Dissapointed from a LED light bulb...

[spec]

... Spec: Perhaps you missed reading 'reporting' that the last attempt was :

No, I didn't. Did you try what I suggested?

 

[ronsimpson]

Try connecting a 3.3V LED rated to a 3.3V supply

At the hobby level you might get away with it.
In production it is very dangerous.
LEDs are not designed to be driven from a voltage source. They are current dependent devices. They need some thing to limit the current.
I have designed LED flashlights that need to work at -40 and survive in the summer. The voltage drop changes with temperature and vary from one batch to another batch of LEDs.
On the forum we see people that put a "12V" LED across a car battery and complain about the results.

 

[chemelec]

Seeing as this circuit does not have a large Voltage Difference between the Rectified Line Voltage and the LED String, why couldn't a person use an LM317 as a constant Current Supply, or a Power Mosfet as a Current Limited supply to protect the LED String.
Might Also add some Extra Filtering.

 

[spec]

Seeing as this circuit does not have a large Voltage Difference between the Rectified Line Voltage and the LED String, why couldn't a person use an LM317 as a constant Current Supply, or a Power Mosfet as a Current Limited supply to protect the LED String.

Hi Chem,

yes, you are quite right, but by the time you had put all the protection circuity in for the LM317 it may be a fairly complex circuit. An ever present problem with the LM317/337 is the temperature shut-down, which seems a great idea, but it severly limits the power the chip can supply without a massive heat sink.

A MOSFET configured as a constant current generator would be an excellent approach though- you need to hit LEDs with a constant current.

I am hoping that the mod that I suggested will do the job.This has the absolute minimum modifications and, like the original acts as a constant current generator, by virtue of the constant energy packets stored in C5 when the mains sine wave swings up and down. The capacitor drop approach used in the original design is lossless in therory, but not quite in practice due to the RMS losses caused by the massive current flow when the rec diodes turn on momentarily at the peaks, negative and positive, of the mains input sine wave.

I don't think filtering is an issue with this circuit.

 

[ronsimpson]

due to the RMs losses caused by the massive current flow when the rec diodes turn on momentarily at the peaks, negative and positive

Is there a design that does not have this diode loss? Out of 160V of LEDs the bridge has about 1.4V of loss. (Less than 1%)
Using 220Vac; here is the current in two of the diodes in the bridge.
This bulb was designed for 220Vac not 120Vac. That is clear.

 

[spec]

Is there a design that does not have this diode loss? Out of 160V of LEDs the bridge has about 1.4V of loss. (Less than 1%)
Using 220Vac; here is the current in two of the diodes in the bridge.
This bulb was designed for 220Vac not 120Vac. That is clear.
View attachment 96536

Hi ron,

I don't think the rec diode Vf is an issue; it is only a couple of volts (2 x Vf at high current). The problem is the 153 volts nominal accumilated Vf of a string of 51 LEDs in series. 120V RMS mains has a Vpeak of 170V so there is only 17V overhead. The VF on LEDS is not accurate and it would only take a string with high bulk resistance LEDs to cause a no go situation. What I am suggesting is to double the input voltage to 340V ( in theory) to set up similar conditions as for a 230V 50Hz input. The 20% increase in frequency going from UK 50Hz to US 60Hz effectively give you 20% more power transfer which is one favourable thing..

I am intigued by the value of C4 (on my schematic) because that attenuates the voltage by 3:1 and thus seems way too high ( C4= C5/10= 150nF max, yes). If my circuit mod does not work, the next move would be to pull C4 (with my mod still in) and replace with a 100nF capacitor.

It almost seems like the circuit designer changed his mind half way through and switched from a constant current approach to a constant voltage approach. Or maybe he just kept flinging components at the circuit until the LED string he happend to have worked OK.

Another approach would be to half the LED strings and provide four mains rec circuits, or a variation of this.

 

[spec]

After reading this thread, I got interested in the topic of LED lighting and tried to identify the SM LEDs used in the 30W lamp.

I could not find the 505SM type on the Cree site but taking another approach Cree seem to have 2 broad ranges of white LED SM chips:
(1) 20mA at a nominal VF of 3V3 (x51= 168.3V)
(2) 30mA at a nominal VF of 3V6. (x51= 183.6V)

**broken link removed**

The peak voltage of a 120V sine wave = (1.414 * 120V) = 169.8V. If you subtract a nominal 2% to cover mains supply variations and local voltage drops you get a worst case effective mains peak voltage of (0.98 * 169.8V) = 166.4V. Then if you subtract a further 2V for the two diodes in the bridge that are in series with the supply at the + and - sine wave peaks, you get 166.4- (2*1V) = 164.4V.

There is not much point in quantifying further, especially in view of the ratio of the values of the two capacitors. Quite simply, as has already been stated, the circuit has no chance of driving the 51 series LEDs at full brightness with a 120V mains supply- yet the advert clearly states that the circuit will operate from 85V to 265V.

 

[spec]

This circuit should do the job and requires only an additional 400V 100nF cap. Any value around 100nF would do.

​

ERRATA
(1) 4K7, 2W resistor should read 4K7, 4W
(2) VF=3V should read VF=3.45V in two instances
(3) C6 schematic symbol should be for an electrolytic capacitor with the positive terminal uppermost
(4) LED2 & LED3 should read LED2-1 & LED2-51 respectively


NOTES
(1) The input capacitor and two diodes in the bridge rectifier are a voltage doubler, generating 170V across capacitor C7 on the negative excursion of the mains sine wave .
(2) Some of the energy stored in C7 is then transferred to the LED string on the positive excursion of the mains input sine wave which adds the positive swing of the mains sinewave to the 170V stored by C7. Thus the LED string is presnted with 170V + 170V = 340V (not quite true, but good enough for illulstration)
(3) The current through the LED string is pulsed, but the average current (RMS) through the LED string is set to 20mA by the series resistor, R9.
(4) D6 & D9 have no function and are only shown because they are part of the bridge rectifier.
(5) First build the circuit with just the input capacitor C7 and bridge. Ensure there is 170V across C7 (there will be no ripple voltage)

 

[spec]

This is a recommended circuit:

​

​

​

ERRATA
(1) R12 should read 4K7, 4W
(2) C8 & C9 schematic symbol should be for and electrolytic capacitor with the positive terminal upward in both instances.

NOTES
(1) The bridge rec and 2 capacitors are plus & minus half wave rectifiers.
(2) The effective PSU as seen by the LED string is 340V across 4.7uF/2 = 2.35uF
(2) First build and test the two half wave recs but do not connect the LED string. Ensure the half wave rectifiers produce 170V and -170V respectively (there will be no ripple)
(3) With the LED string taking an average of 20mA there will be around 340V- 176Vsting= 164V ripple across the two caps. Just in case you measure, this large ripple voltage is correct and reduces the dissipation in the series R.
(4) The current flowing through the LED string is pulsed, but the RMS value is set by the series resistor to 20mA.

 

[Externet]

I got interested in the topic if LED lighting and tried to identify the SM LEDs used in the 30W lamp....

... yet the advert clearly states that the circuit will operate from 85V to 265V.

Same here. Unable to find specs to put it to work properly.

And, do not miss the boldly underlined claim by the vendor "Question B" about the 'driver' circuit :

 

[spec]

You have to be on your toes.

The worst thing I ever did was to reserch for weeks some patio doors for a house that I was refurbishing. I read reviews, checked the Bristish Standards, checked the security, optimised the price. Designed the patio door to suit the wall aperature, checked delivery. Finally all set- placed order. Four weeks later, sure enough, a large lorry (truck) pulled up ourside the house. Two very helpful guys placed the doors just where I wanted them- the doors were big, big and heavy and expensive too. Anyway, I finished the aperature in the masonry- it was widened. Then unwrapped the patio doors- Oh dear, hinged doors not sliding- I hadn't checked the type of door when I ordered: too busy with the details.

 

[chemelec]

SPEC..............Since your in the UK, isn't your power 220 or 240 volts?

Are ALL these LED's Wired as ONE SERIES STRING? (Both Outer LED's and bottom Panel LED's)
Or Are the two Series Strings in Parallel with each Other?
Or is One Supply Operating the Outer Ring of LED's?
And the Other Operating that Bottom LED Plate?

And Your last Power Supply Schematic does not seem logical to me.
Also I Only do Real World Assembles, ever Simulations.

Lastly, Does this Lamp Come Apart Easily and also Re-Assemble Easily?

I will possibly order some of these lamps, but I am going to Chile on a holiday in Jan/Feb, so it will need to wait till I get home

UPDATE: "I Just Noticed it Says these are NO LONGER AVAILABLE"!

 

[spec]

SPEC..............Since your in the UK, isn't your power 220 or 240 volts?

Hi chem,

UK was 240V, 50Hz but a few years ago changed to 230V, 50Hz

Are ALL these LED's Wired as ONE SERIES STRING? (Both Outer LED's and bottom Panel LED's)
Or Are the two Series Strings in Parallel with each Other?
Or is One Supply Operating the Outer Ring of LED's?
And the Other Operating that Bottom LED Plate?

Two separate 51 white LED strings each with their own power supply (bridge and caps)

And Your last Power Supply Schematic does not seem logical to me.

Hmm! not sure what to say. Can you give me a bit more to go on? If you have found a fault I would appriciate a description. As far as I can see, the second circuit is dead conventional.

Also I Only do Real World Assembles, ever Simulations.

I'm with you chem- never used a sim in my life. Mind you, someone else did it for me if needed. I think sims are very dangerous and very usefull- just like guns. In the digital field you could not do a design without a sim- it would be impossible.

Lastly, Does this Lamp Come Apart Easily and also Re-Assemble Easily?

Can't say for sure, but going by Externet's pics, yes

...but I am going to Chile on a holiday in Jan/Feb

Can I come too?

"I Just Noticed it Says these are NO LONGER AVAILABLE"!

I wonder why!

spec

 

[chemelec]

Just that In My Opinion, A Full Wave Bridge circuit would be More Efficient.

And 4k7 Seems like a LOT of Resistance, base on the LED Voltage Compared to the Rectified Line Voltage.

Sorry GOT Confused, I Met to Ask Externet About the Assembly and Dis-Assembly

 

[spec]

Just that In My Opinion, A Full Wave Bridge circuit would be More Efficient.

A full wave bridge would be much better, only problem is it would not turn the LED string on, as previously outlined in the posts on this thread.

The dual half wave bridge circuit charges the effective reservour capacitor every half cycle, just like the the bridge, but it does have the advantage of 1 less diode Vf (1V or more at the high pulse of current into the caps at the plus and neg peaks of the input 120V sine wave). The only difference is that 2 less diodes are needed but an extra capacitor is needed. Also the effective capacitance seen by the LED string is half the value of a single cap (2 caps in series). This doubles the ripple voltage, but the high ripple is an advantage in this application because it means roughly half the dissipation in R7 (4K7): 165V at 20ma with a loss of 1.65W rather than 3.3W (all this is very rough- but in the right order)

And 4k7 Seems like a LOT of Resistance, base on the LED Voltage Compared to the Rectified Line Voltage.

The 4K7 resistor may be to high- I got tired of working it all out so played safe, besides I had a few other things going on at the same time. The effective rectified voltage with both of my circuits is 340V. The diode string forward drop is 176V so that leaves 340V- 176V= 164 volts for the 4K7 resistor to loose- seems about right to me!

Sorry GOT Confused, I Met to Ask Externet About the Assembly and Dis-Assembly

Appologies not needed- I get confused all the time

PS: as a general rule if you are not making mistakes, you are not making anything.

 

[chemelec]

OK, Since I Didn't see a Neutral Line, I Assumed you were running at 120, Not 240.

At 240, I would tend to go with the original circuit, Rather than that POWER Resistor and the Heat coming from it

 

[spec]

Hi Externet,

I have it in my mind that the circuit of post #49 may be capable of driving two LED strings, with their assosciated balast, resistors in parrallel, but that would be a bit further down the line.

 

[spec]

OK, Since I Didn't see a Neutral Line, I Assumed you were running at 120, Not 240.

At 240, I would tend to go with the original circuit, Rather than that POWER Resistor and the Heat coming from it

The rectified voltage is 120V * 1.414 = 170V. thus the total effective voltage seen by the LED string is 340V. Without any ripple the 4K7 would dissipate around 3.28W, with the ripple voltage half of that. Who cares about 1.64W or so, but in principle you are right.

The second circuit is better all around- much easier to characterise too. It could also be configured for 230 volts mains supply quite simply.

 

[Externet]

Sorry GOT Confused, I Met to Ask Externet About the Assembly and Dis-Assembly

Yes, very easy to pry-open the round end, just fingernails. The round end has two series of 8 LEDs, the side cylinder part has two series of 42 LEDs. Half cap 8 + half side 46 = 51
Each 51 series driven by a circuit board. Both 'driver' circuits fed by mains. All the side strips slide out of their slots if pulled. Seems well constructed, but the 'driving circuit' does not match.
Schematic----> https://s588.photobucket.com/user/Innernet/media/P1010590_zpsn1ztzgxb.jpg.html?sort=3&o=0

The use of a bridge rectifier is meaningless. IMAGINE the diodes in the bridge rectifier were LEDs instead. They would still rectify and yield light. That was my last approach connecting the two series of 51 in counterparallel directly to 120VAC.

Edited --- Yup, "No longer available" now. They sold two ! At the transactions listed , mine is the US flagged.
Similar ones :
----> https://www.aliexpress.com/wholesal...124108&SearchText=30+watt+led+white+corn+bulb

 

[Externet]

Needed some fun...

A jar holed its cap, a spent CFL bulb base, a 24W power supply inside and a 5 metres reel of 12V LED self adhesive strip wound on. Now I have light. So bright its ugliness is unviewable.