High efficiency multi mode analogue LED driver
- Thread starter royfellows
- Start date
royfellows
Well-known member
Tangent_tracker said:
I'm sorry, its the wrong answer! You have just lost.............................
As Chris Tarrant would put it.
Tangent_tracker said:
If you read fully what I said, I mentioned "manufacturers data sheet". I for one do not have the temerity to think that I know more about "the chemistry" than the people who make the bloody thing...
Oh Dear
royfellows
Well-known member
ChrisJC said:Tangent_tracker said:
Do tell.I know something about electronics, so I'd be interested to have a read.
Chris.
Yes, do tell.
I rather fancy that if I supplied lamps that drove the LEDs at 5 amps the term "Headtorch" would take on a whole new meaning
royfellows
Well-known member
Chris, how you getting on with the Glowworm 2?
have you decided to go quads or whatever?
have you decided to go quads or whatever?
royfellows
Well-known member
Project update
I have been busy with a few things and had to put it on one side. The original build got accidentally trashed, so had to do a second. This is not as efficient for some reason as the original, running between 87% and 93%, dependant on input voltage. On future builds I will be using a slightly heavier ground plane and attempting to keep components as close together as possible.
I am also not getting the outputs exactly as I wanted on the 4 brightness modes, so will have to do my sums again. I rather expect its to do with the circuit as the values are very finite for the 4 values of Rsense. I wanted outputs of 120mA, 220mA, 1 amp, and 3 amps.
Other results are quite encouraging.
Test LED configuration, 3 series Cree XP-L.
I have tested to two lower, and most important, outputs and get full output, steady as a rock, from 12V input down to 9V, after which it goes into a slow death. At the 8.1V switch off its down to 70mA which is exactly as I wanted. But I will still be installing the flashing red low voltage warning.
Its worth mentioning, as this has been discussed, that with the current state of Li Ion chemistry, I am using cells of NCR and IMR chemistries, both have manufacturer stated max discharge levels of 2.5V, with cut off at 2.7, as already discussed. So this extra margin of safe discharge is a major contributory factor to the slow death cycle really desirable in a lamp at the end of the discharge cycle. Of course, if one cell is down relative to the others, switch off would come in at higher voltage, but this should not happen with high quality healthy cells. My battery management uses the Ablic S 8254 dedicated chip, output 1 is FET gate connection pin for charge control, which is open drain, thereby expanding possibilities a bit.
Crunching the numbers, this design of mine obliterates every digitally controlled buck driver I have previously tested, both on consistency of output and efficiency.
I have been busy with a few things and had to put it on one side. The original build got accidentally trashed, so had to do a second. This is not as efficient for some reason as the original, running between 87% and 93%, dependant on input voltage. On future builds I will be using a slightly heavier ground plane and attempting to keep components as close together as possible.
I am also not getting the outputs exactly as I wanted on the 4 brightness modes, so will have to do my sums again. I rather expect its to do with the circuit as the values are very finite for the 4 values of Rsense. I wanted outputs of 120mA, 220mA, 1 amp, and 3 amps.
Other results are quite encouraging.
Test LED configuration, 3 series Cree XP-L.
I have tested to two lower, and most important, outputs and get full output, steady as a rock, from 12V input down to 9V, after which it goes into a slow death. At the 8.1V switch off its down to 70mA which is exactly as I wanted. But I will still be installing the flashing red low voltage warning.
Its worth mentioning, as this has been discussed, that with the current state of Li Ion chemistry, I am using cells of NCR and IMR chemistries, both have manufacturer stated max discharge levels of 2.5V, with cut off at 2.7, as already discussed. So this extra margin of safe discharge is a major contributory factor to the slow death cycle really desirable in a lamp at the end of the discharge cycle. Of course, if one cell is down relative to the others, switch off would come in at higher voltage, but this should not happen with high quality healthy cells. My battery management uses the Ablic S 8254 dedicated chip, output 1 is FET gate connection pin for charge control, which is open drain, thereby expanding possibilities a bit.
Crunching the numbers, this design of mine obliterates every digitally controlled buck driver I have previously tested, both on consistency of output and efficiency.
royfellows
Well-known member
I suppose this aught to be in Friday Joke, but more relevant to the thread.
So, a maths question
If a lamp has 3 serial connected Cree XP-Ls are seriously over driven at 5 amps, from a seriously overrated DC-DC converter pushing out this current from a 12V input, assuming a lamp surface area of 11.5 cm and composed of an aluminium alloy having a thermal conductivity of 220 watts per Kelvin Metre
How long would it take to fry an average egg before the whole lot went up in flames?
So, a maths question
If a lamp has 3 serial connected Cree XP-Ls are seriously over driven at 5 amps, from a seriously overrated DC-DC converter pushing out this current from a 12V input, assuming a lamp surface area of 11.5 cm and composed of an aluminium alloy having a thermal conductivity of 220 watts per Kelvin Metre
How long would it take to fry an average egg before the whole lot went up in flames?
royfellows
Well-known member
ChrisJC said:That's a joke question - areas are measured in units of area, not distance...
Chris.
My error, should be "11.5 square cm", otherwise as you say, makes no sense. Be a rather funny lamp 11.5 cm long? Mind you, thinking about it, I may know of one with a battery that size.
Cantclimbtom
Well-known member
Well at 220W per Km^2 and an initial temperature of 12C, given that MBF (mean time before failure) is inversely proportional to the inconvenience or negative consequences, I'd say the lamp will melt within roughly 3 minutes (+20seconds -180 seconds) of the wearer entering shit creek.royfellows said:I suppose this aught to be in Friday Joke, but more relevant to the thread.
So, a maths question
If a lamp has 3 serial connected Cree XP-Ls are seriously over driven at 5 amps, from a seriously overrated DC-DC converter pushing out this current from a 12V input, assuming a lamp surface area of 11.5 cm and composed of an aluminium alloy having a thermal conductivity of 220 watts per Kelvin Metre
How long would it take to fry an average egg before the whole lot went up in flames?
royfellows
Well-known member
That is totally brilliant Sir
:bow:
But have you taken into account the decreasing Vf of the LEDs as the temperature increases?
:bow:
But have you taken into account the decreasing Vf of the LEDs as the temperature increases?
Cantclimbtom
Well-known member
No, I'd perhaps wrongly assumed that effect would be smoothed out by the flux capacitor?royfellows said:
royfellows
Well-known member
The project is totally screwed now until Feb next year!
I had overlooked source - drain resistance on some MOSFETs and this prevents me getting full power from my module. I can only find one that fits the bill, and that is back order.
Its the way it goes.
Anyway, in the meantime, some specs
Input voltage range is 3V -18V.
Max output is 3 amps, or will be, ref above.
Switching rate is 850 KHz with very sharp edge rise and fall, 10 ns or less.
Brightness output modes = 4
(Although analogue, its binary logic, 2, 4, 8, etc, cant do 3, 5, 6, or 7 and dont want 8, its ridiculous)
All outputs at whatever mode are consistent and dont fall with decreasing input voltage until it hits the LEDs forward voltage, then goes into a slow death cycle down to battery disconnect. However, on the upper 3 this is academic, as there is a power down feature on low battery to number 1 mode. The slow death cycle is quite short on paper so the power down will activate a flashing red LED.
Suitability, single, parallel, 2, 3, or 4 series Li Ions driving any configuration of LED, single, 2, 3, or 4 in series, or 12V quads in parallel subject to a voltage overhead above that required by the LEDs
Hopefully, it will go into a lamp sometime, I will then report on the lamp. But at the moment its full stop. I might be able to find some bigger MOSFETS that fit the bill but unlikely they will fit the project. Its been like this from the start. I would hit what would appear as an insurmountable obstacle, put the job on one side, then a solution would occur.
I had overlooked source - drain resistance on some MOSFETs and this prevents me getting full power from my module. I can only find one that fits the bill, and that is back order.
Its the way it goes.
Anyway, in the meantime, some specs
Input voltage range is 3V -18V.
Max output is 3 amps, or will be, ref above.
Switching rate is 850 KHz with very sharp edge rise and fall, 10 ns or less.
Brightness output modes = 4
(Although analogue, its binary logic, 2, 4, 8, etc, cant do 3, 5, 6, or 7 and dont want 8, its ridiculous)
All outputs at whatever mode are consistent and dont fall with decreasing input voltage until it hits the LEDs forward voltage, then goes into a slow death cycle down to battery disconnect. However, on the upper 3 this is academic, as there is a power down feature on low battery to number 1 mode. The slow death cycle is quite short on paper so the power down will activate a flashing red LED.
Suitability, single, parallel, 2, 3, or 4 series Li Ions driving any configuration of LED, single, 2, 3, or 4 in series, or 12V quads in parallel subject to a voltage overhead above that required by the LEDs
Hopefully, it will go into a lamp sometime, I will then report on the lamp. But at the moment its full stop. I might be able to find some bigger MOSFETS that fit the bill but unlikely they will fit the project. Its been like this from the start. I would hit what would appear as an insurmountable obstacle, put the job on one side, then a solution would occur.
andrewmcleod
Well-known member
royfellows said:How long would it take to fry an average egg before the whole lot went up in flames?
The solution is to join two separate problems in caving.
Problem 1) a sufficiently bright light generates a lot of heat which is difficult to remove
Problem 2) cavers often get cold
Solution: water-cooled caver jacket
royfellows
Well-known member
Cracked it.
Doubled up on some MOSFETS by mounting a second set on top of the first, some delicate soldering!
Anway, here are some specs for the lamp
Battery is the existing 12V. its 3500 mAh at 10.8V nominal, 12.6V fresh off charge. Weight 250 grams, compare with Scurion 4 cell at 310.
LEDs 3 series Cree XP-L2
So, Mode 1 is 140 mA -output 230 lumens, est run time about 28 hours to slow death.
All other modes are run time to power down, which is to this mode. So this mode is important, its the baseline, and if something else in the lamp fails, it will close down to this one. Power down will show flashing red LED in the background, feint at first but becoming more obvious as the slow death cycle sets in and output falls.
Mode 2 - 250 mA, about the same as the previous "standard" or "walk mode", about 400 lumens about 17 hrs to power down.
Then there is modes 3 and 4, 800 mA and as just tested, 2850 mA - near enough.
There will be some other changes.
I am dumping the XT60 on this one and the X16 in favour of IP68 DIN connectors. The cable will attach by an angled connector at the base with the battery mounted upright Scurion style. Adaptor cables will be available for 1st generation batteries.
The X16 is now the same, detachable battery so you can carry a spare, battery mounted upright Scurion style, no boxes on the side or other clutter, just a slim battery of weight 275 grams - big change!
Charging control, cell balancing etc is now in a charge adaptor.
Still describing though as "Sealed rechargeable", which when the DIN plug lock ring is screwed up, it is indeed.
Doubled up on some MOSFETS by mounting a second set on top of the first, some delicate soldering!
Anway, here are some specs for the lamp
Battery is the existing 12V. its 3500 mAh at 10.8V nominal, 12.6V fresh off charge. Weight 250 grams, compare with Scurion 4 cell at 310.
LEDs 3 series Cree XP-L2
So, Mode 1 is 140 mA -output 230 lumens, est run time about 28 hours to slow death.
All other modes are run time to power down, which is to this mode. So this mode is important, its the baseline, and if something else in the lamp fails, it will close down to this one. Power down will show flashing red LED in the background, feint at first but becoming more obvious as the slow death cycle sets in and output falls.
Mode 2 - 250 mA, about the same as the previous "standard" or "walk mode", about 400 lumens about 17 hrs to power down.
Then there is modes 3 and 4, 800 mA and as just tested, 2850 mA - near enough.
There will be some other changes.
I am dumping the XT60 on this one and the X16 in favour of IP68 DIN connectors. The cable will attach by an angled connector at the base with the battery mounted upright Scurion style. Adaptor cables will be available for 1st generation batteries.
The X16 is now the same, detachable battery so you can carry a spare, battery mounted upright Scurion style, no boxes on the side or other clutter, just a slim battery of weight 275 grams - big change!
Charging control, cell balancing etc is now in a charge adaptor.
Still describing though as "Sealed rechargeable", which when the DIN plug lock ring is screwed up, it is indeed.
royfellows
Well-known member
Latest news
Got it into a lamp and tried it out at Glyn Ceiriog. Very pleasant to use - until the controller played up and I lost the two power modes.
For the teckies, it appears that I had mistakenly assumed my controller was open drain, when in fact it isn't. Fix was to channel the outputs through a couple of MOSFETs.
For what its worth, my controller can also be used to divert output, State = off, State = beams, state = floods, state = both.
After a bit of deliberation I had built the lamp with a toggle one side, as existing models, and mode selection by a push button on the other side. If it goes into production I will fit 2 toggles, the other double throw would be momentary and do modes plus blue battery check. Didn't think of that at the time. Mode selection needs to be separate because the main on and off gives beams + 1 flood on down stroke, 3 floods on upstroke and I didn?t want the modes changing to confuse a user.
Also, before I stuck it in the lamp I did a duration test on the basic 230 lumens output. Battery was my 12V series, 3S Sanyo NCR 18650GA, chemistry is Nickel Cobalt Rechargeable, capacity 3500 mAH at nominal 10.8V. Battery well used, about 18 months old, shutdown level 8.1V . This is 2.7V per cell ~ 0.05V, manufacturers stated minimum is 2.5V. Individual cell monitoring.
My estimation proved remarkably accurate. It went 27 hours full output until the 9.5V level from which higher modes would power down, and then another hour and half before starting to die down. It went for another 20 minutes to about half output and then turned off at battery level 8.086V
I am reasonably pleased with this, but would have preferred and longer 'slow death' with the output diminishing further, but I can?t have everything.
I also think that it?s showing great promise. One can go underground on a longish trip, come out with battery nearly half spent, then next day underground again without a recharge and with no loss of output. But we are nowhere near there yet.
Got it into a lamp and tried it out at Glyn Ceiriog. Very pleasant to use - until the controller played up and I lost the two power modes.
For the teckies, it appears that I had mistakenly assumed my controller was open drain, when in fact it isn't. Fix was to channel the outputs through a couple of MOSFETs.
For what its worth, my controller can also be used to divert output, State = off, State = beams, state = floods, state = both.
After a bit of deliberation I had built the lamp with a toggle one side, as existing models, and mode selection by a push button on the other side. If it goes into production I will fit 2 toggles, the other double throw would be momentary and do modes plus blue battery check. Didn't think of that at the time. Mode selection needs to be separate because the main on and off gives beams + 1 flood on down stroke, 3 floods on upstroke and I didn?t want the modes changing to confuse a user.
Also, before I stuck it in the lamp I did a duration test on the basic 230 lumens output. Battery was my 12V series, 3S Sanyo NCR 18650GA, chemistry is Nickel Cobalt Rechargeable, capacity 3500 mAH at nominal 10.8V. Battery well used, about 18 months old, shutdown level 8.1V . This is 2.7V per cell ~ 0.05V, manufacturers stated minimum is 2.5V. Individual cell monitoring.
My estimation proved remarkably accurate. It went 27 hours full output until the 9.5V level from which higher modes would power down, and then another hour and half before starting to die down. It went for another 20 minutes to about half output and then turned off at battery level 8.086V
I am reasonably pleased with this, but would have preferred and longer 'slow death' with the output diminishing further, but I can?t have everything.
I also think that it?s showing great promise. One can go underground on a longish trip, come out with battery nearly half spent, then next day underground again without a recharge and with no loss of output. But we are nowhere near there yet.
Cantclimbtom
Well-known member
27 hours at 230lm walk is easily 2 long trips underground without a recharge (assuming you don't stop taking lots of photos on some bat-cooker setting).
Why did you say nowhere near yet, sounds like you have victory in sight?
Why did you say nowhere near yet, sounds like you have victory in sight?
royfellows
Well-known member
Bit of bad news, current project is dead in the water.
Main criteria of any caplamp is reliability and over extended testing I have had component failures which indicate that the converter chip is unsuitable for my design.
However, the controller is fine and quite robust, this being my original concept idea. Also the design of the new lamp which combines 2 or the previous models into one single unit is very much alive. The power down module is also fine.
When I spent many hours surfing data sheets etc, I narrowed it down to 3 possibly suitable chips. One was tried and discounted in a short time frame, the second is what has been described, but there is a third. Unfortunately on back order at all suppliers. I wont get the 98% efficiency, probably low to upper 80s, but there we are, if of course it works OK. Might be third time lucky.
Its very much a case of "Watch this space"
Main criteria of any caplamp is reliability and over extended testing I have had component failures which indicate that the converter chip is unsuitable for my design.
However, the controller is fine and quite robust, this being my original concept idea. Also the design of the new lamp which combines 2 or the previous models into one single unit is very much alive. The power down module is also fine.
When I spent many hours surfing data sheets etc, I narrowed it down to 3 possibly suitable chips. One was tried and discounted in a short time frame, the second is what has been described, but there is a third. Unfortunately on back order at all suppliers. I wont get the 98% efficiency, probably low to upper 80s, but there we are, if of course it works OK. Might be third time lucky.
Its very much a case of "Watch this space"
Tangent_tracker
Active member
I do seem to have hit a nerve with you Roy, certainly NOT my intention. Regards low battery voltage, why do you think nearly ALL reasonable kit shuts the circuits down at about 3v? No need to drain to 2.7v, and doing so can further complicate charging if the charging circuit does not attempt to kickstart the protection board into waking up again. Surprisingly common! I have also seen some boards not kick in until below the 2.7v level, which again, is not ideal!Yes, do tell.
I rather fancy that if I supplied lamps that drove the LEDs at 5 amps the term "Headtorch" would take on a whole new meaning
Also, there really is no capacity between 2.7 and 3v. I suppose you could argue you do not want to stop at any voltage with caving lamps, but then again I suppose that is what backups are for!
I forgot to add that the low voltage cutoff could be micro-controller considerations but most these days have quite a wide operating range with low voltage modes down to 1.8v and below.
Not sure what you are referring to with regard to 5A. Did I state something to advocate such a current?
