Jump to content

Core power supply theory of operation.


Recommended Posts

  • 2 months later...

Sorry to cpr this old thread, but what determines what kind of cap and how many to smooth the voltage before and after the regulator?  I have been researching various linear PSU designs and it seems that everyones is very similar (rectifier -> smoothing caps -> regulator -> more smoothing caps) but the number, type, and values of the caps all vary considerably.  Ray Wilsons uses 4 polar 4700 uF with a 1uF polar on the end.  Yours (Smash) has one lytic and one non-polar one each side of the regulator while Seb Francis' bench PSU has one polar and one non-polar plus a load resistor and a reverse protecting diode. 

So I guess my question is what exactly is the theory behind the different cap types?

My other related question is a little more specific to what I'm working on which is a +12/-12/+5 PSU from a 2 amp center tapped 25VAC transformer.  Would it make sense to use a second rectifier behind the first to drop the voltage further before going into the 5V regulator or just driving it with the output of the main rectifier (12.5 VAC) directly the way to go?

Link to comment
Share on other sites

  • 1 month later...

You ask many good questions.

I'll try my best to answer them.

What kind of caps and how many, before and after the regulator:

As you can see from the cool graphics that are at the start of this thread,

the job of the "smoothing" capacitors is to average the voltage across the

cap. To "smooth out" the peaks and valleys. The two most important specs

of a capacitor are the capacitance and the voltage rating. The capacitance tells

how much smoothing you'll get (more is generally better) and the voltage rating is

a maximum for that cap that you must stay below.

Bigger capacitors are usually Electrolytic, these are usually polarized, and so they

should have markings to show which leg is + or -. Another high capacity type

is "tantalum". They have high capacity in a small package, and usually don't

degrade over time like Electrolytics, but on the downside they are more expensive,

and they are not tolerant of voltage spikes.

How much capacitance do you need? That is math beyond my ability. It depends

on the current demand of your device, the properties of your transfomer, and how

smooth you need the voltage to be. Especially with electrolytics it's a good idea

to go a bit over the minimum requirement, as they may vary by 20% right at the start,

and will degrade with time.

As long as there is load on the supply, there will always be some ripple after the first

capacitors. Each regulator requires a certain voltage to do it's job. A typical 7805 may

require 7 volts to assure a good 5 volt output. If the ripple (at it's worst under full load)

ever falls below that minimum, the 5 volt supply will suffer dropouts. See the datasheet

for your regulators to learn about the minimum input voltage requirements. The Texas

Instruments sheet I just read states that the input must be 2 volts over the output

for a 7805, 2 and a half for most other voltages. So if you're feeding a 7812, you'll need

no less than 14.5 volts at the input under worst case conditions.

The total capacitance needed may require a bigger cap than you have room for. In the case

of the example given above by Ray Wilson, there are 4 4700 uF caps. These 4 together are

less tall than one much bigger cap. I expect he's planning for a low profile case, like a 1U rack

case or something.

So if it's about "total capacitance", why include a dinky little 1uf or 0.1 uf ceramic? These

smaller caps are more effective at smoothing higher frequencies. The main ripple in the supply

will be at 100Hz or 120Hz, depending on your country, but much higher frequencies can be stopped

much better with little ceramic caps.

OK, all that for the "smoothing Caps", now on to the regulator. The popular 78xx and 79xx regulators

are "linear" devices. They require a bit more voltage on their inputs in order to provide a smooth, clean

regulated output voltage. As an example, let's say we have about 15Volts DC, and we want regulated

+5 and +12 volts. We supply the mostly smoothed +15 volts to the input of the 7805. The output should

show something very close to +5 volts. We do the same with the 7812 regulator, and we get 12 volts out.

It all tests fine, so we connect our toys, and after a few minutes, it shuts down with a burning smell.

What happened? These "linear" regulators have to "burn off" the difference in voltage between the input

and the output. The greater that difference, and/or the greater the current, the more heat is generated.

In our example, we might be drawing 50ma at 5 volts. The difference between +15 volts in and +5 volts

out is 10 volts. that 10 volt difference at 50ma = 500 mw of heat (1/2 watt) That's a bit warm.

Now try 500 ma at +5 volts, and suddenly you're dissipating 5 watts through your little 7805 regulator.

This is why heat sinks are needed. They are radiators that help to dissipate that heat into the air.

Here's a tip. the positive 78xx regulators happen to have their mounting tab at ground. So you can mount

them directly to your metal project case, which should be grounded anyway. This lets you use your case

as a big, free heatsink. Don't try that with the negative regulators though, their tab is not grounded.

Get a bit of "thermal paste" or "thermal grease" and put a dab between the tab and heatsink. It'll help

carry the heat across to the heatsink.

After the regulator, there are a few more caps, though not usually as big as the previous ones. These help

to smooth the load being placed on the regulator. It is also a common practice for smaller ceramic caps to

also be placed at each chip. These help to reduce spikes generated by (especially digital) signals being


You asked about the "extra" diodes and resistors in Seb Francis' designs. These may be needed

for some systems. Here's what they are about.

Some regulators require a minimum load to regulate well. Many designs have a very light load, especially on

the negative supplies. Adding a load resistor may help the regulator to do a better job.

In some systems, when the device is turned off, the input voltage may fall faster than the output voltage.

This is a bad thing for the regulator. If your input voltage is gone, and you still have a bit of charge in the

caps after the regulator, the regulator may be damaged. Adding a diode from the regulator output to the

regulator input prevents this problem. Another diode that is sometimes added is from the output voltage to

ground, so that if ground ever rises above VOut, the diode will shunt that voltage.

You can see examples of all these things if you download the data sheet for a TI 78XX regulator.

Finally, extra rectifiers to drop the voltage? You'll drop between 0.7 and 1.4 volts per diode, but it all adds up

to the same amount of heat. You're just moving some of it from the regulator to the diode. That works. You can

also get high wattage resistors to drop the voltage. Just calculate for the maximum current you'll be  drawing,

and make sure that the result never falls below the input voltage required for your regulator.

I know these answers are for an old question, but I think power supply basics are useful for a lot of people here.

I hope this response helps someone.


  • Like 1
Link to comment
Share on other sites

  • 6 months later...


Yes, for power supply circuits, you can usually substitute a bigger cap for a smaller one, as long as the voltage rating is high enough.

The reason you usually see two caps. One polarized and the other not, is this: The bigger polarized cap is reducing ripple at the power supply frequency (100 or 120 Hz, usually) and the smaller ceramic cap (not polarized) is absorbing high frequency noise, like you might get from a digital circuit running at 40 megahertz.


Link to comment
Share on other sites

  • 1 year later...

I think this deserves a sticky in the SID section if it all possible.

Out of the 6? months ive being on here, this was my first time noticing this great Diagram.

Would be an asset to all the noobs if it was in there, or even a link in the MB Sid manual.

Just some honest advice.

Thanks smash, that really simplifies things.

Link to comment
Share on other sites

  • 1 year later...
  • 3 weeks later...

Nicely done!  One comment:  wouldn't the voltage level on the output side of the 7805 be lower than the input voltage coming into the 7805?

Sorry folks, I just cannot see any voltage coming into the 7805, where is it please? Can someone else not see it too?  ;)

cheers Johnh

Link to comment
Share on other sites

  • 1 month later...
  • 7 months later...
m working on which is a +12/-12/+5 PSU from a 2 amp center tapped 25VAC transformer

Selectronic.fr sells one at 4,90€ with 5V/1A and +/-12V at 250mA, ELPAC trade


definitly the good stuff to get, and everything is included, you plug it on DIN5 socket !

Link to comment
Share on other sites

  • 3 years later...

Lllehaze, anybody interested in midification;


Great explanation on PSU design!

You bring up an interesting point regarding use of caps for reduction of spikes, especially with operating frequencies above 40mhZ

My midified organ console is built with midibox components from SMASHTV, some dating back to 2004.  Back then, the small caps were not used on each chip as they are now on the recent kits.  Recently I upgraded  to an LPC-17 core merging it with an old 8bit core.  the LPC-17 is used primarily for encoding 3 keyboard, and  64 SAMS contacts  thru 2 attached KB cards.  The 8bit core has 4 DIN  and 4DOUT  which handles pistons inputs, and SAMS coil drivers..

The system functions, but it is experiencing "cross talk", for lack of a better word.  SAMS are being actuated by pistons when not  set .  The gen cancel  clears set SAMS, but also turns on SAMS that were off as well.

The system uses jOrgan and strangely, the behavior described occurs only when using the hardwired devices on the console.  Actuating stops or pistons with a mouse on the display does not produce the same effect.

The power supplys are simple; The LPC through the USB connection to the PC, and the core 8 thru a wall wart module.  The SAMS operate with a 12vdc - 30amp powersupply.  The negative side of the 12VPSU  is common with the negative side of the core8 PS by virtue of the card design, and the ULN2803 drivers mounted on the DOUTS.


Actually, there is a lot of 26ga wire floating around to act as transmitters and receivers, probably not good!


1.  Would addition to small caps on the core8 DINS and DOUTs help?

2.  Would going to a better common, better filtered, regulated power supply for the  LPC and the Core8  Help?

3.  Would putting some type of caps on the SAMS coils help dampen out high freq. spikes and switching noise?

Hopeing for some help, suggestions, etc.


Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Create New...