided to test with my bench supply after uploading the code. Hooked the first one up, the LED blinked some odd pattern, not the basic Blink and not my modified version. Strange. Tried #2 - same thing. Hooked the USB cable up, reuploaded my program, it works. Disconnect the USB cable - it dies. Then I happened to glance up at the power supply...
:palm: :palm: :palm: :palm:
Last time I was using it, I was testing the low current limits of the circuit I was working with, and I set the constant current limit about as low as it goes, sub 1ma (resolution of the readout is only 2 decimal places - HP 3610), and that was obviously too low to drive the Nano, as it had switched to constant current mode and the voltage being applied to Vin was barely over 3V. No wonder it was going wonky. Reset the current limit - each one tested fine, it retained my program in memory. But they still revert to basic blink if I connect the USB cable with no other power applied - with power externally sourced, it retains the last loaded program as it should. But a big self :palm: for having the power supply set for a ridiculously low current setting and not seeing that.
Bonus, got to try out my little fume extractor, and use my soldering station for the first time since setting my bench up.
This reminds me of what happened during testing one of our companies new products, I was not personally involved though.
So several of the units are all set up in the thermal chamber to run and at some point during the run they all get an error light at the exact same time; they were left running so no one was there to observe at the exact time. People are trying to figure out exactly what happened and why; they all get reset and it all appears to be working fine without a clue as to what happened. One of the senior engineers (who had been working there for 15-20 years) shows up and first thing he checks, the power supply and as it turns out, the current control was set too low. When all the heaters in the units kicked on during cold at the same time it drew too much current at that power supply setting. Said senior engineer then lamented about why it was a stupid/bad idea to use a power supply that didnt display the settings; but you know, getting that sorta thing would cost alot of money.