Replacing Neon Lamp with LED

[Conrad Hoffman]

I've found running two LEDs in parallel, with the polarity flipped, provides less flicker and each one protects the other from excessive reverse voltage. Cap drop works fine but is prone to passing line transients so needs some series resistance and small cap. I use this circuit for turntable strobes along with a pair of zeners so the duty cycle of the LEDs is limited to a shorter portion of the line cycle and you get sharper strobe disk bars. Not needed for a power light.

[Benta]

@Zero999, has your account been hacked?

Why do you keep responding to threads like this? It's starting to sound like paranoid delusions. What makes you think his account was hacked? Because he provides some links to the places where he bought some of the LEDs he is describing? There is no rule against doing that, it's not like Rapid Electronics is some random sketchy Chinese company.

Why are you so aggressive?

This thread is about a neon lamp indicator being replaced with an LED.
Suddenly, a known user shows up with pictures of Architectural/Designer/Mood lighting that look like 100% advertising to me.

My question was in no way meant as provocative or challenging, but more a result of confusion, as I know that the said user is very good at staying on topic in a thread, and always provides valuable insights.

Concerning: "Why do you keep responding to threads like this? It's starting to sound like paranoid delusions.", that's perhaps the most unfair I've ever seen.
You should be ashamed of yourself for that statement. Feel free to go through my posts history.

[Zero999]

@Zero999, has your account been hacked?

Why do you keep responding to threads like this? It's starting to sound like paranoid delusions. What makes you think his account was hacked? Because he provides some links to the places where he bought some of the LEDs he is describing? There is no rule against doing that, it's not like Rapid Electronics is some random sketchy Chinese company.

Why are you so aggressive?

This thread is about a neon lamp indicator being replaced with an LED.
Suddenly, a known user shows up with pictures of Architectural/Designer/Mood lighting that look like 100% advertising to me.

My question was in no way meant as provocative or challenging, but more a result of confusion, as I know that the said user is very good at staying on topic in a thread, and always provides valuable insights.

Concerning: "Why do you keep responding to threads like this? It's starting to sound like paranoid delusions.", that's perhaps the most unfair I've ever seen.
You should be ashamed of yourself for that statement. Feel free to go through my posts history.

If you read my post properly, you'll find I did suggest a specific LED which will give a similar colour to a neon lamp, nothing about architectural/designer/mood lighting.

Those LEDs aren't widely available. Another option is to use a warm white LED, which might be acceptable, if the neon has a yellow/orange filter.

[Benta]

If you read my post properly, you'll find I did suggest a specific LED which will give a similar colour to a neon lamp, nothing about architectural/designer/mood lighting.

Those LEDs aren't widely available. Another option is to use a warm white LED, which might be acceptable, if the neon has a yellow/orange filter.

Yes, and if you read the thread, the OP never requested a "neon colour" LED. Your post looked like an advertisement, OK? 'Nuff said.

[mikeselectricstuff]

I've found running two LEDs in parallel, with the polarity flipped, provides less flicker and each one protects the other from excessive reverse voltage.

Pretty sure I've seen LEDs with two die that do this internally

[Benta]

I've found running two LEDs in parallel, with the polarity flipped, provides less flicker and each one protects the other from excessive reverse voltage.

Pretty sure I've seen LEDs with two die that do this internally

Interesting.
I've never seen such LEDs myself, but often optocouplers. Do you have lniks/references? Thanks.

[langwadt]

I've found running two LEDs in parallel, with the polarity flipped, provides less flicker and each one protects the other from excessive reverse voltage.

Pretty sure I've seen LEDs with two die that do this internally

Interesting.
I've never seen such LEDs myself, but often optocouplers. Do you have lniks/references? Thanks.

obsolete but; https://www.elfadistrelec.dk/Web/Downloads/86/20/Lx-bipolar_eng_tds.pdf

if you can live with the mixed color there's crap tons of dual color leds that'll work

[tooki]

If you read my post properly, you'll find I did suggest a specific LED which will give a similar colour to a neon lamp, nothing about architectural/designer/mood lighting.

Those LEDs aren't widely available. Another option is to use a warm white LED, which might be acceptable, if the neon has a yellow/orange filter.

Yes, and if you read the thread, the OP never requested a "neon colour" LED. Your post looked like an advertisement, OK? 'Nuff said.

No, it really didn't. It read like a standard, legitimate "here's a product I'm familiar with, maybe this will help. here's the links" post, the likes of which there are thousands and thousands on the forum.

And your "architectural lighting" is literally the standard way LED colors are shown: a single LED uplighting a white surface. I do see the visual similarity to some architectural lighting, but if you shopped for colorful LEDs you'd have recognized the image for what it is.

I've found running two LEDs in parallel, with the polarity flipped, provides less flicker and each one protects the other from excessive reverse voltage.

Pretty sure I've seen LEDs with two die that do this internally

Interesting.
I've never seen such LEDs myself, but often optocouplers. Do you have lniks/references? Thanks.

See attached datasheets for some currently available on digikey (plus one obsolete Fairchild part).

[SeanB]

Simplest replacement is to use one of the high efficiency LED's , and no capacitive dropper. Simply use the same value resistor as the neon used, but have a bridge rectifier after it. I tend to use 4 12V zener diodes, simply because I have a lifetime supply of them, so they are often used as regular diodes where appropriate. Then a capacitor, around 10-100 uF 16V, and then a series resistor, one tenth of the main resistor, to feed the LED. Safe, and you can use 2 resistors in series on the input side, one in each leg, whose combined value is somewhere between 47k and 220k, which is typical for most neon lamp current limiters. Simple, has no failure mode other than the electrolytic going open circuit, which does not matter as it just flickers, and will have a lifetime equal to that of the LED, or a century in most cases, as the LED is being run at under 1mA. High efficiency LED's in red or green at 1mA are incredibly bright, and are one of the more common type available these days.

[Zero999]

If you read my post properly, you'll find I did suggest a specific LED which will give a similar colour to a neon lamp, nothing about architectural/designer/mood lighting.

Those LEDs aren't widely available. Another option is to use a warm white LED, which might be acceptable, if the neon has a yellow/orange filter.

Yes, and if you read the thread, the OP never requested a "neon colour" LED. Your post looked like an advertisement, OK? 'Nuff said.

He asked for advice about replacing a neon lamp with an LED. It should be obvious, unless you're retarded, that LED should be a neon colour to emulate a neon lamp. No one else thought my post looked like an advertisement. Only you.

I've found running two LEDs in parallel, with the polarity flipped, provides less flicker and each one protects the other from excessive reverse voltage.

Pretty sure I've seen LEDs with two die that do this internally

I'm pretty sure this LED is really three connect in series.
https://www.rapidonline.com/truopto-os5mpl56d1a-4-8mm-warm-white-9v-3000k-22lm-water-clear-130--55-9126

[paul_g_787]

I have taken into account what was posted here and modified my design accordingly. Please see the v3 schematic

I can only get 2.4V Zeners at the moment. Should be close enough? The max rating of the LED is 3V and the continuous is 2V.

I am getting approx 7mA current draw through the whole circuit. (ammeter in series with Live wire).
I am also getting approx 7mA at the LED. (ammeter in series with LED)

By my calculations 7mA at 253V is 1.771W so I will need a 100Ω 2W metal film fusible resistor?

2W seems excessive, the fusible resistors in the commercial lights I have are the small 0.25W metal film resistors? Am I calculating something wrong here?

[TimFox]

The Zener is redundant and can be deleted, so long as the voltage across the LED never goes negative (reverse bias).  The LED, driven from a reasonable source impedance, will limit the voltage across itself.  LEDs, like Zener and forward-biased diodes, should never be driven directly from a constant-voltage source.

[paul_g_787]

The Zener is redundant and can be deleted, so long as the voltage across the LED never goes negative (reverse bias).  The LED, driven from a reasonable source impedance, will limit the voltage across itself.  LEDs, like Zener and forward-biased diodes, should never be driven directly from a constant-voltage source.

If the Zener is redundant then I have a question; Why do they include them in LED light bulbs from the store?

I have taken apart 6 different models and makes so far and they all have a capacitive dropper with a Zener diode across the LEDs in this same configuration.

I thought it is to protect from sudden voltage spikes on the mains?

[TimFox]

Belt and suspenders, maybe?  If a spike gets through the capacitor (which will preferentially pass fast spikes), the Zener might have a higher capability to survive the energy in the spike, but will still (transiently) over-voltage the LED.  Because of spikes, the series resistor is necessary in such simple capacitance-dropper circuits to limit instantaneous current.

[paul_g_787]

OK well it seems like this is a good design then unless anyone else can point out any other design flaws?

Regarding the Wattage of The fusible resistor though. Which will I need?

7mA at 2V - at the LED is 0.014W
However 7mA at 253V (240 +10%) - at the live mains input is 1.771W

So which would it be?

I believe I am miss-understanding something here.

[Benta]

A comment to your latest schematic: you're still treating the LED as a voltage controlled device. It is NOT. You need to control the current through the LED.

[Vovk_Z]

I have taken into account what was posted here and modified my design accordingly. Please see the v3 schematic

It seems to me over-complicated. If LED 50 Hz blinking is ok (usually it can't be seen by eyes), then the circuit may contain only three parts: bright (ultra-bright) LED (1-2 mA), series 1N4007 diode and series 47-200k 1W resistor.
It may contain a diode bridge instead of a singe rectifying diode, so blinking will be 100 Hz, the same as for neon lamp. The main thing is to have a LED diode bright enough with a low (1-2 mA) current (so we could use a series voltage drop resistor instead of a cap).

[TimFox]

OK well it seems like this is a good design then unless anyone else can point out any other design flaws?

Regarding the Wattage of The fusible resistor though. Which will I need?

7mA at 2V - at the LED is 0.014W
However 7mA at 253V (240 +10%) - at the live mains input is 1.771W

So which would it be?

I believe I am miss-understanding something here.

The fusible resistor should be specified as the fuse current:  the current it can safely pass and the higher current that is required to blow it.
The resistor itself should be rated for the power dissipated when the maximum current passes through it, not the entire circuit.  If you know the current through the resistor is 7 mA, then the power in that 100 resistor is only 5 mW.  The power in the other resistor, 1000 , at 7 mA is 50 mW.

[StillTrying]

Simulate it.
It works OK, 9 VDC across the 47u and 6.2 mA through the LED, until some 100V spikes come along.
A 12 or 15V zener across the cap seems safer, that zener would clamp 100s of mA during mains voltage spikes, so probably make it a 1W one.
The zener across the LED doesn't do anything useful.

[Benta]

I have taken into account what was posted here and modified my design accordingly. Please see the v3 schematic

I can only get 2.4V Zeners at the moment. Should be close enough? The max rating of the LED is 3V and the continuous is 2V.

I am getting approx 7mA current draw through the whole circuit. (ammeter in series with Live wire).
I am also getting approx 7mA at the LED. (ammeter in series with LED)

By my calculations 7mA at 253V is 1.771W so I will need a 100Ω 2W metal film fusible resistor?

2W seems excessive, the fusible resistors in the commercial lights I have are the small 0.25W metal film resistors? Am I calculating something wrong here?

I attach the schematic of a cap-drop supply I designed 15 years ago. It's laid out for a bit higher current than your LED driver, and has been operating in many devices with 100% reliability.
Ignore the lower and the right parts of the drawing, they're irrelevant to your application.

I went back over my design notes:
R1: 470 kohm, Vishay VR37 (cap bleed resistor)
R2: 330 ohms, Vitrohm BW234, (wirewound startup current limit resistor)
D1...D4: 1N4007
Z1: 1N592xB, ONSemiconductor (voltage depending on number of LEDs, for a single LED I suggest 5.6 V)
R3: any 0.3 W resistor, value dependent on Z1 and desired LED current and color
C1: 470 nF 305 VAC X2 (this is chosen for 35 mA IIRC, you need a lower value)
C2: 220 uF, 16V

This lists the critical parts. You have to consider the following:
1: at turn on, the only thing limiting inrush current at peak AC voltage is R2, and it will have to withstand >1 A peak current.
2: the same is true for for C2 and Z1. IIRC, that explains the relatively high value of C2 and the high power rating of Z1.
3: the bleed resistor avoids C1 discharging through a person touching the mains plug prongs after unplugging.

R1 and R2 are pure loss, of course, and their values need to be weighted. If you only need 7 mA, R2 can be increased, with lower demands on C2 and Z1 making them smaller and cheaper.

I'm not too proud of the design, I dislike cap-drop solutions. It was only chosen because it was the simplest and cheapest, not due to performance.

Hope this helps, as I said, it's a tried-and-tested, reliable design.

[paul_g_787]

The fusible resistor should be specified as the fuse current:  the current it can safely pass and the higher current that is required to blow it.
The resistor itself should be rated for the power dissipated when the maximum current passes through it, not the entire circuit.  If you know the current through the resistor is 7 mA, then the power in that 100 resistor is only 5 mW.  The power in the other resistor, 1000 , at 7 mA is 50 mW.

Ah I see I am thinking of it back to front.

So for example that means at 7ma, if I were to use a 0.25W resistor it would need to be 5.1KΩ correct?

Or I could use a 2.5KΩ metal film resistor to provide 10mA of fusing at 0.25W?

[TimFox]

A 5.1 k resistor would be just within its power rating of 1/4 W at 7 mA current through it.
In general, it is good practice to derate resistors, typically to no more than 1/2 of the rated power, or do a careful calculation to ensure that the worst-case power in the part will not raise its temperature above what your use requires.
That's not the same as its fusing rating.  For example, the TE series FRN leaded resistors are rated to blow in 60 sec at 16x rated power, or 15 sec at 32x rated power.
The power rating is for < 70^o C ambient temperature, derated above according to the graph.
https://www.te.com/commerce/DocumentDelivery/DDEController?Action=showdoc&DocId=Data+Sheet%7F1773218%7FE%7Fpdf%7FEnglish%7FENG_DS_1773218_E.pdf%7F1625879-2

[james_s]

I appreciate new LED colours- no need to give someone flak for that. Kingbright offers some as well.

Overall, I find mains capacitive droppers simply don't last. For LED nightlights, you can do LTSpice sims and see the peak currents are high which shortens the life of almost all the  components including the LED. I've taken apart quite a few (that have regulatory approvals). You need a fuse, flameproof resistor, dropper-rated capacitor and then a scheme for low flicker. The one where the electrolytic cap exploded stunk up the house.
I've also put three LED's in series for higher brightness.

I have several LED nightlights that have capacitive droppers. The LEDs were cheap and wore out within just a few months so I replaced them with some non-phosphor LEDs I had in various colors. That was probably 10 years ago and not one of them has failed since.

[rfclown]

I've never dealt with trying this before, but when I read this thread I wonder what I'm missing. It seems to me that a current limiting resistor and back to back diodes is all that would be needed. I just grabbed two LEDs out of a bag of miscellaneous parts (didn't know what colors they were). They both measured about 2.7V and I checked that both would light with 1mA of current. Figured that 100k would give me roughly 1mA. Looked in resistor bin and saw 1/2 watt 47k parts, so I grabbed two. Wired together. I don't see any flickering.

[Benta]

People don't have the same eyes.
Tip: try holding the LEDs in your peripheral vision instead of looking directly at them. Odds are you'll see the flickering. There are also people that see 100 or 120 Hz flickering directly in their field of vision.