I just found a huge 18V 1A heavy line-frequency transformer. In comparison, I also have a relatively tiny 32V 3A power adapter from a printer. So I notice the size advantage. but what else? Should I keep my 18V 1A huge power brick?

What are the Pros and cons between the modern SMPS switched-mode power supply and old school heavy line-frequency transformer?

  • 2
    \$\begingroup\$ What did your google search reveal? \$\endgroup\$ – winny Apr 13 '17 at 20:10
  • 2
    \$\begingroup\$ "tiny 32V 3A power adapter from a printer" make really bad boat anchors.... \$\endgroup\$ – Trevor_G Apr 13 '17 at 20:15
  • 1
    \$\begingroup\$ @Trevor That is a good point, but if I have a boat, I will be retired. Won't bother to think about a small thing like a power supply. \$\endgroup\$ – Atmega 328 Apr 13 '17 at 20:19
  • \$\begingroup\$ :) You get my point though. You save a ton of weight, and usually space with a switcher at the cost of increased complexity, reduced reliability and gaining some high frequency noise. Which is a pro and which is a con depends on the application. \$\endgroup\$ – Trevor_G Apr 13 '17 at 20:22
  • 1
    \$\begingroup\$ You cannot compare SMPS power supply to just a transformer. It doesn't make sense. You can compare line-frequency transformer to SMPS transformer. Or you can compare SMPS power source to power source based on line-frequency transformer (including regulation). \$\endgroup\$ – Chupacabras Apr 14 '17 at 4:34

We can compare an SMPS, with a power supply that has a transformer, and uses only rectification and a capacitor, without any regulation, that can output the same voltage at a certain current rating.

Short comparison

SMPS compared to the other is generally

  • Smaller in size and in weight
  • More efficient
  • More noisy
  • More complex

Long comparison


Size difference is basically if you use a higher switching frequency, you can scale down on capacitor sizes and the input transformer size, if you have a lower one, you have to use bigger transformer and a bigger capacitors on the input side.


In the case of a simple transformer-rectifier-capacitor design, because you don't use any switching frequency, if you want a certain power output, you have to use that size of the transformer, and its usually much bigger, because they always work on line frequency and therefore operate on high hysteresis losses.
If you regulate the power supply, you may only get 30-40% of the efficiency.
SMPS designs operate usually at least 60%, but a very good design can achieve 95% efficiency.


Noise is a problem with SMPS designs, and it can be simply solved by choosing a switching frequency that is beyond what our ears can percieve.
If the noise is still causing problem in the equipment, or just simply doesn't comply with the local regulations, then it may be considered to put the whole power supply in a grounded case.


A transformer-rectifier-capacitor design is simpler in design compared to a SMPS, but cost depends on many factors. If you need high wattage, you will need a bigger transformer, and you can usually calculate that if you want 300 W power output, you will need a transformer that is twice the wattage.
In the kilowatt range, its uneconomical to go with this design, because its not as efficient, and a transformer this size could cost significant amount of money.
The most basic SMPS is already more complex than the previous type of design.
The complexity and cost also increases in power but not as linearly. SMPS designs for high wattage applications are still more efficient than the other, but less efficient than the topologies intended for lower wattage output.

Diagram 1

We want to avoid high peak currents that could damage the power supply in some circumstances, and it also depends on the input voltage, but it can be easily seen, that under 100 W, Half-Bridge, Buck and Flyback can be used, while in higher wattage situations, Full-Bridge is preferred.
Efficiency can be further increased using modified topologies, using resonant, or quasi-resonant SMPS topologies. This can increase the design complexity and cost even more.

Diagram 2


SMPS and Linear power supply comparison, Wikipedia

Marty Brown - Power Supply Cookbook

  • 1
    \$\begingroup\$ The other guy said both are efficient, so if he is right, which sounds logical to me, then the power supply with a huge transformer is better? less noisy, simpler design, more reliable? \$\endgroup\$ – Atmega 328 Apr 14 '17 at 1:09
  • 1
    \$\begingroup\$ @Atmega328 Noise as a problem hugely depends on the application. In an audio amplifier, you need to keep great attention to reduce EMI as much as possible, simpler (transformer-rectifier-capacitor) designs are preferred there. Its more difficult to build an SMPS, and easier to repair, but if you designed it, you will make it reliable. Some cheap SMPSs although spread the misconception that they are unreliable. So, which one is better depends on your application and multiple factors. \$\endgroup\$ – domenix Apr 14 '17 at 1:15
  • \$\begingroup\$ @Atmega328 But if you want to build a 2 kW PA audio amplifier, you should use an SMPS, and you should even consider going full-digital; making a D class amplifier which is basically an "SMPS itself". \$\endgroup\$ – domenix Apr 14 '17 at 1:17
  • \$\begingroup\$ @Atmega328 You should keep the power brick for later projects though, it outputs a different voltage, so the two cannot replace each other fully in this case. A type of application would be a low power one, maybe an amplifier, or a simple radio, anything. \$\endgroup\$ – domenix Apr 14 '17 at 1:20

I recall the black-brick switchmode supply of a tablet PC causing interference with a 100KHz magnetic beacon at 2 meters.

On the other hand, the various HighPassFilters of the beacon's receiver successfully attenuated the various 60/120Hz trash generators in the lab.

Evaluate the \$\frac{dI}{dT}\$ of the various supply candidates. Probably 0.1amp/100nS for the black-brick. Probably 0.1amp * 10X peak (rectifying at peak of 120Hz) in 1millisecond, for the big iron transformer. Thus the switcher has 1,000X the \$\frac{dI}{dT}\$.

Now evaluate the vulnerable loop area.

And use

$$ V_{induce} = \dfrac{\mu_o \cdot \mu_r \cdot Area}{2\pi \cdot Distance} \cdot \dfrac{dI}{dT} $$

In our case, we had 10 turns in the Receiver loop, \$0.1m^2\$. Equivalent area is 0.1 * 0.1 * 10 = \$0.1m^2\$. Distance was 2 meters.


$$\begin{aligned} V_{induce} & = 2e^{-7} \cdot \frac{0.1}{2} \cdot 10^{+6} amp/second \\ & = 10^{-8} \cdot 10^{+6} \\ & = 0.01 V \end{aligned} $$

Since the magnetic beacon receiver was expected to work down to 1uV input, we had an overload problem.

======================== added April 9, 2019 =============

Chatted with electrical contractor, couple years ago. Their company performed both the ELECTRICAL and the COMPUTER+MUSIC wiring for expensive homes. Why? The motor sparks (A/C, dishwasher, clotheswasher, etc) and the 2,000 volt rectifiers in microwave_ovens all produce enormous dI/dT within the wiring of homes; yet expensive homes EXPECT the Internet to continue functioning, and the TV_OVER_Internet, and the music system as well, no matter which appliances are operating.

General_contractors for homes were tired of employing separate low-cost contractors for (1) carelessly routed electrical wiring and (2) computer-music wiring that could not co-exist with the carelessly routed electrical wiring.

Solution? has ONE contractor who had learned the isolation methods for co-existence. And the need to specify a more-expensive Microwave Oven with significant power-line filtering INSIDE the oven's power supply.

Why? With 2,000 volts at 60Hz, turning on with a mere 0.026 volts change across a diode (for 2:718 deltaI ratio), if only one diode junction is needed, with 10:1 surge currents assumed thus 10Amps peak, the dI/dT is

(0.026volts diode turnon) / 2,000 volts * 377 radian/sec

0.026 /( 750,000 volts/second) ~~~ 30 nanoseconds.

Thus the unfiltered Microwave Oven has 10 amps/30 nanoseconds dI/dT.


The big difference is L vs I and f and ZL(f) thus secondary voltage ratio. The steel laminated core is efficient at 50/60 Hz and the ferrite transformer is efficient at 20kH to 1MHz depending on core material and SRF.

Neither are efficient at the other frequency due to vastly different secondary voltages, L, C , DCR , hence SRF and impedance vs f.


Not the answer you're looking for?Browse other questions tagged or ask your own question.