This is NOT a repair video
HTML-код
- Опубликовано: 12 сен 2024
- This is NOT a repair video.
Subscribe on Odysee: odysee.com/@ee...
EEVblog Web Site: www.eevblog.com
Main EEVblog Channel: / eevblog
EEVdiscover: / eevdiscover
Support the EEVblog through Patreon! / eevblog
AliExpress Affiliate: s.click.aliexpr...
Buy anything through that link and Dave gets a commission at no cost to you.
Donate With Bitcoin & Other Crypto Currencies!
www.eevblog.co...
T-Shirts: teespring.com/s...
#ElectronicsCreators
It could also just be that this device needed the infamous "turn it off and back on again" repair. :D
A hard reset never hurts, much.
Or the Light treatment. :)
Its the eev touch
Temperature change could also cause an intermittent problem, as can a dodgy ethernet.
I've had exactly that on very similar looking switches.
"Hello, IT. Have you tried turning it off and on again"?
"My phone keeps beeping at me demanding I check notifications, how do I fix it?"
"Have you tried turning your phone on and off again?"
@@dedr4m When you turn your phone on and off again it stops working.
@@Okurka. Well, yeah! That's the plan there :)
Turning the phone off stops the beeping and distractions :-P
4:31 The moment Dave realizes he never actually just tried taking all the cables out and turning the thing off and back on again.
Yeah; I've had a similar issue with one of my TP-Link switches. Specifically I have to unplug and power cycle the WLAN extender I have upstairs, because sometimes it seems to wig out and take down the entire switch (one of these days I'll figure out why...or replace it with something that's not cheap shit).
Even Newton once asked a guy to make two doors one for the cat and one for its kittens
@@hikariyouk I have a ton of TP-Link gear and have had no problems so far. I run the DE variant of the switch in this video (3 switches total) which has better switching hardware, and supports VLANs
@@deelkar To be clear, I don't think this is the TP-Link switch's problem. I think the cheap shitty WLAN extender I'm using. TP-Link are my favoured supply for networking gear at home nowadays.
I think we have all been Dave at some point
4:30 "change it to over here the .. green ... led ...comes ... ... on. Uh..." For anyone new to repair, this happens to everyone at some time. This is real life, not edited life.
Murphy gets you. EVERY SINGLE TIME.
so many times i have been asked to go look at something that is not working, and when i get there it just works without issue.
@@YommyRawr The things know that if they act up around you, you'll take it apart.
Intermittent....failure.... *dramatic music*
I mean, technically you repaired it. You touched it and it started working again. Nobody said HOW a repair has to look like. ;-)
I cannot count how many times I fixed computer issues for my users this way.
"The silkscreen is not marked", says Dave, missing the test points clearly marked 1V2 and 3V3.
;P
Too tired? Well I know Dave is watching to buy a bunker, pay much attention to that.
Hidden on the ground plane. Sneaky bastards.
is not fair! Dave does not have a zoom, playback etc,... you do.
(Dave is paying me 2$ for a favorable comment this month ;-p)
@@EEVblog2 Hidden in plain sight are the worst kind!
@@EEVblog2 to be fair you was a bit flustered with it working all of a sudden
The "turn it off and turn it on again" thing is a meme for a reason. It works infuriatingly often.
Particularly useful for consumer garbage that has poor hardware and or software design. If power cycling actually fixes it, time to buy better gear.
And if that doesn't work, try to show the problem to someone else.
Nothing on the silkscreen? I see 3V3 and 1V2 test points right next to those power supplies!
Dave needs reading glasses.
My guess is a bad solder join which 'fixed' itself when it cooled down while unpowered due to thermal expansion.
I expect a repair video soon after it spends the night heating up again.
Thought the same, which is why it was probably scrapped before. Hard failure to find though
nah not solder just got into a bad state and fixed after a power drain, happens all the time.
Which is probably why it ended up in the dumpster in the first place ;) it's just a cheap crappy switch
@@flipschwipp6572 There aren't a huge number of solder joints relatively speaking, so I'd guess a liberal reflowing would stand a good chance of fixing it.
could also be some memory that cleared itself when he powered it off.
If I may quote Electroboom, "Even I'm surprised by the extent of my genius"
"You won't see this video." 🤭 Priceless.
personally when it comes to networking/pc.. I always just try unplugging and plugging back in the device first to see if that solves the issue.. maybe a silicon chip froze/locked up or is having issues, so a powercycle seems to always be the fix, especially when it comes to networking (i.e. rebooting a modem when having internet performance issues).
As an it guy this makes me laugh a lot.
Only thing missing while we will use this right now because it is an emergency and we will order a new one to replace it.
Fast forward a few years, the old "broken* is still in use and the new one still on the shelf waiting to he implemented 🤣🤣
"Nothing is more permanent than a temporary fix."
It's the new switch on the shelf being a constant remider of how close the old one is to being replaced that has kept it in its toes for the last few years.
Been there. Done that.. Thanks for the inside look. 15 years ago, that case would have been filled with bunches of parts.
I always tend to overlook the simple stuff, like that dual 12V, dual GND - 0.156"-pitch Molex coming from the SMPS into the board. Could be dry solder joints on the rail supplying the right-hand block. I go straight to the complicated stuff, like replacing the caps, whether they seem to need it or not or looking for a faulty regulator. Cheers, mate!
I have the EXACT same model, with the EXACT same problem. Design defect? Maybe. Mine will die after a few days running, so I think the PSU for the 2nd ethernet switch is failing (or intermittent)
Software/hardware bug? Rebooting brings it out of a locked up state.
Maybe you both take part in the manufacturers test to see which of the two designs fails first.
Maybe it's a thermal related problem?
It generates around 20W of heat and has no active ventilation, it can get heat soaked after like few hours or few hours + activity on the network and for example some dry joint disconnects or dry capacitor doesn't make it anymore and one of the devices shuts down, then cooldown + complete power reset brings it back.
I won't be surprised if it fails again.
There's a reason it was found in the dumpster (apart from it being TP-Link), intermittent faults do my head in and TP-Link stuff has caused me more grief than any other manufacturer !
Yep.
Same here, TPlink is junk
I bought a _£~25_ TP-Link TL-WA801ND WiFi access point (basic "300 Mbps" 802.11n thing, though with support for up to four SSIDs, optional per-SSID VLANs, passive PoE, and was even supplied with a passive PoE injector - all of which seems amazing for the price, IMO) to use as a backup because I've had significant reliability issues with my *£300* Asus RT-AC3200 ever since I got it.
I've had a couple of minor issues with the TP-Link AP in the few years I've had it (had to reboot it to let devices connect, but only a couple of times a year on average), but overall it's been _much_ more reliable than the Asus AP 🤷♀️
I've been happy enough with the TP-Link AP that I'm considering getting the outdoor model at some point to improve our WiFi coverage in the garage and driveway.
With hindsight, I wish I'd spent that £300 on a Ubiqiti AP or two. That probably would have been much less of a headache... 😩
I'm also using the 24-port "smart managed" version of the TP-Link gigabit switch from this video as my core home switch (TP-Link TL-SG1024DE).
I got it second-hand from a friend after my previous switch died (probably the power supply; I didn't have time to diagnose and repair it at the time - I just needed the network back online), and that's been totally reliable for the couple of years I've been using it!
I'm not suggesting your experience is "incorrect" or anything like that, but my own experience with TP-Link equipment has been the total opposite :)
@@AndrewGillard TP-Link equipment does work, for the most part but they seem to suffer with with 'interference' related issues which causes them to either randomly stop or just drop the connection at crucial times.
Sure after a reboot everything is fine but sometimes it can require 2 or 3 reboots to get them working again and each time you are going through the trial and error testing etc until you reboot it one last time, so frustrating !
It seems like interference but the router never moves and there is no equipment nearby that changes state or affects the line and the problem can get progressively worse over time ! I think I have 3 or 4 old TP-Link routers in my attic that have all suffered the same result.
I wouldn't class ASUS as being any better though, my experience with them has been pretty sketchy too.
Some are good though. Allied Telesys can also surprise you with these kind of failures.
I totally cracked up when you plugged into the "failed" bank of ports and it lit up! BTDT
Thanks for sharing.
Looks like a standard 2x9 switch chip switch. The 9th port being a MII port that links to the other switch's MII port. Sometimes (especially in a brownout) a switch chip will lock up, and need a reboot which is likely what happened here. If you had already tried a restart, there might be a dry joint that was "sorted out" by being moved and thumped a bit. If that's a case, be prepared for it to play up again. That said, TP-Link have been surprisingly solid for their price range.
Did you try a power cycle reset before pulling it?
My switch on the odd occasion needs one, routing table full or corrupted? who knows.
I was aiming to fix a CD changer for the wife to play our CDs. Couldn't get the damn thing to play. Tried several CDs. Tore the thing apart, checked all the voltages, all the mechanisms, clock signals, etc. Three hours with the thing practically torn down to components, and JUST COULDN'T FIND THE PROBLEM. Put it back together before I forgot how it went together, and noticed a label on the magazine that carries the CDs. I was putting the discs in upside-down... 🤦♂️ As a technology guy, that was not my proudest moment.
4:10 - Dave demonstrates how to use an oscilloscope to troubleshoot a network switch.
The one in the left likely contains the management circuitry in addition to the ethernet controller, which probably explains the extra regulator, and the larger heat sink.
As for the actual problem: I've seen this happen with shorted cables or damaged ethernet ports (the pins go a wandering sometimes, causing shorts). The ethernet controller on the right is probably shutting down for safety. I've also heard of it happening from a nail or push pin being put through a cable router into the wall.
The probability of a hardware failure disappearing is inversely proportional to the distance between the device and the engineer.
Never forget Murphy's laws...
becoming obese solves the problem only to an infinitesimal amount.
Whew. That was a close call. Lucky nobody saw!
I guessing the processor for port 9 to 16 crashed and needed to be rebooted.
This reminds me of when the secondary hard drive controller (controlled the 4 bays on the front) of my old NAS was dying, and would cut out after roughly 800 GB of data read. I had to turn it off, wait for it to cool down, turn it back on, and it continued working. Eventually, port 3 gave out entirely and I removed the mainboard, since it's a custom NAS, and replaced it with a slightly older but more powerful mainboard. I have it hooked up to USB drives temporarily since the replacement board lacks mSATA support and I decided to get a proper NAS rated SSD to use as the boot drive.
you don't need a NAS rated SSD these days, not even if you use it as a cache drive. (I am running about a dozen custom-built NASs on multiple locations, from 2 TB to 40 TB capacity)
I think your 16 port Gigabit Ethernet Switch just need a "POWER CYCLE"...!
If it didn't need a power cycle, then it probably would not have been in the dumpster......
That's about what I'd expect in an unmanaged switch. All these things are done with custom ASICs, as in, FPGAs. (A firmware upgrade can include an upgrade of the FPGA programming as well. On managed devices, you may have multiple modes that have different FPGA programming as well as different settings on the processor-run software.)
port bonding does not increase throughput for a single client, only for multiple machines to one server.
Moreover, not all switches implement needed load-balancing algorithms in their firmware and in the end it increases nothing apart from reliability.
So you can just free up those ports.
I suspect the smaller chip is just a secondary controller whilst the larger one is the primary controller but also processes the data from the secondary, and has the CPU / Memory perhaps that is why it has a 2nd VRM. I had a TPLink that just used to lock up and stop for no apparent reason, a quick off and on again used to solve it, got rid of it in the end and replaced it with a Netgear, which never went wrong once. Maybe whoever threw it out had this happen to them, didn't know to do anything but replace it.
Takeaway protection - love it
It could be an intermittent issue that only shows up when the switch has a bunch of ports loaded. Since both power rails are almost certainly used by both ASICs, the small switch might just be more sensitive to supply noise than the larger one and calls it quits first when power gets dirty.
I have a Netgear 48 port gigabit that does just this. Its heat related and the ports are banked 8 per asic. After about half an hour 16 ports will just vanish. Power down, hit it with freeze spray and it will be fine again for a bit.
I think I can hear Phil Collins singing way in the background noise. Yes, definitely... it's 'Turn it on again' by Genesis... :)
If you look closely in the tranparent rj45 plugs of those yellow or blue cables you notice that they are not just crimped on there,they are melted in place!i myself have found several cables like that in servers and offices that only multimeters could detect the internal shorts,not detectable by cat cable testers,could also be a psu issue like i experienced already also
Maybe the smaller chip is overheating or similar?
You don't get faster transfer speed by bonding synology ports into the same switch (I know this due to doing it myself, with zero increased transfer speeds), you only get semi-redundance... "semi" because redundance is useless if it goes through the same switch.
It is nice seeing the humility of sharing your mistakes.
A lot of snobs sneer at TP-Link gear, but after 30 years running my IT business I have found it to be highly reliable. Unlike Netcomm et al. Always carried TP-Link ADSL modems to replace the inevitable Netcomm failures :) So value for money very good. There are, of course, higher end products that are used when needed, but even Cisco switches that have been running for a few years will not cold restart due to power supply problems.
It might be a temperature issue. I have seen where a switch will over heat and ports will drop. Then if you power it off and wait a few minutes. It will just start working again.
it can also be a Periodic problem over time and heat, where the caps fail over time when they heat up or other thing, it can be hard to find, if you cant stress test the device.
Perhaps its a Daisy fault, as in some days it works and some days it don't. (Sic).
Maybe there is an intermittent connection on FPGA ball grid array. Moving it temporarily fixed it. Maybe that was the reason it was in the dumpster.
do we get a investigation video on finding the issue with cabling
Well, did it still work when you reconnected everything? As you link up more connections (especially Gigabit ones) the load increases. It could very well be fine with one or two ports used but act up when more are active. With my secondhand switch I scoped the 5v input to the buck converters. When I went up to 8 links or so I was seeing about 2 volts ripple. The capacitors looked visually fine but after replacing them it went down to 50mv ripple and has been reliable since then.
Oopsie, looks like Dave just qualified himself as a young player!
"Don't take it apart, turn it off an then on again"
In Germany there is a word for this, wackelkontakt, which means wobbly connection, or loose connection.
It takes 19W under full load, running just half of the ports with not full gigabiot transfer continously it takes between 10 and 13W (had the same switch and measrued it)
Been there done that one I’ve found bad solder joints broken traces that would make contact again once the cover was off. But putting a little stress on the board would cause the problem to return. Then there were times when nothing was found and the unit worked fine reassembled. I would just say it must have been a solder ball. LOL
I had a few of the big brother 48 port TP-Link switches. Ended up dumping them because section of the switch would just freeze (activity lights would go solid or just dark) and had to reboot the switch to bring it back around. You get what you pay for.
You just need to power-cycle these D-Link thingies when ports "go out". Old trick. I've set up wall-timers to power-cycle anything named D-Link at least once a day. (during the night)
Thats usually my way of fixing things. Start by opening the thing then finding out I just needed to replace the cable.
Connection problem on a joint? Could have been a reason it was in the trash. It could maybe be there is like a rail stability problem due to caps or bad transistors that is booting the 2nd micro some of the time.
Reminds me of when I took apart my dishwasher to repair it only to find out after a couple of hours troubleshooting that it was the wall socket that was faulty.
For dumb-switches, rather us netgear pro switches - lifetime warranty. For managed, Mikrotik is great & cheap, just a bit of a learning curve.
There is the possibility of this being a load issue and therefore it would only manifest when there are multiple connections present.
I had several "cheap" conusmer switches fail on me before.
It has always been one of the switch fabric asics that died, the power supplies were always fine in my case.
This made it impossible to fix any of them.
The failure looks exactly like what you got, with all the ports of one ASIC being dead.
Looks like you got lucky and it fixed itself. :)
maybe reflowing them does help extend the life a little bit
If you get problems again it seems to have a serial uart you can poke to see if perhaps you get sensible errors on the console
Had a smaller rackmout, also 16 port TPlink gigabit ethernet switch at work that the status LEDs got stuck, no blinking for activity and didn't update for what devices were plugged in either, power cycling it resolved the issue.
Ports die all the time. I just cover the bad ones with tape and write bad on the tape. eventually, they get replaced.
Maybe it just was needing a power recycling. Or we have a periodic heat fault.
you don't actually need another device to test if a switch port is linking, you can just patch two ports of the very same switch. This works fine as long as DUT is a dumb switch (like yours, without STP which would kill the loopback disabling some port) and you have no other devices connected (which would get a broadcast storm)
Isn't the first rule of troubleshooting is to check the cables first? :P
I guessed as much. Problem? First try unplugging all cables from the bad side, then plug a cable that lights up the other side on the bad side.
Do not be surprised... you don't have to have a "blown" cap to have a blown cap. I've seen the capacitance drift just far enough to go out of spec. Could also cause a heat-related failure when the cap heats up and changes it's range. I've seen D-LINK switches blow caps, and not have a visible popped can... Just food for thought...
Many things could have triggered the fault, and chances are that it will happen again. If it was some sort of protection of the right hand side, then it could have been caused by faulty wiring. Time will tell if the real repair video will either come up soon or never.
You may have a Odd thermal Expansion issue with a few dry joints
I would give everything a little hot air gun blast with some fresh flux and re-seat everything a sort of component & PCB De-stress
then check all the Ethernets for shorts + broken connections One by one
This is why you always try turning it off and on again :P Could of also swapped an active port to a non active port to see what happened before taking it out of the cupboard. TP-Link are cheap garbage IMHO still surprised it looked cheaper on the inside than I thought. It was probably in the dumpster for this reason. However if you don't need 100% up time you can just power cycle it when needed.
Maybe the devices that were previously patched on the right weren't powered ON, i.e. that is why there were no link lights?
I have a 16 port switch that I got from a business that replaced it because it was "dead". I got it home and it was working fine. I'm guessing a transient heating issue.
Maybe you built a loop and the switch turned off that port group to protect the other one?
check the caps, they're likely low in value and poor ESR. pretty common on all those cheap to midgrade networking gear. values drop rails get noisy and dirty and they get flaky, usually worse with more ports in use(higher current draw)
The fear of being probed can make anyone camera shy.
Murphy stopped by. Oh well. I guess we will wait for a drive to arrive for the NAS.
That's a $60 switch, new. You got your money's worth.
My money says your goof was not power-cycling it before declaring it faulty. That said, based on previous experience with units like this, I wouldn't be surprised if this turns into a "It dies every week" scenario....
Yeah I've had netgear switches that would stop working until power cycled.
I hate when I can't find what caused the defect, because the problem might come back soon.
It's just refreshing to know that for all of your big brain education videos that I have learned so much on that you are actually human and make "unexpected findings" just like the rest of us. Thanks for Sharing Dave. It makes me (at least) feel like I'm not as dumb as I probably am lol... :-)
LOL Roy would be proud.. Could also be thermal compound on that chip - if it cooled down while you pulled the cover off.
Wait is this a repair video?
No.
😲
Soon there will be when that port group hangs next time. Did he test to cross ports when failure was there
I've had two of these switches - one died the other behaves badly...reboot fixes it temporarily. After a couple of temporary fixes, I fixed it permanently by replacing it entirely. There, fixed! I think I still have one, will take a look inside, but I suspect the problem is with firmware or the cpu/ga or whatever they use.
Dave, by your definition, everything is "reliable" until it fails. Technically "reliable" is a statistical measure derived from many units (population >> 1), so, technically you can't claim anything is reliable unless you have the stats for thousands and can show the level of reliability, you can however claim yours has been working for 5 years. Its probably an outlier given my experience (< 2years). Oh, the replacement for the TP-Link is 32 port compaq 100Mbs switch...slow yes, but connected to our super fantastic, forward thinking, extravagant, over spec'd NBN FTTN it makes bugger all difference.
Didn't mention, the Compaq switch has a fan and has been operating for over a decade. They must be worth $2.50 now, but I think one can guess that they probably qualify for "reliable".
My past experience is TPlink stuff is crap. Had a router, couple of wifi adapters and network extender, i ended up having to switch them all for not crap ones.
Exact same issue with a 24 port TP-link.. Right most 8 ports gave up. .Left it like that for a few hours, then rebooted and the whole thing failed then. Waiting for RMA right now. Had 2 months of warranty left. Guess their bomb timer is off LOL. Typical of microcontrollers with an internal crystal.
EDIT. LOL watching to the end.. Mine is actually broken..
I appreciate the cable management
Looks like you need to get yourself a patch panel and a small wallmount rack.
You forgot the magic trick “ turn it off wait 30 seconds and turn it on again 😂🤣
The opening is the trick to fix problems
Don't worry Dave happens on my Netgear router as well - occasionally it goes off with the fairies and requires the good old power off and back on.
Mind you it twas a cheap repair hey :-)
The tupperware did the trick. Half the electrons were spraying out of the PS board, but that bowl collected them so they ran back into the board.
This is not a repair video.
This is just a tribute.
At the beginning you mention you bond the NAS interfaces together. Not sure if this would actually be working on an unmanaged switch normally both the Switch and Nas need the be configured for this to work. You are probably still only getting gigabit speeds to your NAS.
Have you tried rebooting it?
I wonder if they are ASICS or FPGA'S? I notice this sort thing happening to some managed switches such as Netgear (I know this is not managed) from 2011 that take time to boot up whereas it didn't happen the previous models ex GS724V2 and that one just come on in seconds compared to about a minute with the V3. Until a later firmware in 2015 some ports use randomly and intermittely down power/don't come back on or no packets went through and it would eventually do it on all pots and hang until cycling the power. Setting things like SMTP and SYSLOG seems to make it happen sooner. I remembered your video EEVblog #496 - What Is An FPGA?
and you saying something about when they make mistakes.
Maybe a re-paste of the heatsinks and some fresh electolytics for the hell of it, or just buy a 2nd hand Cisco off Ebay and be done with it.
This is why toilet paper should never be used to link devices.
Perhaps it's time related, i.e. the second half enters some sort of failure mode after the switch has been up a number of hours. Maybe that's why the original owner dumped it?
Maybe your right it got too warm, may replace the heat sink goo it may have got dry over the years, that what I would do.
Before you opened it, you did try hitting it a few times, right?
Percussive maintenance is one of the most heavily underrated tools in our arsenal.