The case of the mismarked diodes

11/1/2010 11:08 AM EDT

Same here with diodes, also aluminum electrolytic cap with + at wrong end (the case crimp was the clue), mylar caps with value stamped wrong by an order of magnitude. It happens.

Also annoying were those axial resistors whose paint markings were a cross between brown and red and orange.

Errors not just on components, also occasionally in data sheets. Wasn't it the LM7905 that first appeared with a couple of leads swapped in the lead ID graphic? And a few other data sheets with u and m mixed up on graph scales. Microamps vs milliamps.

Always expect the worst of the unexpected. Then you will not be disappointed.

11/1/2010 5:39 PM EDT

What's that saying? Never assume... it makes an ASS out of U and ME. One piece of advice I seem to ignore with monotonous regularity.

Talking of 7905's...I once generated a PCB with LM317's and assumed (there's that word again) that the leads were the same as 78xx regs. Won't do that again in a hurry.

A wrongly marked electrolytic would be fun....did it let the smoke out??

11/1/2010 9:27 PM EDT

I have never encountered before, but mismarked diode sounds like a dangerous affair! Ahh regulators, yes a few times. Not all regulators are the same as 7805.

11/2/2010 1:17 PM EDT

I can imagine that mislabeling may have occurred more often in the past, when automation was not as prominent, but it occurs as infrequent as a runt coin is minted. Rare, but these occur.

In my past experiences, I remember an occasion of a DIP resistor network with a common pin 16, that had minimal labeling and it was miss-installed in the assembly process, and generated a interesting bug in an address bus, that was not detectable during early test efforts.
Many hours were spent due to a tiny ink dot!

11/2/2010 11:22 PM EDT

I once had the pleasure of watching a huge board full of ASTTL "cook off" all of its electrolytics, each a flaming little fireball, because they were all marked backwards. Yeah, I caught on after the first few, but what the heck, they were toast anyway, let's enjoy the show! One of the many Heathkit color TV's I built over the years had a backwards-marked diode supplied with it. It was a credit to Heath that even as a high school kid I could use the included VOM and their remarkable documentation and figure that one out. Got it working and celebrated by watching Planet of the Apes on it. Ah, good times.

11/5/2010 12:37 PM EDT

Mnay years ago, I had a part time job in high school doing electronic test and calibration for an instrument manufacturer. On my first day, the first unit I tested blew the line fuse. I traced the problem to a mis-marked diode in the power supply rectifier. There were no spares at the test bench, so I had to ask my boss how to get a replacement. He wouldn't believe me that the diode could have been mismarked, and assumed he had hired an idiot. However, after a demonstration with a multimeter, he was convinced. When I came in the next afternoon, there was a big package of spare diodes on the bench. Apparently the morning shift had been replacing quite a few, indicating the assembly house had gotten hold of a batch of diodes with a significant percentage of mis-marked parts.

11/5/2010 4:06 PM EDT

Oh yes... 1N914 diodes. I've seen this multiple times over the years.

11/5/2010 4:57 PM EDT

Not just components! I had designed a mutichannel scanning board for monitoring components during thermal testing (they were special copper wire wound resistors for the Minute Man III missiles which were having intermittant shorts or opens at random). I had designed the circuitry but since we were in a hurry, the schematic was sent off to the PC board manufacturer for them to do the layout. When the boards came back, I wasn't there that day, one of the techs started assembling a board, which had over 30 CMOS digital chips and some analog circuits plus the regulators for dual supplies. He wired it up according to my drawing and fired it up....literally! ICs were popping like popcorn along with the regulators and electrolytics before anybody bothered to shut it off. They tried to do a post mortum but couldn't find anything, finally calling me and telling me that my board had gone up spectactularly.

The next day when I got back in, I took a look at the board, the ICs (what was left of them) appeared to have been put in correctly as was the regulators and capacitors. I checked the connections from the power supply to the edge connector on the board, the wiring seemed to be correct as well. ???? What the hey!

Checked all the PC connections on the board for the regulators and the IC buses, nothing, then I checked the PC traces from the regulators back to the edge connector....bingo! The PC board people had switched the plus and minus connections for all the buses! No wonder the Fourth of July came early!

11/5/2010 7:27 PM EDT

I worked for a company developing a laser based fiber channel communication interface. Our vender who supplied our ECL based interface IC provided the specification sheet with pin out so that we could design the PCB for the interface.

When I assembled the PCBA and began testing it, the very expensive ECL interface IC immediately burned up. I was horrified, as the part cost over $400. I carefully reviewed the PCB layout and the interfacing components. Everything looked correct. I replaced the IC and tried again with a low current limit on the power supply powering the ECL chip. When power ws applied, the part burned out before the power supply went into short circuit shutdown.

I contacted the vendor, who we were partnered with to develop the design, and they agreed to send one of their engineers to investigate the problem.

After he reviewed the design and the PCB layout, he removed the IC from the PCBA and replaced it with a new part that he had brought with him. It worked perfectly!

We were both confused. Upon closer inspection, he realized that the chip that he had brought with him was a current production part. I had received preliminary parts. His part utilized a ceramic cavity down package. The parts that I had utilized a ceramic cavity up package. They both had the same die but by using a a cavity up package, it effectively flip the die. The pin out for the cavity up mounted die was reversed from the pin out for a cavity down package. This resulted in reversed power and ground connections.

11/5/2010 10:14 PM EDT

Diodes with the bar on the wrong end used to be more common 40 years ago. It was a known problem when I worked for some potted module manufacturers.

11/7/2010 4:54 PM EST

My father was a technician in electronics, and once he told me that he had suffered a problem with a mismarked diode. Then we both thought it was a rare case.
After some time, when I was working in the first company I worked, I found again a mismarked diode in a board I was developing, for my surprise.
I am sure that, even knowing now that they exist, nobody will think of mismarked diodes the first time one of them causes a problem in a circuit.

11/8/2010 6:07 AM EST

David, can you share some stories of how your colleagues got their revenge on you for giving them Surprise Diodes?

I once mixed up several hundred 1N4148s and 2.4V zeners. The ink markings on glass were virtually invisible. When I got in to work the next day the entire workshop had been wallpapered with home-made cartoons bearing slogans such as "Armpits sweaty with embarrassment? Try new Zener Dioderant!"

11/8/2010 8:24 AM EST

Somewhere around 1990, I was working in Tucson, designing a 200W power supply. The supply quit, so my technician repaired it. Flipped the switch and the fuse went like a flashbuld. Pulled out the bridge rectifier, and applied AC. All ok. Applied DC to the switching section. All ok. What? Replaced the bridge with another new one. Repeated above. I checked the bridge with a DMM, and all four diodes were good. Scratched head. Paced. Had the occasional flash of inspiration, and got the data sheet out, since I can never remember which diode connects how. Turned out that 25% of the bridges in our new part bit had apparantly been rotated 90 degrees before potting, or before marking. Fortunately, all the same date code.

11/8/2010 12:39 PM EST

About a year ago our manufacturing engineer needed -5 volt ECL voltage controlled crystal oscillators (VCXO) for a production run of optical fiber receiver boards. These were a legacy design from an acquired company, and the package style, negative Vcc, and frequency response polarity of the VCXO were getting hard to obtain through regular suppliers.

He was overjoyed to find about a hundred of these in the warehouse, leftover new stock from the original company. All were marked 40.960 MHz and the devices and packaging were labeled with the correct part number.

Fortunately the first test run was for only 15 boards. We quickly found out why that batch of VCXOs was stashed in the warehouse instead of regular parts stock. All had been cut 10 KHz too low, measured about 40.950 MHz at center tuning voltage. Somebody years ago had received these mis-marked parts and held onto them instead of throwing them out.

11/8/2010 1:45 PM EST

I had an experience with mis-made parts, only it was with the lowly one-watt carbon composition molded resistors. As a high-school experimenter I frequented the local electronic stores and surplus junk stores. One day I spotted a bin of resistors advertised for "a penny a piece." They were painted white-white-blue-silver, gee, 88megohms, I thought they stopped at 20megs. So I got a few and experimented a little, the meter did not even twitch (a real analog Heathkit MM-1, digital had not yet been invented) and the first graduation would have been for 20megohms anyway. So, how do they make an 88meg resistor anyway. Cut one open with the wire cutters and saw nothing, the molded case had no carbon in the center. Grinding one lengthwise to a semicylender showed the same result. Apparently the little length of carbon slurry was not injected into the case and it was just the insulator case and the leads. They had essentially made a non-part. I went back and bought a big scoop and kept them as a conversation piece and as a trick part for the past fifty years. They did make good coil forms for RF work.

11/11/2010 10:59 PM EST

My experience is a recent one with a voltage reference ic ADR03BR IN soic-8 package. When this IC was announced ( 3ppm/c tempco was attractive) by ADI, we had waited for nearly 3 months to get them delivered through our vendor in US (not an authorised agent of ADI). One fine morning, when I started assembling these ICs on the PCB and tested, 1st IC was OK,further assembling on PCBs things did not work. Desoldered and wired with new IC and continue to have this problem. Did not want to spoil the entire day, assembled the IC other way ( pin#1 to pin #5..) and when tested it was OK!!! Wrote this incident to ADI concerned engineer, who straight away disowned responsibility in his/her email as they were not sourced from authorised vendor!.
Where is the mistake..? Giving negative feedback is not a good experience....Spoil them with positive feedback?

11/18/2010 11:42 AM EST

I once remembered we had virtualy 100% failures after a new disk drive design was released to pre-production overseas.
Lots of phone calls and emails that week.
At the end a very smart hands-on-engineer over there discovered the MOSFETS in an H-Bridge configuration were marked wrong from the supplier and the N and P channels were swapped (both TO-252 surface mount).
After that expererience, I learned that suppliers can (very rarely) make mistakes.
Up until then, I use to think suppliers never made mistakes, only we did.

...Made for lot's of smoke on every board.

1/25/2011 4:32 AM EST

The strangest one I've seen was a single 2114 memory normally in a DIL16 that had 4 extra pins like a DIL20, but the plastic was DIL16. For fun we tested it, and if the extra 4 corner pins where ignored, it worked fine.

2/8/2011 2:43 AM EST

I also had the 7805 regulator syndrome quite a long time ago. But much more recently (last week) I was hit by what could be the "revenge" syndrome.
My colleage PCB router handed me the long awaited PCB board that he routed for me. He told me that the SMD components were not soldered because of some organization failure with the external PCB maker (it's always like that when you're late in your projet).
I soldered the power supply components and a few other that are critical for the project and needed early testing before the rest of the PCB batch arrives with all components on them (due for end of this week).
Didn't work at all.
It's strange how my colleages look funny when they come in the lab.
AAARGH !
one of the internal layer of the 4-layers PCB was resersed by the PCB maker. You could see it in the usual 1234 little square boxes in one corner of the PCB. But I didn't pay attention to it.
The PCB maker had informed my colleague when he receptioned the failing board, but my dear colleague didn't inform me... Maybe because of all the "few" components I asked him to add to the design after he had already routed 90% of it...

2/8/2011 2:45 AM EST

read "reversed" internal layer

2/8/2011 3:10 AM EST

every I was very encouraged to find this site. I wanted to thank you for this special read. I definitely savored little bit of it.
.
www.healthmantra.co.uk