Variable power supply

[DavidDLC]

I want to connect a LM317 with an input of a 24 Volts SMPS.

Has anyone tried this ?

Is there any implications ?

I want to use this as a general variable voltage power supply, the reason for the SMPS is to have the unit more compact.

Please comment.

[Zero999]

The output voltage will be limited to between 21V and 22.5V depending on the current draw, temperature and voltage regulation requirements.

When the output voltage is set very low the output current will be reduced due to the LM317's safe area protection (the current limit typically reduces when the voltage across it is >15V) and the fact that it might be overheating anyway.

Have you considered hacking the SMPs to give a variable output voltage? That's what I would do. Mose SMPSes use the TL431 to regulate the output voltage, which can be tweaked by replacing a fixed resistor with a pot. It should be possible to reduce the voltage down to 4V and up to over 24V, although be careful not to increase the voltage too much because some of the components on the secondary side won't be able to handle higher voltages, you should be fine to 30V though.

The reason why you probably won't be able to go much below 4V is because the TL431 has a minimum voltage setting of 2.5V and feedback is typically taken via an opto-coupler such as the 4N25 which will have a photo diode with forward voltage of 1.5V.
http://www.fairchildsemi.com/ds/TL%2FTL431.pdf
http://www.micropik.com/PDF/4N25.pdf

[alm]

The output voltage will be limited to between 21V and 22.5V depending on the current draw, temperature and voltage regulation requirements.

And the lower limit will be something like 2V.

When the output voltage is set very low the output current will be reduced due to the LM317's safe area protection (the current limit typically reduces when the voltage across it is >15V) and the fact that it might be overheating anyway.

You need plenty of cooling, and might consider a high power cousin.

Have you considered hacking the SMPs to give a variable output voltage? That's what I would do. Mose SMPSes use the TL431 to regulate the output voltage, which can be tweaked by replacing a fixed resistor with a pot. It should be possible to reduce the voltage down to 4V and up to over 24V, although be careful not to increase the voltage too much because some of the components on the secondary side won't be able to handle higher voltages, you should be fine to 30V though.

I would be worried about the design (controller, magnetics) being able to cope with the very different duty cycle.

Something else to consider is that many SMPS's have a specified minimum load for proper regulation, depending on power, this might be something like 100mA (that's one huge power LED). What will the SMPS do if the output of the LM317 is shot circuited? The initial short circuit current is something like 2.5A (should be a graph in the datasheet), will the SMPS survive that? Not all SMPS designs are that stable in all circumstances, since they were not developed as lab supply. There is at least one documented case of someone who build a lab supply with an LM317/78xx and an SMPS, and the SMPS committed suicide, blew up the linear regulator (lots of voltage, since these are basically fool-proof) and whatever was attached to that. Not saying all SMPS's are unstable, but it's not as easy as a linear supply. The SMPS feedback loop is not inherently stable. Computer power supplies are even more fun, since these can make a decent welder (tens of amps continuous, even more peak).

[DavidDLC]

Looks like is not a good idea at all.

[Zero999]

I would be worried about the design (controller, magnetics) being able to cope with the very different duty cycle.

Yes, you're right, although lowering the voltage should be lower risk than increasing it, there's still the possibility that reducing the gain could destabilise the feedback loop, although that might be mitigated by adding an extra compensation capacitor.

Something else to consider is that many SMPS's have a specified minimum load for proper regulation, depending on power, this might be something like 100mA (that's one huge power LED). What will the SMPS do if the output of the LM317 is shot circuited? The initial short circuit current is something like 2.5A (should be a graph in the datasheet), will the SMPS survive that? Not all SMPS designs are that stable in all circumstances, since they were not developed as lab supply. There is at least one documented case of someone who build a lab supply with an LM317/78xx and an SMPS, and the SMPS committed suicide, blew up the linear regulator (lots of voltage, since these are basically fool-proof) and whatever was attached to that. Not saying all SMPS's are unstable, but it's not as easy as a linear supply. The SMPS feedback loop is not inherently stable. Computer power supplies are even more fun, since these can make a decent welder (tens of amps continuous, even more peak).

Shouldn't you be able to minimise all that by choosing a high quality SMPS with over current protection and no minimum load requirement?

The minimum load requirement can be fulfilled by choosing the value of R1 in the LM317 circuit carefully, for example if 12R is used the quiescent current drawn by the LM317 will be 104mA. The trouble then becomes the lower resistor (good luck finding a reasonably priced 2W pot.) so you'll need to use a rotary switch with power resistors. If the minimum load current is too high for R2 then a more practical solution is to use a separate LM317 current regulator to fulfil the minimum load requirements.

[alm]

Shouldn't you be able to minimise all that by choosing a high quality SMPS with over current protection and no minimum load requirement?

Sure, that would help. But the point of this type of exercises is usually to reuse an existing power supply, which is typically one build as cheap as possible (with all protection components removed from the design), designed for one specific load. If you're going to buy a brand new, high quality power supply from a company like Meanwell, you might as well build a linear supply, unless you want lots of current (but the LM317 isn't going to do much good in that case either). It's not like a ~18V/2A transformer is that large or expensive. It will be easier to understand, more stable and have a cleaner output.

The minimum load requirement can be fulfilled by choosing the value of R1 in the LM317 circuit carefully, for example if 12R is used the quiescent current drawn by the LM317 will be 104mA. The trouble then becomes the lower resistor (good luck finding a reasonably priced 2W pot.) so you'll need to use a rotary switch with power resistors. If the minimum load current is too high for R2 then a more practical solution is to use a separate LM317 current regulator to fulfil the minimum load requirements.

That would work, although 2W is cutting it a bit close in my opinion, it will get pretty hot (not comfortable). The 100mA was based on a different supply I had lying around, since David didn't give any information about the power supply, so it may be lower or higher.

[Abhinav]

Hello all  ,

I want to charge 12V 150Ah lead acid battery from my old 400W computer SMPS. but i need to raise output voltage to 14.5 V for charging the battery. as hero999 suggested can i change Tl431 reference voltage to raise the output voltage for charging my battery?
please guide me can i try this hack 


Thank you so much

[Zero999]

It depends on the SMPS, you'll have to reverse engineer the output section.

Unfortunately, the feedback is more likely to be taken from the 5V rail and the 12V rail will be poorly regulated so it might not be possible,

[NiHaoMike]

http://www.qrp4u.de/docs/en/powersupply/

[Abhinav]

Thank you Hero999 and NiHaoMike  for your replies.

Bad news for me that it is not possible to improve output voltage 

Sir NiHaoMike thanks for link but it is very hard for me to hack SMPS by that method 

Sir Hero999 how to check that my SMPS will support TL431 reference changing method or not.


Thank you all.

[Abhinav]

Hello all,

can i use two SMPS and take 12V output from one and 3.3V output from one and add them in series to get 15.3V output for charging batteries. i think we can add 3 high current diodes in series with output to limit voltage around 14V.

Please guide me that am i right or not.


Thank you .

[JohnS_AZ]

I must be missing something here.
Why wouldn't you just build a simple linear supply for battery charging??

[Zero999]

Hello all,

can i use two SMPS and take 12V output from one and 3.3V output from one and add them in series to get 15.3V output for charging batteries. i think we can add 3 high current diodes in series with output to limit voltage around 14V.

Please guide me that am i right or not.


Thank you .

No because they share a common 0V rail.

[Abhinav]

Hello sir hero999 and Johns 


Thanks for your replies.

Sir JohnS here in my city low voltage is a big problem (145V only) . transformer based chargers are useless. and branded SMPS battery chargers are too costly . so o was looking to modify the PC SMPS.
I think i can not use PC SMPS to charge my battery as all of you suggested .

Thank you so much all of you for giving your valuable time 

[JohnS_AZ]

Ahh. That makes sense.
Does your voltage fluctuate, or is it a pretty constant (low) voltage?

[Zero999]

find another higher Vo say 36V, and use LM317 to regulate to your required V.

The trouble with that is that when the input-output voltage is over 15V the safe area protection will kick in and the output current will be reduced. You'll also need a very large heat sink.

If you want DIY, use a 24VAC transformer, 10,000µF capacitor, bridge rectifier and a suitable buck regulator to get the voltage down to the appropriate level. The battery will be destroyed if connected to 14.5V continuously, you need to  to reduce the voltage setting to 13.8V when the voltage hits 14.5V, you can probably use a comparator or use an MCU to do this.

[Abhinav]

Sir JohnS here in my city low voltage is a big problem (145V only) . transformer based chargers are useless. and branded SMPS battery chargers are too costly .

find somewhat higher output voltage charger, say 24V DC. with 145V mains, it will goes down to 14.5V, then you are good to go. to be more safe, find another higher Vo say 36V, and use LM317 to regulate to your required V.

Sir JohnS the mains supply is not stable at all here.
Sir Shafri i am willing to modify the PS SMPS as charger because dedicated ready made chargers are very costly approx 50$ (13.8V 15A) and PC SMPS are available at very low rates 9$ for 400W SMPS.

Thanks to you all for replying.

[alm]

Sir JohnS the mains supply is not stable at all here.
Sir Shafri i am willing to modify the PS SMPS as charger because dedicated ready made chargers are very costly approx 50$ (13.8V 15A) and PC SMPS are available at very low rates 5$ for 400W SMPS.

I'm not familiar with the computer power supplies sold in your country, but I would be wary of any PC supply costing $5 or even $50 to be able to deliver 400W. There are lots of stories of cheap power supplies with high rated power that blow up well below their rated power, modifying low-quality power supplies is not a good idea in my opinion. Building such a supply for 15A is unlikely to be much cheaper than $50 unless you already have most of the parts.

[Abhinav]

Hello sir alm 

Thank you for looking my post.

Sir PC SMPS in india starts from 8$ (sorry i wrote 5$ in my last post) one of the 480W SMPS at http://www.zebronics.net/zeb_480_sata_plus.asp  is available only in 12$ is this very cheap?
sir my English is not so good so i can not understand your last line fully
"Building such a supply for 15A is unlikely to be much cheaper than $50 unless you already have most of the parts."

please explain if you feel comfortable.

[alm]

Sir PC SMPS in india starts from 8$ (sorry i wrote 5$ in my last post) one of the 480W SMPS at http://www.zebronics.net/zeb_480_sata_plus.asp  is available only in 12$ is this very cheap?

As I said, I'm not familiar with the power supplies available in India, but if I paid $12 for a power supply, I would expect something like this 650W power supply that died with 184W load, with no protection and very bad output (although the latter is probably not an issue for a battery charger).

sir my English is not so good so i can not understand your last line fully
"Building such a supply for 15A is unlikely to be much cheaper than $50 unless you already have most of the parts."

please explain if you feel comfortable.

I would find it hard to build a 15A power supply from scratch for under $50, if you have to buy a case, PCB, cooling, transformer, power transistors/MOSFETs and capacitors. So $50 for a 13.8V/15A power supply doesn't seem that expensive to me. But maybe parts are much cheaper over there.

[Jon Chandler]

I've used an LM317 regulator with a 30-something volt wall wart for a computer printer or some such.  Seems to work great, with a minimum voltage of around 1.5 v as I recall.  The switcher in the wall wart seems happy over the range of voltage and current.


Jon

[ddrl46]

I've build a power supply running of a 24V 2.4A laptop charger, worked great. The laptop charger was a switch mode too judging by the size of it.

[Abhinav]

Hello friends 

So finally we found that modifying PC SMPS is not a good option . Well i am not good enough in electronics to design a new SPMS , so i am planning to build a transformer based charger with buck regulation method . but unfortunately this i haven't used buck regulator IC's before . So friends kindly help me building buck regulator circuit.

Thank you 

[alm]

Well, a buck regulator is a type of SMPS . The efficiency would probably be somewhere between a real primary-side switcher (like most commercial designs) and a linear regulator. If you don't have much experience, I would use an integrated solution with as few external parts as possible, like the National Simple Switcher series (there are similar parts from other vendors like Linear Technologies and Maxim). National has a somewhat useful power supply designer that lets you input your requirements and generates a schematic, but it does tend to recommend expensive controllers in small form factors. The LM2575/2577 ('75 is buck, '77 is boost) is a fairly cheap chip, but might not satisfy your current requirements, if it's anywhere near 15A. A linear supply at that current would be quite expensive because of the large number of power transistors and cooling, but a buck converter is more complex. Building a lower power one first to get some experience might not be a bad idea.