at about 12 years old the caps in the PSU power supplies dry or split. the relays then latch in random states at power up. I have 8 of my 12 that have not died.
We actually still use these at work! (Not new mind you, as I believe they are not end of life.) Very solid devices and we've been lucky enough to never have one fail. As another commenter mentioned the piddly little CPUs isn't up to doing modern TLS - HTTP or SSH - but that's fair given the design is about 15+ years old now. Some of their transfer switches though have a particularly nasty bug - during a firmware upgrade, they say the outputs could start oscillating at 50/60Hz! Great video as always.
Yeah, I still use these in other production environments, they are definitely slow when it comes to TLS though! That's a pretty nasty bug with the transfer switch, thankfully the only ones I use don't have any sort of network module so I shouldn't need to worry about that, definitely worth bearing in mind though!
I've used some of these that are 15 years old. The only problem is the processor is too slow to support https and ssh using the latest firmware. Older firmware uses poor encryption. I note you're not on the latest firmware as it's now green and branded Schneider. You can flash the firmware, or reboot the management card, whilst the unit is in production; the status of the sockets doesn't change. I assumed the relays were latching, and pulled open on power up. I have had a few of these where the network port had died. But generally they run and run.
The display board looks to be the same as on the full rack 3 phase pdu. the 3 leds will light showing you what phase the current monitor is selected and displayed on the 7 segment display.
So I see 3 current transformers. Also in the UI one can only see the total power for entire PDU. Weird. For the price like that I would want to have power monitoring on every single outlet, plus a separate for entire PDU (including the management board, and relays, and losses in PSU). Why there are 3 tho? Is it to detect current leakage? (GFCI style). Oh. I see, this is to also support 3 phase power with the same board. My bad.
What's really bad with PDU manufacturers is that they subcontract out the software and design super pretty graphical interfaces to appeal to what I imagine is their principal audience, but ignore how painful this makes them to use from automated scripts. When you have a board under test that you need to power cycle 1000 times over 3 days and ensure that each time it boots without problems, these GUIs have to be reverse engineered to do the power off and on. The industrial PDUs that you can still control through simple telnet or serial port commands are more and more rare.
Yeah. Lack of open source firmware for these devices, and crappy Java or web browser interfaces with fancy graphics is just absolutely killing me. All I want is some simple SSH or HTTPS interface where I can do GET and POST and do simple things quickly. Plus some quick HTTP monitoring where I can get a status of entire device quickly, and my monitoring system can plot it or alert on weird conditions (like PDU not accessible). SNMP is also good option, but obviously more horrible than some Prometheus metric endpoint.
They do work off the hall effect but in a less direct way than "Hall effect current sensor " and is in no way what is call a "Hall Effect sensor" as that is a sensor for magnetic fields not AC current.
No, if you crack one open it will have a lot of windings, the single winding is the wire passing thru the middle of it. Also It is still a transformer with one winding, two windings will double the rated output, three will triple and so on. current transformer en.wikipedia.org/wiki/Current_transformer hall sensor en.wikipedia.org/wiki/Hall_effect_sensor edit changed from google img search to Wikipedia
I typed this before your last post showed, but I will leave it as general info for anyone who may find it helpful. looking again at 6:53 you can see they are marked CT on the board backing me on them being current transformers as well as the dot that shows phase of the output only needed on the CT as a hall effect current sensor would have at least power, ground and signal pins instead of the two sides of the windings.
Dragan Milivojević OMG you don't know where Kyrgyzstan is, that's common knowledge.... Resistive shunt measurement is a lot closer to common knowledge but to know about current transformers you need to know about isolated analog AC current measurement. Maybe in your field or area of interest in the hobby they are used often but I would bet 80% of people doing this as a hobby don't have anything in there house that has a current transformer unless they have a Kill a watt. Also telling some one they are a dumbass for never hearing about something is just a dick move.
Sure, just disconnect a power to it or from it to main board. I guess the PSUs are connected to common bus bar to relay drivers, opamps and CPU board via some diodes. Desoldering a diode is another option.
It will show Er on the 7 segment displays if an internal power supply fails. Also, even if both power supplies fail, it will still stay “latched” in the current state till power is removed from the pdu. At that point you need to repair/replace both power supplies.
I think flash was probably just more expensive than battery-backed RAM when this was new. (Some?) GBA carts used it despite flash being available, but it was cheap enough by the DS that they switched. Since you mentioned this being an old supply, the production design is probably unchanged since its inception, even for a seemingly small swap like that.
The redundancy could be solved easier by using bistable relays, instead of two PSUs. bistable relays require less power to keep in one state, and remembers the state even if the entire device looses power and is recycled, or software crashes or PSU is faulty. One could argue that it might be good to start of PDU with all relays in off position, so one cay do a delayed start on each outlet (to limit total inrush current to servers). Also redundancy in PDU is not that important. If you are powering critical infrastructure, then your servers and switches will have redundant PSUs itself, and they will be powered from two independent PDUs, and failure of one is fine! Sure, having redundant PSU is better, but this adds to cost, and in a lot of applications is unnecessary.
They use two PSU in every 'Switched' type PDU model, and every PSU output voltage level is checked by MCU and reported to event log and further. There is a comment on reddit telling how it shows up when two PSUs are dying and the relays go out for a walk.
It definitely does seem unusual although for something like this I'd like to think APC would know what they are doing, they definitely know way more about this sort of stuff than me!
6:00 If I were to hypothetically do this with the grounding wire for a DIY box, I'd be obliterated by comments from people saying "I've been an electrical engineer for 87 years, you're an idiot!" or something. Hilarious to see a multi-hundred $ enterprise product doing something like that. 😂
I think you might be right, they support NTP although for some reason mine had the time specified manually and it seemed to keep time while it was off. The time is mainly used for timestamping logs. Would be interesting to know if the battery is purely for the RTC or if it also uses it for storing settings.
off or unplugged? There may be an rtc but it looks like battery backed ram and I don't see any crystals for an rtc but it may be mounted on the back of the pcb or hidden under the battery. Backed ram is cheaper than flash "or was in 04" an RTC again is added cost you don't need. RTC or not I stand by my bet that the battery is for backed ram for keeping network and start up info..
Unplugged - as in that it kept the time despite being dismantled in this video :P - I suppose they have to cater for environments without NTP servers and in a device like this that costs over £620, the additional cost to fit RTC hardware is likely not an issue.
![Felix "Unfair" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/Oswujxm2Ag0/mqdefault.jpg)
at about 12 years old the caps in the PSU power supplies dry or split. the relays then latch in random states at power up. I have 8 of my 12 that have not died.
Yes, they do detect failures in one of the supplies and alert you, including by SNMP. :)
Thanks for this teardown. I have one of these in production and i'm glad to find out that it's a pretty nice bit of kit.
I’m a fan of apc there stuff always seems to be top notch
We actually still use these at work! (Not new mind you, as I believe they are not end of life.) Very solid devices and we've been lucky enough to never have one fail. As another commenter mentioned the piddly little CPUs isn't up to doing modern TLS - HTTP or SSH - but that's fair given the design is about 15+ years old now. Some of their transfer switches though have a particularly nasty bug - during a firmware upgrade, they say the outputs could start oscillating at 50/60Hz! Great video as always.
Yeah, I still use these in other production environments, they are definitely slow when it comes to TLS though! That's a pretty nasty bug with the transfer switch, thankfully the only ones I use don't have any sort of network module so I shouldn't need to worry about that, definitely worth bearing in mind though!
At 8.40 the component you identify as a ceramic capacitor looks more like a metal oxide varistor for transient suppression.
I've used some of these that are 15 years old. The only problem is the processor is too slow to support https and ssh using the latest firmware. Older firmware uses poor encryption. I note you're not on the latest firmware as it's now green and branded Schneider. You can flash the firmware, or reboot the management card, whilst the unit is in production; the status of the sockets doesn't change. I assumed the relays were latching, and pulled open on power up. I have had a few of these where the network port had died. But generally they run and run.
I hope some day you get a few more from different vendors to compare them (ie Raritan, Vertiv, Rittal)
The display board looks to be the same as on the full rack 3 phase pdu. the 3 leds will light showing you what phase the current monitor is selected and displayed on the 7 segment display.
So I see 3 current transformers. Also in the UI one can only see the total power for entire PDU. Weird. For the price like that I would want to have power monitoring on every single outlet, plus a separate for entire PDU (including the management board, and relays, and losses in PSU).
Why there are 3 tho? Is it to detect current leakage? (GFCI style).
Oh. I see, this is to also support 3 phase power with the same board. My bad.
What's really bad with PDU manufacturers is that they subcontract out the software and design super pretty graphical interfaces to appeal to what I imagine is their principal audience, but ignore how painful this makes them to use from automated scripts. When you have a board under test that you need to power cycle 1000 times over 3 days and ensure that each time it boots without problems, these GUIs have to be reverse engineered to do the power off and on. The industrial PDUs that you can still control through simple telnet or serial port commands are more and more rare.
Yeah. Lack of open source firmware for these devices, and crappy Java or web browser interfaces with fancy graphics is just absolutely killing me. All I want is some simple SSH or HTTPS interface where I can do GET and POST and do simple things quickly. Plus some quick HTTP monitoring where I can get a status of entire device quickly, and my monitoring system can plot it or alert on weird conditions (like PDU not accessible). SNMP is also good option, but obviously more horrible than some Prometheus metric endpoint.
movax20h SSH will work perfectly on most of these modern APC PDUs, the commands are fully documented. It works really well for me, no GUI needed!
SNMP write option saves your time
surprised there is no any form of thermal management inside
they are not inductors, They are called current transformers.
They do work off the hall effect but in a less direct way than "Hall effect current sensor " and is in no way what is call a "Hall Effect sensor" as that is a sensor for magnetic fields not AC current.
No, if you crack one open it will have a lot of windings, the single winding is the wire passing thru the middle of it. Also It is still a transformer with one winding, two windings will double the rated output, three will triple and so on.
current transformer
en.wikipedia.org/wiki/Current_transformer
hall sensor
en.wikipedia.org/wiki/Hall_effect_sensor
edit changed from google img search to Wikipedia
I typed this before your last post showed, but I will leave it as general info for anyone who may find it helpful. looking again at 6:53 you can see they are marked CT on the board backing me on them being current transformers as well as the dot that shows phase of the output only needed on the CT as a hall effect current sensor would have at least power, ground and signal pins instead of the two sides of the windings.
It's a bloody current transformer, common knowledge.
Dragan Milivojević OMG you don't know where Kyrgyzstan is, that's common knowledge....
Resistive shunt measurement is a lot closer to common knowledge but to know about current transformers you need to know about isolated analog AC current measurement. Maybe in your field or area of interest in the hobby they are used often but I would bet 80% of people doing this as a hobby don't have anything in there house that has a current transformer unless they have a Kill a watt. Also telling some one they are a dumbass for never hearing about something is just a dick move.
Could you perhaps disable one of the power supplies to see if the error is reported?
Sure, just disconnect a power to it or from it to main board.
I guess the PSUs are connected to common bus bar to relay drivers, opamps and CPU board via some diodes. Desoldering a diode is another option.
It will show Er on the 7 segment displays if an internal power supply fails. Also, even if both power supplies fail, it will still stay “latched” in the current state till power is removed from the pdu. At that point you need to repair/replace both power supplies.
I think flash was probably just more expensive than battery-backed RAM when this was new. (Some?) GBA carts used it despite flash being available, but it was cheap enough by the DS that they switched. Since you mentioned this being an old supply, the production design is probably unchanged since its inception, even for a seemingly small swap like that.
I believe flash was also perceived as less reliable compared to battery backed SRAM back in the 90s when this would have been designed.
this battery is just for RTC, and flash is over there
Красота!!! Хочу такой. В понедельник куплю.
The redundancy could be solved easier by using bistable relays, instead of two PSUs.
bistable relays require less power to keep in one state, and remembers the state even if the entire device looses power and is recycled, or software crashes or PSU is faulty.
One could argue that it might be good to start of PDU with all relays in off position, so one cay do a delayed start on each outlet (to limit total inrush current to servers).
Also redundancy in PDU is not that important. If you are powering critical infrastructure, then your servers and switches will have redundant PSUs itself, and they will be powered from two independent PDUs, and failure of one is fine!
Sure, having redundant PSU is better, but this adds to cost, and in a lot of applications is unnecessary.
They use two PSU in every 'Switched' type PDU model, and every PSU output voltage level is checked by MCU and reported to event log and further.
There is a comment on reddit telling how it shows up when two PSUs are dying and the relays go out for a walk.
Very interesting , but sending the earth via the case is a big no no. The case does have to be earthed ofc but you should run a direct earth too.
It definitely does seem unusual although for something like this I'd like to think APC would know what they are doing, they definitely know way more about this sort of stuff than me!
That's actually standard for metal back boxes and earthing in general here in England.
Cool! I want one of those LOL.
6:00
If I were to hypothetically do this with the grounding wire for a DIY box, I'd be obliterated by comments from people saying "I've been an electrical engineer for 87 years, you're an idiot!" or something.
Hilarious to see a multi-hundred $ enterprise product doing something like that. 😂
The battery must be for a rtc probably
Unless this has a timer function it has no need for an rtc, I would bet $10 its to store the IP and power up delay "can be worked out without an rtc"
@@chevsev I don't really know, but given the pdu has an internal web and telnet servers, it may be logical to think it has a way to keep time..
I think you might be right, they support NTP although for some reason mine had the time specified manually and it seemed to keep time while it was off. The time is mainly used for timestamping logs. Would be interesting to know if the battery is purely for the RTC or if it also uses it for storing settings.
off or unplugged? There may be an rtc but it looks like battery backed ram and I don't see any crystals for an rtc but it may be mounted on the back of the pcb or hidden under the battery. Backed ram is cheaper than flash "or was in 04" an RTC again is added cost you don't need. RTC or not I stand by my bet that the battery is for backed ram for keeping network and start up info..
Unplugged - as in that it kept the time despite being dismantled in this video :P - I suppose they have to cater for environments without NTP servers and in a device like this that costs over £620, the additional cost to fit RTC hardware is likely not an issue.
Thank you. Nice video.
Cool video!
I picked this up at goodwill for $20
Ground is the green/yellow one.
I'm from Newcastle and lived in Glasgow for a few years, but I struggle to understand such fast talking. Try to slow down!
can u talk more faster? my trains gonna miss if u talk any slower
First!
No one cares