Techie solved supposed software problem by waving his arms in the air Welcome once again to On Call, The Register's reader-contributed column that each Friday serves up your stories of biting into half-baked tech support problems. This week, meet a reader we'll Regomize as "Charlie" who once worked as field service engineer for a company he told us "installed computers for various large …
You nearly certainly use the ICAO/ITU Radiotelephony/NATO Spelling Alphabet (Alpha, Bravo, Charlie, Delta, etc), not the WWII Allied military phonetic spelling alphabet (Able, Baker, Charlie, Dog, Easy, etc) that was referenced in the post you're replying to.
Oh please ....
Anton, Ärger, Berta, Cäsar, Dora, Emil, Friedrich, Gustav, Heinrich, Ida, Julius, Kaufmann (not Konrad as many use, since that can be written as Conrad as well), Ludwig, Martha, Nordpol, Otto, Ökonom, Paula, Quelle, Richard, Siegfried, Schule (Sch), Theodor, Ulrich, Übermut, Viktor, Wilhelm, Xanthippe, Ypsilon, Zacharias.
Anyone here knowing the French, Italian, Spanish, etc versions?
There used to be different ones between the Royal Navy, the British Army, and the Royal Air Force until they standardised on the US one for convenience during WW2, and later the NATO phonetic one which is well known to most these days - the NATO one is designed to be said understood in many languages on frequencies with lots of "noise" on the line so should be readily understandable.
This post has been deleted by its author
This post has been deleted by its author
When my daughter was about four or five years old, she was very fond of a Richard Scarry book (large picture book with cute anthropomorphic animals) with a few pages describing the adventures of a mouse named Able Baker Charlie.
Oddly, he did not have a dog named Easy, nor was he pursued by a fox named George.
We had a problem that when someone sat in front of a computer, the screen had problems. It was due to a floor tile squashing a cable. When you wheeled the chair over the floor tile - the screen went funny. When people came round to fix it - they didn't use the chair, but stood around the floor tile, and so it worked.
I've never understood people who run wires/cables under floor tiles, especially carpet tiles. I suppose it's in "out of sight, out of mind" sort of mentality, or maybe they think are being "good" by eliminating a trip hazard. I couldn't count the number of offices where the networking "issues" were caused by Ethernet leads under carpet tiles, being walked over etc and both outer and inner sheaths being shredded. I've seen mains cable run that way too for extension leads. I always point out verbally and in my service report, which they have to sign, that I've noted the safety incident and reported it. It's then 100% on them to deal with as an urgent matter. I have, on rare occasions, refused to work on kit connected that way until they sort it out because there's bare copper showing, carrying 240V. I've had a few sarcy comments but as a 3rd party I just point out I can't do anything other then my legal obligation under the Heath & Safety At Work Act to report it.
We had a problem once with an airport-based customer complaining that when they were printing from our software (using Okidata dot matrix printers) it was fine a lot of the time, but was intermittently inserting garbage into the printouts.
They swapped out the PC, the printer, the cables, but still no joy, until we sent someone to site to diagnose it.
Turns out the serial cable from the PC they were using was running next to a baggage injection belt at a check-in desk - and if someone activated the injection belt while a print job was on the wire, that was the cause of the data corruption. Bought a network card for the printer and chucked the serial cable in the bin.
Twisted pair has an inherent immunity to such things - the induced voltage applies to both legs and as such cancels out. The network interfaces are also galvanically isolated (each pair drives a small transformer)
It's not completely immune, but it's a lot better than you think without any shielding.
> but the shielding has to be grounded to be effective
Which raises another problem: Grounding / Ground Loops via LAN cable shielding, instead of the fat cables normally used for that. Coworkers found suspiciously warm LAN cables, and a clamp-meter confirmed the issue. One of the big big advantages of optic cables: No weird grounding and other electric effects to care about.
Which is technically correct and incorrect at the same time. Now, get CAT6a/CAT7 cables, hundreds of them, with plugs missing the metal contact - best only on one side. Then a few switches, at several places of the building. Have fun diagnosing why it is so weirdly slow.
The only way to get this right is to have all rooms, switches and racks involved a proper grounding throughout the building. And when not possible FC, which you need then anyway since those with different groundings are usually bigger distances anyway.
Several employers ago, I had to support some test gear for a high voltage product. The control and sense lines were high impedence, of course. The tester would flake out randomly, sometimes causing the HV side to destablize in a puff of magic smoke.
The intermittent nature of the problem was made worse by the fact that this tester wasn't located at our main campus, but at a rented building some miles away. The building wasn't on airport property, but it was fairly close. That fact wasn't interesting until we noticed that if the test was running around the same time a flight was arriving or departing, failure was more likely.
This was a small airport with 2-3 commercial flights per day, so our theory was that the ILS wasn't always running.
I remember a case on Victors VASAviation. Emergency near such an airport with low hanging clouds at night and lost radio (could send, but not receive, defective ILS on the airport). Another chessna-type plane currently near the airport sent such a ping signal to turn on airstrip lights, so the mayday plane could see the airstrip. I think either Mentor Pilot or Kelsey mentioned such a case too, with ILS.
So this is true for low traffic airports.
... to fix a similar problem after a similar fix had been applied.
Burned out one of the network cards the instant the "ground strap" was connected. And then a second network card, which is when they called me.
The motherboard catching alight tipped the geniuses off that there might be an issue, or their may have been three dead cards. Or six.
Be very, very careful when tying equipment grounds together, even within the same building. Things are not always as simple as they seem.
powered from different phases of a three phase supply... yeah... not good
YMMV depending where you are, but here in blighty it makes SFA difference - most large offices have stuff spread across multiple phases. The biggest risk is actually to maintenance people since it means that (for example) there may be 415V between two wires in the back of a multi-circuit light switch rather than just 240V. The earthing is shared, as is the neutral (but the neutral is considered a "live" connection anyway in terms of insulation etc. required by regulations).
What DOES matter on a large site is earth differentials - the bigger the site, the more "earth" at one point can differ from "earth" at a different point. Within one building it tends to be limited as they'll share a common MET (main earth terminal), but when you get to multiple buildings, they can have separate feeds from the incoming switchboard. So, within one building, a fault on a circuit will momentarily put voltage onto the earth for that circuits and also the MET - but as everything shares the same MET then that's not a problem. But when there's a fault in one building, all the (e.g.) computers in that with serial links into another building will see that MET voltage imposed onto the serial lines - and "poof", the magic smoke is let out.
For best results, have a lightning strike to ground close to the site - that creates massive voltage gradients, and allows the magic smoke to be let out of equipment within the one building as well. That's where (as mentioned) optical isolation is "a good idea".
It's also worth remembering that when lightning strikes, the rate of rise of voltage and current is so fast that thick cables become high impedances. So it's no good fitting a load of surge protectors - and having a bonding cable from them going to the MET at the other end of the building, the inductance will stop it doing much when lightning strikes.
Differential drivers and receivers are useful, bot without optical isolation the magic smoke escapes if the common-mode voltage exceeds the chips' limit -- typically, 25 volts if I recall correctly. Grounding and bonding are great, but all bets are off when the building gets hit by lightning, or even a near miss.
Once lived in a long narrow house. Had a couple of long runs of CAT cable from a switch at the front of the house to devices towards the back of the house. One day a BIG thunderstorm came, sheets of water running down the windows (Asia where they know how to do thunderstorms properly). Then there there was a mighty flash and a simultaneous boom. If the house wasn't hit then something very nearby was. The devices at the back of the house were taken out. No smoke but no more comms so dead. Annoying because one of them was the internet router and the other was a first gen ATV, expensive at the time. The switch lost the use of those ports.
I learned about ground loops the hard way in high school basic electronics class when the mains-powered AM radio I'd brought in from home to fix, blew up in my face as I touched the oscilloscope probe to a trace.
BANG!/FLASH!/smoke. When I could see again, I observed that the radio now had an additional problem: about an inch-and-a-quarter of the PCB trace I'd touched with the probe had blown free from the underlying printed circuit board.
We'd not yet covered ground loops in class, and as the quote goes, "I feel I have been denied critical, need-to-know information."
I learned about ground loops back in the Xbox 360 days. I used to mod the DVD drive firmware through the SATA of a Shuttle PC. When you did this you had to ground the 360's metal chassis to the PC case (in my case using a metal coathanger) as the Xbox used a floating ground inside its PSU, there was no ground pin on the kettle lead to the mains socket. The DVD drives had to be powered from the console as they used a proprietary connector for power. One day I hadn't realised the coat hanger had slipped out. The SATA ports on the Shuttle I used for modding stopped working permanently, only the IDE worked from then on!
The Shuttle soldiered on for a good 6 years after that, even without working SATA!
Me too when on the Rockwell Collins factory floor in Texas, I attached my ground probe to our box. I knew that we used a simple power resistor to drop the 48v to 12v, (this being a positive ground in telco installations). But, yeah did it before I could think, effectively bypassing the dropping resistor. Blew up the box o' CMOS.
Back in the 80's my company used to design and manage all sorts of serial interface convertors (The one in the Black Box Catalogue were ours). I would have recommended RS232-485 convertors at each end as a proper solution. Ours were also opto-isolated, which would have been perfect in this case.
We also used to do protection barriers should the buildings suffer a lightening strike. I would have been tempted to upsell a few of those too!
"Be very, very careful when tying equipment grounds together,"
There have been a few instances of musicians being shocked or even killed by grabbing a "hot" microphone whilst holding their guitar. In the case of IT equipment, the comm cable takes the place of the lead singer. And many I/O ports are no more robust than they are.
The only band members that seem to fare worse are Spinal Tap drummers.
This is a well known issue with any process that generates large amounts of dust with rapidly moving materials. Printing presses are very similar. People are often surprised at the amount of paper dust that is created. Quite substantial cross-bonding and a good earth is pretty much obligatory.
Oh, and you also get the double whammy of elevated fire risk, and breathing problems for the workforce.
This, kiddies, is why you never shove important data down a wire without some form of active flow control - even if it's something as simple as breaking the information into 128 byte blocks (or whatever is convenient) and then sending a checksum.
Smarter protocols may combine that with some degree of built in error correction, but whatever, assuming that what goes in matches what comes out is a rookie error.
Yep, which is where things like CAN bus come in handy - it can handle ground differences and all transfers are validated and re-tried if there are any errors (so the receiving end only ever gets valid data*).
* Nearly - it is theoretically possible for the CRC check to pass a bad frame, but it's so unlikely that only a few system need to worry about it.
FYI CAN the bit stuffing (to aid in clock recovery) can destroy the CRC characteristics such that what was supposed to be something that detects all 5 bit error can only detect all single bit errors. Slide 56 https://users.ece.cmu.edu/~koopman/pubs/koopman14_crc_faa_conference_presentation.pdf
So CAN is not a great example as it is arguably inferior to RS-485 with parity detection. CAN does have advantages in terms of acknowledgement and bus arbitration.
Kermit FTW.
Ah, happy memories of cat'ing Kermit source over serial to another machine, fixing the few glitches in transmission by hand, compiling and then using it to pull (or was it push?) all of the the actual project's code & data onto what then became the new platform.
I recall returning from a DECUS* meeting in central London on the train, clutching a precious tape with Kermit for VAX and PDP-11! It was my go-to for several years, initially for DEC - PC transfers and later for PC - PC before more sophisticated programs became available.
*Independent DEC user group
I can think of one very popular backup package which relies on TCP acks to provide the totality of flow control between the backed up device and the tape spooler - including when the spool is full - meaning that data is sent until the network buffers are choked and that in turn can trigger a Linux memory leak (which nobody seems interested in fixing) if you've set the buffers LARGE to get maximum throughput (Tape drives are FAST, faster than disks)
Suggesting that they use actual end to end flow control handshaking was dismissed as "unnecessary" because "It works well enough"
Swiss-German managers can be extremely pig-headed
would know what a multimeter was let alone how to use one.
Most software types I had to deal with happily dived into their expensive proprietary workstations without a wrist strap/ground plane. Offering one of the Sun carbon coated paper straps from my hoard invariably was met with blank stares.
More recently I imagine those also into the "maker" culture are likely to be a lot more cluey about this stuff.
(Dust explosions aren't unknown in bakeries. :)
Bread factories are notorious dusty, helping sparks to develop and reach further than without. So I expected a lot of static, depending on how and when the flour is handled you get your little fun spark into your signal.
Grounding works, but I expected switching to optocouplers as separation here, on both sides :D.
very very frightening *
This is the sort of issue that used to plague my life... where the PC was grounded via the 240 power supply system and the machinery was grounded(if at all) by a grounding post or using ground in the 3 phase supply.
Now a decent set of opto-isolators in each RS232 cable worked wonders... however , this is el-cheapo industrial control land we're talking about.
And so when a lightning bolt hit the nearest lampost , there'd be a ground surge... so the machinery is at a different ground voltage to the PC.... and that does'nt do the RS232 interface any good at all.
If you were lucky it would be the PC that took the shock, so it was off to Maplins to get another plug and pray RS 232 card... if the machine took the hit, then you'd get the engineer out and then he'd get the bill out to change out the main control board.... **
* if you sang that we can be friends
**At this point the rest of the lyrics spring to mind...
He's just a poor boy from a poor family
Spare him his life from this monstrosity
Was the equipment, both client's and telecom's, keep breaking and there were lots of theories about lighting strikes as it was often found system wasn't working during thunderstorms.
After digging through the specs, it turns out that yes V11 (RS422) can run on a 1km cable, but the bit they'd overlooked in setting up this particular link was no more that 7v ground variation - one end in a manned building, the other in an an unmanned building several hundred metres away with rather extreme earthing. Needless to say chances that 7v being exceeded was high, telecoms switching to fibre and it went away.
Why people thought thunderstorms had much to do with it - because that is most likely when they'd need the system, so was finding it broken.
Back in the 80's when I was writing 8 bit games we were close to release of a big coin-op conversion but there was a persistent report of it crashing randomly from the testers. Very occasional but they could always eventually get it to crash but with no discernible pattern. However, we absolutely couldn't reproduce the problem. Went back and forth like this for days, maybe weeks, without resolution. Xmas was approaching. We tried to convince the publishers that it was bad electricity at the testing site but they weren't having it. Eventually did get to the bottom of it and indeed it was a software error.
If its the one I think it was then it was the B register getting overwritten as an occasional side effect of a subroutine call. Solution to temporarily store the value in a memory location with the very appropriate label of BSTRD,
On summer break while in college, I worked for a company that made filter tow for cigarette filters. Specifically, I worked for their customer support group which had 1 of each machine their customers had to make the actual filters in a room. Sometime before I worked there, a salesman had sold them a strain gauge so they could measure from the elevated boom when the machine picked up more of the tow due to how it was packed into the box. They had abandoned the equipment, removing the machine part the salesman had installed the strain gauge on from the machine and replacing it with an identical part. They did so because they were getting nonsensical data on the computer, and asked me to determine the problem. I re-installed the part with the strain gauge on it and attached the serial cable to the PC where the associated software was installed. I then turned on the machine and saw the same nonsensical data they had seen. There were spikes in the strain gauge data the weren't associated with actual issues with the filter tow begin pulled over the boom. I looked around the room where some of the other machines were also running. Then I went into their shop and got a heat gun, plugged it in across the room where I could see the computer screen and triggered it. Sure enough, every time I triggered the heat gun there was an associated spike on the screen. I then got on a ladder and removed the electrical tape wrapping the connection from the strain gauge to the cable that went to the computer and saw that the salesman had used unshielded cabling and that the gauges shield connection was attached to nothing, making the cable a really great antenna picking up the AC spikes in the room when the many motors on the various machines started during their cycles of operation. I rewired the equipment using shielded cable readily available in their shop and the problem went away.
My late father worked for a large department store which had a mainframe payroll computer. Data input was normally done by a rather well brought up young lady. There were intemittent problems which only occurred when this particular lady operated the terminal.
It turned out that the lady in question had a penchat for silk underwear which was causing static electrical spikes when she moved arond on the chair.
I do not know how they found this out (unfortunately) but I believe that the solution was to ask the lady to change to a different underwear material!
One of the things about being an engineer is having a wider range of understanding than just your job specialty or what courses you took at school. I've met similar issues many times and as 'the firmware guy' everything is firmware unless proven otherwise.** Often, even if the problem is obvious, explaining it is difficult because people don't realize just how much they can abuse a link and get away with it. On one remote session involving me trying to explain the various ways signal crosstalk could be induced I ended up asking if they had a radio amateur on the staff, someone who may have been just a lowly tech but actually knew what on Earth I was talking about.
(**Often reinforced by the fix being a tweak to the firmware because that doesn't involve any hardware rework.)(We also get tapped a lot because, as the QA manager said to me once, "We're the only ones who is willing and able to actually look at a problem rather than just pointing a finger and walking away".)
I worked for a mainframe company that made “IBM compatible” machines. While I was away on a course, the other guys installed one of our machines to replace an old IBM one. On my return, I was asked to hustle over to the site, there was a problem and the customer was refusing to accept our system. The site was a library, not your suburban book lender, but a main library a large building with 4 reinforced floors to take the weight of all the books. The system had a lot of terminals, the classic IBM setup a controller connected to the mainframe and coaxes from there to the terminals. Our system was “Compatible” (identical), and one of our terminals had been located on every desk where there was an IBM one.
The problem was that when our terminal was turned on, the IBM one died, not a problem in the long run but the customer would not accept our system until it was fixed. A couple of things struck me, our coaxes had been laid alongside the old ones, and looking at the circuit diagram of the controller signal ground was not connected directly to ground ground. They were connected with a small value capacitor, I rummaged in my “Come in handy” box, found a bigger one, and, with a couple of croc clips put it in parallel to the original. Instant fix, everybody happy, my brownie point score increased, but the customers engineering manager wanted an explanation. “Electromagnetic radiation between the outer of the coaxes” l offered, he blanched at that “If you say radiation the whole staff will walk out” he said. So we agreed on interference between the terminals.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
