What is the effect of switching noise?

A long time ago I remember a series of adverts for domestic cookers with the tag line "what do pros use when the ingredients cost more than your cooker ?" .

Well, a phone or a laptop can easily cost more than £500 - so we are in that territory.

You need a power supply you can trust with a current limit that works and where you don't get overshoot  - this is the minimum you should contemplate.

But phones are susceptible to interference and switchers (especially the kind in the video) make lots of noise - I've measured 500mV pk-pk and more from Chinese 3780 based designs - expect odd results testing a phone with a very noisy supply.

Even very basic linear supplies tend to do better on current limiting than a 3780 based design (the 3780 was no designed for this purpose !).

I assume that you have an oscilloscope (can't fix phones and motherboards without) so you could test a power supply - some of the Farnell Roadtests show you how to do it. What goes wrong with the bad ones is that the the current limit can take a long time to start operating and that the output voltage will overshoot a lot if you remove some load so that the supply comes out of current limiting. This can happen easily with a mobile phone - the transmitter starts up and it draws a load of current, then it stops transmitting and the power supply overshoots by a volt or two and almost anything can happen next.

Good fast and precise current limiting is almost impossible with a pure switcher like the board in the video - good commercial switch mode bench supplies use either very fancy switchers or a linear regulator after the switching regulator.

If you need to save the money by far your best bet is to buy a second hand linear supply - I recently bought some  AimTTi PL155-PPL155-P 15V 5A supplies for about £60. You could find something cheaper if you look around.

MK

Potentially plentiful. You can think of noise as disturbances in what should otherwise be constant. Devices generally expect power regulated to within a certain limit. Outside these limits, your device could malfunction, become overstressed or even damaged. What happens really depends on how bad your noise/ripple is and how tolerant your device is, so there's no way to tell for sure. Despite this, there is almost one universal truth - you can't supply power that is "too good" for a device ... so depending on how much you care for your devices, you will need to spend accordingly. Or in other terms, how much will killing a device (or several) cost you?

When it comes to power supply, ripple and noise is not everything - michaelkellett has mentioned the current limiter behaviour. If you rely on such functionality to diagnose faults, something which is cheap with low ripple may have a very large capacitor bank at the output terminals, thus cannot regulate current well. This may result in a large current spike into a device on start-up which could well permanently damage semiconductor based components. High quality supplies don't come cheap for a reason, as they work to actively regulate the power to your device.

To give an analogy - most regular people might be familiar with the concept of a power surge and have surge protectors plugged in at home on their AC outlets to protect against large fluctuations in voltage (and mostly clamp them down to a safe level). They're not perfect though, and there are still short term "swells" and "sags" in voltage that pass through and eventually take a toll on our devices as well. Well, switching noise is similar in concept - it's a fluctuation in voltage. Make it too big and sensitive devices will have issues. If you're using the supply for diagnostic work, is the fault in the product or in your power supply? How much time are you willing to waste to find out?

You might be able to see this with low-quality power banks and wall chargers - some of them put out a heap of noise onto the USB output. What is the result? Sometimes crashing phones, noisy headphone outputs, phantom touches on the screen, jittery touch-screen response, complete lack of touch-screen response, problems with Wi-Fi/Bluetooth/Cellular radios through to potential downright failure of the device to recognise the power input as being valid (or even damage to the power management IC). Many electronics devices are sensitive - a modern PC's 5V supply is specified by the ATX standard to have just 50mVpp of ripple and many chargers will achieve about 150mVpp or below. To use something that could put out 500mV is a bad idea. Just think of it as a ratio - 500mV on 5V = 10%. That means that the voltage is basically going from 4.75V to 5.25V constantly (assuming it has been set to 5V) which is the entirety of the USB power specification standard. If it has any regulation deficiencies (I'm sure it will) then a change in load will move this voltage up or down slightly and then you won't even meet the voltage window for USB at all.

You have been warned. I'm not saying it's impossible to use low-end equipment for testing ... but if you're not highly skilled and don't know your equipment's abilities and limitations well (and the powered device well), it probably won't end well. Having proper tools makes it easier, safer and (many times) quicker.

- Gough