Offline Smps breadboard

[dr pepper]

Its been mentioned before that you cannot build an smps on breadboard, while that is good sense it can be done.
This circuit it built completely from salvaged parts except the 1/4w resistors, freq is 50kc.
A 600v fet and an electrolytic gave their lifes for the project.
In order to keep turns down the supply has a low reflected voltage and a high secondary reverse voltage, requiring a 200v rectifier, but it seems to work Ok.
This kind of project isnt for beginners.

 

[schmitt trigger]

First of all, I love your messy workbench! That is what a real workbench looks like.

Second, most everything can be done in electronics, if you have the skill and knowledge like you do.

But the reason that in this forum, and other forums I also lurk, is that providing that advice for a newbie who doesn't know squat about SMPS construction is a sound advice. The opportunities for failure vastly increase for them.

How many times have you seen a post with the following or similar question: "I have a 3524 chip and a pair of 2N3055 transistors. I would like to build a 12v to 230v, 3 kw sinewave inverter. How do I connect them?"
Imagine such a newbie attempting to do it on a breadboard.

 

[AnalogKid]

Its been mentioned before that you cannot build an smps on breadboard,

Not by me. I did a 2 kW power factor corrector on a combination of solderless proto-board, perf-board, and a salvaged heatsink. The caps all survived, but I popped a few semis. Love that smell...

ak

 

[AnalogKid]

First of all, I love your messy workbench! That is what a real workbench looks like.

No ****. I start each project with a pristine work space, and if things go well it looks like that at the end.

L.O.V.E. the giant notepad as the work surface; I might borrow that idea.

ak

 

[JimB]

A "SafeBloc" , I have not seen one of those in ages.

JimB

 

[schmitt trigger]

I also think that the notepad is an excellent idea. It allows you to scribble notes and some calculations that one does as one is building/debugging/testing the circuit.

 

[dr pepper]

Its a real workbench, and it didnt cost a packet, the whole thing is made from 3 sheets of Mdf.
I dont claim to be an 'expert' just well seasoned in engineering.
I concur with the noob thing, mainly 'cause my first attempts at smps's exploded, literally in fact I wear safety glasses now I still get fets liberating their cases occasionally, plus 320v without an isolation transformer, a work area without easily reached grounds and a healthy fear of being zapped can be dangerous.
I have a paper desk tidy and a notepad on the bench, as I get older I need them more not so much to remember to get it clear what I'm doing, to the right of the breadboard you can see the transformer pinouts with the starts of the windings marked.
Yes safeblocks are still made, but they are different now, I got that one for a cheap price at a yard sale, I had to spend some time on it as it was worse for wear.
The supply works, and doesnt oscillate, well not in an unintended way, being current mode the gain & phase margins are good, only I got some ferrite loss hiss on light load caused by the supply riding discontinuous mode causing audible noise just a little, I thought about slowing down the error amp but that'll introduce a phase shift so I opted for a min load resistor.

 

[dr pepper]

Wasnt well worded, hiss at low load is caused by magnetostriction, basically the edges on very low pulse width are so close that the magnetic domains in the transformer have to try and align very quickly and the core vibrates at the osc freq(and/or harmonics of that, what you can hear) and dissipation goes up, your basically running the trans at too high a freq.
Fixed that now, I rewound the trans will less turns on the secondary and the aux widings, and put 1/2 the primary on first and the rest on top, no hiss at all now, theres enough pulse width lighting an Led not to cause any issues.
Getting an Smps right is not as easy as a linear supply.

 

[dr pepper]

Added some niceties, soft start & loss of feedback shutdown.
Not sure whether to disable the built in op amp on the control chip and let the reference do all the regulating.

 

[dr pepper]

I ran some spice sims, when you make something like this real world parasitics can make a real mess of things.
Of course spice can sim those too, but you need to build the circuit and measure them.
The blue & green trace is the output, blue pre o/p choke, green post choke, the light blue is the o/p of the '3843's error op amp, the grey is the opto emitter, and the pink is the current sense pin on the '3843.
Stabilization is done by creating a dominant pole in the error amp at 1.7kc, at 8ms theres a 300% step in o/p current.
Soft start isnt sim'd, takes to long.
All in all not so bad this time.

 

[tomizett]

This is very much my kind of thing Dr P.
Thanks for sharing your experiences!

 

[AnalogKid]

1.7kc,

Geezer-speak ! ! !

 

[dr pepper]

Glad you liked it.
Might continue posting my results, I'm not going to make a pcb just going to chuck it on spot board.
Been messing with Ltspice I was thinking of taking the feedback direct from the output of the Ps, but the delay in the o/p filter is so bad I'd have to slow the whole thing right down so I'll leave as is for now.