0
\$\begingroup\$

I am modifying an existing UPS to allow for a longer backup time. I plan on adding a cooling fan along with an external current controlled power supply to charge the larger batteries. There are some appliances in the my setup that run on 12v (monitor, desk light, ONU etc). I planned on using these buck converters to have five or six 12V 2A ports. As I would be connecting these directly to the 24V battery, I was also planning on adding an over discharge protection for this, as well as an auto-switch circuit so that it switches to another SMPS when mains is available.

Now I am wondering if this will be worth it. Using the wall adapter the 12V appliances came with on the AC output of the UPS would keep things much simpler, but I am worried if there would be a lot of power loss in switching. My goal is to maximize battery life as much as possible. I am using two 12V 45Ah batteries in series. The maximum consumption of my current setup is 620W (going by ratings on the labels of devices.) The maximum load capacity of my UPS is 720W. It had two smaller 12v batteries in series.

Should I add the DC-DC ports or just use the AC output of the UPS?

\$\endgroup\$
1

3 Answers 3

1
\$\begingroup\$

From a 24V to 12V DC-DC you can expect 85-90% efficiency if it's not synchronous, 90-95% if it is.

Going through the inverter, you can expect 80-85% efficiency if it uses a cheap push pull converter for the 12V to high voltage conversion, and more like 90% if it's resonant, but for a cheap 12V inverter it probably isn't. That's at high load ; at low load efficiency will be lower due to idle losses.

Then another 80-85% for the cheap flyback AC-DC converter in the wall wart.

So efficiency will be much higher with DC-DC converters, but since a significant part of the losses will be idle losses from the inverter, and these don't depend much on load, adding a few DC-DC converters is unlikely to help much.

LM2596 has terrible efficiency because it uses a NPN switch instead of a MOSFET. It's an obsolete chip. The fake LM2596 modules you linked in the question are even worse, they do not honor the specs, the inductor saturates, and the high-ESR caps will die quickly.

If you want DC-DC converters, then you should get ones from good manufacturers. They are not expensive.

There is another issue: isolation. Using the original AC-DC wall warts (or isolated DC-DC converters) will isolate all your devices from each other. Using non-isolated DC-DC converters will not.

So if there is any chance of ground loop issues, for example if this involves audio or analog of any kind, if you use non-isolated converters it's pretty much guaranteed you'll get noise problems.

\$\endgroup\$
4
  • \$\begingroup\$ Thanks a lot for the detailed explanation. I was planning on using a separate buck converter for each 12V output for isolation as they are pretty cheap. Getting original ones might become too expensive for where are I am. Thus, I think for now I will stick to using the wall adapters. That way I could also utilize all the built-in protections the UPS offers without making them separately. \$\endgroup\$
    – Hanzala19
    Jan 20 at 16:39
  • \$\begingroup\$ Yes it'll cost 4-8€ per channel for good quality readymade 24V 2A non isolated DC-DC modules. If you need a lot of them, it'll probably be cheaper to get a bigger battery instead! Really, do not buy the fake LM2596, if you use them at the rtated current they WILL fail, and when that happens it is possible the input voltage ends up on the output, which would destroy your 12V gear. \$\endgroup\$
    – bobflux
    Jan 20 at 17:17
  • \$\begingroup\$ I guess there's a little confusion... Looking at a DC-DC converter you can name it by its function (in this case "Step-down" since you want to lower the voltage) or by its topology like "Buck". The buck topology is always a step-down converter and it is not isolated. If it's isolated, it's not a buck converter! So if you want isolation you need DC-DC converters based on transformers, which are more expensive. But using the wall warts will work fine, you already have them and it costs nothing. \$\endgroup\$
    – bobflux
    Jan 20 at 17:18
  • \$\begingroup\$ Back in the day I did the same, got an old inverter for free (a big server rack one) and wired 2 car batteries on it. It worked fine and lasted quite a while running a PC. Note that your "500W" PC power supply is only going to use that much if you put big gamer GPUs in it and push it to the max playing games, mining bitcoin, 3D rendering, that kind of heavy stuff. If you have a PC that is less than 10 years old it probably uses less than 50W doing easy stuff like word processing. \$\endgroup\$
    – bobflux
    Jan 20 at 17:20
1
\$\begingroup\$

For any switch-mode voltage conversion you can expect around 90% efficiency but PSUs vary from 80 or so to 95 or more so it’s important to check out the specific devices you’re looking to use. It’s not a given that a small voltage ratio will be more efficient than a large one, or that DC-DC will be more efficient than AC-DC. It’s probably more important to minimise the number of conversions but again thats not a guarantee. In particular if the load is highly variable (2 or 3 orders of magnitude) then choosing a converter with good low-power characteristics is key.

\$\endgroup\$
1
\$\begingroup\$

As a rule of thumb, ratios closer to 1:1 are more efficient because it's like the converter has less work to do. And less converter stages are better, too. Converting 24V to 12V is better than converting 24V to 220V and then 220V to 12V.

On the other hand, the double conversion path is easier because you don't have to modify any of your equipment. You don't have to void your warranty to do it this way. Your choice.

If you have long wires and high currents, the reduced losses from the long distance might outweigh for the losses from the extra conversion.

\$\endgroup\$
1
  • \$\begingroup\$ Though note that very wide ratios like 220 to 12 can be "effectively" 1:1 as far as the converter is concerned if you use a transformer-coupled topology with an appropriate transformer. But transformer-coupled converters tend to be less efficient in general as it makes synchronous switching more complicated. \$\endgroup\$
    – Hearth
    Jan 20 at 17:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.