Abusing the LM386

No, I mean I know what it is, and it's the _opposite_ of what you do in a current-fed inverter. It's a voltage-fed inverter technology.

I said SCR, because that's the only current-mode component we have handy. You short out the supply, so that no transistors get overvolted. (This only works if your PP output has a transformer with low leakage, so that shunting the CT to GND also shunts the load inductance to GND through the co-pack diodes. For uncoupled inductors, you need an SCR across each transistor, which probably won't work -- or, if it does, you might be using them in the first place, no IGBTs or desat needed!)

It's not a crowbar, but a fuse. A current source at zero volts delivers no power!

This also assumes an actual current feed, which some power supplies do, but I forget if yours does. I think you have a buck in front of it, for power control? But I don't remember if it's a current regulator, or what.

Ah, that's where the energy is going, right.

Which, ironically -- /is/ a crowbar, because maximum voltage is maximum power draw from the CCS. :)

Easily managed with a modest size inductor and a buck converter, of course; just as the same is true of a voltage-fed inverter, with modest size bypass capacitors, an SCR crowbar, and a current-limiting supply.

Which, there you have it. :-)

Static tests don't mean much, unfortunately:

- How many surges, and what amplitude, did you measure in that time? (If you didn't log it, _it doesn't count_!)

- How many load conditions did you go through? How quickly (dLoad/dt)? Shorts, opens -- transients? (Crunchy connections and loose metal makes for interesting sparks on the work coil.)

- Temperature swings?

Not to say static run tests are useless, but they're much more informative (data per time) when at least done under worst-case conditions at high temperature (usual intent being accelerated aging). (You didn't mention, so Idunno. 260V is a pretty high line voltage, which is a good idea. Low can be a good idea as well, what with SMPSs around.)

But yeah, by design if nothing else -- it sounds good.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/

How can it not blow up? Surely a little current flows through both LED's at the same time?

Where does FB go?

The whole unit functions as an op amp, the "actual" feedback connections are not shown but you can do the usual stuff with the whole block as you would an op amp. It will surely "blow up" (that is to say rail itself) as drawn because the circuit as drawn is just an amplifier with very large gain at DC and the RC setting the "dominant pole" rolloff.

** Can't get much simpler than the inverters used in electronic flash units built into disposable 35mm cameras.

FYI:

I once took three of the above, mounted the flash tubes evenly around the end of a close focus lens on my Pentax as a "ring flash". Got some very clear pics of the chips inside TO3 transistors.

... Phil

How fast does it charge the output cap, though? I'm more interested in strobing so need to push a lot of current quickly and dump it repetitively.

It's been a long time since I've used a film camera with a flash, I seem to remember it took several seconds for the flash to charge up, you could hear the inverter "whining" sometimes

Phil Allis>

** Depends on the size of the cap and the final voltage - the camera flash shown uses 120uF and about 270V. With a 4.7uF, charge time would be in the vicinity of 200mS, maybe less if you used a NiMH battery. ** My post was about how simple and effective the circuit was.

You can learn from it.

.... Phil

Last strobe I did used about 150w of fluorescent tubes, and simply switched where the starter usually goes to flash it. There's more to it than that, but that works. If you know your fluorescents you can figure out the rest.

NT

The strobe tubes I'm playing with are like, 20 watt/sec aircraft xenon strobes. 4.7uF charged in 200mS isn't going to cut it. The circuit I showed can push a couple amps through the primary of the xfmr on startup into a flat capacitor

If you persist in not telling us your max rep rate... shrug, we can't know which energy storage approach suits you best. At least we now know your energy discharge.

NT

** The LM386 has no such ability.

..... Phil

Please don't make excuses for it - it's lazy enough as it is.

Have you tested one? A glance at the internal schematic indicates it has no overcurrent protection.

NT

It will gladly let you shove amps thru it for a few seconds until it goes up in smoke. I've done it many times! ;-)

Oh how many watts for how long? That would be useful to know. Mostly dip packages?

George H.

I do not understand the need to overstress components to the point of failure. Why demolish a defenseless resistor in the name of superior design. Why push an ic to the melting point and say it's OK.

Proper design principles means never overstressing components or systems. Many switching ic's have soft start capability. These are put there for a reason. Learn from it.

** A lot more than one.

** So what ?

Did you look at the OPs schem at all ?

.... Phil

Yes. Having now blown it up so I can actually see the part values, 2 things are noticeable

1. with 2x 4.R7, as you say there won't be 2A anywhere
2. There is no capacitor to charge on the primary side, and the 2ndary caps certainly won't see 2A, so the claim made no sense to begin with.

NT

Phil shifts the goalpoasts as usual to talk about particulars of the schematic/component values unrelated to the properties of the LM386 itself. It's silly to take his criticism seriously he never designs shit, just a habitual drama-creator.

I asked on JL's thread if there's a commonly accepted timing procedure for failure of things that fail pretty quickly like the thing he was doing. Looks like he was capturing some parameter on the scope - at what instant is a part considered no more?