0
\$\begingroup\$

I would like to generate ca. 12.5V, 8-10A, DC power from an incoming voltage between 50 and 400 DC that has been rectified from 3-phase AC coming from a brushless motor acting as a dynamo. I will smoothen out the rectified DC with some capacitors. I'm not worrying about filtering the various stages yet.

I have 10 of these very cheap voltage regulator modules: http://www.aliexpress.com/item/Free-Shipping-10pcs-LM2596-LM2596S-DC-DC-4-5-40V-adjustable-step-down-power-Supply-module/1475560768.html

Datasheet here: http://www.ti.com/lit/ds/symlink/lm2596.pdf

I have already tried setting them up with serially connected inputs (to split the high input voltage). When serially connecting ONLY the inputs of 10 LM2596S modules, applying e.g. 140V to the whole chain of modules and then measuring individual outputs, it seem to work in that each module gives the desired output voltage.

But both from further experimentation (I now have 8 functioning modules left ;) and subsequent perusing of other Q&A here (doh!), I learned that parallel wired outputs of an non-isolated buck regulator, such as the LM2596S, is a no-go for several good reasons.

My (first) question then becomes: What about serially connecting the outputs in stead? E.g. 10x 1.25V (it is the LM2596S adjustable version which go down to 1.2V on the output) to get 12.5V? I have read somewhere that serial connecting the outputs of buck regulators when adding schottky diodes between them might be possible(?).

But even if this works, It seems that I might also hit an possible input/output voltage differential limit of the LM2596. E.g. assuming the range is 4:1 (the TI data sheet does not specify, IFAIK), the modules will only produce my desired output voltages if the input voltage is between 1.25V+1.5V= approx 2.75V (or a bit more) and 4x1.25V=5V, meaning the incoming rectified DC must stay over 27.5V and below 50V in the case of 10 serially connected modules. Otherwise the output voltage will start to rise above 1.25V from each module, if each get more than 5V individually, and as a consequence also the total serially added-up output voltage will rise above 12.5V (which I don't want). Right? Anyone knows if the LM2596S has this voltage differential limit?

Is there another way to re-use the LM2596S modules I have by e.g. putting them in some cascading or interleaving configuration with or without additional components? Or am I better off discarding the LM2596S modules and finding a completely different solution to the problem of adjusting down the variable 50-400V DC input to constant 12.5V DC output?

\$\endgroup\$
1
  • \$\begingroup\$ You wish you had LM2596... They are probably not. Try checking the datasheet, take some given Vin and Vout with a given current, and see if the efficiency you measure is comparable to the one from the datasheet. Or the one from the seller description. In my case datasheet and seller claimed 80+% efficiency, I measured (Pout/Pin) 55%. Fake chip, maybe LM2576 or another completely different. Therefore do not trust any rating from the datasheet. But even if it were original chips, cascading is not a solution. \$\endgroup\$
    – FarO
    Aug 4, 2015 at 13:13

2 Answers 2

Reset to default
3
\$\begingroup\$

Those regulators are not designed for such high-power applications. You can't connect them serially because they were not made to be poly-phase. If you try to connect them in this manner, the feedback loop in the converter will freak out and can cause very unexpected results which may damage whatever you are trying to connect it to.

Your required large input voltage range poses kind of a problem. I'm having a hard time finding a converter that can accept up to 400V.

Regardless, you are going to need a much more suitable solution. The converters you have are not fit for this in any way. You will want to be looking for buck switching mode power supplies

EDIT: Generally you would convert the 3-phase AC down using a transformer. Common DC voltages output by the transformers are around 24v, 12v, among others and this approach is much more easily usable by other switching mode power supplies.

\$\endgroup\$
1
  • \$\begingroup\$ Yeah, those regulators probably aren't fit for purpose. 'Will look at transformer option, although weight is an issue as well. Thanks! \$\endgroup\$
    – user49124
    Jul 10, 2014 at 18:16
1
\$\begingroup\$

Why not use a transformer to step down the mains first, say a 10:1 ratio, then rectify and use a buck/boost module to create 12V.

\$\endgroup\$
5
  • \$\begingroup\$ Yep, seems to be the way to go. Only to find those light-weight transformers... :) ...and thanks! \$\endgroup\$
    – user49124
    Jul 10, 2014 at 18:18
  • \$\begingroup\$ Based on your inputs I think one of these: farnell.com/datasheets/1736663.pdf, 1:40, max 10A transformers which should transform my input voltage to 1.5-10V, plus a simple booster to 12.5V, could do the trick! Thanks again for pointing in the right direction! (hopefully). \$\endgroup\$
    – user49124
    Jul 10, 2014 at 18:48
  • \$\begingroup\$ The only potential problem I see with a 40:1 ratio is your lowest voltage could be 1.25V. I think it will restrict you to a few boost converters than can accept 1-10V and output 12V. I could be wrong as I haven't looked, just going by past experience. \$\endgroup\$
    – ACD
    Jul 10, 2014 at 19:07
  • \$\begingroup\$ hmm, good point. BTW when you say buck/boost, do you mean a module that will output same voltage regardless of whether the input voltage is above or below the desired output voltage? or is it either buck=step-down or boost, but not both functions in same module? \$\endgroup\$
    – user49124
    Jul 10, 2014 at 19:38
  • \$\begingroup\$ The former. They're usually a little more expensive and a little less efficient though. \$\endgroup\$
    – ACD
    Jul 14, 2014 at 12:31

Your Answer

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

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