Setsuden?

[Wage Slave]

Right. Mystery solved. SMD components may be dirt cheap but that isn't how it was done except for one resistor. We were right in our surmises except even more brutally simple capacitative dropping, straightforward bridge rectification and no attempt at all at smoothing. Gets round the problem of electrolytic capacitors I suppose but at the cost of the LED being at half brightness. In this application it doesn't matter too much except the light is not very nice. It has that sort of "partial eclipse" quality to it if you have ever seen an eclipse of the sun. Flicker will not be seen though it is of course there at 120 times a second here in good old Kansai.

Build quality was better than I thought it would be. Acceptable but definitely not neat and tidy. Even the cheap Japanese stuff back in the day was made far more neatly than this but its not too bad I suppose. Safety wise - I am no expert but there is nothing in there to stop anything that might go wrong which I am sure will not impress some folk.

High tech it ain't - that's for sure.

P1000252.JPG

P1000254.JPG

P1000255.JPG

P1000256.JPG

led cct diag0002.jpg

[Coligny]

Brutal...

The Yellow core in a round led bulb is not too common, usually it's only SMD territory...

[yanpa]

A bit late to the party on this, and my electrical knowledge extends as far as knowing that there's a +, a - and sometimes something which looks a bit like a 〒, but I do remember converting to CFLs in the late 90's in Germany, and they did last a decent length of time (to the point when changing bulbs in the dark was a once-a-decade event) and theoretically the power savings should have at least partially offset the additional power drained by all the wall-wart powered gadgets one started acquiring around that time. Not that I'm sure if they're an absolute win-win from an environmental point of view.

The new-ish Chez Yanpa is fitted out with LED lights for the main ceiling lights, which can be dimmed and have the "warmth level" changed to an extent I don't think was possible with other forms of lighting. 15+ months and no evident issues.

[Russell]

Wage Slave, I see four diodes, two resistors, a small capacitor, and that 151-thing resistor. This appears to correspond with your schematic.

For a 0.5 W bulb, brightness is not too important, methinks.

BTW, where in Kansai are you?

[Wage Slave]

The green wire for the hot leg of the AC mains is a nice little touch. Perhaps it was cheaper than brown or black. Russell, I'll PM you with that little nugget. No big deal but this is public space after all.

[Yokohammer]

Wage Slave,

You put some effort into it ... I am suitably impressed ...

But that circuit does not inspire confidence. In addition to the fact that it's off-the-shelf parts, it looks like it relies on the overall resistance of the circuit to drop the line voltage to a level that's somewhere in the ballpark for the LED, which is a pretty iffy way to go about it. There's no regulation, so if there's a bit of a spike or surge the LED (and possibly the diodes) could quite easily go pffft.

High quality LED lightbulbs are a bit more complex. There's some good information at the link:

http://spectrum.ieee.org/energy/environment/the-subtle-circuitry-behind-led-lighting

The drive electronics must include two basic functional elements: a power conversion circuit (essentially a transistor switch that rapidly turns on and off) and a sensing circuit, which monitors the average current through the LEDs and provides a feedback signal to regulate the proportion of time that the power-conversion switch remains turned on. In many cases, a transformer is used to change voltages and to isolate the LED from the high-voltage electrical mains.

[Russell]

Hammer, thanks for the link.

Looks like driving LEDs is more complicated than one would imagine at first.

There is also a more technical article for IEEE subscribers.

[Wage Slave]

Wage Slave,

You put some effort into it ... I am suitably impressed ...

But that circuit does not inspire confidence. In addition to the fact that it's off-the-shelf parts, it looks like it relies on the overall resistance of the circuit to drop the line voltage to a level that's somewhere in the ballpark for the LED, which is a pretty iffy way to go about it. There's no regulation, so if there's a bit of a spike or surge the LED (and possibly the diodes) could quite easily go pffft.

High quality LED lightbulbs are a bit more complex. There's some good information at the link:

http://spectrum.ieee.org/energy/environment/the-subtle-circuitry-behind-led-lighting

The drive electronics must include two basic functional elements: a power conversion circuit (essentially a transistor switch that rapidly turns on and off) and a sensing circuit, which monitors the average current through the LEDs and provides a feedback signal to regulate the proportion of time that the power-conversion switch remains turned on. In many cases, a transformer is used to change voltages and to isolate the LED from the high-voltage electrical mains.

Well thank you for the acknowledgement. It was all that complexity and the cost that accompanies that piqued my curiosity to work out how it was possible to make something so cheaply. In fact, the circuit doesn't rely on conventional resistance to drop the voltage, it would get far too hot that way. Rather, it relies on capacitative reactance, which looks like resitance to AC, to drop the voltage. That x rated capacitor is doing all the dropping and the two low value resistors are there to provide some cushioning for the circuit from transients while the big resistor is there to give a little cushion to the capacitor against inrush.

Actually, I have wondered for a while how it was possible to make such cheap and light charging units/PSUs for very small devices like ipods and phones. This is basically how it is done I think but with some smoothing, regulation, surge protection and a fuse as a last resort built into the circuit. This kind of thing is much more proper and still very simple and cheap. Best of all it avoids the need for a mains rated transformer - they cost money and are comparatively heavy:

[Yokohammer]

So your next project is to sacrifice an expensive full-size LED bulb!

You'll probably find that they, and all those teensy PSUs, use switching regulation. Higher efficiency, lower parts count, fewer connections to fail, makes more sense for mass production.

Looking forward to pics.

[Wage Slave]

Not sure about the parts count but certainly switched mode supply is a lot better and they can be made very small, light and pretty cheap that's for sure. The giveaway is usually whether they can accept 100V - 220V. Switched supplies can, the cheapest ones can't. I am curious now about those full size Panasonic bulbs I bought - For 400 Yen a bulb have they managed to fit a switched supply or have they done something like the circuit above?

I don't see simplicity and low tech as necessarily a bad thing at all depending on the realistic demands of the application. If it means more economical and less things to go wrong then I am all for low tech as a rule. The same applies to controls for things like cookers. I vastly prefer knobs and switches to push-buttons, LED screens and menus. Maybe I do have a Luddite streak but is that always a bad thing? Are soft start appliances really an improvement worth implementing?

[Yokohammer]

Low tech is great if it's reliable (and compared to complex "high-tech" stuff it often is), but in this case I see a couple of problems with the circuit you traced for us.

My analysis is a little different from yours, as follows: the cap across the resistor is a low-pass filter that lets the 50/60 Hz AC through but squelches sharp transients. That's their spike filter. It'll work OK for isolated transients that are well into the cap's cutoff band, but slower surges will promptly blow the circuit.

Perhaps that's why it's labeled as a 装飾用電球 (decorative lightbulb) rather than something more ... um ... functional.

Anyway, you get a bad line glitch and that thing is a goner. That's why transformers for step-down and isolation and responsive switching regulation are essential if you want a bulb that's going to be reliable. And thus the expense.

But these have come down in price so much lately (economies of scale) it's amazing.

You got full-size bulbs for 400 yen? That's an example. I don't think I've seen 'em that cheap around here.

So is a small sacrifice in the name of science worth 400 yen to you? Oh go on ...

[Russell]

I agree with Wage Slave that the capacitor is doing most of the voltage dropping. Isn't the big resistor across it just to discharge the capacitor when power is switched off?

[Yokohammer]

I agree with Wage Slave that the capacitor is doing most of the voltage dropping. Isn't the big resistor across it just to discharge the capacitor when power is switched off?

Trouble is (you know the resistance of a capacitor to AC is frequency dependent, right?) that series 0.47µF cap is going to be pretty much transparent to the AC at normal line frequencies. It depends on the impedance of the load, but I'm pretty sure it's only going to start attenuating the AC much higher frequencies.

0.47µF is a common value in audio tone control circuits. I.e. it works nicely at audio frequencies (once again, impedance dependent), but 50/60 Hz is getting pretty far out.

We'd need to either know the load impedance or actually measure the thing to be sure, but I think that's what's happening. Or at least that was the design intent.

Wage slave ... do you have an oscillator and an AC voltmeter handy?

Or ... I suppose I could see if I can find one of those bulbs at my local Daiso and do it myself (I can't believe what I'm getting myself into ...).

[Samurai_Jerk]

This is officially the must boring thread in the history of FG.

[Yokohammer]

This is officially the must boring thread in the history of FG.

SJ! I am shocked!
You mean to say you're bored by the subtleties of electronic circuit design?

(EDIT: This is why FG.com needs a geek section similar to the one I set up at the Café ... Yanpa?)

[Wage Slave]

Wage slave ... do you have an oscillator and an AC voltmeter handy?

You read my mind.

I will put a voltmeter on it but an oscillator I don't have. Nor an oscilloscope - I have almost bought one several times - perhaps I will but I don't really have enough use for one.

Where do you want to measure? Voltage drop across the capacitor?

[Russell]

I agree with Wage Slave that the capacitor is doing most of the voltage dropping. Isn't the big resistor across it just to discharge the capacitor when power is switched off?

Trouble is (you know the resistance of a capacitor to AC is frequency dependent, right?) that series 0.47µF cap is going to be pretty much transparent to the AC at normal line frequencies. It depends on the impedance of the load, but I'm pretty sure it's only going to start attenuating the AC much higher frequencies.

0.47µF is a common value in audio tone control circuits. I.e. it works nicely at audio frequencies (once again, impedance dependent), but 50/60 Hz is getting pretty far out.

We'd need to either know the load impedance or actually measure the thing to be sure, but I think that's what's happening. Or at least that was the design intent.

Wage slave ... do you have an oscillator and an AC voltmeter handy?

Or ... I suppose I could see if I can find one of those bulbs at my local Daiso and do it myself (I can't believe what I'm getting myself into ...).

If the capacitor would be transparent at 50/60 Hz, you could replace it by a wire, and the resistor in parallel with it would not make sense.

But it isn't transparent. Using this capacitive reactance calculator, I get an equivalent resistance of 5.6 k Ohm at 60 Hz, and 6.7 k Ohm at 50 Hz, so let's say we are talking about 6 k Ohm here. This is much less than the 470 k Ohm resistor in parallel with it, so this resistor can basically be ignored. I maintain my opinion that it is just for discharging the capacitor when power is switched off.

What the two 150 Ohm resistors are for is becoming more and more a mystery to me. The same current goes through both of them, if Wage Slave's schematic is correct (and I think it is, judging by the pics he posted), but one is likely rated for a higher power than the other, given their size differences. That does not make sense at all. If they have a function at all, they could as well be replaced by a single resistor.

[Russell]

This is officially the must boring thread in the history of FG.

SJ! I am shocked!
You mean to say you're bored by the subtleties of electronic circuit design?

(EDIT: This is why FG.com needs a geek section similar to the one I set up at the Café ... Yanpa?)

SJ is probably more interested in a difference kind of circuit...

[Yokohammer]

Wage slave ... do you have an oscillator and an AC voltmeter handy?

You read my mind.

I will put a voltmeter on it but an oscillator I don't have. Nor an oscilloscope - I have almost bought one several times - perhaps I will but I don't really have enough use for one.

Where do you want to measure? Voltage drop across the capacitor?

1) Voltage across the output of the two 150Ω input resistors (before the resistor/cap).
2) Voltrage across the output end of the cap and the output of the other input resistor (i.e. across the bridge input).

... and ...

3) No 2) above with the output of the cap/resistor disconnected from the bridge.
That should give us a few clues.

The reason for using an oscillator rather than the AC mains (other than safety) is to watch for voltage change as you sweep from below 50/60 Hz (say about 10 Hz) up to a few kHz.

Oscilloscopes ... yeah, I have a really nice Techtronics that hardly gets any use. Invaluable when you need 'em, but a waste of space most of the time.

[Yokohammer]

If the capacitor would be transparent at 50/60 Hz, you could replace it by a wire, and the resistor in parallel with it would not make sense.

But it isn't transparent. Using this capacitive reactance calculator, I get an equivalent resistance of 5.6 k Ohm at 60 Hz, and 6.7 k Ohm at 50 Hz, so let's say we are talking about 6 k Ohm here. This is much less than the 470 k Ohm resistor in parallel with it, so this resistor can basically be ignored. I maintain my opinion that it is just for discharging the capacitor when power is switched off.

What the two 150 Ohm resistors are for is becoming more and more a mystery to me. The same current goes through both of them, if Wage Slave's schematic is correct (and I think it is, judging by the pics he posted), but one is likely rated for a higher power than the other, given their size differences. That does not make sense at all. If they have a function at all, they could as well be replaced by a single resistor.

Yes, the resistor is probably just a bleeder resistor for the cap. If my theory is correct it wouldn't have much effect at bypass frequencies.

Having built numerous audio and guitar amplifiers in the past, I'm not proud to admit that I have had direct physical encounters with 100 VAC coming through 0.47µF caps (we use 'em for hum cancelling in the power switch sometimes). It hurts like hell. You don't get much current (or I'd probably be dead), but the voltage sure is there. The only thing that stops me from proclaiming that I'm definitely right in this case is the unknown load impedance. There's no way to say precisely at what frequency a cap will start attenuating AC unless you know the load impedance (which makes your reactance calculator results kinda suspicious, methinks).

[Wage Slave]

Low tech is great if it's reliable (and compared to complex "high-tech" stuff it often is), but in this case I see a couple of problems with the circuit you traced for us.

The Diaso circuit is definitely problematic. No smoothing, no regulation, inadaquate transient protection and no fail safe protection. Although to be completely fair on that last point one on the 150 ohm resistors , perhaps the SMD one, is a fusible resistor.

The other diagram comes from here.

http://www.electroschematics.com/5678/

He explains quite well how capacitative dropping works. As he includes a varistor for spike protection, an electrolytic for smoothing, a zener diode for regulation and a fuse for failsafe protection I think it is a simple, cheap and efficient design for certain applications.

The big worry remains, as you say, the lack of isolation from the mains. For an application like a led bulb that is acceptable though because there is no way that the user can touch any part of the circuit while it is in use. For a charging device that is not true of course so on further reflection this design would be a lousy idea at any price.

I think the cheaper charging units probably aren't switched mode but do have a transformer to provide isolation.

You got full-size bulbs for 400 yen? That's an example. I don't think I've seen 'em that cheap around here.

Look a few posts up. KS Denki is the only place I have seen them. (In the bins by the checkout.) I'm very tempted to sacrifice one.

[Coligny]

I have one of those:

sku_39753_1.jpg

extra limited, some only goes up to 80 volt. But good for signal watching on the whips and cheap enough to stay in the car...

MOAR
http://dx.com/s/Oscilloscope

[Yokohammer]

The big worry remains, as you say, the lack of isolation from the mains. For an application like a led bulb that is acceptable though because there is no way that the user can touch any part of the circuit while it is in use. For a charging device that is not true of course so on further reflection this design would be a lousy idea at any price.

This reminded me of that woman in China who was recently electrocuted to death by a phoney Apple iPhone charger (<- see what I did there? I know it's not a laughing matter but I couldn't resist that one).

Yup ... gotta be careful with that stuff.

[Yokohammer]

I have one of those:

sku_39753_1.jpg

extra limited, some only goes up to 80 volt. But good for signal watching on the whips and cheap enough to stay in the car...

Nice.
Oh shit, when the geek talk starts around here all hell breaks loose.
Related but a bit off topic, I just picked up a neat little OBD II code reader for the Beamer.

I'll stop now before the thread becomes so diffuse nobody can figure out what the heck is going on, even us geeks.

[Russell]

If the capacitor would be transparent at 50/60 Hz, you could replace it by a wire, and the resistor in parallel with it would not make sense.

But it isn't transparent. Using this capacitive reactance calculator, I get an equivalent resistance of 5.6 k Ohm at 60 Hz, and 6.7 k Ohm at 50 Hz, so let's say we are talking about 6 k Ohm here. This is much less than the 470 k Ohm resistor in parallel with it, so this resistor can basically be ignored. I maintain my opinion that it is just for discharging the capacitor when power is switched off.

What the two 150 Ohm resistors are for is becoming more and more a mystery to me. The same current goes through both of them, if Wage Slave's schematic is correct (and I think it is, judging by the pics he posted), but one is likely rated for a higher power than the other, given their size differences. That does not make sense at all. If they have a function at all, they could as well be replaced by a single resistor.

Yes, the resistor is probably just a bleeder resistor for the cap. If my theory is correct it wouldn't have much effect at bypass frequencies.

Having built numerous audio and guitar amplifiers in the past, I'm not proud to admit that I have had direct physical encounters with 100 VAC coming through 0.47µF caps (we use 'em for hum cancelling in the power switch sometimes). It hurts like hell. You don't get much current (or I'd probably be dead), but the voltage sure is there. The only thing that stops me from proclaiming that I'm definitely right in this case is the unknown load impedance. There's no way to say precisely at what frequency a cap will start attenuating AC unless you know the load impedance (which makes your reactance calculator results kinda suspicious, methinks).

It is just a standard formula that underlies that calculator. And the frequency is constant, so the reactance also is. Not much magic in that.

Regarding your shocking experience with caps at 100VAC, I understand the feeling. Ahhh, those memories as a kid with a big capacitor from a washing machine motor used at 240VAC. Ouch!!!

[Coligny]

I have one of those:

sku_39753_1.jpg

extra limited, some only goes up to 80 volt. But good for signal watching on the whips and cheap enough to stay in the car...

Nice.
Oh shit, when the geek talk starts around here all hell breaks loose.
Related but a bit off topic, I just picked up a neat little OBD II code reader for the Beamer.

I'll stop now before the thread becomes so diffuse nobody can figure out what the heck is going on, even us geeks.

I have this one on the whip:

header_scangaugeii.jpg

http://www.scangauge.com/products/scangaugeii/

Even give me the fuel flow in liter per hour so I know how long I can idle without fear of getting dry...

[Wage Slave]

This is much less than the 470 k Ohm resistor in parallel with it, so this resistor can basically be ignored. I maintain my opinion that it is just for discharging the capacitor when power is switched off.

What the two 150 Ohm resistors are for is becoming more and more a mystery to me. The same current goes through both of them, if Wage Slave's schematic is correct (and I think it is, judging by the pics he posted), but one is likely rated for a higher power than the other, given their size differences. That does not make sense at all. If they have a function at all, they could as well be replaced by a single resistor.

You are right about the 470k resistor. It's just a bleeder. R1 in the better design is 150 ohms and provides protection from inrush current when switched on. Exactly how I am not sure. I suppose just a bit of resistance will reduce the flow enough to give the capacitor time to reach charge?

I agree regarding the Daiso circuit. Why then have two resistors when one would do it? Perhaps I was right and one, (the SMD device for preference) is fusible and the other conventional? But still, why not a 300 ohm fusible in that case? Maybe the maker just got a ridiculous deal on those specific two resistors?

The diagram shown below is a simple transformer less power supply. Here 225 K(2.2uF) 400 volts X rated capacitor is used to drop 230 volt AC. Resistor R2 is the bleeder resistor that remove the stored current from the capacitor when the circuit is unplugged. Without R2, there is chance for fatal shock if the circuit is touched. Resistor R1 protects the circuit from inrush current at power on.

Regarding this type of design being too crude for a LED lamp. If the ability to be dimmed is important then clearly it isn't any good but if dimming is not needed and the mains supply is reasonably stable then I think it is OK and dirt cheap with it. Apparently the ability to be dimmed is important in the US market but it isn't in many others.

Another point to consider is that we have been considering a single LED in circuit. I saw another teardown where the array of LEDs was being run at about 80VDC. All the maker did was put the 30 or so LEDs in series rather than parallel. In that case the dropping capacitor could be a lot smaller and the current available will be higher. Of course if one goes pop that's your lot but .......

[Wage Slave]

1) Voltage across the output of the two 150Ω input resistors (before the resistor/cap).
2) Voltrage across the output end of the cap and the output of the other input resistor (i.e. across the bridge input).

... and ...

3) No 2) above with the output of the cap/resistor disconnected from the bridge.
That should give us a few clues.

Agreed. That should tell us a fair bit. I'll do these readings later tonight when I have an hour or so to spare. Grabbing a few minutes here and there at present. I'll probably pull the diode and see what DC we get across that bridge as well.

[Wage Slave]

I have one of those:

sku_39753_1.jpg

extra limited, some only goes up to 80 volt. But good for signal watching on the whips and cheap enough to stay in the car...

MOAR
http://dx.com/s/Oscilloscope

Very tempting. I had considered similar but wondered if they were just a toy to be honest. Vot ist "whips" and could I be enjoying them too?

[Coligny]

Works with cars that support obd2 protocol. All euro and US cars.
In japan, it's a bit more difficult. Better ask the dealership what protocol is used.
As for the usefulness... I can't answer...
I really love to have the instant fuel flow/h , fuel use/100km engine load in% and oil temp. But i could drive without...
Unlike my 10-ish optionnal mirrors who allow me to squeeze anywhere without second guessing anything. I once got stuck in some ricefield backroad with 90° turn requiring a 5 point manoeuvre, without then, i would have lost at least 2 body panels instead, I perfectly knew that I had less than 2 cm of clearance, but that where enough to finish my turn safely.