I think it's ok. I thought something like that at first, until I realized he was using a P-channel MOSFET.

What is a P-fet doing with the positive supply on its drain?
You say it works the same backwards?
The shunt diode doesn't.
Better buy a mosfet that doesn't have a shunt diode.

 

Are you saying my FET is backwards?

Edit:
Guess it is. So can a N-Channel be used for this application?

 

Are you saying my FET is backwards?

Yes. A P-MOSFET must have the source more positive than the drain.
The substrate parasitic diode has the anode connected to the drain and the cathode connected to the source (note the direction of the arrow on the P-MOSFET symbol, which indicates the substrate diode connection).

 

What is a P-fet doing with the positive supply on its drain?
You say it works the same backwards?
The shunt diode doesn't.
Better buy a mosfet that doesn't have a shunt diode.

Good eye, I take back what I said about it being ok. Could the drain and source just be exchanged in the schematic? In a simple teaching schematic, it would seem to be unnecessarily complicated to have a high-side drive for an N-channel mosfet in that location. The reasons for wanting an N-channel there could be a later lesson.

 

There is no flyback recovery diode next to the inductor...

 

Yep to all the above. Learned something here. A redraw is in order.

That's why I like to start these discussions.

 

Your best solution is to decide which chip you want to use for Smps, look at the datasheet and make the circuit they have, then modify and play around with it,
its the only way to learn,

look at other Smps circuits on the web, like the Ucm3845 series or Tl494.

 

You are missing the point. I want to do this with simple parts, the kind in my junk box. While it might be used for something, it is not being designed for a specific application. One of the core concepts is it is something a beginner might build in the Vol. 6 of the ebook.

A little like making the circuit that will make comparator go both ways, + and -. Otherwise I might as well build from a datasheet and be done with it. This how I let my creative juices flow. I also occasionally learn something new in the process.

 

Well you're on the track with pwm chopped transformer or inductors, and monitor the outputs to alter the pwm cycling for voltage control, same as buck converters, you may suffer with regulation when a load is put on,

what software do you use to draw the circuits..?

good luck...

 

OK, now for the redraw...

As for the drawing software, I use M/S Paint along with a set of templates I created. The template package can be had on my blog, it is in the public domain. I have quite literally written the chapters on the 555 in Vol 6 of the eBook, along with quite a few other chapters.

Bill's Blog

Introduction and PaintCAD

 

Cr2 diode wrong way round....

 

So it is. I'll just repost on the previous thread. Thank you.

 

Where is the pwm coming from to pulse the mosfet?

 

This is an old topology, a ripple reglated SMPS regulator. As the voltage drops below the setpoint the FET is turned on. When it goes above the setpoint the FET is turned off. The gain of the comparators and the delay of the LC filter set the frequency. All SMPS have some ripple, in this circuit it is fundamental to the operation. The actual ripple is quite small.

I tried something similar in the past using op amps, but they were woefully too slow. Got it to work though.

Now to draw a protoboard.

 

This is an old topology, a ripple reglated SMPS regulator. As the voltage drops below the setpoint the FET is turned on. When it goes above the setpoint the FET is turned off. ...........................

Also called a hysteretic or bang-bang converter.
One advantage is that it requires no loop compensation for stability.
But one problem I've noticed in simulation with these is that they tend to have a very large overshoot in the output voltage upon startup.
What happens is that the FET is turned on until the output capacitor is charged to the set voltage and, at that point, the inductor current is high.
By the time the inductor current stops and all the inductive energy from this current is dumped into the output capacitor, the voltage can reach near twice the set voltage.
One way to minimize this is to slowly bring up the reference voltage with an RC filter during startup to give a soft start.

Note that you need a pull-up resistor to Vcc at the output of the LM393.

 

Again, look again. U1a has the pull up resistor, which turns on Q1 when U1a is on, but U2b doesn't need it. I am only interested in grounded or open, with Q1 providing the Vcc voltage when U2b is off, It is a push pull configuration. I tried to prevent shoot through, we'll see.

The LM393 in this config has a true tri state output potential.

Thanks for the suggestion, I'll probably use it.

 

I think he is talking about the base of Q1. It needs a resistor to Vcc or it will be slower than my old dog.

 

Again, look again. U1a has the pull up resistor, which turns on Q1 when U1a is on, but U2b doesn't need it. ......................

Yup, I missed that. Putting a positive voltage at the bottom of the schematic tends to confuse me.

 

Maybe this was discussed earlier and I just missed it, but it appears that you are using a forward biased PN junction as the voltage reference. The voltage will have a temperature coefficient of about 0.4%/°C. Will that be a problem?

 

Probably, if this were a practical circuit. For proof of concept though, it will work.