A Simple Linear Power Supply, Done Well

When reaching for a power supply design it’s normal here in 2022 to reach for a switching design. They’re lightweight, very efficient, and often available off-the-shelf at reasonable prices. Their benefits are such that it’s become surprisingly rare to see a traditional linear power supply with a mains-frequency transformer and rectifier circuit, so [ElectroBoy]’s dual voltage PSU board for audio amplifiers is worth a second look.

This type of linear power supply has an extremely simple circuit consisting of a transformer, bridge rectifier, and capacitors. The transformer isolates and steps down the AC voltage, the rectifier turns it into a rough DC, and the capacitors filter the DC to remove as much AC ripple as possible. In an audio power supply the capacitors have the dual role of filtering and providing an impulse reservoir for the supply in the event of a peak in demand imposed by the music being played. Careful selection is vital, with in this case a toroidal mains transformer and good quality capacitors being chosen.

The choice between a linear power supply such as this one and a switching design for high quality audio is by no means clear-cut, and may be something we’ll consider in our Know Audio series. The desirable properties are low noise and that impulse reservoir we mentioned, and it’s probably fair to say that while both types of power supply can satisfy them. With the extra expense of a toroidal transformer a linear supply is unlikely to be the cheaper of the two, but we suspect the balance tips in its favour due to a good linear supply being the easier to design.

43 thoughts on “A Simple Linear Power Supply, Done Well

  1. I thought the cost of transformers factored in. Even decades ago, I’d hesitate to put a transformer in a project, because those generally cost money. You could scrounge, but consumer equipment often had transformers with low current, or odd voltages.

    1. I can order a 300VA toroidal transformer with any voltage from 10V to 50V, one or two secondary windings, for 167PLN, which is now about 39,18USD. These are made to order. How much would you have to pay for similar transformer?

      For a bit of extra I could get higher voltages, higher powers, or even a custom one with multiple windings.

    2. For audio projects, I always use transformers, and toroids to boot. There really is no better option for low noise esp. in tight quarters. Antek transformers are well made, and their prices are pretty good. They still run $25-50 for most of my projects, but still worth every penny.

  2. It starts out as a good project but indeed, the end sort of feels like an advertisement. Kind of weird. But at least the files are available so you don’t have to use them.

  3. Isn’t it a good thing that individuals can get a shot at seeing their work promoted, and maybe gain a few sales?

    People need to eat, heat, and have a roof over their heads, and if they can do that by inventing/designing and selling products that people want, all power to them I say (no pun intended)…

    … or do you expect everyone to work for free without even the hope of compensation – even if that’s just someone picking to buy their open source project kit/product from a vendor that kicks back a commission to the creator?

  4. Probably because this is the internet and even you are allowed to post “projects” on .io that go against what some other stranger across the world would consider stupid and not allowed? I really don’t get comments like this, no one is holding a gun to your head to read these articles. Either ignore them or if you do accidentally read them then realize that you and your time aren’t as important as you think.

  5. Wonderful. Kudos. ❤️

    Finally once again a *real*, proper PSU with clean, smooth output and fuses. 😃👍

    If only these primitive switching PSUs were banned!
    At least the cheap ones without shielded chassis, heatsink, fuse, protective ground,
    ferrite baids and 1nF/10nf/100nF/1000nF caps.
    80% of world’s RF pollution would be gone.
    Just like regular pollution went away for a brief moment when Corona forced us in quarantine in 2020/21.

    1. There’s a place to linear/transformer supplies and there’s a place for switching supplies. Imagine the device you are using to post this message and how inefficient/large it would be if it had to use linear supplies for the multiple required internal power rails. Ironic you bring up pollution, because replacing all switching supplies with linear transformer designs would result in lower efficiency and thus higher pollution production for an equivalent power output.

      1. Well, yes and no. It’s true that ATX PSUs and professional laboratory switching PSUs with banana plugs and a metal chassis are sorta.. Ok.

        However, the majority of switching PSUs come from China and are of questionable quality. I mean, the clones and official smartphone chargers look alike, anyway. These devices do cause a lot of RFI, just check with an SDR.
        Even these switching PSUs in 4K TVs and BD players are noisy as heck.
        They pollute our precious communication bands tgat we need more than ever.
        Okay, a bit of noise can be lowerd by adding ferrite clips to the power chords, but the average Joe doesn’t know that, even.

        The core problem is: Quality (proper!) switching PSUs are very complex to build. They are literally RF generators. All the drawbacks – that linear PSUs don’t feature- must be fixed by heavy filtering. And filters (caps, coils, etc) are expensive, so they’re omitted. Like fuses. By contrast and by design, linear PSUs can’t be build to be as bad. Unless you really try. ;)

        Efficiency.. Efficiency is an illusion, I think. And a drug. Our uncle Dolfie used it to brainwash an entire generation. :(
        Personally, I think, efficiency is often brought up by biased and short-sighted people that think that modern technology is superior no matter what and that economy will always rise.
        There’s even a term that describes that way of thinking (blind faith in progress).

        In reality, though, resources are limited and age doesn’t matter for something to be advanced or not.
        It’s all relative. Ancient aqueducts seem primitive but needed no power, no pumps or engines. Just gravity. They were very nifty in their own way thus.

        Equally is tube technology seen these days..
        It looks so primitive and power wasting, yet tubes can handle high amounts of energy, are robust against radiation/EMPs and do go softly into saturation.
        If correctly operated (protected against brute force), they last very long and are environmentally friendly. Just glas, a vacuum and metal. No rare or poisonous materials.
        Also, tube technology is more recent than solid-state, actually.
        Crystal detectors (diodes) were used long before them.
        But people forget so quickly.

        Even CRTs from 70 years old B/W TV sets are still working fine these days.
        And the weak ones often can be regenerated, even.
        If people would still care for their appliances they could last decades.
        Our cheap high-tech, by contrast, made from plastic, lasts 10 years at best at the average. Because free market economy and the current mindset of being “efficient”.
        What ever that means. Maximum profits are perhaps part of it, dunno. Omitting fuses snd safety circuits is part of it. Cost efficiency..

        Longevity is much more important perhaps.
        It’s time that we people change our minds in a way that we don’t just selfishly think about our present, but also future generations.
        Quality devices that last need less repair, don’t end on the dumpsters and don’t have to be recycled.
        Instead, they would be fixed or used as parts.

        Or in short: It would be a positive step going back a little bit back to pre-industrialism, when people were more than just consumers still and still fixed their appliances. Anyway, that does not imply being backwards minded.
        Maybe it will mske people and society smarter even, if they learn to do things independently.
        New technology are still worth to be discovered. But used in harmony with the old one and the environment.

        And that’s were, ironically, primitive light bulbs and transformers have their advantage. They are easily understood, can be fixed with hand-made tools and are very robust. They are very efficient in their own ways, if properly made. Light bulbs with protective gas could last 100 years ideally..
        So what we need rather than efficiency is a proper power source. Like fusion energy etc.

        Best regards,
        Joshua

        PS: Sorry for my poor English. I’m writing this on a horrible screen keyboard on a smartphone with a weak battery.

        1. Dismissing the topic of efficiency by resorting to personal attacks on those who would disagree with you and without a technical underpinning to your point greatly undermines your stance imo. Stick to facts and not personal biases and emotions.

          I actually agree with you that reliability and quality should be the most important aspect, but you are conflating a reduction of those with modern technology itself. There’s no inherent technical reason that modern tech itself should be unreliable, irreparable or of low quality. It’s more a symptom of capitalism, outsourcing production and bottom line profit being king that’s to blame, not the technology itself. I have a feeling the reason why you have such strong faith in vintage tech is a form of survivorship bias.

        2. So get good at building less noisy switching supplies, rather than hiding in the past.

          Imagine the Apple II with a linear power supply. It would be bigger, heavier and the weight would be in one corner. Its use of a switching supply was a new thing for consumer electronics at the time. Nobody had tackled them before because they were “too hard to design”, but that’s changed because switching supplies have become common.

          Sure, anyone can throw together a linear power supply, which you seem to think makes them better. It’s easy to avoid technology if you live in the past.

          Fifty years ago, the average home had a radio, a tv set, maybe a stereo. The cost was high, people put time into their decisions. And it was primitive equipment. People paid to repair because the initial cost was high.

          Five years later, there were a lot more possibilities. Calculators, digital clocks and watches, computers, the beginning of video recorders in the home. None of that came from tubes, they came from ICs, specifically digital. It became complicated overnight, but you got so much more. And with time, it got cheaper, because consumers demanded it.

          Don’t attack advances in technology because you don’t understand it.

          1. The BBC micro computers were initially built with linear power supplies at the insistence of the BBC, who wanted zero interference from switching ones. It turned out that the heat generated didn’t do the computers much good, and so the power supplies were changed to switching.

    2. Yeah so you don’t know what you are talking about. How is a linear PSU any more “real” than a SMPS. SMPS are generally more efficient, smaller, cheaper, adjustable etc. You talk about pollution but there would be much more pollution if everything was still linear supplies due to them having lower efficiency. Most devices you interact with now will use a SMPS and for good reason. Linear supplies with a transformer are very limited in their uses, they only work with AC input, they are large, etc. With SMPS you can do DC – DC conversion as well as AC – DC very easily. Linear regulators are used for DC – DC but are very inefficient, if you want to half the voltage then you only get 50 % efficiency, if you used linear regulators in any mobile device you might lose potentially up to 60-80 % battery life depending on the phones/laptops/etc core voltage.

      I’ve only ever seen a few SMPS without heatsinks and they are very low power ones so they don’t need them, so I don’t get what your point is, also most SMPS need caps to function so I don’t know what ones you’ve been looking at that don’t have caps. If you are complaining about noise problems then linear supplies can have noise issues too, they will always have some ripple but if the output caps aren’t big enough then it can be quite a large ripple and if mains powered the noise will be 50 Hz so not really as noticeable but it does still exist. I don’t know where you are but RF pollution isn’t a major issue for most devices or people and if it could be an issue then there are steps that can be taken to mitigate it. Also most power cables here in the UK are fused and often the device is too and there are breakers as well, so linear supplies aren’t any safer than SMPS. Then there is the adjustable output voltage of SMPS, I know linear regulators can be adjustable but AC – DC linear supplies aren’t easily adjusted, this makes SMPS much more versatile and you can potentially use the exact same converter design for two different voltages with minor changes, or have a potentiometer to adjust it, they can also provide constant current functionality which isn’t as easy or common with a linear supply, without SMPS technologies like solar panels and most electronic devices wouldn’t really be possible.

      There is a place for both linear power supplies and SMPS and in many cases they will be used together, the SMPS for the large voltage conversions and the linear regulators for small voltage drops to reduce noise. By wanting to go back to just linear supplies then you are really just showing you don’t know enough about SMPS, or electronics in general.

      1. “Yeah so you don’t know what you are talking about. ”

        I’m sorry, but I stopped reading after that.
        You don’t know me personally and you thus can’t judge me properly.

        An attitude like this shows me, however, how unwilling people in general are to accept unpopular ideas, no matter if they contain truth or not.

        Or even worse, thex tend to get upset if someone dares to even question their worldview.
        On such a basis, productive discussions can’t be held, sadly.

        If you think that I don’t know what I’m talking about, that’s fine, just say so.:

        “Yeah so *so I think* you don’t know what you are talking about.”

        Would have had been perfectly fine, imho.
        And perhaps it would have been correct, also. Not sure.

        Anyway, so long.
        73, Joshua

        1. In order to make a judgement that you don’t know what you’re talking about, he doesn’t have to know you, he just has to read what you wrote.

          My judgement of what you wrote is that it’s not really an engineering comment. Engineering is about making design choices based on constraints, and evaluating design tradeoffs. The core of your comment is that linear PSUs are better, and switching PSUs should be banned. This is obviously completely insane. Probably more than 80% of the electronic devices in your home use switching supplies, and a fair portion of those devices wouldn’t even be able to exist without switching supplies. Do you know what voltage a typical modern laptop CPU might use? Its around 1.1V. Please tell us, oh great one, how you’re going to deliver 15 amps at 1.1V from a battery with a linear supply in any reasonable way. What about a modern desktop gaming PC, running 150W at 1.3V? If linear supplies were the only option, your PC is going to be huge and crap.

          Is RF pollution an important problem? Yes. Is your opinion reasonable and not insane? No. Does it sound like nostalgic rambling? Yes.

    3. Kind of funny because the amount of power wasted by linear supplies’ would add a lot of unneeded real pollution to the world.

      Analog supplies have their place, but it is not universal.

  6. On transformers: It’s tedious, but not all that hard to rewind transformers for any voltage you want. Newer transformers have plastic bobbins, which makes it easier than ever.

    I have a junk box full of salvaged transformers; so they are “free”. When I need a transformer for a project, and can’t find one with the right voltages, I find one of the right weight (so the right power rating), and rewind it as needed.

    I just did this for a vintage tube amp project (Ondes Martinot). Plate transformers are rare as hen’s teeth; so I rewound a 60-0-60vac transistor stereo amp for 300-0-300vac to do the job.

  7. This is “pretty” but not really well done IMHO.
    First off, soldering CAR Fuses to the board is not a good choice as far as I am concerned.
    Only the 24V circuit is fused, the 12V is not.
    The holes for the wires are too small as pointed out elsewhere.

    Have Hackaday sunk soo far that a simple rectifier circuit is news worthy?
    If so I should dig out my old 7812 based 10A regulator I used back in the day, even without a commercial PCB it was a better setup than this. 12.8v out at no load and no drop when hit with 5A load. Never hot it with more as it was just an overkill PS for Dad’s CB radio to be in the house.

    1. I’m totally with you. One would expect that a power supply for a 400W amp, would be capable of well over that to keep the peaks from distorting. A 10A transformer at 24V is only 240W. I don’t think that that amount of capacitance is anywhere near enough to make up the difference.
      Really odd using a integrated full bridge rectifier for one side of the power and discrete diodes for the other.
      As others have said, there is no voltage regulation, so you can be assured that 12V isn’t what this is going to put out.

      Basically, I call this a mediocre unregulated power supply. Pretty similar to basic wall warts, although a little higher in power.

  8. Aside from cost, the size and weight of transformers were always a pain but I never did a DIY with a toroid. 50/60Hz transformers are just plain big. Some of the old S100 computers, like IMSAI 8080, are nearly too heavy to lift. The transformers for switchers are 100kHz and up, not only tiny, but with frequency way outside the audio band.

    I wonder which method has the lowest noise. In fact I have asked around and not gotten a good answer to the question of what is the lowest noise supply for very low noise high gain op-amps and such. Is it battery or extreme filtering? Is a battery with a supercap the best? Is an amp with more current consumption better in order to have less electron noise in he supply? Etc.

      1. A pretty reliable source. You remind me of one of the craziest double tragedies ever. Bob Pease driving into a tree on the way home from Jim Williams’ memorial!

      2. I fully acknowledge: When it comes to SMPS noise Jim Williams’ App Notes are one of the best sources you will find out there. He shows: If you really know what you are doing and spend the time to do proper measurements, you will get SMPS noise nearly as low as you want.

  9. 2 problems with old linear design. One they use electrolytic capacitors which eventually break and sometimes catch fire when they do – I have personally witnessed this in someone’s living room – quite frightening , fortunately there were people around to see to it or the flat may have burned down. It was a Maim amplifier, not a cheap one.

    Two when they get a bit old the transformer can start to hum. Maybe not so much with a toroidal design though.

    1. You can’t make switching supplies without electrolytics. High voltage on the primary side, low voltage on the secondary that have to deal with frequencies higher than 60Hz.

      With a linear supply, a bad cap means morehum, somewhat covered after ic regulators. But in switching supplies, you may not get an operational supply if a cap goes bad.

      I’ve never seen an electrolytic explode, except when I put one in backwards. But if they did, the issue is there in switching supplies too.

      1. In SMPS capacitor that goes bad usually causes supply to stop operating, as either output voltage can’t be maintained, or input current across the switching elements is too high. Best case scenario: the supply just switches on and off trying to maintain the voltage(-s). Worst case scenario: one of the switching transistors gets shorted, which will trip either the mains fuse or fuse inside the supply.

    1. Toroidal transformers above 300VA and high power transformers might require a soft-start circuit on primary side. It all depends on capacitor bank size, too, and on power demand of the powered circuit. There is no reason for using more capacitors than project requires.

      1. When I was at electronic school a roommate built a PS with a huge toroidal transformer. The lights (incadescent at the time) of the building dimmed for half a secomd each times he turned it on.

    2. It’s the toroidal transformer to worry about, because there is no little air gaps like on an EI core, the inrush can be 15 times or more rated full load current.

      Depending on the size relative to the upstream CB/Fuse this can start causing real headaches to not have problems and still get it on while complying with your local electrical code.

      If it’s marginal you can try a few times until you get lucky and catch near a zero crossing, but no one really wants to do that and it’s not much fun with fuses.

      I’ve had to fit D curve CB in industrial settings before, which have higher magnetic trip ratio multiplier (to full load), but you have to ensure all wiring associated is rated accordingly.

  10. Single point earthing might be nice. I have had interesting problems with hum in sensitive circuits because of circulating (ripple) currents through power caps before in linear PSUs. Maybe use a single point at the center of the capacitor cluster, then bringing this point independently to input and output connections. Anyway, nice job.

  11. Look at power supplies for early home computers. Capacitors the size of Coke cans, and a bunch of them. The Altair 8800 had that, and soon there were upgrade kits from other companies to improve power supply operation.

    High power amplifiers were like that too.

  12. It is actually hard to filter out the ripples without bulky parts (or active regulator). For 100/120Hz (full wave from 50/60Hz AC), the capacitors need to have enough charge until the next cycle. In the old days, some would use a PI filter with L and C.

    I have tried using the old transformer + bridge + 3000uF cap for a Class D amplifer. It was more quiet with a switch mode brick as the switch noise is outside of audio frequency.

  13. I have an IMSAI. To be precise, the transformer is 5.25″ x 4.5″ x 4″ (133 x 114 x 100mm) and weighs about 12 lbs (5.5kg). The main filter capacitor is 3″ dia x 4″ tall (76 x 100mm). The 45-year-old transformer is silent, and the filter caps are still good.

    Transformers hum when they are poorly made. Low-quality electrolytics, and ones that are run too close to their max ratings go bad early. The IMSAI uses industrial-grade 15vdc electrolytics on its 8v supply; that’s the main reason they are still good.

    Electrolytic capacitor technology has also steadily improved over the years. A new part with the same capacitance and voltage rating today is about 1/4th the size of an old one.

    A good linear supply can actually be more efficient than a cheap switcher. The majority of consumer-grade switchers are only about 80-85% efficient. A constant-voltage transformer regulated linear supply can top 90% efficiency.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.