Driving a led with mains. / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Driving a led with mains.

first previous 6 7 8 9 next last

Move Thread

LAN_403

Dr. Slack

Fri Sept 27 2013, 12:12PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

MURS320 diodes seem fine, but are slower. It seems any 15, 25 or 50 nS diode rated for a few amps will do it

Given that you are powered by 50Hz mains, do you really think any diode much quicker than 1mS makes much difference? A 1N4004 bridge will be fast enough, the losses and charge storage, when correctly modelled, would surely make irrelevant any effects of modelling capacitor self inductance.

Remind me what the purpose of the 10uF cap to ground is?

What's the effect of removing the series surge limiting resistor?

Ash Small

Fri Sept 27 2013, 01:13PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Dr. Slack wrote ...

MURS320 diodes seem fine, but are slower. It seems any 15, 25 or 50 nS diode rated for a few amps will do it

I think that's pretty much what I said, Neil, the self inductance of the capacitors, and associated ringing, doesn't appear to present any problems as the diodes seem to block it, or, at least, isolate the LED from most of it, and that the speed of the diodes doesn't seem to make any difference either.

I've learned a lot from this. This morning I've been thinking about the way the diodes actually work, and the similarity with transistors, ie the diodes need a certain amount of charge before they start conducting, much like the gate on a MOSFET needs a certain amount of charge before the MOSFET starts conducting. I think I'm beginning to understand the mechanism of how semi-conductors work, in terms of charge, which is an approach that I find easier to understand than most explanations that I've heard. I was wondering if diodes are more similar to 'fets or BJT's, ie voltage controlled, or current controlled?

Dr. Slack wrote ...

Remind me what the purpose of the 10uF cap to ground is?

It appears that the 10uF capacitor isn't required, I've attached a screenshot of the waveforms without it, I suppose I originally thought the more self-inductance I could add, the better, as it would 'impede' the surges, and reduce the frequency of the ringing. It would appear that it's the low ESR value of the 47uF capacitor (0.002 Ohms) that really matters, and I think that more self-inductance does help here, as it reduces the frequency of the ringing further. The only effect the 10uF capacitor has is to reduce the frequency of the ringing.

Dr. Slack wrote ...

What's the effect of removing the series surge limiting resistor?

If I understand the question properly, the effect of removing the surge limiting resistor is that it no longer consumes any power. Resistors just waste energy, and should be avoided wherever possible.

I think the value of the 0.5uF capacitor could be further reduced, this would reduce the overall losses in the circuit.

EDIT: The ringing in the above screenshot is virtually eliminated with ~100nH extra inductance

Dr. Slack

Fri Sept 27 2013, 07:01PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Dr. Slack wrote ...

What's the effect of removing the series surge limiting resistor?

If I understand the question properly, the effect of removing the surge limiting resistor is that it no longer consumes any power. Resistors just waste energy, and should be avoided wherever possible.

Well, the hint was in the question. The effect of removing the surge limiting resistor is that the switch-on surge is no longer limited. Possible burnt switch contacts, heavier fuse and diodes needed, just generally an icky thing to do. Like removing the suspension from a car saves weight, which is great until you go over a bump in the road.

Ash Small

Fri Sept 27 2013, 07:44PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Dr. Slack wrote ...

Dr. Slack wrote ...

What's the effect of removing the series surge limiting resistor?

If I understand the question properly, the effect of removing the surge limiting resistor is that it no longer consumes any power. Resistors just waste energy, and should be avoided wherever possible.

I agree, but the burnt switch contacts could be avoided using a capacitor, aka the kettering points ignition system.

I'd probably use a neon bulb for this, but it has been interesting.

I agree that heavier diodes and fuses are required.

I'm currently looking at various other capacitor/inductor/diode solutions to this, I think there is a much better solution, I just haven't found it yet

Edit: The ringing is definitely reduced with the second capacitor in the curcuit (10uF). I've tried adding some 1mH inductors (I tried various values), the results are below. While the current through the LED is too high in the simulation where it is switched on mid-cycle, after 10mS or so the voltage will change polarity, so I think it will survive. I haven't worked out how to model this in LTSpice yet, though. I also haven't input figures for ESR for the 0.33uF and 10uF capacitors yet.

EDIT: Node1 is between the 0.33uF and 10uF capacitors, before the diodes drop the voltage.

EDIT: It looks like this is about as good as it gets with the values etc that I've entered. Self-capacitance of the 0.33uf cap is 10nF and self inductance of the 10uF cap is 100nF. (I think I had some errors in the above screenshot, the 'correct' waveform maybe somewhere between the two).

EDIT: I've found some mistakes in the above simulations. I'll post an update after I've sorted it.

Ash Small

Sun Sept 29 2013, 12:41PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

I'm fairly certain I have 'realistic' values for everything in this simulation, apart from the resistance of the 1H inductors.

I'm looking into toroidal inductors at the moment. I'll post more details later.

Mattski

Sun Sept 29 2013, 05:33PM

Registered Member #1792 Joined: Fri Oct 31 2008, 08:12PM
Location: University of California
Posts: 527

wrote ...
I was wondering if diodes are more similar to 'fets or BJT's, ie voltage controlled, or current controlled?

I'd say that diodes are more similar to BJTs since a BJT is simply composed of two back to back PN junctions with a shared base. But I would say that the diode is voltage controlled because it is the potential energy difference between the P and N side which determines how many minority carriers can diffuse across the depletion region.

You can describe the BJT as a voltage controlled device, because physically the currents are controlled by the potential energy barrier from emitter to base which is controlled by the base emitter voltage. The collector current and base current both depend exponentially on the base emitter voltage. Since the collector and base current are somewhat linearly proportional this is often a better way to treat the BJT in circuit design, and it makes it easier to control the device by setting up a base current because the exponential response to voltage is very sensitive and susceptible to device and temperature variation.

Just to mix things up, if you wanted you could even view the FET as a current controlled device where the gate current charges the gate capacitance, so the drain current is proportional to the integral of gate current.

Steve Conner

Sun Sept 29 2013, 06:41PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Yup. A diode obeys the same exponential current-voltage characteristic as a transistor. It just can't do any amplifying since it only has one junction.

Ash Small

Mon Sept 30 2013, 11:45AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Mattski wrote ...

You can describe the BJT as a voltage controlled device, because physically the currents are controlled by the potential energy barrier from emitter to base which is controlled by the base emitter voltage.

This is a lot easier for me to comprehend than simply saying that one is voltage dependant, and the other is current dependant.

Now I just need to understand what the 'textbooks' refer to as 'the movement of holes.....etc'

Mattski

Tue Oct 01 2013, 04:14AM

Registered Member #1792 Joined: Fri Oct 31 2008, 08:12PM
Location: University of California
Posts: 527

Ash Small wrote ...

Here's a quick analogy for holes and electrons in semiconductors:

Think about a water bottle that's nearly full with a bit of air at the top. Flip it upside down, and the bubble moves to the bottom and then floats up. The bubble flowing up is the same as the water flowing down. There can only be a net downward flow of water when there is a bubble. It's much easier to think about the movement of the bubble because there's a lot more water than there is air, and different bits of water will flow in to take the bubble's space but it's the same air molecules always flowing upward.

The valence band of a semiconductor is nearly completely filled with electrons, but there are a few holes where electrons were but they got excited up to a higher energy band by thermal energy, and it left behind a "hole" which is a bond site where an electron is allowed to exist. These holes are like the air bubble and the electrons are like the water. When the electrons vastly outnumber holes, it makes most sense to talk about the holes rather than electrons.

Dr. Slack

Tue Oct 01 2013, 07:10AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Holes/electrons - it's far more complicated than that, or perhaps far easier.

More complicated - when you do quantum mechanical sums on holes, for how they behave in a semiconductor, they turn out to act like real fully paid-up card-carrying particles, with effective mass, momentum, they scatter, they have wavefunctions, as they have the same charge they can repel etc etc.

A lot easier - I think your problem is that you think you understand electrons as being nice particle, whereas a hole manifestly isn't. However, you don't really *know* what an electron is, do you. Not really. You can treat in it models as if it was a particle, with charge and mass. But do QM sums on it, and it just isn't. So just do the same for a hole, treat it, in models, as if it was a charged particle with mass. The only difference is that, with the electron, you have a delusion that you understand it. With a hole, you are delusion free. Sounds like +1 for the hole!

first previous 6 7 8 9 next last

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob