Power supply design, can someone review my schematic ?

[bleubleu]

Hi!

I am building a regulated power supply inspired by the "Smallbear SmallWart" and the ones on GGG. I just want to know if the whole design makes sense. My electronics skills are very humble and I want to be sure everything is OK before I solder everything.

My requirements :

• Capable of powering quite a few boxes (most of them 9V)
• 9V outputs.
• 15V outputs.
• 18V outputs.
• Dead battery mode (Something like 7-8V).

I bought 2 FP24-250 flat pack tranformers from Mouser. One of them will be wired in series to get 24V and one will be in parallel to get 2x12V. From those, I plan to have four LM317 and some trimpots to get the voltages I need. I guess I could use non variable voltage regulators for 9V, 15V and 18V (7809, 7815 and 7818 respectively) but I want to have the option to change them.

Does this work I did not calculate the exact resistor values, I just want to know if the general design makes sense.



Mat

[CGDARK]

Looking at the schem , I think it may work (depending of the real values used) , but the capacitors orientation are wrong , they must go the other way.

CG

[smallbearelec]

The regulator circuit looks right, but I would not have used those expensive Triad transformers. Rather, I would have gone with cheap, 12 Volt/60 ma. parts as shown in R. G. Keen's Spyder at GEOFEX. There are lots of useful ideas and information in that article, and I suggest that you look it over before going further.

Beyond that, I don't think that the 250 ma. fuse would pop under a short-circuit condition, because the transformer primary draws so little current. This is why I used a resettable fuse for current-limiting in the Small Warts.

If you really want to pursue this, do get some experienced help. Even once you get the circuit right, you have to sweat the mechanical details in order to avoid shock hazards.

[R.G.]

Adding to the previous comments, which are solid advice;

You will not get 12V and 24V from your rectifiers. You will get 1.414 times that, minus two diode drops. Rectifier/capacitor setups detect the peak of the AC waveform they're fed, and with nominally 24V going in, you'll get 24*1.414 = 33.9V and 16.97V.

That's at full loading on the transformers. At low loading, small transformers usually have a considerably higher voltage than their specification. That's because the resistive losses make them sag to the spec voltage under load, so the makers set the no-load voltage higher. This can be 10-15%. So your "24v" may be as high as 37.6V at no load, perhaps more at high AC line voltage. This forces you to buy 50V capacitors, and the high voltage that must be dropped across the regulators means that the regulators must dissipate a lot of power as heat.

Most of this info is in the "Power Supplies Basics" article at GEO, by the way. Good reading for beginning power supply designers.

If you can return your transformers for credit, I recommend that. There are better choices for the transformer voltage.

For 18V out with an LM317, you need a minimum of 20.5V DC going in. The rectifiers need 1.4V for a bridge, and you'll get some sag from the filter caps. Let's choose that at 2% and make it true by capacitor selection later. 2% high on 20.5V is 1.02 times 20.5V = 20.9, then the diodes so you need a minimum DC of 22.31V.

The AC rms voltage from the transformer for that is 20.9/1.414 = 15.78V. We can call that 16V. A 16V transformer would have been ideal. Everything above that gets wasted as heat. If you had the FP16-375 transformers, you would get 16V*1.414 = 22.6V minus 1.4V in the rectifiers or 21.2V minus ripple. This is just about ideal for an 18V output. 15V is fine from this as well.

You'd get half that for the 9V outputs, which is not quite ideal.  8Vac times 1.414  minus 1.4 is 9.9V, too low for the regulators. For the 9V outputs you need 11.5V dc at least, so add 2% ripple (11.27V) and 1.4V for diodes (12.67Vdc) to get a peak voltage on the AC of at least 8.963Vac. You'd need a transformer producing about 18Vac (17.927) to make the 9V outputs work.

Looking back at mouser, there is no 18V device in that family, but there is the FP20-300, giving you two 10Vac/300ma outputs. This is going to be really good for the 9V outputs; same price, more current.

Then there's the matter of safely wiring the incoming AC power. Please - unless you already know how to wire that up so that it works essentially forever without becoming a hazard to someone through the most likely failures DO NOT do this on your own. You could die wiring it up and testing it, and that is a tragedy in itself. But bad wiring technique could kill someone else later after you have sold, given, or junked the supply.

[bleubleu]

Wow, I learn so much on this forum!

Very interesting post RG, i didnt know transformers/rectifiers behave like that. I never considered the diode voltage drop aswell. You and smallbear are right, I did pay WAY too much for those transformers and I probably went a little too high on the transformer ratings, I will probably need to install those heat sinks afterall. I will probably be able to heat my room this winter with my power supply alone!

Thank you very much everyone.

Mat

[PerroGrande]

Another quick note --

You've got a bypass capacitor on the Adjustment terminal.  When selecting your values, keep in mind that the LM317 is only internally protected from the discharge of this cap for voltages less than 25V and capacitance less than 10mfd.  You're okay on the voltage side of things, but if you go larger on the bypass cap, you'll need a diode to protect the LM317. 

The same holds true if you go with large output capacitors (>25 mfd).  Two diodes protects against both -- the LM317 datasheet contains a sample circuit showing the protection diodes. 

For the price of two inexpensive diodes ($0.04 each) (1N4002 is what is shown on the datasheet), you get protection for your more expensive regulators ($0.43 each)...   (Okay -- so the financial argument wasn't so strong...)  You might prevent some downtime, though.  OTOH, you can keep the cap values small and not worry about it at all.   

However, the *best* way is to actually compute the right value of output and bypass capacitor for your intended application.  It is far too common in power supply design to merely throw massive amounts of filter/output capacitance and call it good.  However, this is not always necessary (or efficient, cost-effective, etc).  If the *right* output/bypass cap requires diodes, then you use diodes.

[George Giblet]

The rectifying and filtering looks far too complicated.

Do you really need isolated outputs? for the 18V and 15V and if so why make the 15V not isolated from 18V and the "Battery" not isolated from the 9V.

You need to decide how much current you need from each rail.  Since you will mainly be using 9V and Battery the transformer for 15V and 18V is under utilized.

For maximal simplicity I suggest running 15V and 18V from DC input and 9V and Battery a second DC input (no need for separate inputs).

Here's the simplest I can come-up with:



The 0V (bottom) is common to all.
The centre is for +9V/Battery
The top is for +18V and 15V.


The number of transformers drops from two to one.
The number of rectifiers drops from three to one.
The number of filter caps drops from three (which is drawn as four on your ckt) to two.

[bleubleu]

Hi!

Thx George, that is really clever. I will really look into it. I knew there was a way to simplify things, but I didnt know how! Maybe an updated shematic later this week.

Perro : Yeah, I saw an LM317 circuit with a diode the other day, but could not figure out the its purpose. Thanks for the info, I will dig into the datasheet for the sample circuit you are talking about. I do have a truckload of 1N400x here!

Mat

[bleubleu]

Hi George.

I was just looking at your schematics, can you explain to me why a second rectifier is not needed for the 12V part of the circuit ? Wont this part still be AC ? Can you walk me through it ?

I have just redraw the circuit just to make just I am getting this right.



Mat

[oskar]

I was just looking at your schematics, can you explain to me why a second rectifier is not needed for the 12V part of the circuit ? Wont this part still be AC ? Can you walk me through it ?

But you do need a second rectifier...

skar

[George Giblet]

> But you do need a second rectifier...

You don't.  It works as is.

>Can you walk me through it ?

BTW your drawing looks fine.

I'll explain it two ways.  Believe me, i's not *that* clever.  It's a normal circuit that I've made a small spin on which aids the cap choice for the 24V rail.  It's more a matter of choosing the right tool for the job!

- 24V: Imagine removing the centre-tap winding from the pic.  That means you can also remove the "12V cap".  What you are left with is a normal bridge rectifier and filter cap for the "24V" supply.  That has to work.

- 12V: Imagine removing the 24V part.  If we take out the 24V rail that means we can take out the 24V cap, and the two diodes that feed the +24 rail.  What you are left with is a full-wave rectifier circuit and filter cap.  You will usually see thus circuit drawn with the two diodes feeding the +ve fail and the centre-tap feeding the 0V rail but it works perfectly well with the diodes on the -ve rail.  That has to work.

- Both circuits share a common 0V rail so the two part can work without interferring with each other.

For an example of a full-wave rectifier circuit checkout figure 4 of this page.

http://sound.westhost.com/power-supplies.htm

Now here's how I thought it up:
- You basically were building a power supply which needed two DC rail, one half the voltage of the other
- A very common "DC centertap" circuit is the Full-wave centre circuit, it's used in most of the larger solid state power amplifiers.  You can see this in figure 5 of the page I gave.
- The full wave circuit is usually drawn with with a centre 0V and +V and -V rails but you can simply rename the -V terminal to 0V and get +V and +2*V rails.  Even that form of the circuit would work and you could use a smaller cap for the top if you wanter to.
- What I have done is moved connection the top cap's -ve connection to the 0V.  This means you can tune the caps sizes independently.  It's perfectly valid to do this (and has been done before) and as I've shown above it is basically two common circuits connected with the same 0V rail.

As a side note the Full-wave centre tap circuit (figure 5) can also be considered as common-ed circuits that don't interfere.  The top half is a normal single rail full-wave rectifier circuit and the bottom half is another full-wave rectifier circuit except with the reversed diode connection (like I used for the 12V rail).

[bleubleu]

Oh i get it now! That is awesome. Thanks for the link, I realised that wikipedia also has a great article on Full Wave Rectifiers (which is didnt know about) with schematics very similar to what you are describing (http://en.wikipedia.org/wiki/Full_wave_rectifier).

Cool, now I just need to select my caps and resistor values, create a PCB and I should have a working power supply soon. I'll post the final schematic for review before I create the PCB. I started building pedals in three months ago and I am still running on batteries, i cant way to have this thing running.

Thanks a lot.

Mat

[bleubleu]

Here we go.

The caps values comes from the datasheet (http://www.angelfire.com/electronic/hayles/downloads/lm317.pdf) except for the input caps which I copied from geofex and smallbear.  I added the protection diodes as recommended by the datasheet.

The A/B path will give from anything from 9.25V to 19.5V (so I can do 12V, 15V and 18V) and the C/D path will give me anything in the 7.5V-9.25V.

Comments ?

Mat

[George Giblet]

Good stuff.

The only comment I have is a new 9V battery tends to put out 10.5V or so.  When I design power supplies for effects I tends to make them in the 9.5 to 10.5V region.  I think some of the commercial power supplies do the same thing.

[oskar]

> But you do need a second rectifier...

Got it... finally! 

[bleubleu]

You are right, ive got a brand new battery here. 10.45V!

Mat

[bleubleu]

Another last hypothetical question...

If I put 2 or 3 transformers right next to each other real close, like on the picture, all in the same enclosure. Would there be any side effect or strange voltage fluctuation ? I mean, how far does the magnetic field of the coils/inductors goes ? Can the magnetic field of one transformer affect its neighbor ?

Thanks!

Mat

[George Giblet]

No problem with magnetic fields.

They will run a little hotter pack together, probably OK though.