Community discussions

 
sihubert
just joined
Topic Author
Posts: 9
Joined: Wed Feb 13, 2019 1:22 pm

mUPS with two different DC power inputs?

Wed Feb 13, 2019 1:47 pm

Hi everyone,

I wonder what are the differences between PoE-in and DC-in in the mUPS?

Could I use both in parallel, possibly with slightly different DC voltages? If yes, what would be the "priority" between those two inputs?
Do both inputs feed power to PoE-out? If yes, I don't understand why MikroTik would ship an accessory PoE injector...

For a bit of context, I would like to use two redundant DC power inputs, PSU1 and PSU2. That is:
- use PSU1 to feed PoE-out and charge battery
- if PSU1 is down, use PSU2 to feed PoE-out and charge battery
- if both PSU1 and PSU2 are down, use battery to feed PoE-out.

Is this scenario possible using the two power inputs of the mUPS (PoE-in and DC-in)? Or do I need external hardware?

Thanks in advance!
 
mkx
Forum Guru
Forum Guru
Posts: 2930
Joined: Thu Mar 03, 2016 10:23 pm

Re: mUPS with two different DC power inputs?

Wed Feb 13, 2019 5:04 pm

I can't say how things work on mUPS in particular. I have read an explanation by @Normis that when device (a RB device with PoE in) has connected both PoE in and barrel-plug, it'll use which ever power source has higher voltage by some margin (makes sense as it only involves a pair of diodes to make this happen). If one intends to use both power supply options, it is best to have some systematic difference in voltages so that power source used won't change too much.

The scenario of changing power source is the following: let's say that PoE in has 0.5V lower than barrel. Then device will start using barrel power source. As power draw increases, the power adapter will drop voltage slightly and it might decrease to 0.5V below the PoE in. So device will switch over to using PoE in. This will drop voltage on PoE in and allow barrel to increase again, causing barrel voltage to rise significantly above PoE in voltage. And RB will change power source over again.
This changing doesn't harm much the device itself as voltage fluctuations won't be critically high. However it might upset both power adapters due to constant switching between full power draw and idle states.

One of intended installation places of a mUPS is also directly on tower / antenna pole, where AC power might not be available. Hence possibility to power device over ethernet cable using provided PoE injector. If, however, AC power is available near the mUPS, it is always advisable to use barrel-type power input due to lower power drop.

Your scenario is quite feasible if mUPS behaves the same way as other RB devices. What I'd do is the following:
  • feed barrel-type power input using provided 24V power adapter
  • feed PoE-in using provided PoE injector and non-standard 18V, 2A adapter (this gives same wattage as provided power adapter).
    Possibly use short ethernet cable between PoE injector and mUPS so that voltage drop (and power loss) is kept at minimum. This might not be feasible if you need long ethernet cable to connect to some switch ...

Accroding to mUPS brochure, mUPS regulates input voltage to 24V if input voltage is lower than that. Hence mUPS-powered devices won't notice the switch-over from barrel power source to PoE in.
BR,
Metod
 
sihubert
just joined
Topic Author
Posts: 9
Joined: Wed Feb 13, 2019 1:22 pm

Re: mUPS with two different DC power inputs?

Thu Feb 14, 2019 1:27 pm

Thanks a lot mkx for your detailed answer! Very useful information indeed.
I'll test your suggested procedure and come back to share the results. My only worry is if power inputs from PoE-in and barrel are simply shorted together, but I don't expect such dangerous design from MikroTik...

In the meantime, anyone having more accurate information is more than welcome to share :wink: .

One of intended installation places of a mUPS is also directly on tower / antenna pole, where AC power might not be available. Hence possibility to power device over ethernet cable using provided PoE injector. If, however, AC power is available near the mUPS, it is always advisable to use barrel-type power input due to lower power drop.
I perfectly understand the usefulness of having both RJ45 PoE-in and barrel-type connectors on mUPS. However, since barrel DC-in is there, I don't understand the need for an accessory PoE injector included in the mUPS package. That is, except if two power inputs are tolerated - or better, if such redundacy is considered by design.
 
mkx
Forum Guru
Forum Guru
Posts: 2930
Joined: Thu Mar 03, 2016 10:23 pm

Re: mUPS with two different DC power inputs?

Thu Feb 14, 2019 5:07 pm

I perfectly understand the usefulness of having both RJ45 PoE-in and barrel-type connectors on mUPS. However, since barrel DC-in is there, I don't understand the need for an accessory PoE injector included in the mUPS package.

Why does this surprise you? If installation demands PoE in as power source (which is quite likely, the same reasoning says voltage up-regulation performed by mUPS is a good thing), then one would need to purchase PoE injector. So why not include it? Price of the injector is quite low, shipping costs can easily surpass it. As almost all (if not all without exception) RB devices support passive PoE in, this PoE injector can come handy anywhere so no harm done is it's not needed for particular mUPS installation.
I'm using one (had to purchase it separately) at home where RB is placed near power receptacle ... but I want to use PoE to connect RB to UPS (which happens to be on the other end of ethernet cable). And would use another one if I had one lying around. Hint: if you'll purchase mUPS and won't use PoE injector, please kindly send it to me ;-)
BR,
Metod
 
sihubert
just joined
Topic Author
Posts: 9
Joined: Wed Feb 13, 2019 1:22 pm

Re: mUPS with two different DC power inputs?

Thu Feb 14, 2019 6:30 pm

Okay, I understand now your comment about mounting it directly on a tower. Thanks for clearing that up!
Anyway it was definitely just a minor remark, and a bit off-topic :roll: I'll come back with news about the tests.
 
sihubert
just joined
Topic Author
Posts: 9
Joined: Wed Feb 13, 2019 1:22 pm

Re: mUPS with two different DC power inputs?  [SOLVED]

Mon May 13, 2019 4:12 pm

So I took the time to do some tests, using 2 different DC power sources : the provided 24V power adapter, and a solar panel giving 19.6V in open circuit and approx 19V in full load. I'm actually measuring this voltage drop on the solar panel to determine which power source is used.

My observation is that
when device (a RB device with PoE in) has connected both PoE in and barrel-plug, it'll use which ever power source has higher voltage by some margin
is true. No matter which power source is plugged where (barrel plug or PoE-in), the wall charger always wins over the solar panel due to its higher output voltage. Which means it is not possible to "force" mUPS to use either one of the DC inputs without additional hardware.

I also checked
Accroding to mUPS brochure, mUPS regulates input voltage to 24V if input voltage is lower than that. Hence mUPS-powered devices won't notice the switch-over from barrel power source to PoE in.
and it's true, though when mUPS uses the battery (no DC in available), the voltage drops from 23.7V to 19V. So by monitoring DC voltage input on the PoE-powered device, we can determine if power comes from the battery or fom another power source, but we can't discriminate which one.
 
mkx
Forum Guru
Forum Guru
Posts: 2930
Joined: Thu Mar 03, 2016 10:23 pm

Re: mUPS with two different DC power inputs?

Mon May 13, 2019 4:54 pm

... though when mUPS uses the battery (no DC in available), the voltage drops from 23.7V to 19V. So by monitoring DC voltage input on the PoE-powered device, we can determine if power comes from the battery or fom another power source, but we can't discriminate which one.

Which is also according to brochure. Well, brochure states 20V as output voltage when on battery ... but the it wouldn't surprise me if output voltage was not exactly DC due to DC-DC voltage conversion which is probably realized by high-frequency switching and if output would be observed with a decent instrument (e.g. oscilloscope) some AC component might become visible and most of consumer-grade instruments get slightly confused when they're presented with (relatively small) AC component when they are requested to measure DC.

So to the "used source" dilemma: the only way of prioritizing one power source (e.g. solar panel) over another is to choose the "non-preferred" to have lower voltage. In your particular case, using 18V wall power adapter would do the trick, mUPS would use solar power until voltage dropped below 18V (or whatever wall PA output) and would switch over to the wall PA. But if solar power voltage is not stable (and I guess it'd rise sharply after disconnection of load) it would start the ping-pong between both power sources (as mentioned in post #2 above). Not nice to both PA electronics at all.
BR,
Metod

Who is online

Users browsing this forum: No registered users and 17 guests