That is all
Friday has come around, reliable as ever, to mark the end of that most dreaded thing, the first full week after New Year. To celebrate and get you through the day, here's On Call, our regular instalment of readers' tech conundrums solved. This week, our tale comes from "Charles", who was working at a small business that sold …
Story set in Early '90s.
So NTP was just getting under way (RFC 1305 in 1992), but far from widely established. A bleeding-edge try-me-and-debug protocol might seem a less-than-obvious choice for such an, erm, immediate problem. Even if Our Hero was himself a Networking Man and had heard of it.
Not just "early implementations"... NTP is a method to correct for small drift of the internal clock, not to set the correct time under all circumstances. If your clock is wrong by more than some maximal amount (which may be implementation-dependent, but usually equals 1000s) then NTP will not correct the clock.
This even if NTP were available at the time of the story whether or not it could be effective would depend on what exactly happened to the clock under the influence of the kitchen machinery in question.
Besides, changing NTP settings in the organization might or might not be possible/acceptable just to fix one single computer under a vendor support contract (surely you know what I am talking about).
"if the clock was too far wrong"
I remember 2 problems
- firstly they wouldn't touch the clock if it was off by more than 10 minutes IF you could get a time from the server
- secondly if no servers were available they'd set the time to the start of the epoch
And then there was the issue of sensitivity to whitespace (tabs vs spaces)
Yeah, NTP is no good if the computer isn't networked - no indication that this one was, or, if it was, it was using an IP stack that included UDP - or if the time jumps further than the permitted clock skew that can be corrected by the service (which is a configurable setting in the NTP client implementations I'm aware of).
Back in the 90s, I lived in an apartment, and ever day at 6 PM, my TV and cable reception went to hell. Couldn't figure out why, until I also noticed that the guy downstairs came home each day at that time. He had one of the cheaper PC clones (with almost no shielding on the case) that he'd overclocked a lot, and it was emitting enough RF at just the right frequencies that it would hose every TV in the vicinity.
The problem went away when he lined the case with aluminum foil.
Even if you affect your real time clock (RTC), virtually all operating systems (even MS-DOS) have a separate software timer which is used by the software and only synced on bootup.
I am not aware of any operating system that ever directly passed it's RTC values to application software.
If any outside interference would actually cause your OS to count backwards instead of forwards, the time would probably be your least problem.
So I call BS on this.
As described at http://www.beaglesoft.com/mainfaqclock.htm the operating system's time is based on a count of the CPU timed interrupt signal. The interrupt is a function to pause application processing and handle some device requiring attention, or in this case just to tell time. However, the web page is to support a utility which deals with early and inaccurate timing runtime hardware by using either the CMOS RTC or "The U.S. National atomic clock" to provide time. In the circumstances, something like that might have been already installed on the machine.
Or, on the lines already suggested, something cues up the time to be added, 00000001, but the microwaves cause this input to be read as 11111111, which is minus 1 approximately ish.
Running forwards faster than expected due to outside interference from a strong electromagnetic field would be possible,
That's interesting. Why would it run faster? The reason I ask that is that the other day we noticed that a clock in our house, which is supposedly regulated by the DCF77 radio time signal from Germany, was 15 minutes fast, and getting faster.
I had a look online to see what could cause this, but found only articles suggesting replacement of batteries, which wasn't the problem here.
Is specifically getting faster a known phenomenon, or is it that the interference could cause the clock to run either fast or slow?
Because the interference from an external source would provide extra clock-pulses to the (very) sensitive input of the crystal oscillator...
As for your DCF77 clock... it probably has bad reception due to a low battery and thus not recieving the correct time...
(It olny tries to synchronize it's internal clock every hour...)
The internal clock that keeps the time is the one that's running a bit fast
"Is specifically getting faster a known phenomenon,"
When I was in college, there was a major power outage. The college helpfully cranked up some HUGE generators to keep the dorms running. In the wee hours of the morning, my roommate's alarm clock went off, saying it was time to get up. A few minutes of observation (the next morning, after getting the rest of the night's sleep) showed that it was, in fact, running MUCH faster than it should, something like double time. When the building was restored to normal power, the clocked worked fine again.
I've always wondered if it was such a cheap clock that it used the power-line frequency as its oscillator - and the generator produced "noisy" power.
@AC . More likely the generators were poorly regulated and running fast due to low load. Mechanical clocks have a synchronous motor. Many 7segment clocks have a zero crossing detector and divide by 50 or 60 depending on the country. both these solutions produce clocks which are highly accurate at least once per day, as opposed to a cheap MHz range XTal on an RTC which may lose or gain a couple of seconds a day.
It is sensible to use the power line as a sync source. The 50/60 Hz power line frequency is held to very tight tolerances, traceable to atomic clocks..
The internal crystal controlled oscillators, not so much. How often do you have to reset your digital watch, or kitchen clock because the time is too far off?
"The internal crystal controlled oscillators, not so much. How often do you have to reset your digital watch, or kitchen clock because the time is too far off?"
I have to reset my car stereo's internal clock regularly. I believe this is due to a slight drop in power every time you start the car, from ignition. Over time this half a second builds up and it's a few minutes out.
One scenario I can think of is that the clock is sensitive to the AC cycle of the power supply. For why it runs faster, I can think of one plausible scenario- you live in the UK and the clock is attuned to 50Hz power, but the generators were running at 60Hz for some reason- this makes more sense if the clock is analog and depends on spinning motors (for example, one of those old alarm clocks with a flip-over display). Alternatively, it could be that the generators were putting out a little more than 240V which is well above the acceptable range of some equipment, but still not enough to cause them to blow up immediately in a spectacular fashion. I've seen this happen with poorly made generators in the past, they'd pump out 260v of power despite being listed for 230v, and yet all the equipment connected to them seem to be running fine despite the power being 20v above acceptable range. Solution for the latter issue is to plop a AVR between the clock and the power supply. I don't know if there's any solution to the former issue except to build a converter to change 60Hz AC to DC and then to 50Hz AC.
Much like US visitors bringing their mains-powered clocks to countries running at 50Hz. They can usually handle the different voltage, but they soon notice their clock is losing 10 minutes every hour. (Or gaining 12 minutes, depending on if your reference is the clock in question or a local clock).
It said they installed "specialised accurate time cards" - it could the microwaves was changing the card's memory/registers and the computer time was being updated according to these new values. It would have to impacted something in the computer, probably clock/interrupts - but that may not have been meant clock going backwards
Maybe if the clock time is held in a value where max=N, and it receives N-1 pulses in a very short time, it will count up and wrap around, and the result is to appear to be counting down rather than up.
Like a wagon wheel (not the biscuit) appearing to roll backwards when it's actually going forwards.
Hardware clocks can't run backwards because the core BCD counters don't generally have the hardware to do it. What can happen is the hardware time chip might have been programmed to provide 60 interrupts a second and the real time was calculated based a interrupt counter but the the microwave was blocking some of those interrupts so it calculated a time that went backwards. That was common on early multi-processing systems.
Thats true of several TV broadcast relay stations.
The mast at Llanddona on Anglesey uses two sets of microwave dishes for the backhaul to the mainland, one for use during normal and low tides, the other for high tide.
Not sure if the same effect applies with tides, but (warning, long story from an Engineer follows):
Back in my college days, we had an upper division class on telecommunications. Our instructor was an adjunct faculty member (in other words, he had a "real job" in industry and wasn't a tenure track professor). One of his projects involved long distance microwave networks, including links over desolate low-lying swampy areas. In the swampy areas, certain atmospheric conditions (like calm and cold nights) could cause the air to stratify into layers that would diffract the microwave, effectively "bending" the signal. IIRC, they would transmit on one antenna, and they had a pair of antennas for reception. He said that you could sit in the communication shack just before dawn and watch the signal strength fade from one dish to the other as the sun came up.
Have heard of similar things in early mobile phone systems - one story in particular claimed that under certain conditions phones would connect to towers in different cities/towns/rural areas (not sure which propagation environment it was, suspect rural) and play havoc due to the timing offsets required to hit the framing of the phone signals over those distances (100's of km)
"certain atmospheric conditions (like calm and cold nights) could cause the air to stratify into layers that would diffract the microwave, effectively "bending" the signal"
You can see that some times to. We could see a long way from our old house. Pilot Knob was about 9 miles away and we could see much of the land in between.
One day I noticed that things near the horizon were stretched vertically. It was really surprising as I'd never seen anything like it before.
Back in the 80s I was told that during the Vietnam War, CMF soldiers on an exercise at Appin (just inland from Wollongong,NSW,Australia) were surprised to have a real contact in progress involving US troops come over their net.
All received on the classic "Pr*ck 25" (AN/PRC 25)
On a school trip in the 1980s our coach broke down in the Western Australian outback (east of Carnarvon). The driver got on the radio, and wound up being picked up by a truckie in New South Wales, who relayed the message to the coach company.
Not the same range as NSW to Vietnam, but wow!
"One day I noticed that things near the horizon were stretched vertically. It was really surprising as I'd never seen anything like it before."
You'd be surprised how frequently this causes UFO reports if there happens to be a light in the "things near the horizon" and it's dar or twilight, as it will appear to be bouncing around merrily with atmospheric distortions
"In the swampy areas, certain atmospheric conditions (like calm and cold nights) could cause the air to stratify into layers that would diffract the microwave,"
This used to regularly happen to microwave links across the English Channel and Irish Sea with fog layers on calm nights. It wasn't just diffraction. The layers of fog could act as quite effective waveguides.
Far less of a problem with fibreoptic cables of course - until someone drags an anchor.
Why doesn't the high-tide dish work at all states of the tide?
You can get fading due to interference between direct and reflected (off the water) signals, so that when one path is ok the other isn't. We used to have that problem getting TV across a sea path between NI and Scotland, reception varied between great and almost unwatchable as the tide rose and fell.
ISTR that back in the '60s Divis was relaying a signal from the IoM which was relaying a signal from Holme Moss.
That seems unlikely, Divis started in the mid-50s, and there was no IoM transmtter until long after that. People on the IoM west coast used to get their TV from Divis. Perhaps the microwave backhaul links to Tele House in Belfast followed that route? I have a vague memory of a mirowave relay on Snaefell.
Presumably you were in a location where you couldn't get Divis direct.
Indeed, although it was the 80s, not the 60s. There are some hills outside Newtownards that used to be mined for lead, they made a very effective RF screen if you were in their shadow, but the Cambret Hill relay on the Mull of Kintyre usually produced a watachable signal from Border TV. It was at fairly low altitude, hence the tidal fading Some relay transmitters were eventually installed around the area to fix the problem.
Back in the day, a convert to CTV from B&W found at night the colour faded away & returned some hours later.
Much heads cratching was done by the repair engineer on multiple visits & bench tests (Leaving her with a B&W loaner).
One day she muttered perhaps it's the tide while the engineer was there, who had his "Eureka!" CMOA.
Her TV aerial was aligned to a mast across a estuary & the received signal strength would drop dependent on the tide\skip conditions, realigning to a different relay mast solved the issue.
We had something similar where a radio link became unusable in the months following its installation. It turned out there was a tree inbetween the line of site that impacted the link during the summer months.
We classified the issue as a .... spanning .... tree problem ....
I deployed a chain saw to fix my home Free Sat reception. The prevailing wind eventually managed to push a bit of a fir tree in the way. It was felled at 2230ish and slightly annoyed my neighbours (I couldn't hear much through my ear defenders and was only mildly inconvenienced.)
"The sea is as flat as the flat earth we live on!"
The distance between Scotland and the Antrim coast is ideal for seeing that the Earth isn't flat. Looking from either one to the other you can see the other side but from near sea level you can't see the other coastline, just the tops of the hills. The big bump of sea gets in the way. In fact anyone living near the sea will be familiar with vessels disappearing hull down as they sail away.
A belief in a flat Earth is only sustainable by someone who's never seen the sea or even a sufficiently large inland body of water.
"The link passed across the bay she lived near and would fail at high tide."
This is why (in a previous life) when I was a comms tech we used vertical receive diversity on such links. The signal might drop out on one antenna but it wouldn't normally drop out on both at the same time.
Very similar story did the rounds when I was at IBM back in the '80s; the S/370 mainframe would suddenly going haywire every couple of months for no apparent reason. System software reloaded and parts replaced to no avail, the problems would appear, and then a week or two later disappear with no explanation. Until a tech looked out of the window on the day that the problems started, and noticed that the aircraft carrier USS Enterprise was in port. Turned out the mainframe was susceptible to radiation from the carrier radar array, and installing shielding in the server room fixed the problem.
I've heard this story in a number of variants many times since then, I used to think they were all retellings from the same source. But then; if one mainframe is susceptible to radar frequency radiation, it stands to reason that others are too. So it's definitely possible that multiple variants of this situation actually happened.
"But then; if one mainframe is susceptible to radar frequency radiation, it stands to reason that others are too. So it's definitely possible that multiple variants of this situation actually happened."
That's why propper data centres typically have at least some amount of shielding. Even building your building out of cheap aluminium helps a lot with that.
Back in the early 80s the company I was with had a datacentre (multiple mainframes) in an industrial unit backing on to the West Coast Main Line.
They had to install a Faraday Cage to keep the mainframes happy, because the electric locos heading out of Euston could produce some hefty electromagnetic fields.
Luckily I was never there when the fire alarm went off - the emergency exit took you straight onto the tracks!
There was an early episode on the Starship Enterprise which led to the ship generating a space-time warp field that made the ship's clock run backwards 71 hours. At the time it was "worry about that later" (er, how?) Later on they became increasingly proficient at time travel. The title for the original event is slightly NSFW so I'll just let you look up stardate 1704.2.
Reminds me of a story I was told about the local phone company. They had their IBM mainframe in a room on the ground floor of an office tower, that is located half way up the slope from a busy civilian and military harbour. One day a service rep from Univac approached the IBM site service rep and asked if they had any problems with the system being knocked down with radar from passing ships. Univac apparently had a system on another floor of the building and was having all kinds of problems with it being knocked over by radars. The IBM rep replied that they did not have any problems like that, however he failed to mention that when the computer room was built someone had apparently thought this might be an issue, especially since a harbour traffic control radar is installed on a small island opposite their building, and they had screened the room as part of the construction.
"Until a tech looked out of the window on the day that the problems started, and noticed that the aircraft carrier USS Enterprise was in port. Turned out the mainframe was susceptible to radiation from the carrier radar array, and installing shielding in the server room fixed the problem."
I live quite close to a river that, once upon a time, had a number of shipyards. If there was a ship parked at the bottom of the street, some, but not all, of the terrestrial TV channels would ghost due to the signal bouncing off the big wall of metal.
My partner and I lived in an apartment above our elderly and very deaf landlady. In the evenings she would use the remote to crank up the volume on her powerful Grundig TV to 11. It was so loud that we could not hear each other speaking. Her fusebox was outside of her door so we'd go down and briefly switch off her entire apartment, the TV would reset to normal volume and she'd be asleep and not notice the lights going off and on.
Have a nice weekend!
Taking a look at what a Radarange might be I found an interesting piece of history - that the first commercial microwave ovens came from Raytheon, who were looking at finding a new market for their magnetrons, no longer selling well after the end of the war dramatically lowered the demand for radar equipment.
And to think that in the military one of the demonstrations about the power and dangers of radar was to put a chicken (dead, de-feathered, etc.) on a long wooden stick and hold it in front of the air strip radar. Everyone got a piece of well done chicken to nibble on while contemplating the power of
this battle station the radar system.
I worked on a marine facility for boat repair. They upgraded their systems and specified a shed load of shielded VT220 terminals. They needed them shielded because radar was tested in the dock.
So, the equipment was installed, but somebody somewhere along the line changed it. I was an external contractor, so I don't know if the dockyard cheaped out or the supplier thought they could pull a fast one and supply unshielded for the price of shielded.
Anyway, long story short, an American naval ship was in port and tested its radar and... Zap, all the new terminals were dead.
Back in the early 80s we had a problem reported from one of or satellite offices in France. A girl's terminal connected to a mainframe in the UK was suffering repeated retransmits. The mainframe would send a screenfull of data followed by a checksum. Her terminal would verify the checksum and if it didn't match would request a retransmit. She said it was happening several times an hour. After a couple of days she stopped reporting it so we didn't pursue it.
But then the week after another girl in the office reported the same issue. She was more vocal so myself (programmer) and a comms engineer planned to fly out there on the following Monday with a datascope so we could view the traffic and try to work out what was going wrong.
When we got there on the Monday morning she sheepishly told us that the problem had gone away. Just in case, we hooked up the datascope on her line and waited. Nothing happened.
On the Tuesday a third girl in the office started reporting the problem. We attached the 'scope to her line and sure enough the screen data was frequently arriving corrupted.
One of the girls then red faced announced that she knew what was causing the problem. She had notice that the problem affected any girl in the office who was menstruating. Anti-static mats under all their chairs solved it.
I'd heard of computers causing gynaecological disturbance but not the other way around. Particularly through electricity.
So I am wondering about, let us say, a convenient product for the circumstance, that might either be wrapped in a plastic polymer with disagreeable electrostatic properties, or contain such a polymer.
Alternatively I'm thinking about French underwear, but I do have work to do, so I'll save that reverie for later.
Military Ship Radars are very powerful things, they should never be activated in port and being anywhere near the main array when active requires some serious lead underpants. Circa 1988 HMS Invincible is in Devonport Dockyard, the main radar array was accidentally activated for 2-4 full sweeps, at the time I worked for a Mini Computer company as a systems engineer, almost everything in line of sight that got a dose of the HMS Invincible radar sweep either crashed or had memory corruption. A very busy day, initially confused as to the cause of the issue, but as first system call was from Devonport Dockyard and the Radar Array activation was being talked about it didn't take long to connect the two events (And all the other calls coming in). I seem to recall there was some very serious disciplinary action taken on a number of the naval crew from HMS Invincible, it was a very serious matter and could have been worse for the crew and people nearby.
Back in the early 90s I worked at a fibre comms supplier and we had to run stability tests to check that the gear was working without errors. I don't know how it was originally found, but it was discovered that a plastic bag of coins, when shaken about a bit, would cause errors to appear.
Even now I'm unsure of the physics that made that happen, but suffice to say that subsequent generations of equipment are much better shielded!
I read "a bag of plastic coins", which raised many questions, such as can you get middle-age onset dyslexia!
I believe some people jingle coins in their pockets, and at a certain time in the past, the pockets might be all nylon or whatever. I don't intentionally... but I do use a plastic box in a flattened egg shape as a pocket coin purse - which along with a case of unusual design for my keys, continues to prevent wearing holes in the fabric - and at a brisk walking pace, it rattles.
I remember working in a datacenter in ~2002, in the telco room. Working on SDH kit that in normal operation had an integral metal cover that sealed it all in. I was connecting a new fibre when I got a call on my mobile, Nokia 6310 or similar.
I took the call and continued working with the phone cradled against my shoulder, and watched in horror as first it failed over from one processor to the second (with a audible clunk), then rebooted the entire device, all cards reset.
I was 6 days into a 7day acceptance test on a circuit at the time.
I guessed that metal cover was for a reason, so when I got back to base I replaced the ~15 covers that had been permanently left off for ease of access.
I am a Structural Engineer by trade but family and friends ask my advice when their computer misbehaves.
My brother who needed to access to a professional website could not stay logged in, it defeated both their support desk and his own IT support. He rang me and I remoted in and became puzzled by the dates on some files. He corrected the date and voila.
I deduced that the log in set a cookie with the system date, but as soon as he accessed the online database it deduced that the cookie had expired.
You asked so I told you
Some years ago, I worked at a secure (DoD contractor) site. WiFi and laptops were a new thing and some engineers thought it would be convenient to plug an access point into their network drop. To be able to walk around the office with their laptops. An absolute no-no as far as IT security was concerned. So they (security) contacted our electronics lab and had them sweep the building. The lab built up a contraption made up of microwave horns, waveguides, local oscillators and mixers. And a top of the line HP spectrum analyzer. All on a cart with a portable power supply. Probably a few hundred thousand dollars in precision equipment.
I watched on the day they did their sweep. No offending WiFi access points were found. But they did locate every microwave oven on the premises.
Years ago, in a career far away, I managed a restaurant. The point-of-sale system would gain about 5 minutes a week. The company ran a very tight labor percentage. Every other Friday (busiest night of the week) I had the closing shift. When the peak number of employees were on the clock, I would reset it back to accurate, thus gaining about two extra hours of labor on my shift.
Was I screwing employees? Maybe, but keep in mind it had been creeping forward for two weeks while they were on the clock getting paid.
I ended up working for that Point-of-Sale company as my lead in to tech. Oh, and I married one of my employees.....so yeah, I guess I was at least screwing one of them.
It was built like a tank and is probably still running somewhere. The same basic design is still available I think for commercial cooking, albeit with updated digital timers and controls.
Wouldn't the oven be leaking RF or EM, causing RFI or EMI, not EMF?
A company I used to work for had an office in Lagos, Nigeria and they needed to expand and move from one area to another. The ISP proposed a microwave link across the city (cable infrastructure is insanely tangled and unreliable in the city) and offered two alternative prices. Officially licensed/registered with local government, or not. Licensed/registered was more expensive (of course) but offered the benefit that if anyone planned to build a high-rise across the line of sight of the link then we'd get some warning. In what seemed an unusual burst of candour the sales-bod told us licensed/registered was probably not worth it because high-rise buildings were going up anyway and existing buildings were getting additional floors added without any pre-registered plans.
Time waits for no man, except for the Captain. I commissioned a DCS control system in the late 80's, on a well-known UK gas terminal. One day I found the client's Instrument/Elec Engineer plonked in front of a display, (a Captain's Chair, hence the nickname) watching the bottom right corner. Every once in a while he'd jump out of his seat, point at the screen and say "There it is!, Done it again!". After a while he came to me to complain that the system clock was wrong. Except it wasn't, it was spot on. So he got me to watch too, and sure enough the on-screen digital clock would get to e.g. xx:00:00 then jump to xx:00:02 or so. I had to explain that the screen display was just a task, like all the other 100+ or so tasks, and its priority was pretty low and due to that it occasionally re-synced itself with the system clock. I'm sure he was still there when I left....
It's likely that the emergency generator had harmonic content that was tripping the cheap clock's zero-crossing detector multiple times per cycle. Or perhaps the generator's output was not as "stiff" (low impedance) as the mains source, allowing harmonics generated by some other load (switching power supply, perhaps?) to propagate when the generator was sourcing power.
My wife's digital alarm clock runs crazy fast when the washer runs. It started when we replaced the old top-loader with a new front-loader. The old one had a mechanical timer and an induction motor. The new one has electronic controls and a direct-drive motor for the drum -- not sure whether it's DC or variable-frequency AC. But whatever it is, the drive electronics must be putting spikes on the power line, causing the clock to run crazy. Our solution was to ignore the clock and use the alarm function of our mobile phones.
When I was in college it was impossible to sleep in on the days they would test the dorm's emergency generator. The exhaust pipe ran up through interior partitions to exit through the roof, and it was so loud that I doubt that it had a muffler. It made an awful blatting noise if you were outside, too.
I client of mine - a school - bought two of the horrible Power Macintosh 4400 machines back in the late 90'ies. They were to be used in a computer room for the students. I stacked the two machines on top of each other to save space and had the monitors on each side of the stack.
The students experienced regular but strange crashes on both machines. I was called and looked into the problem. Both machines was indeed quite unstable and crashed in places, I haver never seen the old MacOS crash. I began testing on one machine, while a student was using the other. When I experienced a crash, the student cried out in frustration. His computer crashed as well. Well that was interesting. And it was possible to recreate the event. Each time one of the 4400s crashed the other followed. Could this be mass hysteria? Was the machines so badle build, that they would influence each other?
I moved one computer away - to the other side of the monitor. After that both machines ran smoothly and the crashes disappeared.
At another client, I installed a WiFi. It ran perfectly, and I left. But over the next couple of days my client called several telling me, that the WiFi was gone. Each time a power off/power on solved the problem. But this network were very unstable. I went to my client, and the problem couldn't be reproduced. Then I moved the router to a desk to work. I reinstalled firmware and tested and tested and tested. Everything was fine. The uses connected without problems, and I thought that the newly installed firmware had solved the problem. Then I moved the router to its location (in a window). I booted it, the users connected and ... crash! The I noticed that below the window (my client was on the 1st floor) there was a taxi. As a matter of fact it was a parking spot for taxis. Could the radio bursts from the taxis knock out the router? I moved the router to a dfferent location in the office, and the problem disappeared.
Amana (Raytheon) Radarange oven, circa sometime after 1967 – the precursor to the microwave oven
The Radarange line was a line of microwave ovens. They were not "precursors". They used microwave-generating magnetrons, just like other microwave ovens.
And by 1967, the Radarage line had been in production (by Raytheon) for about 20 years, so this wasn't even an early model - though 1967 was when Amana started selling consumer (as opposed to commercial) units.
Biting the hand that feeds IT © 1998–2019