6
\$\begingroup\$

I have this crazy idea of using a computer SMPS with active PFC boost to take high voltage DC battery banks (144V+) and drop it down to 3.3V, 5V and 12V.

Here's my thinking: the power supply internally rectifies the AC to DC, and the PFC boost should then boost the 144V to an acceptable 350V-400V for the power supply. The 144V input is okay for it because it falls in the 100VAC range, and most are rated down to 85VAC if not lower.

I'm not looking for a guaranteed solution - it's a one-off problem I'm trying to solve, but I think it could be a cheap and viable solution.

\$\endgroup\$
4
  • \$\begingroup\$ Depends if there is an isolation transformer on the input. Bypassing one that's there would be...dubious. \$\endgroup\$
    – Nick T
    Nov 16 '10 at 3:24
  • 2
    \$\begingroup\$ I've never seen an isolation transformer on the input to a computer SMPS. Maybe a filter choke, but never a transformer. \$\endgroup\$
    – Thomas O
    Nov 16 '10 at 23:45
  • \$\begingroup\$ @barsMonster, Can you please let us know how this worked out in the end? \$\endgroup\$
    – AndrejaKo
    Mar 21 '11 at 19:52
  • \$\begingroup\$ Sure thing )............ \$\endgroup\$ Mar 22 '11 at 0:53
4
\$\begingroup\$

I would not rely on the AC fuses in the PSU - you probably have some HVDC fuses kicking about with that much battery to hand anyway :-)

\$\endgroup\$
2
  • \$\begingroup\$ I don't have the batteries yet, but it's an idea for a future project. I'd have some proper fuses on it I suppose. Would 250V fuses not be good enough? \$\endgroup\$
    – Thomas O
    Nov 15 '10 at 20:54
  • 4
    \$\begingroup\$ You must make sure the DC current interrupting rating for the fuse will be acceptable. The AC rating is usually ~50% higher than the DC rating as the zero-crossings can help to extinguish the arc. \$\endgroup\$
    – Nick T
    Nov 15 '10 at 21:58
4
\$\begingroup\$

For a one-off you could just open it and remove the rectifiers. If you put a high enough voltage on the input then even an active PFC should not complain (it should make the average current proportional to the instantaneous input voltage while maintaining the average output voltage; i don't see a reason for failure here). It it does not have an active PFC then you should be completely safe.

\$\endgroup\$
3
\$\begingroup\$

Technically, any DC power supply powered from a PFC stage is a DC-DC converter, so your idea has some merit.

It mainly depends on the PFC controller IC. Simple ones (which look only at bulk DC for UVLO) should run. More complex ones which sample the input AC for waveform shaping may not work if 'correct' waveforms are absent. YMMV.

The input bridge AC rating shouldn't be far off from it's DC rating. You'd be passing all the power through two diodes, mind you, so they may get hotter than you'd expect (under AC, each pair of diodes in the bridge gets a break every half-cycle).

The warnings about DC fuse ratings are 100% correct. Batteries can be nasty.

If there's some exotic auxiliary supply (quasi resonant) that is expecting sinusoidal AC in, the whole thing may not start. Most ATX supplies aren't exotic, so I expect that the aux converter would be powered from the bulk DC voltage which should be fine.

Good luck!

\$\endgroup\$
2
\$\begingroup\$

Someone once claimed that it was possible to supply 170V DC directly to a computer power supply (switch-mode) and it would work as normal. I can only see this succeeding if the power supply has no transformer (I don't know if this is common for SMPS computer power supplies). Otherwise if it's isolated the DC won't do anything and the power supply won't work.

But if it does work then you have nothing to worry about. Fully-rectified AC is about 170VDC with ripple. It won't hurt to remove the ripple and the power supply will work as normal. I'm fairly sure that 144VDC would be sufficient as well, but the power supply might have to work a bit harder.

You should have no problem with this as long as the supply has no transformer But put some fuses in - I'll bet your battery pack can supply some insane current.

\$\endgroup\$
5
  • \$\begingroup\$ Most PSU's I've found have a voltage doubler option to select between 115VAC/230VAC. In that case, the voltage may be as high as 400V (dc.) \$\endgroup\$
    – Thomas O
    Nov 16 '10 at 15:51
  • 1
    \$\begingroup\$ It's rare for a computer power supply to have an isolation transformer, but afaik, they are all isolated. \$\endgroup\$
    – Thomas O
    Nov 16 '10 at 15:52
  • \$\begingroup\$ A PSU with a voltage doubler switch may work with DC, but only with the switch in the "230" position. \$\endgroup\$
    – davidcary
    Feb 25 '11 at 17:44
  • 2
    \$\begingroup\$ The "isolation" transformer is the main DC/DC converter transformer, which is expecting DC input and is chopping it as required. Even though it's stepping down the voltage, it is providing isolation from primary to secondary. \$\endgroup\$ Mar 21 '11 at 20:47
  • \$\begingroup\$ @ThomasO: The supplies all have to be isolated because the chassis earth and output DC GND are bonded in the supply as well as many devices inside the computer that connect to the supply, which is in turn connected to mains earth and ultimately mains neutral in the switchboard. If the supply wasn't isolated you'd see excessive output voltages. \$\endgroup\$
    – Malvineous
    Jun 2 '20 at 12:30
2
\$\begingroup\$

Dont know if this topic is still alive, but i have been using a bridge rectivier and a capacitor in between my offline ups and the computer. This is just to prevent system restart when there is voltage fluctuation. So, the smps can handle input DC voltage... Hope this helps.

Kripal

\$\endgroup\$
2
\$\begingroup\$
  1. If the power supply has a range switch on the back (115/230), it most likely does not have PFC and you'll have a better chance of powering it up directly from DC.

  2. Make sure that all fuses are rated for the DC voltage you intend to apply.

  3. Do not exceed the expected RMS AC voltage with the DC input - if it expects 208VAC go with 208VDC. RMS is the 'equivalent' of DC as far as power dissipation goes, after all. This will make it less likely that you're going to damage anything downstream from the input.

  4. Yes, you're only using two rectifiers in the input bridge, so they'll see higher power (since they won't be off half of the time) - you may have to upgrade the bridge rectifier to a higher-rated part, or simply jumper the low-side diode and install a single, large (with heatsink) diode on the HVDC side. I wouldn't recommend not using a diode.

  5. Make sure you have some sort of switch that can cut off the HVDC if something goes south, and make sure it's rated for the DC voltage you're applying.

  6. This goes without saying, but: HVDC is VERY effective at killing people. Insulate everything 2-3x more than you think you need to before even considering turning anything on. If the power's on, keep your hands off.

\$\endgroup\$
8
  • 1
    \$\begingroup\$ Point #2 is incorrect. The AC is rectified. 230V is rectified to 325V dc with some ripple. Therefore you should feed it 325V dc. It may not work with a lower voltage. \$\endgroup\$
    – Thomas O
    Mar 21 '11 at 21:09
  • \$\begingroup\$ "Some ripple" is undefined. The average voltage on the caps will be less than 325V, especially if you're doubling 110VAC input. If the converter will work at the lower, stiffer DC input, why push it? \$\endgroup\$ Mar 21 '11 at 22:04
  • \$\begingroup\$ @Madmanguruman, the ripple will be almost zero with no load on the PSU, so you can't say that the average voltage will be less than 325V. Also, don't forget higher AC inputs like 250V which gives you ~350V. I'd imagine that due to the wide input range, ripple is going to be around 50Vp-p. \$\endgroup\$
    – Thomas O
    Mar 21 '11 at 22:06
  • 1
    \$\begingroup\$ #6 is also debatable - DC will not drive your heart into fibrillation. The main risk from DC are burns. AC On the other hand also causes tetanus, or extended involuntary muscle contraction, while DC just causes an initial jerk. \$\endgroup\$ Mar 22 '11 at 5:52
  • \$\begingroup\$ Any voltage source above ELV rating, AC or DC, is hazardous. You cannot 100% say that AC throws and and DC jerks. (I don't intend to test it either.) \$\endgroup\$ Mar 22 '11 at 17:24
1
\$\begingroup\$

I just tried injecting some DC voltage into a 5 V wall-wart power supply (100-240 VAC, switch mode), suitable for powering a small computer like a Raspberry Pi.

The supply did nothing until the input reached 30.2 VDC and then it switched on and operated normally, only drawing 20 mA (it is rated at 2 amps output). Once it was on, the input voltage could drop back but once it got below 25 VDC the output started dropping below 5 V, and the input current increased up to 60 mA before it switched off due to too low an input voltage.

However even at 30 VDC input, the device would shut off (and switch back on a second later) if you tried to draw anything more than a few mA through it, even though it's rated to deliver 2 A. I was surprised it worked at all as low as 30 V, but clearly it needs a higher voltage to deliver any sort of useful current.

Since my bench power supply only goes up to 30 V, next I tried a 96 V battery pack, and I had much more success with this. This time the device drew closer to 50 mA with minimal load, and it increased to 80 mA (at 96 V) when successfully delivering 5 VDC @ 1 A.

As a bonus this device also appears to be isolated, as there was less than 1 V potential between GND and either terminal on the 96 V battery bank.

So for me, it looks like I don't have to find an expensive power supply that can convert 96 VDC to 5 VDC to run an RPi for monitoring my battery bank, I can just use an ordinary wall wart to do the job.

\$\endgroup\$
0
\$\begingroup\$

Is cost a factor? If not, then how about using the batteries in series-parallel to get 24VDC and using a PSU that is made for 24VDC? I had this setup some time ago when I had 5 machines running in a 19" rack. Worked quite well as a very efficient UPS. (Although the 24VDC PSUs are more expensive than standard ones)

I just ran a very simple step-down/rectify/regulate circuit to trickle-charge the batteries.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Problem is with this is that the batteries will discharge at different rates and it then doesn't allow me to power other high voltage equipment. \$\endgroup\$
    – Thomas O
    Nov 16 '10 at 7:55
0
\$\begingroup\$

If you are going to power it with DC, then why not just get a direct DC-to-DC power supply for your computer? This would be much more efficient as well as much less "risky".

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Well, there are also monitors :-) Universal & non-orthodoxal solution would be much more interesting. \$\endgroup\$ Mar 20 '11 at 21:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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