Blog

It’s Switch Mode, But Not As You Know It | Hackaday

The switch-mode power supply has displaced traditional supplies almost completely over the last few decades, being smaller, lighter, and more efficient. But that’s not to say that it’s a new idea, and on the way to today’s high-frequency devices there have been quite a few steps. An earlier one is the subject of a teardown video from [Thomas Scherrer OZ2CPU], as he takes a look at a 1960s HP power supply with a slightly different approach to regulation for the day. Instead of a linear regulator on its conventional transformer and rectifier circuit, it has a pair of SCRs in the mains supply that chop at mains frequency. It’s a switch mode supply, but not quite as you’re used to.

In fact, these circuits using an SCR or a triac weren’t quite as uncommon as you might expect, and could at one point be found in almost every domestic TV set or light dimmer. Sometimes referred to as “chopper” supplies, they represented a relatively cheap way to derive a regulated DC voltage from an AC mains source in the days before anyone cared too much about RF emissions, and though few were as high quality as the HP shown in the video below, they were pretty reliable. gfs 60w series

If older switchers interest you, this is not the first one we’ve shown you from that era.

“The switch-mode power supply has displaced traditional supplies almost completely over the last few decades, being smaller, lighter, and more efficient. ”

Not at my home. And not in all labs, either. Or in the shacks of CB radio operators and hams.

Linear power supplies are large, inefficient and heavy. But they’re also clean and well built.

The amount of investment required to create an equally pure switching-PSU isn’t to be underestimated. It requires good shielding, good filter stages and quality parts.

For the same (or less) money and work, someone could just build things right from the start by using a transformer, a linear regulator and a big heat sink.

The whole problem with TVI, RFI happened after switching PSUs became mainstream. When people still had incandescent light bulbs and linear power supplies, the radio spectrum was much cleaner.

It wasn’t so long ago, by the way. In the 90s, stabilized/regulated power bricks with a transformer were still around and sold in super markets.

Paradoxically, the appliances and toys could work with the little power these “inefficient” PSUs provided. How comes?

Could it be that technology back then wasn’t as power hungry as we may want it to believe ?

Another good moment to think about it.

Where I worked during the 1990s, we had a couple of large inductors in our building. The many switch mode power supplies on our computers wreaked havoc (distorting the power) for our power company that we added those to compensate. Their cabinets were always warm to the touch.

Hmmm. That is pretty strange. The front end of switchers is a rectifier and caps. Then chopping at 200kHz in old ones and higher in newer ones. Then the much smaller and more efficient high frequency transformer and then rectify and filter. Regulate by PWM of the chopper or similar. How could that affect the power source? Maybe it looks like a capacative load and changes the VI phase? Like balancing caps on inductive loads?

I pilled the monster linear supply in the large Ohio Scientific box. I think it weighed over 30 pounds. I replaced with a then very new switching supply with more current, less heat, 1/5 the space probably 1/15 the weight. IIRC it also had the +-12V we needed for op-amps and ADCs.

There was never any interference with photon counters or anything else.

A lab where I worked in the 1980s had a Tracor Northern signal analyzer which was built around a Data General computer. That rack had a switcher you could easily hear a couple of hallways away because it chopped at a few KHz and it was incredibly loud.

I also worked for a while on a DEC PDP-11/03 which was in low profile cabinet with a fairly loud switcher which was annoying to the point where I considered using ear plugs. That switcher hunted around in frequency which made it particularly difficult to mentally tue out.

I was having difficulty explaining the problem. But switchers (back then?) would draw their current only during a portion of the sine wave. With a few hundred switchers all drawing around the same instant from the utility, what should have been a sine wave looked more like a square wave. I hope that explains it better. By installing the inductors, we were trying to be good customers. (Or maybe the utility was going to hit us with a $urcharge if we didn’t “clean up our act” B^) )

Really? Your phone has more than one in it. And its charger. And your computer, laptop, screen, TV, modem, anything with a USB port or lithium battery; your car has dozens of dcdc smps’s, most LED and fluorescent lights… id bet a dollar that there are ten SMPS’s with 20 feet of you right now.

“Not at my home” insofar that here are still a number of linear power supplies at work. Switching PSUs have not completely displaced my conventional PSUs, thus.

Note that the article said “almost completely”.

Also your comment makes little sense, you say not in your home because you still use them at work?

Every single computer power supply since at least the 80s is a switching power supply. They reason why they work so well is because they are expensive compared to crummy wall-warts, linear or switching.

And leave your “it’s a conspiracy” stuff at the door.

That’s a special case, though.

XT and AT PSUs were Switching-PSUs since the early 80s and they are built like tanks. They had to provide lots of amps to power big Winchester drives, after all.

And I think most of you know that, so it shouldn’t be misused as an instrument to prove a point of view.

AT/ATX PSUs are far cry from cheap Chinese wall plugs with a switching PSU.

But PC PSUs aside, many power supplies were still being based on a linear design in the 90s (and beyond).

Like many of these power bricks used everyday electronics. Say, a base station of a cordless phone, an answering machine, a typical game console (Euro NES/SNES had a simple AC power brick with just a transformer; the Mega Drive had a big DC power brick).

Similarly, a radio/cassette recorder made in late 80s early 90s or a radio alarm clock etc had used a transformer inside. And that was okay.

In the 80s, the popular Commodore 1701 monitor was being based on a linear PSU design, too before being replaced by the equally popular 1702.

But again, such an early switching PSU is not comparable to a modern equivalent.

Nowadays switching PSUs suffer from cost-cutting measures through and through. All the necessary filtering is being omitted, because the hard regulations (FCC etc) nolonger exist these days. And that’s the problem here.

A really good switching PSU needs a lot of high quality parts. An anachronism/no-go in todays mindset.

By contrast, even the cheapest linear design is cleaner out-of-box. A transformer with a linear regulator simply is more foolproof by design. It cannot be messed up so easily. That’s the point.

Another example. Let’s take the the common lab PSUs with 13,8v. They were widely being used by CB radio operators and hams, too. To power a linear or a mobile radio. It was being used by the electronic hobbyist in general, too.

To what I know, they’re still being made and sold as professional equipment. The medium-big ones can do 20A at 13,8v, I think.

Unfortunately, they’re do cost more than their recent switching PSU cousins. – Or rather, cheap switching PSUs in their plastic enclosure have become cheaper than traditional lab PSUs.

But money isn’t everything. In the long run, paying a bit more for something reliable and clean isn’t a bad investment.

You talk about everyday electronics but then talk about electronics devices from decades ago, do you know what year it is?

The fact that electronics devices like SNES or radios or old computer equipment didn’t use SMPS is irrelevant for the statement that SMPS has almost completely displaced other power supply types. Why would what was used decades ago apply to what is used now?

At MY house, 2, 20, 30, 40, 60 and 80 meter bands are just fine around my LEDs and switching power supplies, although my POE Ethernet switch does barf on 20 meters here and there. So there are offenders. But, is it really that bad?

The problem are the neighbors, rather, I’m afraid. We have no saying in what they can use and what not. Unless we call the authorities, which in turn may cause a bad relationship with the neighborhood.

Let’s imagine one distant neighbor with a solar cell on the roof and a bad voltage regulator.. It might be enough to ruin the hobby.

Then there are those PSUs used by flat-screen TVs or DSL/cable modems which do radiate a bit of RFI, too.

I’ve installed ferrites to block a bit of sheath current and it lowered the noise floor (checked with an SDR).

If the PSUs were being properly galvanic insulated, this wouldn’t be necessary in first place. Sadly, many switching PSUs don’t provide this. Their grounds aren’t being fully insulated, even. They do induce noise to the mains.

Amen brother…you and me both. :-)

Except for my PC’s and (very) low freq inverter, linear all the way…and to the cellphone comment, I even dropped that out of my life almost a decade ago (life is bliss without the little noisemaker – RF and otherwise).

Not my primary reason, but as a radio amateur, you’ll be surprised how much sensitivity returns to a radio front-end just by removing the switching hash in your direct 10-20m vicinity.

I think everyone just cares less about power usage, before CFLs and now LEDs it was hundreds of watts to adequately light a room, so inefficient PSUs weren’t really a consideration.

You really can’t disagree that switched mode power supplies have replaced almost all other types. Yes there may be niche areas where they are still used but just look at everyday objects, almost all of them are powered by switched mode power supplies, practically every TV, PC, phone charger, laptop charger, etc, all use switched mode power supplies.

Just because you don’t like them or don’t use them it doesn’t mean that they haven’t displaced almost all other power supply types.

You think they will or can make a comeback now that electric cars seem to be the way of the future? What benefits would they provide? – – – ———————- —————————— – – –

Never did I consider toys from back then to be power hungry. They’re actually quite efficient. Todays toys can just do more since battery technology has been improving, putting more AH in the same volumes.

Battery operated toys have always been a luxury. Some can require up to a dozen batteries for all components.

Can’t imagine spending so much money on a constant supply of batteries. Without rechargeable batteries, yeah, they would seem power hungry.

A good point of comparison is server power supplies. Server racks full of incredibly bad EMI sources wouldn’t make the servers much happier than radio equipment would be, and computers keep their voltage rails steady across very widely varying loads so no problems there either. You’d have to see if it has a large enough voltage adjustment, since hams are too good to build equipment to operate at full power on any more or less than the design voltage, whether it be 13.85 or whatever their favorite number is. But while an excellent linear supply can try for lower and lower noise, a decent switching supply that has even the slightest attention paid to noise will be fine. It’s worth getting away from the cheap crap that dumps a rectifier into a capacitor and puts all kinds of crap onto the mains waveform.

Old stuff consuming low amounts of power doesn’t make it efficient; it makes it lower power. Have you seen headlights nowadays? They’re led so they get much more light per watt, but they’re so powerful and bright I almost get a sunburn from driving at night… Sometimes we’ve just started doing things that take more power, e.g. it takes more power to stream music over wifi than to receive FM.

So inefficient though. Save some energy. Build a hybrid supply for those sorts of uses. Mount the switching part inside a metal box.

I have both linear and SMPS in the ham shack. The SMPS has a tunable feature in case the HF frequency I’m on has a birdie on it.

I run my HF rig with a 600W PC switching power supply. No birdies. But if I connect a cheap 2m/70cm Leixen transceiver to the same supply, it makes HF unusable. Looks like the Leixen has an internal switcher, that leaks to its power input even in receive mode.

I guess the one thing about SMPS most can agree on is they have made finding the culprit in any complex setup more difficult.

Beware of the huge current spikes the linear power supply generates, if the rectifier feeds large capacitors without also large inductors in series with it (then all the power is drawn during the short time the AC voltage exceeds the voltage on the capacitor). Therefore a linear power supply is not necessarily quieter than a SMPS.

It sounds like your are mistaking high crest factor (which happens at low currents) with diode turn-off switching harmonics (which happens at higher currents).

High crest factor causes extra heating of the rectifier diodes, not RF noise, but not usually a concern since the average heating is not very high.

Diode turn-off noise is easily mitigated by bypass capacitors. High power ones can use a full snubber (capacitor + resistive element). These were common in the thyristor-based switched regulators. See some design methods here: http://www.hagtech.com/pdf/snubber.pdf

no, the diode ringing (as shown in fig. 9) is another problem up in the MHz range. The crest factor noise happens in the audio and kHz range, may couple into the amplifiers, gets worse with better capacitors and to lower frequency with higher currents. My point is, a linear power supply isn’t inherently noise free, I think we can agree on that. Thanks for the pdf on snubbers, it is that kind of document that I just can’t find when I need it :-)

I think it is a good idea to compare smps with linear, know the goods and bads about all methodes, clearly this supply from 1960 is not for ultra silent mA lab experiments, but for high power to be used many hrs, this way efficiency (powerloss) become the dominant problem to solve, assuming the high power load dont case much about ripple and higher frequency harmonics. This HP is actually impressivly clean, how ever its regulation loop is a bit slow, clearly due to the ultra large caps and very low switching and regulation speed.

Yes. It is a high power DC supply not designed for fast changing loads but then due to slow regulation loop, the output is quite clean.

“Tiny little heat sinks here in the back.”

Everything is relative, I suppose. :-p

There’s switch mode on the starboard bow..

SMPSU are fine if well designed. It’s the (many) crap ones that give a bad name.

You should check out the Carver M400 amplifier which used SCR switching at 60 Hz switching saturable core magnetic amplifier multi-voltage power supply. It was a little cube about 10 inches with 400 watts per channel. It was pretty cool.

My PSU from 1982 built in socialist Poland (Unitra-Unima ZT-980-2M) has a similar solution: 2 SCRs instead of 2 diodes in a bridge rectifier. Schematic: https://i.imgur.com/mMRJiry.png parts D8 and D9.

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.

9v 1a adapter By using our website and services, you expressly agree to the placement of our performance, functionality and advertising cookies. Learn more