Go Down

Topic: Two power supplies in series? (Read 729 times) previous topic - next topic

Professor Chaos

I have made two 5v linear supplies to power two RFID readers, each consisting of a standard bridge rectifier + filter capacitor + linear regulator + output capacitor.

Both are being powered from a 14V AC bus.  So the regulators get pretty hot dropping the full-wave rectified voltage.

I thought of putting the supplies in series:   AC+ ---  linear supply --- linear supply --- AC-
on the theory that then each supply would have to drop much less of the input voltage.


Does that make any sense at all?  I tried it and the supplies both work as expected, but I'm not sure if I'm really being more efficient.

xolroc

I suppose that could work, in theory, but linear regulators are not designed for that and this could easily make an unstable power supply.  What I would suggest is just use one of them as normal and get a better heatsink.

majenko

Or switch to using a switching regulator (pun intended) and get 85%-95% efficiency.
Get 10% off all 4D Systems TFT screens this month: use discount code MAJENKO10

MarkT

I'm pretty sure that can't work without shorting the AC input directly across two diodes.  A full diagram would be needed to check.

[ I won't respond to messages, use the forum please ]

majenko

The normal accepted way of spreading the voltage drop of linear regulators is to have multiple stages.

For example, a regulator that takes your 14VAC and drops it down to say 9V.  Then a second one which drops the 9V down to 5V.

It doesn't reduce the amount of wasted power, or the total amount of heat, but it does spread it around reducing the need for massive heatsinks.

You still can't beat a switching regulator though...
Get 10% off all 4D Systems TFT screens this month: use discount code MAJENKO10

jraskell


You still can't beat a switching regulator though...


Quoted for emphasis. ;)

jaydie

How about just converting it to a half wave? Can't you just pull out some of the diodes or replace them?

winner10920

Why do you need two supplies not just one?

Professor Chaos

#8
Jun 18, 2012, 08:23 pm Last Edit: Jun 18, 2012, 08:27 pm by Professor Chaos Reason: 1
These are power supplies for some ID-20 RFID readers;  the installation is in a model railroad system where I have a 14V AC line running around the layout.

Quote
Or switch to using a switching regulator (pun intended) and get 85%-95% efficiency.


I used linear supplies (a) for the obvious simplicity of building one, and (b) I seemed to be having some issues with interference from another switching power supply connected to the system, so I wanted to avoid the noise of a switcher.

Quote
Why do you need two supplies not just one?


I actually tried powering both readers from the same 5V supply, but for some reason it didn't work.  I had built a module that provides optically isolated serial ports and a 5V supply to avoid the noise problem mentioned above.  It works with either reader plugged in, but with both plugged in the readers don't work - not even activating the LED that registers a card read.

It ought to work with just the module's power supply - it's a 1.5A TL7805-5 with the proper capacitors, and I think that should be ok for two readers.  But I threw together another 5V supply, powered the second reader with it, and everything seemed happy.  I know I should try to figure out why that's the case, but the ID-20 data sheet has no info on power requirements, and there is already a lot of voodoo involved.  I had to optoisolate them because the read range would drop to nothing if a monitor's power supply was also connected to the system.

Quote
I'm pretty sure that can't work without shorting the AC input directly across two diodes.  A full diagram would be needed to check.


I don't think it shorts;  the power supply rectifiers are simply in series:

                       +                           +
AC -----  ~rectifier~   ------  ~rectifier~ ----- AC
                       -                            -

Quote
How about just converting it to a half wave? Can't you just pull out some of the diodes or replace them?

I'm not entirely clear on this.... if the filter capacitor on the input of the regulator is big enough, don't you still wind up charging to peak voltage?


So.... still haven't figured out my original question:  is there any drawback/advantage to doing it this way?  Yes, I could built a 9V supply for each if I was worried about heat, but to my simple-minded way of looking at it, it seems more efficient to put the supplies in series than to put them in parallel, each with an extra board to dump more heat.

Docedison

How about just going to a design better suited to power supply efficiency, every watt of power used to produce heat is a watt of power that won't turn wheels... unless perhaps you are really designing a coffee warmer? Pre and post regulation done with linears is just spreading the losses. Remember that a linear regulator is essentially a voltage controlled resistor. A Switcher fills a coil with just enough energy to meet it's set output voltage that is the reason why that topology is so very efficient. Use a separate PSU for the Heavy loads and switchers for the regulated supplies that are critical. A motor doesn't care if the voltage swings a volt or so but an analog measurement or a digital process can be affected by swings and things like the transients or spikes created by servo's, motors, relays and solenoids. A final thought is that IF you really need to use the same transformer and rectifiers then use really big output filter capacitors in the power suppply. The capacitors store energy and when a heavy, short duration load is present the output filters will take up the slack for the main motor or servo supply I personally would use two or three 4700uF caps or more on the motor/servo power rail or source again just to 'Take up the slack" and provide clean or at least relatively clean power. My suggestions have worked well for me in my endeavors for many years.

Doc
--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

majenko

One thing I will mention... if I want a power supply that is both high efficiency and nice and clean, I will often use SMPS down to say 6V, and then use an LDO linear regulator to take it the last volt.  The linear regulator helps to suppress some of the noise of the SMPS, but with it only dropping a volt, it's not wasting too much power, nor getting too hot.
Get 10% off all 4D Systems TFT screens this month: use discount code MAJENKO10

Docedison

2 22uF caps and a 100uHy choke will do the same thing without the extra volt especially the LM2596 @ 150 Khz... real easy to filter and good to 3A uses a 33 uHy choke and a pot and some other small stuff ...

Doc
--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

Grumpy_Mike

You can't put linear regulators in seriese like that. They need a ground and you are not supplying one. You would need a center tapped winding and a negitave regulator.

Go Up