EEVblog #90 - Linear and LDO regulators and Switch Mode Power Supply Tutorial

[jimmc]

Can anyone point me to a good reference for how voltage references work, using a Zener or normal diode doesn't give results nearly as good as stand alone units which is to be expected but why? How do they work then?

Consult this guy! http://www.national.com/rap/Horrible/czar.html

See http://www.national.com/rap/Application/0,1570,24,00.html for a very good explanation of the intricacies of designing bandgap references.

Jim

[s3c]

Thanks Jim, exactly what I was looking for.

[RayJones]

With regards to the lack of stability of the LDO regulator, it occurred to me whilst watching the Vblog that an emitter follower would have a very low source impedance to the load.

The collector based output of the LDO version however would be somewhat higher in impedance.
Perhaps this leads to this configuration being more likely to be unstable (more load sensitive)?

[KTP]

I don't know if anyone pointed out to the OP that even though current cannot change instantaneously in an inductor, voltage can.  When the transistor switches off, the voltage across the inductor instantly flips polarity across it such that the diode's cathode (negative terminal) is low enough such that current can flow through the diode (so it would be somewhere between -0.6V and -1.0V depending on the diode type and current).

hey so interestingly, I had found this simple 3 transistor switching power supply application note a few days ago.  It is actually a current regulator but still very nice and simple:

http://www.nxp.com/documents/application_note/AN10739.pdf

[Pyr0Beast]

I don't know if anyone pointed out to the OP that even though current cannot change instantaneously in an inductor, voltage can.  When the transistor switches off, the voltage across the inductor instantly flips polarity across it such that the diode's cathode (negative terminal) is low enough such that current can flow through the diode (so it would be somewhere between -0.6V and -1.0V depending on the diode type and current).

hey so interestingly, I had found this simple 3 transistor switching power supply application note a few days ago.  It is actually a current regulator but still very nice and simple:

http://www.nxp.com/documents/application_note/AN10739.pdf

Wooow. That is really a nice find. I just love how you can do switchers with discrete components

[Simon]

anything can be done in descrete components, IC's only make it simpler and cheaper and faster to implement, you know we had computers long before IC's but instead of an IC you had a room or more full of panels full of relays or valves

[NiHaoMike]

Simple switchers that only use a few discrete components are fun to design. I even built one that would "period skip" at low load to increase efficiency.

[Simon]

I'd like to see a switcher that ajusts frequency to load but maybe that would have issues with inductor values, but then if your switcher leaves a period dead that amounts to the same thing ? although a fixed frequency

[NiHaoMike]

Period skipping will not cause the inductor to saturate. The basic idea is that in a power supply that never period skips, the pulses become narrower with decreasing load. At some point, the pulse has to be so narrow that a very significant amount of it is turning on and off the transistors. Efficiency drops as a result.

Now imagine if every other period is skipped, meaning that the output pulse is suppressed. Now the pulses are twice as large as before and the effective operating frequency halved, reducing losses. The pulses are still narrower than when the power supply is operating at full load, so inductor saturation does not occur. The low output current also means that the filter capacitors will be able to cope with the low frequency.

In reality, a 1:2 skip ratio is very low. My discrete design (2 transistors, but I forgot the details) had a skip ratio on the order of 1:1000, since it operates at about 25kHz and slows down to a pulse a few times a second at no load (only the load of the voltage sense circuit). It only pulls a few mW from the rectified AC line in that condition, not bad for a simple circuit! (If designs like that were used in the standby circuit of modern electronics, standby loss would be a lot less significant!)

[Simon]

sounds like a very good design

[Pyr0Beast]

Indeed it does