Alonefire H45 "45W" dissection and analysis

[ulf]

Some time ago I bought an Alonefire H45 UV torch that turned out to be defect.

 

I filed a Dispute at Aliexpress and got a rather substantial refund while allowed to keep the torch.

I wanted to open it up to learn more how it was designed and why it failed and had a small hope of a possible repair.

This is topic is about the trouble it have and what I found out taking it appart.

---

 

The problem:

My H45 overheats and fades very quickly when turned on.

A proper intensity fading comparison between the H45, the Nemo and the Convoy S2+ give this result:

All three torches had fully charged batteries and a measurement from the integrating sphere was taken every 3s. during 3min.

After ca 1 1/2 min the H45 stabilises at the low intensity level without fading anymore. 

As the light pattern of the H45 is much wider than the pattern from Nemo the actual intensity appears much weaker from the H45 after a while.

The rate of the fading is also slow enough to make the eyes adapt to the fluorescence light making a visual judgement very difficult.

 

I measured the chip temperature at the end and it was over 160°C.

This has led to a partial fail of one of the LEDs and a slight beginning of failing of another.

The light from the three LEDs now differ with the failing one emitting less UV and a bit longer wavelengths.

Overheating a LED way beyond its Maximum Rated Temperature is not a good idea.

 

The problem with my H45 is not due to a lower selection grade of LEDs, but due to bad thermal design and assembly.

Exactly as aphalo wrote "I think it is impossible that a torch this size can dissipate even 15W without quickly getting hot."

---

The dissection:

The Lamp head is opened by unscrewing the ring, holding the main PCB:

It has oddly placed tool holes, but was not very tightly locked.

 

The PCB can then be wiggled out by angling it to get the USB-C connector free from it's opening.

After removing the front ring, UV-Pass filter and a silicone O-ring the LED cooling block can be unscrewed forward.

Observe the threads on the block and inside the housing.

Here was the main problem with my torch. The block was not tightened at all.

Due to a very loose fit there was almost no direct thermal connection to the outer housing.

Even when tightened the thermal contact area in such a thread is rather small. This is a design flaw.

 

The thickness of the block together with the LED's copper based PCB are reasonably thick, more than 6mm and reasonably OK.

There were sufficiently with thermal paste between the PCB and the Aluminum block.

The PCB was secured against the block by the reflector that was held in place with the central screw.

The three LEDs are connected in parallell and also have their four chips in parallell.

They should be driven with constant current getting a forward voltage around 3.7V

 

---

The Electronics:

The main PCB have three functional blocks beside the switch and indicating LEDs.

I have not seen any sign of a constant current control for the LEDs.

The only thing that limits the current is the row of resistors by the red cable where two has been removed.

The current is defined by the voltage of the batteries and the voltage over the LEDs and serial resistances in resistors, cables and contact points to the batteries.

This is extremely primitive!

 

The function of the board is controlled by a small 8-pin processor at bottom right, above the six resistor positions in a row.

 

The battery charging is reasonably advanced with a switch mode charger circuit.

 

The processor controls two 13A Mos-FETs to turn the UV on and off and likely by pulsing for the lower light position.

 

Interestingly there is plenty of room between the cooling block and the main PCB.

There would be enough room for a constant current controlled driver,.

Then ideally the UV-LEDs should be connecter in series to get a bigger difference between the LED driving voltage and the battery voltage.

Such a design would be more costly.

 

 

 

 

[colinbm]

Thanks Ulf
I haven't taken mine apart to compare the PCB.
Are you able to find the LEDs or other suitable ones ?

[Stefano]

Nice analysis. 160 °C is insane for an LED, you can't expect them to last long at that temperature. And I agree, dissipating 30 W of heat (if the LEDs have a ~33% efficiency) is not easy in a small size torch. The heatsinks I use for my 10 W LEDs are quite big and they reach almost 50 °C after a while.

[ulf]

1 hour ago, colinbm said:

Are you able to find the LEDs or other suitable ones ?

No, the ones used in the H45 is physically small, 5 x 5mm.

I have not found anything that would fit that footprint on the PCB.

 

I think I will put the torch together again tightening the LED module well with some thermal paste in the threads and see if it still fades the same way.

It is likely that the worst LED has lost much of it's conversion efficiency and mostly work as a heating element making everything a bit worse.

 

[ulf]

1 hour ago, Stefano said:

Nice analysis. 160 °C is insane for an LED, you can't expect them to last long at that temperature. And I agree, dissipating 30 W of heat (if the LEDs have a ~33% efficiency) is not easy in a small size torch. The heatsinks I use for my 10 W LEDs are quite big and they reach almost 50 °C after a while.

You are still assuming that the lamp is powering the LED module with 45W in your calculation.

I do not think that is true, but it is still generating much thermal energy and the cooling surface is not three times as big as the one on a Nemo.

 

The Nemo was advertised to be consuming 15W, but in reality draws ca 8W.

The Nemo dissipates ca 5-6W and get really hot when driven for longer periods with a stable DC-source.

It still keeps the temperature within a safe range for the LEDs

 

The thermal design of the H45 is unrealistically optimistic for continuous usage and the current control is non-existant.

Connecting the LED block better thermally to the cade will increase the life expectancy of the H45 torch.

 

[ulf]

I reassembled with thermal paste Hy-510 between the LED Block and case.

The retest show a marginal improvement of the final intensity, but changes the fading speed.

 

The fading is slower. it stops after 3 minutes instead of after 2 minutes.

The fading is also slightly less severe. Instead of reaching 51% of the start intensity it reaches 55%.

The fading is  uneven over time and the first 90 seconds the intensity loss only goes down 10%

Before the modification the torch had lost 30% intensity after 90 seconds and stayed above 90% half as long as after the modification.

To improve the stamina further some more aluminium could be added to the LED Block.

As the small button board extends over the main PCB and there must be space for the cables the added material could not be that big if the shape should be kept simple.

A more complex shape could couple the heat better to the case, if it had a reasonably good fit and more thermal paste.

 

I have no idea why the result from the modified torch is more noisy.

[colinbm]

I don't have fading issues with my H42, yes it gets warmed up quickly but not to hot to hold.

[ulf]

17 minutes ago, colinbm said:

I don't have fading issues with my H42, yes it gets warmed up quickly but not to hot to hold.

Are you quite sure?

I would not trust my eyes for this looking at fluorescence as they adapt to light intensities.

Your fancy illumination spectrometer is perfect for checking this.

 

If you decide to measure the power output from start and then after five minutes at max power, take care keeping the exact same distance and position of the spectrometer and torch.

If your H45 is without problem you should still be able to detect a slight fading from the hotter LEDs. They always loose some efficiency when hot.

 

It might be that the really sick LED in my H45 creates secondary problems, causing the fading in my unit.

[dabateman]

Or it might be that Ulf got a cheaper knockoff H45.

Spending all the money on the Led and then using bubble gum and duck tape to hold it all together. 

That happened to a friend of mine sister with a stereo.  It was held together inside with only black electrical tape.  She made all the critical soldiering points and got it working correctly. 

[ulf]

5 minutes ago, dabateman said:

Or it might be that Ulf got a cheaper knockoff H45.

Spending all the money on the Led and then using bubble gum and duck tape to hold it all together. 

That happened to a friend of mine sister with a stereo.  It was held together inside with only black electrical tape.  She made all the critical soldiering points and got it working correctly. 

That might be the case, but from a technical point of view I have problems make those ends meet as it obviously is a thermal problem in the design, If overpowering the LEDs in the first place did not cause the problem.

 

It would be really interesting to know if the Main PCB is of the same type in Colin's torch.

There are space for something better in the shell for electronics. 

[Guest]

I thought that some flashlights have a feature that lowers the output over time so to keep it running for a longer duration. I know my Petzl head lamp advertises something like that. I thought that the UV torch I have mentioned that as well in the instructions. Obviously related to Lithium batteries as the power source.

[Andy Perrin]

15 hours ago, ulf said:

The only thing that limits the current is the row of resistors by the red cable where two has been removed.

The current is defined by the voltage of the batteries and the voltage over the LEDs and serial resistances in resistors, cables and contact points to the batteries.

This is extremely primitive!

For real?!!! That’s hilarious. That’s like how I would have designed things as a ten-year-old. 

[colinbm]

I will start by saying I don't think it helps the Forum or its members with a report of a broken H42 light when it was purchased at a bargain price from China.
Other forums that I belong to that use this H42 light have had great success with it.....just saying....

Here is my contribution with a fully functional H42 light.

This is the H42 PCB.....

 

SRI 200UV spectrometer report immediately the H42 light was turned on......

 

SRI 200UV spectrometer report after the H42 had been on for five minutes......

 

The H42 light was comfortably warm to hold, about 40 degrees Centigrade.

 

[ulf]

1 hour ago, colinbm said:

I will start by saying I don't think it helps the Forum or its members with a report of a broken H42 light when it was purchased at a bargain price from China.
Other forums that I belong to that use this H42 light have had great success with it.....just saying....

Here is my contribution with a fully functional H42 light.

This is the H42 PCB.....

 

SRI 200UV spectrometer report immediately the H42 light was turned on......
 

 

SRI 200UV spectrometer report after the H42 had been on for five minutes......

The H42 light was comfortably warm to hold, about 40 degrees Centigrade.

 

Congratulations Colin that your H45 is working for you!

 

Thank you very much for opening it up to show us the PCB.

Now we know that your expensive H45 also use the same crappy electronics design without any active current control, as I have in my defect torch.

 

Unfortunately your measurements do not confirm that your torch is OK.

I hope it is a measurement error and not a real problem with the torch.

 

Please look at the spectral diagrams and observe the amplitude value.

The first diagram say 0.544 mW/cm2, while the second one say 0.149 mW/cm2

0.544>>0.149!!!

 

The difference is far too big to be caused by any faulty torch due to fading alone.

 

The original reason for the problems with my torch was that it was not assembled correctly, possibly in combination with some problems with the LEDs.

The H45 is really a thermally marginal design with primitive LED-driving design. I will elaborate more in detail about that later.

 

I really do not understand why you have any problem with my report of the analysis of the defect torch. 

It was bought at a bargain price to see if it was OK.

It was not.

That I think is a valuable information for the members too.

 

Please redo the measurements with fresh batteries and both the torch and spectrometer in really fixed positions.

When working with position sensitive setups sometimes "duck"-tape is very good to use. 

[Andy Perrin]

Colin, the peak value in yours declined from 0.544 to 0.149 in five minutes, which means it fell by a factor of 3.7 or so. Ulf’s looks like it fell by about a factor of 2. 
 

ETA: and I see Ulf pointed out the same thing. Yeah, it looks like you are both getting the same results here. 
 

I think it’s useful to know that the torch is poorly designed. Thank you both for your investigations, which are confirming each other. 

[colinbm]

OK I will redo the tests again soon.
Yes I was shocked to see the drop in 5 minutes too.

 

[colinbm]

Ulf
You say you have the Alonefire H45 UV torch.
Mine is the Alonefire H42 UV torch.

[ulf]

On one side it is marked Alonefire® and below

H45.

The text on the opposite side states:3x LED UV 12 Core and below a row of

approval logos for FCC, CE, RoHS, recycling and overcrossed thashbin.

I am convinced that the marking do not affect the performance. 

 

[enricosavazzi]

My specimen of the Alonefire H45 UV comes from an EU-based seller on Amazon. I have been unable to unscrew the retaining ring of the PCB even with a proper spanner screwdriver, but all the external markings as well as all the plated-through holes in the rear of the PCB correspond exactly to the pictures published earlier in this thread.

 

The reflector was also loosely screwed in from the front of the torch, as described by Ulf. The only difference at this point seems to be that the retaining ring of the PCB in Ulf's specimen came loose and could be unscrewed, while mine might be glued in place.

 

It seems unlikely that my specimen, or other specimens purchased in the EU, contain any internal components different than the specimen studied by Ulf.

[colinbm]

This is the one I have, the price has increased.....
https://www.amazon.com.au/Alonefire-Flashlight-Rechargeable-Ultraviolet-Blacklight/dp/B08RHTN7RV/ref=sr_1_12?crid=23OKU57G41FV3&keywords=alonefire&qid=1640863202&sprefix=alonefire+%2Caps%2C233&sr=8-12

[ulf]

I think the H45 they sell are the same configuration as the H42.

I think that they had all six "current control" resistors mounted from the beginning and got too much reclamations due to overheating.

It might be that they modified the original boards and lowered the power, but still had material looking like the "45W" H45-version

 

6000mW optical power "illumination" as they state might even be reasonable for a more realistic 25W input power.

 

 

[colinbm]

Are you able to know the LEDs Vf & Cf so I can run the torch head from an adjustable power supply & test them again please ?

[ulf]

On Alonefire's official store on Aliexpress there is also a H42:

https://www.aliexpress.com/item/1005002133384717.html

It is cute to see how much they trust the life of the LEDs, stating 100 000 hours.

That is 11.4 years of continuous usage!

[ulf]

2 hours ago, colinbm said:

Are you able to know the LEDs Vf & Cf so I can run the torch head from an adjustable power supply & test them again please ?

No, i have no idea of the  specifications of the LEDs they have used. We actually do not know their maximum ratings.

If you have fully charged batteries, preferably charged outside the lamp, but at least with a cell voltage , un loaded, of ≥ 4.1V they would work fine for several five minute tests.

No need for an external powers supply.

I would even advice against powering the lamp head that way as the current limiting serial resistance partially comes from the batteries internal resistance.

When driving the head from a power supply there is an increased risk of destruction!

 

I have been thinking about doing the same, but as one of my three paralleled LEDs is not quite OK the risk is even worse.

 

If you want to redo the intensity measurement, just start over with fresh batteries and before measuring tape the torch and spectrometer in a good position for measurement. Even a small shift in position or alignment will affect the measurements. 

Good luck measuring!

Do not forget that the torch has two intensity settings: High - Low - Off...

Last time you might accidentally have shifted to Low for the second measurement.

[colinbm]

Thanks Ulf
I am cycling the batteries. They are discharging now on my OPUS BT-C3100, ten I will recharge them fully & redo the test.
I had the light mounted on a heavy adjustable arm & in a fixed position for the test.