Skip to main content

Just over a year ago, I purchased a 25 pack of direct replacement LED's from a well-known seller of these bulbs who attends York regularly. They were replacements for the 1445 bayonet bulb and are rated for 18 volts. The problem is that 16 out of 25 bulbs have now failed in just over a year.

I decided to open one up and see what's inside.  There are 3 components: bayonet base, 8mm "straw-hat" type LED, and a 300 Ohm resistor.

What's making these bulbs fail? They are installed in lamp posts running at a steady 10 VAC using a postwar ZW.  Is it just my bad-luck?  Is it a defective lot of bulbs? Is there an inherent problem such as reverse-voltage breakdown, overheating on pulsed AC, or an incorrectly sized resistor?

Has anyone else had similar experiences with direct replacement LED's?

IMG_8049IMG_80508MM LED

Attachments

Images (3)
  • 8MM LED
  • IMG_8049
  • IMG_8050
Original Post

Replies sorted oldest to newest

@Ron045 posted:

You could make your own.  I re-did my bayonet bulbs with Christmas lights and they are still running bright since 2016.  Some might also suggest a diode when using AC.

https://ogrforum.ogaugerr.com/...or-lionel-260-bumper

Have Fun.

Ron

Thanks for the good advice, Ron. I love recycling LEDs' also. I recently did an ERR upgrade on a loco using LED's scavenged from one of Harbor Freight Tools free flashlights. Although I enjoy doing LED upgrades on some accessories to improve performance, I just don't have an interest in custom building 50 of these bayonets from scratch. I suppose I could crawl under the layout and insert some diodes in-line at various locations, but it may be easier to pop in the old incandescents bulbs again.

@shorling posted:

I would suggest contacting the vendor and recant your experience.  They have been most helpful with any of their product issues I have encountered.

When the bulbs first arrived, 4 were dead right out of the box. The vendor was very helpful and sent replacements right away. They have always been good to work with. But with  these bulbs dying off like flies, I don't want to make an ongoing nuisance of myself by asking them for more and more free replacements. Ultimately, choosing to go in a different direction may be appropriate. 

I'd try adding a general purpose rectifying diode ("normal") (1/2 wave/pulsed dc) or a bridge rectifier (dc).  LED are dc devices only. Any AC LED is rectified internally or has more resistance to ACs negative wave, it still functions on DC internally.Some LED can handle a certain amount of ac, but dc is easier on them in general.

  Over voltage will burn an LED near instantly too.  A surge can wipe them out. 

Plus with future replacement, you can use a wider variety of LED on 1/2 wave or dc.  Incandescents can still be used if a diode is on the circuit.

 Voltage will drop by about 0.75v after the diode..no big deal really. (diode amps have to exceed the total amp draw of bulb(s)/LEDs. Voltage rates should be higher too (much higher= safer from surges... say 35v or more)  Làrger amp diodes stay cooler as well. (A Brdge Rectifier is just 4 diodes in one package, already arranged to make dc from ac).   

  Price them all in small bulk from an electronic supplier & you may save quite a bit.

I like incandescent's look best, but it is hard to argue with the life and low draw of LED done right.

ANY car should be able to have LEDs added. Very few conditions would rule it out.

@GregR posted:

...

I decided to open one up and see what's inside.  There are 3 components: bayonet base, 8mm "straw-hat" type LED, and a 300 Ohm resistor.

What's making these bulbs fail? They are installed in lamp posts running at a steady 10 VAC using a postwar ZW.  Is it just my bad-luck?  Is it a defective lot of bulbs? Is there an inherent problem such as reverse-voltage breakdown, overheating on pulsed AC, or an incorrectly sized resistor?

That you identified the resistor as being 300 Ohms tells me you probably have a meter?  Your photo suggests the resistor is in good shape (not charred/toasted) but a simple meter test will confirm that.

I'd say the failure is because they did NOT install a 1-cent diode to protect the LED from reverse-voltage breakdown from applied AC voltage.

In this recent OGR post regarding a wedge-style plug-and-play LED replacement for an incandescent, they include the diode.  I'd contact the manufacturer and refer them to this thread!  In bulk the diode will cost them less than a penny.

Separately, and I realize you're running at "only" 10V AC, but a 300 Ohm resistor is suspicious.  Your photo shows a 1/2 Watt LED.  Let's say it's operated at half of that since it's AC...so 1/4 Watt.  The math is a little more involved with AC operation but if operating at its rated 18V and dropping that down to ~3V as per a typical white LED...that means the resistor is absorbing several times the power of the LED and that looks like only a 1/4W resistor.

I did have a couple of issues with the subject direct replacements.  I had incandescent lamp posts powered by 14 VAC through a dimmer.  I changed to direct replacement LEDs and some of the LEDs didn’t work because I didn’t pay attention to lamp post wiring polarity since it was AC.  The dimmer output was chopped and basically DC.  But it looked like I had bad bulbs at the start.

I also had what looked like shorted direct replacement LEDs.  Turned out it was the lamp post socket.  On some sockets the center contact was slightly off center.   The direct replacement LED base created a short with the offset socket contact.  No bad LEDs here either just bad lamp post sockets.

 

 

 

@stan2004 posted:

That you identified the resistor as being 300 Ohms tells me you probably have a meter?  Your photo suggests the resistor is in good shape (not charred/toasted) but a simple meter test will confirm that.

I'd say the failure is because they did NOT install a 1-cent diode to protect the LED from reverse-voltage breakdown from applied AC voltage.

In this recent OGR post regarding a wedge-style plug-and-play LED replacement for an incandescent, they include the diode.  I'd contact the manufacturer and refer them to this thread!  In bulk the diode will cost them less than a penny.

Separately, and I realize you're running at "only" 10V AC, but a 300 Ohm resistor is suspicious.  Your photo shows a 1/2 Watt LED.  Let's say it's operated at half of that since it's AC...so 1/4 Watt.  The math is a little more involved with AC operation but if operating at its rated 18V and dropping that down to ~3V as per a typical white LED...that means the resistor is absorbing several times the power of the LED and that looks like only a 1/4W resistor.

Yes, I did meter test to confirm the resistor value (which matched the banding). I also suspected the resistor was undersized. The whole bulb casing may well be acting as a heat sink. As you and others have suggested, I think reverse-voltage breakdown is the culprit.

I have a sleeve full of 1.0A diodes. I just crimped one onto the end of the wire that powers the light in my 445 switch tower. It was easy enough and I just trimmed and inserted the end of the diode into the spring clip.

Now for my string of Lionel No. 70 lamp posts: Will a 1.0A diode be good for a string of 3 LED bulbs, or would each lamp require a 1.0A diode? I'm the wrong kind of engineer for making these calculations.

IMG_8052

Attachments

Images (1)
  • IMG_8052
Last edited by GregR
@shorling posted:

The vendor must have sold tens of thousands of these direct replacements.  If there was a fundamental design issue I suspect they would have been out of business a long time ago.  My issues in my previous post were all application related.

Just because they've sold thousands doesn't mean there isn't a basic design flaw.   Lots of products have major flaws, yet massive quantities of them are shipped before the flaws surface.

The diode is a basic precaution running LED's on AC voltage, the reverse voltage rating of most LED's is around 5V peak.  A 10 VAC source has a peak reverse voltage of 14 volts to the LED, almost three times it's rating.

Once again, though you should expect it, Stan is spot on with his analysis, they needed the diode.

@GregR posted:
Will a 1.0A diode be good for a string of 3 LED bulbs, or would each lamp require a 1.0A diode? I'm the wrong kind of engineer for making these calculations.

Yep, unless these are REALLY high power diodes, the 1A diode is plenty.

conundrum1

By happenstance, GregR photographed what appears to be some kind of marking on the outside of the bayonet that might indeed be at the same point where a few seconds of huge-heat is applied to the exterior wall. 

However your answer is incomplete.    To earn a gold star, you must articulate whether they apply (1) a blob of solder to the end of the LED lead, or (2) a blob of solder to the inner wall of the bayonet, or (3) both 1 and 2...  before final assembly.  Or identify some other method.  I'm nominating GregR to be the arbiter.   To be clear, I do not know the answer!

I guess my point is if they indeed go thru such machinations to install a resistor-LED combo, it would be only a minor incremental effort to add the 1-cent diode.  A 64% failure rate (16 out of 25) after 1-year is no way to run a railroad...

Attachments

Images (1)
  • conundrum1
Last edited by stan2004

I await GregR's input on HOW he disassembled it before awarding the highly coveted gold star.

For example, he might have chosen not to photograph a short flexible strand (i.e., a tether) of, say, #30 wire that is attached between the inner wall of the bayonet and the LED lead.  Then the assembly is inserted and the solder blob at the base of the bulb is heated to allow the resistor lead to make the "hot" connection.

Last edited by stan2004
@stan2004 posted:

I offer a gold star to the first person who explains HOW they assemble these.

The bottom part of the resistor appears to be firmly attached (soldered) to the bottom of the bayonet bulb...but HOW is the other other lead of the LED securely attached to the inner wall of the bayonet cylinder?

Devil's advocate - The outside lead is very poorly attached to the bayonet cylinder, and that is the source of the failure.........

Seeing that these are made in China, there could be no solder at all on the sleeve. Just "spring tension" on the lead holding it against the sleeve.

Last edited by PLCProf

I have added a few more photos to reconstruct the pieces. The cathode lead is actually hooked under the plastic sleeve which is then wedged (friction fit) tightly into the bayonet base. It seems to be an adequate electrical connection. The plastic sleeve takes some fighting to remove. It is not the connections that have been the issue. The LED's have been failing. I have been testing the failed ones.

From what I can tell, Stan2004 is correct that the resistor on the anode side is spot soldered through application of heat at the base of the bulb.

Although there is space for a diode, it would add some complexity to the manufacturing process. Frankly, I doubt there is enough of a market to justify the cost of making LED bulbs that are truly meant for an AC power source.

For now, I'm not giving up on the LED's. I have a ton of 1.0A silicon diodes laying around that can be soldered and shink-tubed in place where needed.IMG_8054IMG_8055

Attachments

Images (2)
  • IMG_8054
  • IMG_8055
Last edited by GregR
@PLCProf posted:

Devil's advocate - The outside lead is very poorly attached to the bayonet cylinder, and that is the source of the failure.........

Seeing that these are made in China, there could be no solder at all on the sleeve. Just "spring tension" on the lead holding it against the sleeve.

gold star small

@GregR posted:
..

Although there is space for a diode, it would add some complexity to the manufacturing process. Frankly, I doubt there is enough of a market to justify the cost of making LED bulbs that are truly meant for an AC power source.

For now, I'm not giving up on the LED's. I have a ton of 1.0A silicon diodes laying around that can be soldered and shink-tubed in place where needed...

I realize in your layout application you can externally add 1A diodes "before" the LED bulbs.  But for anyone who is ambitious and wants to add the diode inside the bayonet base, consider the 1N4148 which is rated at 0.2 Amps which is plenty for the application.  This is the diode used in the linked thread earlier where the LED-replacement bulb has both resistor and diode.

The 1N4148 is easily found for less than 5 cents in small quantity.  But it's physically much smaller.  Albeit the leads are closer to 0.02" than 0.03" diameter for the 1N400x.  Thus it's a bit more flexible which may not play well with advocates of friction-fit.  Maybe put the 1N4148 in-series with the resistor...this leaves the stiffer LED cathode terminal as-is.

1N400x vs 1N4148

Attachments

Images (2)
  • gold star small
  • 1N400x vs 1N4148

Just because they've sold thousands doesn't mean there isn't a basic design flaw.   Lots of products have major flaws, yet massive quantities of them are shipped before the flaws surface.

The diode is a basic precaution running LED's on AC voltage, the reverse voltage rating of most LED's is around 5V peak.  A 10 VAC source has a peak reverse voltage of 14 volts to the LED, almost three times it's rating.

Once again, though you should expect it, Stan is spot on with his analysis, they needed the diode.

Yep, unless these are REALLY high power diodes, the 1A diode is plenty.

As you have correctly noted this waould be a major design flaw.  However this vendor has been successfully selling this product for many years.  They are the major supplier of this product which is why you see them at York.  It is not like they are a startup and released a new product.  I run over 100 of their direct replacements on my layout with zero failures or degradation over many years..  That performance is not possible without addressing reverse voltage in the design.

@shorling posted:

As you have correctly noted this waould be a major design flaw.  However this vendor has been successfully selling this product for many years.  They are the major supplier of this product which is why you see them at York.  It is not like they are a startup and released a new product.  I run over 100 of their direct replacements on my layout with zero failures or degradation over many years..  That performance is not possible without addressing reverse voltage in the design.

Without a diode, they did not address a major design shortcoming.  They're exceeding the ratings of the LED whenever the applied AC voltage goes over around 5 VAC, if some of the product is not failing, they've just been lucky with the LED's taking the abuse.  I don't know if the LED's you are using have the same design, but the item in this thread we're looking at is a defective design, there's no if's, and's or but's about it.  Any design that violates the absolute maximum ratings of a component is a bad design, even it it happens to work a majority of the time.  This is electrical engineering 100, it doesn't even rate the 101 class!

@stan2004 posted:

I realize in your layout application you can externally add 1A diodes "before" the LED bulbs.  But for anyone who is ambitious and wants to add the diode inside the bayonet base, consider the 1N4148 which is rated at 0.2 Amps which is plenty for the application.  This is the diode used in the linked thread earlier where the LED-replacement bulb has both resistor and diode.

For that matter, you could also use an SMT diode, even smaller.  There's no reason you need leads for this application, just stick it in line with the bulb and put the assembly back together.

Last edited by gunrunnerjohn

If I were to attempt building these little "ships in a bottle", I woudn't try to modify the existing bulbs. It's very difficult to get these apart without destroying the mounting sleeve and/or the annular glass insulation around the bottom contact. I would start with new components.

Empty BA9s bases are available relatively cheap. For constructability, I might try soldering a couple of  segments of thin magnet wire into the base leaving about a 1/2" portion sticking out of the base. The LED/resistor/diode assembly could be built separately "out of the bottle", then soldered to the exposed magnet wire ends. Then the entire LED assembly could be simply inserted and hot glued into the base. I think there is plenty of room to work with to install both the diode and resistor in the 8 to 10 mm space available.

Sockets

Attachments

Images (1)
  • Sockets

The BA9S bases are even available thru Amazon and qualify for the Free Shipping w/ $25 order.

ba9s

The listing even suggests this is for DIY LED bulbs!

Amazon has tons of BA9S LED bulbs in countless styles.  It's interesting how there's some word-play going on.  Some LED bulbs use the term "AC/DC" and others use "non-polarity".  I was reading some comments/complaints that some bulbs that say "non-polarity" only work on AC or positive-DC (positive being on the center pin).  In my opinion, if it says "non-polarity" it should accept negative-DC which of course means it has to have a bridge rectifier inside.

 

Attachments

Images (1)
  • ba9s
@GregR posted:

Just over a year ago, I purchased a 25 pack of direct replacement LED's from a well-known seller of these bulbs who attends York regularly. They were replacements for the 1445 bayonet bulb and are rated for 18 volts. The problem is that 16 out of 25 bulbs have now failed in just over a year.

I bought the 50 pack split order 25 red/25 green 1445 glass-globe style LEDs several years ago from a well-known seller for use in K-Line switches. The 25 red are all fine to this day, every single one of the 25 green were dead within a year.

@ADCX Rob posted:

I bought the 50 pack split order 25 red/25 green 1445 glass-globe style LEDs several years ago from a well-known seller for use in K-Line switches. The 25 red are all fine to this day, every single one of the 25 green were dead within a year.

Did you happen to pry open a green to see what's there?  I assume you wouldn't have opened a red since they are working!  It would be curious to know if the manufacturer figured it out and changed their ways...

Add Reply

Post

OGR Publishing, Inc., 1310 Eastside Centre Ct, Suite 6, Mountain Home, AR 72653
800-980-OGRR (6477)
www.ogaugerr.com

×
×
×
×
Link copied to your clipboard.
×
×