silly circuit

But you are posing a different kind of problem - not one that I would have any trouble solving, since I once set up a 20A constant current drive for xenon arc lamps and a bunch of them worked for years - but not one that has much to do with this thread, which did start with John Larkin bitching about the EDN circuit.

Applying power to a breadboard with a wire will generate the capacitive spikes to start it up.

But transistors aren't specified for beta at femtoamp currents.

It's really a dumb circuit for several reasons. An LM317 and one resistor would work fine. It always starts and one can use a dpak or TO-220 and get good thermals.

LM317s thermal limit too. Those dinky TO-92 things will smoke.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Take a look at the cute little TI switchers, like TPS562208. Simple, spread-spectrum, cheap. We pay 21 cents for that one. There are higher voltage parts in the family.

I think one could design a switchmode 500 mA LED current limiter with that chip and maybe 5 cheap passives.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Do any similar parts exist which have LTSpice models?

What transistors do you know of that have useful betas at picoamp or femtoamp currents? That would be good to know.

There is basically no data on super-low-current bipolar transistor behavior.

Everybody should have a lot to learn. If they are lucky, they do.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

TI has Pspice models of many of their parts. I don't have much luck importing them into LT Spice, but I know that it can be done.

I'd like to find someone to do that for me now and then, pick up something like the TPS chip and deliver a working LT sim.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I got TPS54302.LIB and made a symbol. The pin names don't match but presumably PH is SW and VSENSE is FB.

I then tried to make Fig 7-1 in the data sheet but although it runs it does not so far do anything useful.

News just in. The alternate solver produces 5V at 3A into 1.7 ohm. Switching looks fine.

Be careful - LTSpice can have lots of problems with pspice syntax which are not always apparent. Simulations can appear to run but in fact generate bollocks. LTSpice 26 is a great improvment over earlier versions in this regard. Attached is a TPS562208 sim - it looks OK in

Ok thanks for that. I have 26.0.1

Piglet's evidence shows that the BC557 and BC548 have enough to let the circuit work. I've no idea what he means by "47megohm B-E mirror leakage" - it's not clear to me where he put the 47megohm resistor.

So generate some, rather than posting guesswork.

The LM108 used a specialised input transistor which had a current gain of 5000 at 1uA.

R. J. Widlar, ‘‘Super Gain Transistors for IC,’’ IEEE Journal of Solid-State Circuits , Vol. SC-4, No. 4, August, 1969

It was probably still pretty respectable at 1nA.

That isn't what piglet did. "It would startup even if I cranked the bench PSU up very slowly."

This isn't a femtoamp circuit. Transistors aren't specified for current gain at low currents because it would take too long to measure it, but some parts are known to have current gains of about 10 to 50 at 1nA.

You don't actually list any particular reason,

Only if you ask them to deliver too much current. 45ma is quite a lot of drive for a regular indicator LED. Edward Rawde decided to drive an illuminating LED at 500mA, which is more than any of EDN transistors could have survived.

Thank you for sharing. The cheap LD24AJTA [1] constant current driver is powered by a ?popular? PT4115 [2] IC deployed in the style of a buck converter.

Allow me to transition this topic to 120 VAC LED COB filaments. This picture shows such specimens, with the topmost powered-on and the bottommost powered-off:

The protoboard's black wire connects to 120 ACV Line while its white wire connects to Neutral. My goal is to discover the silicon secrets in such a simple filament.

Note:

I put the 47meg resistor between the NPN mirror bases and one of the emitters to simulate the effect of humidity or dirty pcb or just slightly elevated B-E leakage.

But no specific part numbers.

Say, that sounds famaliar. Who said that recently?

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Excellent! Your circuit suggests some silicon secrets pertinent to the

Allow me to transition this topic to 120 VAC LED COB filaments. This picture shows such specimens, with the topmost powered-on and the bottommost powered-off:

The protoboard's black wire connects to 120 ACV Line while its white wire connects to Neutral. My goal is to discover the silicon secrets in such a simple filament.

Look up NUD4011 and it's alternatives, it'll be something that operates in a similar fashion.

Thank you.

That would need the NUD chip and a bridge rectifier and maybe a resistor and a varistor.

I'd expect simpler and cheaper in that tiny gadget.

What do you call that axial lead LED thing?

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics