Dual Supply for Low Power but Sensitive Circuit

[miallen]

I need +12V and -12V for a bench measurement device but the power will also make it into the signal path so I'm concerned about what parts I use for the supply and regulator.

For the transformer I was wondering if a small toroidal transformer like this would be good:

http://www.newark.com/multicomp/mcfm32-12/miniature-toroidal-transformer/dp/38K4645

However it's not crystal clear to me if the secondary can be used to get +-12. Can this be wired so that it's like a center tapped 12-0-12?

Is there really any advantage to using a toroidal vs. a simple cheapo 100mA center tapped transformer?

Also, can someone recommend good positive and negative low power voltage regulators? The circuit only needs ~4mA. But again power is going to be coupled directly into the signal path so I need great ripple rejection. I was thinking there are probably some little TO-92s that are used for this sort of thing all the time. If yes, what are they?

Mike

 

[paulb]

Yes, the transformer will work. You'd wire the secondary windings in series to get a total of 24V (12-0-12). Read the datasheet warning about mounting arrangements. It doesn't give information about how to connect the leads so you may have to experiment - if you connect the secondary windings with the incorrect phase you'll get a total of close to 0 volts instead of 24.
TO92 regulators are 78L12 and 79L12. You'll probably get a lot of posts about how they aren't very good. They are, especially the 'A' versions.

 

[paulb]

This article and the other articles at the same site are worth reading:
http://sound.westhost.com/articles/power-supplies2.htm

 

[miallen]

I've been looking at 78L12AC/79L12AC and the typical ripple rejection is 42dB where as if I were to use LM317/LM337 with an extra cap I can get 77dB. That's a pretty significant improvement.

Given I don't care about wasted space and using an extra cap, should I just use the LM317/LM337? Does it matter than I'm only loading it with ~4mA?

Also, as a side question: Regarding LM317/LM337 I read that the output voltage of the transformer secondary should not be a lot more than the output of the regulator or it will be forced to dissipate heat. But does that mean maximum output voltage or the output voltage actually being output at the time. Meaning if my secondary put out 36V and I adjusted the regulator to put out only 5V would it dissipate more heat than if I it adjusted to put out 30V? I assume the answer is "yes" but I just want to be sure (not that it matters in this particular case because I can just use a 15V secondary).

 

[richie00boy]

if my secondary put out 36V and I adjusted the regulator to put out only 5V would it dissipate more heat than if I it adjusted to put out 30V?

Yes. Because power is voltage multiplied by current.

 

[MondyT]

Hi

Data for the 1.6VA transformers can be found here and shows a schematic for the range...

http://www.farnell.com/datasheets/79592.pdf

The secondaries on these transformers can be wired for series and parallel connection and if you require a centre tap then Black and Yellow leads should be connected together to form this.

One thing to note here is that the primary is a tapped wind and so if you apply 115v input to the Blue and Grey leads, then the Brown lead will need to be insulated well as this will then be at 230v potential

Hope this helps
Ray

 

[MondyT]

Sorry I pasted the wrong link!!!

Here is the correct one...

http://www.farnell.com/datasheets/320644.pdf

Cheers
Ray

 

[miallen]

Yes, the transformer will work. You'd wire the secondary windings in series to get a total of 24V (12-0-12). Read the datasheet warning about mounting arrangements. It doesn't give information about how to connect the leads so you may have to experiment - if you connect the secondary windings with the incorrect phase you'll get a total of close to 0 volts instead of 24.
TO92 regulators are 78L12 and 79L12. You'll probably get a lot of posts about how they aren't very good. They are, especially the 'A' versions.

I just built a supply using 78L15AC / 79L15AC and a torroidal with 22V secondaries and it doesn't work. Output is +1.246 VDC and -0.776 VDC.

Here's the data sheet on the transformer:

http://www.farnell.com/datasheets/320644.pdf

I connected black and yellow together to make the virtual ground.

I use the circuit from Practical Electronics for Inventors p607 figure 10.10b. I don't have a schematic drawn up because I've been working from the book but it's just a rectifier followed by two 4700uF caps between the input and gnd, then the 78L15AC / 79L15AC, then 10uF caps between out and gnd, and finally two diodes between out and gnd. It's a trivial circuit. I don't see how I could have messed it up.

One thing that is strange is that when I measure the secondaries, I get 27.66 VAC and 27.72 VAC. The output of the rectifier is +33.95 DC and -33.08 DC.

So everything is a bit higher than I thought it would be. Why am I reading almost 28 VAC from the secondaries of a transformer that clearly shows on it's side that it has 22 VAC secondaries?

Also the rectified voltage of 33 VDC is a bit higher than I expected. Is this just a false reading from the volt meter because it's not really flat DC?

The 78L15AC / 79L15AC input voltage is supposed to be between 17.5 and 30 so maybe I'm overloading those regulators? The datasheet says they can accept a max of 35V:

http://www.st.com/stonline/products/literature/ds/2145/l78l15ac.pdf

Any ideas would be appreciated. I'll try to provide a schematic tomorrow.

Mike

 

[MondyT]

Hi Mike

If you look at the data sheet for the transformer you will see that the rated voltage for each secondary is 22v *On-Load* This means that the voltage on each secondary will be 22v only when supplying 0.159A into a resistive load (A rectifier with capacitor smoothing is anything but a resistive load!). You can see in the data that the Off-Load volts states 27.0V and this is the voltage of that the secondaries will read with out any load connected and this is what you are in fact reading. The difference between the off load volts and the on load volts is termed the regulation of the transformer

These voltages all assume exactly 230v or 115v input on the primary, your mains input maybe fractionally higher than this and therefore you get 22."something" volts.

When connected to a capacitor filtered rectifier, the load is capacitive and therefore the voltage on the secondary will rise above the rms rating to a maximum of 1.414 x the rms voltage, or the peak value. The load seen by the transformer is complexed, but around +/-33v is correct for the no load condition and I would expect to see around +/-29v or so when you load the rectifier to 100mA DC. The transformer you chose will be overloaded in this condition however as the rms current pull from the secondaries will be in the order of 230mA and the transformer is only rated at 159mA!!

You also seem to have quite a high difference between your positive and negative DC voltages. They will rarely be exactly the same but you have nearly a volt difference here so a check on your rectifier/wiring may be worth while.

I would need to see your schematic and wiring to determine why you may not be getting any output from your regulators. If you only require a +/-15v DC supply then the 15+15v transformer will be a better choice as this is rated at 233mA and the dissipation in your regulators will be be much better

I hope this helps

Cheers
Ray

 

[miallen]

I would need to see your schematic and wiring to determine why you may not be getting any output from your regulators. If you only require a +/-15v DC supply then the 15+15v transformer will be a better choice as this is rated at 233mA and the dissipation in your regulators will be be much better

Hi Ray,

Thanks for the help. Here are a few pics including a schematic, layout diagram and pictures of wiring:

http://207.192.69.113/~miallen/OpAmpSupply15V/

I found a missing solder joint between positive out and it's diode.

With that fixed now I'm reading a 33.0VDC and -0.777. So it seems all of the potential has fallen over to the positive side with nothing for the negative side. Strange.

Regarding the transformer, I was under the impression that it needed to put out a little more than +-15 to drive the 7815/7915. There is another circuit in the book (on the same page) that shows +-18 to put out +-12 so I extrapolated and came up with a 22V transformer. But I did not fully understand how RMS vs peak factored into everything so I will try a +- 15VDC torroidal (although that will take a few days to get that of course).

Still, it seems this circuit should work even though I'm putting in 33V rectified?

I wasn't sure what specific sort of rectifier I should get so I just went with this 2N254:

http://www.vishay.com/docs/88532/3n253.pdf

I used some solder brade to heat sink the voltage regulators when I soldered them but could I have fried them?

Any ideas?

Thanks,
Mike

 

[MondyT]

Hi Mike

If your wiring is correct, first thing to check is that when the red and orange transformer leads are disconnected from the rectifier, see that you have 54v rms off load between the black and orange, and 27v rms from either of these leads to the black/yellow centre tap. If not try reversing *one* of the secondary connections.

Cheers
Ray

 

[pacificblue]

That transformer is too small to charge 2x4700 µF capacitors to ~29 V and may just be sufficient to maintain them charged with no load at the regulator output. To compensate for missing power it has too high voltage. That means you can reduce those 4700 µF to 100 µF and achieve a good regulated voltage. But the power dissipation will be much too high for the 78L/79L.

Your two options are
- replace the 78L/79L with the TO-220 version of the 78/79 regulators and add heatsinks, replace the 4700 µF capacitors with 100 µF and keep the transformer.
- replace the 78L/79L with the TO-220 version of the 78/79 regulators and add heatsinks, if the load draws more than 50 mA and/or the ambient temperature around the regulators is more than 25 °C, buy a transformer with 2x15 V secondary voltage and at least 30 VA to cope with the 2x4700 µF and 100 mA load at the output and keep the capacitors.

 

[MondyT]

Good point PacificBlue!

I doubt if it would drag the negative rail down to 0.7v so I would think there is still a problem somewhere.

Ray

 

[miallen]

Problem solved.

The VRs were in backwards. My make-shift diagram is a top-view but the pin layout in the datasheet is a bottom view and my brain was stuck on top-view mode so the pins ended up reversed. Lesson learned.

But there were other problems as well. It seems that after meddling with it so much, I fried at least one of the chips and probably a capacitor because I had no luck trying to resurrect the circuit. I just started over with a new circuit, with all new parts, tested frequently and it worked like a champ the first time.

I have not tried the existing 22V torriodal with the new circuit though. I have just been testing with two 9V batteries in series because I'm worried that the 33V coming out of the rectifier a little too close to the 35V limit of the larger caps. I'll just wait for the 15V torroidal. I'll use the 22V one for a little bench supply maybe.

Anyway this was all mostly just an exercise so I think I got what I was looking for out of this. Ultimately I want to see what sort of ripple I get from this 78L15AC/79L15AC circuit as opposed to an LM317T/LM337T supply. The LM317T/LM337T supply will probably be the one that I ultimately use for the device in question. I also have a single set of LT1962/LT1964 chips and some surfboards are on the way but I haven't tried SMT soldering yet so that will have to wait until stop making stupid mistakes.

Thanks to all who replied. If you haven't noticed, I'm not an EE so I appreciate the help.

 

[MondyT]

Well done Mike!

Remember, as PacificBlue said, either drop the value of the capacitors or up the VA of the transformer to say 15 or 30VA as the 7VA transformer will not cope with the circuit as it is.

Cheers
Ray

 

[miallen]

Remember, as PacificBlue said, either drop the value of the capacitors or up the VA of the transformer to say 15 or 30VA as the 7VA transformer will not cope with the circuit as it is.

I did switch to 1000uF capacitors and I will be getting the 2x15V / 0.233A torroidal.

However, I must admit that I did not understand PacificBlue's comments. The power supply only needs to put out maybe 5mA. So why do I need all of that power?

 

[pacificblue]

The transformer needs to charge the capacitors. A discharged capacitor is very similar to a short-circuit. The inrush current would be several tens of amperes, with a bigger transformer. The 7 VA transformer cannot deliver that amount of current. The voltage will collapse, when you switch that circuit on. The capacitors charge to that lower voltage, their impedance rises, the transformer voltage does the same. If nothing else but the capacitors were in the circuit, the voltage would probably reach 29 V after sufficient waiting, during which the transformer is severely overloaded. In your circuit, however, the voltage regulators are discharging the capacitors, while the transformer is still trying to charge them. Therefore the full voltage will not be reached. One of the regulators is obviously faster to clamp the voltage, so the secondary voltages are unbalanced. That is a possible explanation, why you get the full voltage on one regulator and next to nothing on the other.

http://hyperphysics.phy-astr.gsu.edu/Hbase/electric/capchg.html

 

[miallen]

I understand (mostly).

So given the size of a capacitor and a voltage, how do I roughly compute how much power will be required to charge and sustain that voltage?

For example, if I have a 15V torroidal (instead of 22V) and 2x1000uF capacitors, will 7VA be sufficient?

You stated I need 30VA for 2x4700uF caps with 100mA load. So if I use 1000uF that is a factor of 0.21 times 30 is 6.3VA and the load will only be ~10mA.

Regarding the TO-220 vs TO-92, is it also a problem that a larger capacitor can deliver too much current to a TO-92 and that's why I need a heat sink?

It seems to me if I do not draw that much power from the voltage regulator and provided that the regulator input voltage is not incorrectly high (forcing it to dissipate the extra power as heat), then should the TO-92 be ok?

I'm trying to get low ripple in less space but it seems that large filter capacitors (low ripple) without heat sinks (less space) are incompatible?

 

[AndrewT]

One thing that is strange is that when I measure the secondaries, I get 27.66 VAC and 27.72 VAC. The output of the rectifier is +33.95 DC and -33.08 DC.

So everything is a bit higher than I thought it would be. Why am I reading almost 28 VAC from the secondaries of a transformer that clearly shows on it's side that it has 22 VAC secondaries?

transformer regulation and mains voltage variation.

 

[pacificblue]

So given the size of a capacitor and a voltage, how do I roughly compute how much power will be required to charge and sustain that voltage?

You don't need a lot of power to charge it. The point is that you need it in a very short time span, usually in the first half of half a mains cycle. The formula is 1 F = 1 As/V, so to charge 1000µF to 20 V, you only need 0,02 As. The first half of half a mains cycle is 1/240 S, where you live, which means, the transformer would have to supply an average of 4,8 A during 1/240 s. It can probably not do that, so it takes several cycles at overload to achieve the task.

To sustain the capacitor voltage you need to refill, what is taken out by the load and the regulators. Also not much power. P=U*I, e. g. 30 V * 10 mA = 0,3 W.

The problem could be that the regulators are trying to build up a stable voltage and draw more current from the capacitors than the transformer can supply to fill them up. You could make two tests. 1) put a switch between the capacitors and the regulators. Measure the voltages. When (if) they reach ~30 V on both capacitors, switch the regulators on. Maybe it works like that. 2) Replace the capacitors with 100 µF and see, if it works without the switch.

For example, if I have a 15V torroidal (instead of 22V) and 2x1000uF capacitors, will 7VA be sufficient?

Probably.

Regarding the TO-220 vs TO-92, is it also a problem that a larger capacitor can deliver too much current to a TO-92 and that's why I need a heat sink?

The current is determined by the load. The heat dissipation is a result of (Vin-Vout)/Iin, where Iin is Iload+Iregulator. If you assume 30 V in, 12 V out and 100 mA the regulator has to dissipate 1,8 W of heat. With a thermal resistance of 140 K/W that means the TO-92 will be ~250 K above ambient, while the maximum temperature is defined as 150 °C. The TO-220 has 50 K/W -> 90 K above ambient, which could work up to 60°C ambient. If that circuit lives inside a case, 60°C can easily be reached and surpassed, which means heatsinks are needed.

If your assumption about 10 mA maximum current draw is correct, we have to add 6 mA quiescent current of the regulator to that and only get a disspation of 0,288 W. That means 40 K above ambient, in which case TO-92 is okay with no additional heatsinking.

I'm trying to get low ripple in less space but it seems that large filter capacitors (low ripple) without heat sinks (less space) are incompatible?

The low ripple is provided by the regulators with the small capacitor at the output. The capacitors at the regulator input are there to maintain the voltage high enough, so that the regulator can work, i. e. above 12 V + what the regulator needs to maintain a stable output. Therefore you only need 100 µF, when you have ~30 V DC, but 4700 µF, when the voltage is below 18 V DC, to maintain the lowest point of the ripple voltage before the regulators high enough.