Regarding the 2-pin terminal: on this device, the 2-pin connector and the standard DC barrel jack share the exact same internal power input rail. Because they are wired in parallel, applying power through the 2-pin connector will behave identically to using the DC jack, so switching between them will unfortunately not bypass the internal power failure. Given that a replacement power supply did not resolve the issue, the unit will need to be processed via RMA for technical inspection and repair. Please note that in accordance with standard MikroTik RMA procedures, our technical team will inspect the internal board for electrical overvoltage or surge damage during intake. If physical inspection confirms an overvoltage event, warranty coverage may be affected.
So now I have a delima ... do I send the unit back to Latvia or do I eat the loss ... Why its a dilemma for me is due to the fact that my configuration is stored in memory and that configuration has very sensitive information that cannot be exposed ... so eating the loss is my only option ... BTW, all my gear is protected against electrical overvoltage or surge damage via high quality UPS systems incorporating Active Power Factor Correction ...
The Block Diagram posted by MikroTik for this model does not show an Integrated PSU Failover mechanism ... ???
Another Question: on the CCR2004-16G-2S+ the specs state Number of PSU slots = 2 and the block diagram show an Intefrated PSU Failover <> BUT <> does it share the exact same internal power input rail ?
If they do share the exact same internal power input rail Then this is a unacceptable point of failure for a data center device ...
Anyway power surges should (normally) fry only the external brick and overvoltage are only possible if somehow the brick fails (the overvoltage on the AC line may damage the brick, but it is not like if you give 50% more voltage - let's say 160 instead of 110 on the AC line the brick will produce 72 volts instead of 48).
There is also only a single CPU, RAM, Flash, ethernet controller, etc. Is that unacceptable as well?
When you want that much of redundancy, you can get 2 routers and use VRRP for failover between them. The use of 2 powersupplies covers 2 failure scenarios:
the failure of a powersupply (AC to 24V DC), which happens more often than other failures
the failure of the power source, e.g. you can have one on the UPS and one directly on the mains, when the UPS fails but the mains power is on, your router still works.
Apparently in your case the second regulation that converts from 24V to the actual supply voltages has failed, or at least it appears so. With some electronics knowledge you could troubleshoot and possibly repair that. There are workshops who do that kind of thing, but it will not be financially attractive.
If he wanted redundancy and cared that much about redundancy, then he would know better than to get a single powersupply but 2 ways of powering the the same device, theres routers and switches with double powersupplys that are even hotswappable let alone redundant.
Other simple solution is double everything, as u said VRRP and two routers and double links between everything, this is not meant for a datacenter the datacenter gear has AT least 2 psus and theyre both hotswappable powered by straight power cable and screaming fans.
Hes just complaining because hes mad his router died
So hes just looking for a way to get his frustration out for bad network design
I would not call it bad network design. It is a prosumer/small-office device and it is not usual to design such networks with full redundancy. Apparently a component has failed in the router and it no longer works, I presume he has a backup of his super-important-and-super-secret configuration so he can just buy a new router (it is still sold) and load the configuration and continue working.
A company that provides consultancy/maintenance to some customers with small offices would usually keep a spare device that they can immediately deploy in such cases, to have the failing device repaired or replaced at the timespan that usually takes. Of course a home user normally does not have that and he has to replace or repair it and it will take some time.
Odd that they didn't at least provide diode isolation . . . It's not like that is either complex or costly, and would keep one supply from pulling the other down. (True dual supply would have regulation in the supplies, and only become common at the rails.)
OK, then I would not expect a supply failure to cause a second supply/feed to not work then . . . . Seems inconsistent with the issue being discussed here.
The issue being discussed here is the second regulator having failed. Most MikroTik devices are powered at 12-24V, sometimes 48V, by a first powersupply that is either external or built-in. There can be 2 of those as well, in which case their outputs are combined via diodes.
The resulting supply rail is then fed to a second DC-DC regulator that converts that down to the supply voltage(s) needed on the board, like 3.3V, 5V(sometimes), 1.8V, etc. Apparently that is what has failed in @mozerd 's device. There is only one of those, it would be technically more difficult to have that redundant as well. In practice the first supply fails more often, it works at higher voltages, is isolated, etc.
Hence my point above about true dual supplies - they would each carry thier own regulators, and combine the rails afterwards. I would consider this more "dual feed" than "dual power supply", but that is somewhat nitpicking . . .
As someone who has designed, repaired, and worked with this kind of stuff for decades, I understand it exceptionally well. (Servers, data centers, telco power systems, etc.)
It's essentially the difference between a true multi supply (server for instance) deployment with each supply providing all voltages directly with it's own regulators, vs. dual intermediate voltage feeds and a single bank of regulators, which is far less redundant. Makes sense in a cost-contained environment such as Mikrotik, but is hardly "enterprise class" . . . .
Another Canadian drama.
If the unit is dead it doesn't mean that the internal regulators failed, other things can fail causing the internal regulators to shutdown/not start.
Also that claim that if overvoltage is detected you lose the warranty is crap. If YOUR power brick that got shipped with the router CAUSED the said overvoltage you can't deny my warranty.
Interesting approach, I will probably check tomorrow. However, it is typically totally OK that the device has plenty of local filtered or DCDC converted (or locally stabilized) supply because of interference and noise (and stability as well). It is far easier that spread all voltages across the PCB from single power supply. The main power supply face transients, interference from other sources, it can be thermally more stressed etc. This shorten its life. The conditions inside the device are more stable....and also rAM, CPU, ports...they are not redundant anyway.
