benz, the sag at higher voltages is likely due to heat. The driver is floating in space with no thermal contact. In a flashlight, the driver ring is sitting on the pill and this conducts heat away from the chips. The thermal tabs (big Gnd pin) is right there on the edge which will make a huge difference in board temperature.
With the drivers floating in the air there is very little cooling for such a small pcb, I believe that it overheated in maybe 30 seconds.
Mounted in a flashlight with prober contact to the aluminium, it will be much better.
This is one aspect that I cannot test properly. Even if I mounted a piece of aluminium on the driver, it would not be a prober test, because the actual cooling will depend on the actual flashlight and how it is mounted in it.
I have ordered some drivers, see post #2.
It will probably be a month before I am ready to do the reviews.
I did include the fasttech sku number, you can check with it.
Yes, I know, but your 7135’s are marked 38P (380 mA) in your photo. Fasttecs photo shows 350 mA type, giving 2.8A as stated in the specs.
The photo (and test) is of the one I received from them (I do not have any other drivers laying around at the current time).
I do not know if 38P is for 380mA, datasheet says 340mA is marked with A.
38P is 380 mA. That’s why you get around 3.0A in your measurements.
I wonder if all of their stock of that sku number is in fact 3.04A drivers. That would be an opportunity.
They do not have an A printed on them, that could also mean 380mA or rather between 340 and 380mA.
I would have expected the 38P to be a datacode, i.e. week 38 in year P.
Do you have any documentation saying that 38P is the current?
Don’t want to trust others experience without proof - eh?
To find specs you have to know the maker and find his specs sheets which I can’t.
But look at the V2 driver 3.04A at Kaidomain. You can’t see anything from the pictures but the 7135’s are marked “38K” the same way as yours “38P” (I have one in my hand right now). This is a common way to mark the bin and the letter is not relevant, but often a package code.
If you don’t trust me now that you have a 3.04A driver then it is your loss.
I do tend to trust datasheets more than random Internet postings (I do not know your level of experience). But I can see the one on FastTechs website is marked 35F, this does kill my date guess.
It is not really a loss for me, just some other numbers in my test. The "low" current version has the advantage that it runs slightly less hot, with many drivers close together it might allow it to run longer at full brightness.
I agree with HKJ, no one is saying your out to spread misinformation, but it does happen, there are many posts by people who did not understand something correctly and explained it to a new member with mistakes, or people who came up with wrong conclusions based on many factors, with no malicious intent or occasionally with intent.
When you have built a reputation as a respected member and a reliable source of information then people are more likely to take you at face value (there are many members who people trust for technical advice, including HKJ), but the adage trust but verify does also come in handy, and a datasheet is something that a company is using to drum up business so they are generally used as a source of verified information.
Correct, I have not rejected his specification, I would just like to see some documentation for it. In my first post about the AMC7135 I did link to the datasheet from ADD, where it specifies the A for low current version.
ADD might have changed the way they mark the chips (Datasheet is from 2006) or the 7135 might be from another manufacturer.
When I start to test all the drivers from intl-outdoor, I might be able to see what is correct (If some of the uses the low current driver).
When I get started on testing these drivers, it would probably be a good idea with some sort of index or table of the tested ones.
I was thinking something like this:
| Name | Type | Min. input V | Max. input V | Leds | Led current | Max. modes | PWM |
|---|---|---|---|---|---|---|---|
| 3 to 4.2 volt 800mAh driver (East 060A) | Buck | 3.3 | 4.2 | 1 | 1 | 1 | none |
| AA/AAA battery 550mA driver (Nanjg 102) | Boost | 0.9 | 1.3 | 1 | 0.37 | 1 | none |
| 3 to 4.5 volt 2800mAh driver (Nanjg 105c) | Linear | 2.9 | 4.5 | 1 | 3 | 5 | xxxHz |
Types:
Buck: Driver uses an inductor to decrease the voltage, input current is less than output current.
Boost: Driver uses an inductor to increase the voltage, input current is larger than output current.
Linear: Driver uses a linear chip (like 7135) to reduce the voltage, input current is the same as output current*.
Direct: No current regulation (except resistance) is present, input current is the same as output current*.
All driver types can use pwm for brightness regulation, the first 3 can also use linear regulation.
*Except for current for the control circuit.
It is possible to copy the table to a spreadsheet and sort on different columns.
Any comments?
Great info here, HKJ. Thanks for making the time to do it all! :)
You should be able to cut and paste it straight to Excel. Just tried your table and it worked on my computer.
Great work, very interesting, thank you. Is there a way to monitor the driver temperature during testing? I would be interested to see what performance drop there would be due to high temps, also the performance gains when heatsinked well. Or maybe point out which drivers are more temperature sensitive than others.
No, it would require the driver to be mounted on a heatsink.
I try to do the test rather fast, to avoid heating the driver. Looking at the efficiency you can estimate how many watts is lost in the driver.
Now I am just waiting for the drivers to arrive, the led test modules are ready:
The wires and current test resistor adds about 11mOhm resistance, i.e. at 3A I will have an error of 0.033 volt. I expect there are more variations between leds than that.
Drivers has arrived:
Any last minutes ideas, before I start testing?
Yes, check if they have modes as advertised (number and order L~~H or H~~>L)…
Thanks