Building a Linear Power Supply - Capacitor Value Question

[norman]

I decided to bite the bullet and build, not buy, a transformer-fed 12 volt DC regulated linear power supply with half wave rectification for my phono stage to drop the noise floor as far as I can, and get away from the wall wart that came with it.  The design is from the "Short Circuits 3" book.

 

I decided to try a kit for the solid state regulator component, and a nice multi-tapped transformer as the AC source.

 

The trafo provides AC +12 V positive, and the rectifier needs at least 2.5 volts more than this to work.  How do we get the extra volts out of a 12 volt transformer? 

The answer it seems is that the AC voltage from the "12 volt" transformer is of course RMS, not peak voltage, which is actually 1.414 x the RMS voltage, so around 17 volts AC at its peak.

 

I have bought a biggish case to fit it all into - plenty of air space and ventilation, and lots of room to keep the high voltage and I presume noisy transformer away from the regulator board.

I have already drilled out and fitted stand-offs, and fitting bolt holes, and supply and output fitting openings. Tonight I will start soldering the board and components in place.

 

The regulator only has a few components including a rectifier, 2 diodes, 2 capacitors, and a resistor.

 

I do have a question - the supplied smoothing capacitor is a 1000 uF 25 volt polarised electrolytic in the design.

Would I be better to substitute a bigger capacitance but same 25 volt value smoothing capacitor?- I can get a 2200 uF 25 volt low ESR cap or a 4000 uF ordinary cap that will fit right in? I understand that regulator circuits like low ESR components best, so is the 2200 uF part better? Or should I stick to 1000 uF and get a low ESR cap? The lower the ESR, the lower the ripple and therefore noise. I have seen similar pre-made regulator boards\ with 4000 uF caps, hence my question.

 

Another question - is it OK to twist wires pairs together?  For example the wires that will lead from the regulator board to the output connection?  Would this have any impact on noise?

 

Thanks in advance.

 

Edited November 20, 2019 by norman

[acg]

@norman, the smoothing capacitor/s is important in a linear power supply, especially when there are not enough AC volts in comparison to the DC volts required.  In your case, where you are looking for 12VDC but are using a 12VAC mains transformer, have 2.5v loss across the rectifier (which is a lot for low voltage, you should be able to get this to less than a volt with better rectifiers) I would be looking for at least 5,000uF smoothing capacitor, preferably 10,000uF so that the ripple after the rectifier will not cause the LDO to turn-off every cycle when the level of the sine drops below the output voltage plus the dropout voltage of the LDO.  The more smoothing capacitance you have the less ripple you have going into the LDO which means that the sinewave troughs are higher and the LDO is less likely to turn off.

 

Then you need to consider the regulation of your power transformer, which is basically a number that lets you know how much the AC voltage it supplies will drop under higher load conditions i.e. when pushing plenty of current.

 

EDIT:  My advice, twist all wire pairs together regardless of whether they carry AC or DC

Edited November 20, 2019 by acg

[aussievintage]

14 hours ago, norman said:

12 volt DC regulated linear power supply with half wave rectification for my phono stage

 

You should use a full wave, not a hald wave recitifier.  Much less ripple after filtering.

 

14 hours ago, norman said:

The trafo provides AC +12 V positive, and the rectifier needs at least 2.5 volts more than this to work.

Rectifier, or regulator?  The rectifier diodes will drop an amount, maybe 0.7 volts each (two if using full wave)  but the regulator drops even more voltage (to reduce the raw filtered DC to the desired regulated voltage),  and depending on it's design it may require a minimum voltage difference to work - often 2 or 3 volts.

 

14 hours ago, norman said:

I do have a question - the supplied smoothing capacitor is a 1000 uF 25 volt polarised electrolytic in the design.

Would I be better to substitute a bigger capacitance but same 25 volt value smoothing capacitor?- I can get a 2200 uF 25 volt low ESR cap or a 4000 uF ordinary cap that will fit right in? I understand that regulator circuits like low ESR components best, so is the 2200 uF part better? Or should I stick to 1000 uF and get a low ESR cap? The lower the ESR, the lower the ripple and therefore noise. I have seen similar pre-made regulator boards\ with 4000 uF caps, hence my question.

It's probably OK to put a bigger cap in there, but in some cases it can cause problems.  The bigger cap will have bigger inrush current so the circuit must cope with that.  The best thing to do is to use an LC or RC filter stage after the capacitor already in there.  However this also cause voltage drop, which you might have a problem with because of the above concerns you mentioned.

 

14 hours ago, norman said:

is it OK to twist wires pairs together? 

Yes.

 

46 minutes ago, acg said:

Then you need to consider the regulation of your power transformer, which is basically a number that lets you know how much the AC voltage it supplies will drop under higher load conditions i.e. when pushing plenty of current.

This may be a concern, but hopefully the transformer voltage spec is at the rated load.

[norman]

Hi all. Thanks for the advice provided. The transformer I am using is well specified for the proposed use. I have checked and I think I will be using full, not half, wave rectifcation, so my mistake. The voltage my phono pre-amp uses is 12 v DC, so it's appropriate. I will twist all internal wires, plus as I have decided to use a metal case I will  use an IEC C14 input connector and I will earth the case using crimped, not soldered attachment rings. Not sure if I can shield the trafo better. Cheers, Norm

Edited November 21, 2019 by norman

[stereo coffee]

Hi Norm

You will attain 16.9v with full wave rectification and a 1000uf capacitor.  The rule of thumb is 1000uf for every amp of current the circuit draws, exceeding the capacitance rule of thumb, does little, other than risking rectifiers breaking.  

 

i suspect the parts you have for achieving 12v + single rail is the 7812 regulator. It is not a good device for any audio circuit. See the earlier thread here that discovers internal resistance between the device output and ground, the opposite of what audio circuits require. .   

Instead of the 7812 lets design a good audio purpose regulator. First off what is the circuit- do you have a link to it 

showing its schematic ?  post that in this thread, and I will design you a better one. 

 

Edited November 21, 2019 by stereo coffee

[norman]

Hi Stereo Coffee. Greetings from across the ditch.

 

Yes, it is a 7812 regulator. Thanks for the link, I will read it when I get a chance.

I have completed it, it delivers 12.1 volts steadily according to my Fluke 87 true RMA multimeter.

Here's the circuit, noting that it is powered with a 12-18 volt (18 VA), 1 amp multitap transformer.

I made a small mod - I have also used a 4,700 uF capacitor in place of the 1,000 uF shown in the circuit. Should I go back to the 1,000uf unit instead as shown in the design below??

And I am still in error - it is only half wave rectification I now see, not full wave. It would be great to have full wave instead, but I couldn't find a circuit....

 

I would greatly appreciate any assistance you can give me with a better circuit design. I'm trying to cut down power source noise arising from using a wall wart

 

Many thanks, Norm

 

 

 

 

 

 

Edited November 21, 2019 by norman

[stereo coffee]

Let's begin our quest for a better 12v regulator with a

1. Plan to improve the rectification 

2.Provide current regulation to the exact circuit requirements

3.Provide Voltage regulation to 12v DC+  

4 Provide High impedance good for audio circuits. 

 

Sometimes in circuits it is better to regulate current rather than voltage, but your requirement is for 12v DC, so let's do both.   

 

The LT4320 ideal diode bridge Controller in a DIP package   https://www.analog.com/media/en/technical-documentation/data-sheets/4320fb.pdf    is our starting point, we can see it uses just 6 additional parts to provide a DC output. 4 of those parts are N channel mosfets. Lets choose the IRF540 as our mosfet , although many other N channel varieties will also work like the AO4408 

 

Next we need a small capacitor located close to the positive and negative output of the bridge controller suggested as 1uf ( 105) , and our last part is a 1000uf capacitor rated at 35v or more. 

 

The result will surprise you to just how much DC is developed, as the LT4320 defies the usual divide by .707 relationship of converting AC to DC and will typically provide 18 volts or more when 16.97 suggests as being the maximum achievable- such is the efficiency of this marvellous circuit. 

 

Next we need to regulate current , you need to observe how much current your circuit needs,  our choice is the LM317 arranged with its adjustment leg as the output and a resistor calculated as Vref/R between output and adjustment. We can see Vref is 1.25, so divide that by the resistance shown in the attached schematic as 220R. with 220r Yes we achieve 57ma of current but your requirements may be more or less . If you do not need current regulation a variation is shown with a common base transistor which should provide a very high impedance at the adjustment leg output. The capacitor at the base  acts as a capacitance multiplier, and receives a tiny voltage from the anode of a thyristor sufficient to keep the capacitor charged.    

 

Next we provide voltage regulation, once again we use the LM317, using the formula Vref x R2/R1 +1  we use 1900 ohms for R2 and 220 ohms for R1  to get 12.03 volts DC   

 

The thyristor turns ON via its gate from the adjustment leg of the voltage regulator, its anode connects the load  and its cathode then to the negative of the rectifier

 

The attached schematic shows the parts involved and their connection. AC input is on J1 and DC output on J2 

 

 

Edited November 21, 2019 by stereo coffee

[aussievintage]

10 hours ago, norman said:

And I am still in error - it is only half wave rectification I now see, not full wave. It would be great to have full wave instead, but I couldn't find a circuit....

 

It's full wave.  It requires a centre tapped transformer however, the centre tap goes to 2, and each end of the winding go to 1 and 3.  If your transformer has no centre tap, simply add 2 more diodes to form a bridge rectifier arrangement.

[norman]

I am in awe of the fantastic minds here that provide help and assistance, it's wonderful.

 

I am only a beginner at this type of build, so it will take me a while to digest and draw up a board arrangement for the design that Stereo Coffee has provided. Please indulge my ignorant questions that I have.

 

Q1) The spec for my Musical Fidelity V90 LPS pre-amp existing power supply is " 12v 500mA DC power supply" so I imagine the V90 uses somewhat less than 500mA so if I guess say 450 mA, noting the maths above the 220 ohm resistor for 57mA current needs to be varied.

 

Q2) Which 220 ohm resistor is it that allows 57mA to be developed? Is it R2, R3, or both? And what value do I need to use to get say 12 volts, 450mA? I see that vRef/R if I use a 27 ohm resistor produces 460 mA - is this thinking correct?

 

Q3) Also, U3 seems to have only 2 pins connected - is this correct? Pin 3 (In) appears to be not connected anywhere.

 

I will draw up a physical layout (I am away over the weekend so it won't be until next week) and post it here for review if possible.

 

In the meantime, I have another question regarding my existing LPS - how do I implement "adding 2 more diodes to form a bridge rectifier arrangement" using my existing transformer which is not center tapped? Or would I be better just using the existing circuit with a center tapped 12 volt trafo?

 

All help is greatly appreciated!

 

Many thanks, Norm

 

 

 

 

Edited November 21, 2019 by norman

[stereo coffee]

33 minutes ago, norman said:

I am in awe of the fantastic minds here that provide help and assistance, it's wonderful.

 

I am only a beginner at this type of build, so it will take me a while to digest and draw up a board arrangement for the design that Stereo Coffee has provided. Please indulge my ignorant questions that I have.

 

Q1) The spec for my Musical Fidelity V90 LPS pre-amp existing power supply is " 12v 500mA DC power supply" so I imagine the V90 uses somewhat less than 500mA so if I guess say 450 mA, noting the maths above the 220 ohm resistor for 57mA current needs to be varied.

 

Q2) Which 220 ohm resistor is it? R2, R3, or both? And what value do I need to use to get say 12 volts, 450mA?

 

Q3) Also, U3 seems to have only 2 pins connected - is this correct? Pin 3 (In) appears to be not connected anywhere.

 

 

 

 

R3 in the schematic sets the current. You are right the supply current and actual current draw are quite different - if choosing 450ma lets go for 27 ohms as R3 providing 467ma. If the current draw is appreciable you will need to use a heatsink on U1 and U2 .  I suspect in reality the circuit draws no more than 150ma - rather 500ma supplies are readily available and is what MF chose. . If designing to 150ma  make R3  82 ohms.  

 

Follow the diagrams shown in the LT4320 application document for pin allocation of the 8 pins involved. My schematic program lacks the LT4320 in its parts bin. the image shows the pin orientation. 

 

I do have all of these parts in stock if you want me to build one if that would help.  

 

You need to contact musical fidelity to advise the orientation of their DC connector which does NOT show DC polarity on the casework.  

 

R1 and R2 set the output voltage which the maths shows is 1.25 x 1900/220 +1  = 12.03v 

 

re question 3  The alternate circuit is shown as U3,  view it as entirely substituting U1  - hence its input connects where U1 presently is shown and its output to U2   Sorry force of habit of mine to show alternate circuits on schematics presuming where connection substitutes.   

 

It will create a low impedance at U3 output - like dragging everything it can out of the output of U3, but then calm down and  present a high impedance less than unity current to the adjustment terminal which controls the overall circuit.  I can put one through its paces and see how much current results if your current needs are specific.  

Edited November 21, 2019 by stereo coffee

[Gordon Macfarlane]

You can get drop in replacements for the 78xx series on the site that I linked to in your other thread on this matter.

LT3045-78XX Ultra Low Noise LDO Voltage Regulator

WWW.LDOVR.COM

---

 

 

Edited November 22, 2019 by Ozcall

[norman]

Such fantastic help!

 

Thanks for the pointer on the optional bit (U3) in the diagram, I doubt I would have worked that out!

 

Are you located in NZ? Or in Australia? Your build offer is very kind, I will see how I go with struggling with a board layout before I take up your kind offer. Saying that, I bet that I will be asking you soon enough

 

If I go for 150 mA as the power demand, do I need to use a heat sink?

 

I am assuming my existing transformer can be re-used? It provides a +12 v and 0 v taps, or would I be better replacing it with a center tapped +6.3 v / 0 v / -6.3 volt unit?

 

I have already checked that my existing wall wart provides positive on the center of outlet.

 

Many thanks, Norm

 

 

[stereo coffee]

At 150ma  a heatsink is advisable, as it improves the longevity.  your 12v -0v transformer will be fine. You will not need a centre tap 

unless using the prehistoric arrangement of two diodes that you previously were thinking of.   I am in NZ 

 

For audio try to use the best available parts at all times, your reward will be smiles when the needle hits the groove. 

Edited November 22, 2019 by stereo coffee

[aussievintage]

2 hours ago, norman said:

I am assuming my existing transformer can be re-used? It provides a +12 v and 0 v taps, or would I be better replacing it with a center tapped +6.3 v / 0 v / -6.3 volt unit?

That would only give you 9 volts.  You need need +12/0/-12.  see 

Full wave rectifier - Center tapped full wave rectifier

WWW.PHYSICS-AND-RADIO-ELECTRONICS.COM

---

A full-wave rectifier is a type of rectifier which converts both half cycles of the AC signal into pulsating DC signal.

  for how it works

 

and here

Bridge rectifier - Definition, Construction and Working

WWW.PHYSICS-AND-RADIO-ELECTRONICS.COM

---

A bridge rectifier is a type of full wave rectifier which uses four or more diodes to efficiently convert AC to DC.

for a bridge rectifier setup (to work with the single 12V ac transformer.

 

2 hours ago, norman said:

If I go for 150 mA as the power demand, do I need to use a heat sink?

You can figure out the power it dissipates by multiplying the voltage dropped across the regulator by the supplied current.  So, roughly 17 volts (the raw DC) minus 12 volts (the regulated DC) times 150ma, i.e. 5 x 0.15 = 0.75 watts.  So it may be OK without a heatsink, but why not add a small one just in case.

[norman]

So, As I was away for the weekend I only got to try my first attempt linear power supply tonight. No magic smoke in evidence, so that was a good start.

The hiss noise is lower than from the wall wart, however there is a hum, probably 50 Hz that isn't there with the wall wart. I can fuss about with shielding with my existing build, or jump in and just build the circuit that Stereo Coffee has provided, so I think that is the direction I will go. Incoming PM Stereo!

[stereo coffee]

I have started on design of a pcb with online pcb design platform Easy EDA  

EasyEDA - Online PCB design & circuit simulator

EASYEDA.COM

---

EasyEDA is a free and easy to use circuit design, circuit simulator and pcb design that runs in your web browser.

for a LT4320 based rectifier.  Whenever projects like this commence they always invite further interesting questions, like  can the 4320 be used in a dual rail supply ? It appears Yes, as long as the AC supply is not centre tapped. rather it requires separate AC windings - which most toroidal transformers provide

 

The rectifier board will be used later in a amplifier design, where a board will be used for positive and another for negative. For the moment though one rectifier board will  serve purpose for the rectifier section of Norm's linear supply.  The two regulators for Norm's project is best on a separate board.  

 

More as it develops

[stereo coffee]

16 hours ago, thesongofice said:

Amazingly described, I don’t understand much, to be honest, and your work is amazing. Please tell me, how much time did you spend on all this training?

About 45 years. I began by studying at the reference section of a library, where i would devote most Saturdays and Sundays to reading as much as I possibly could.