AMC7135 with single 18650 battery

[Vikas Sontakke]

Is it true that even though AMC7135 has operating voltage of 2.7 ~ 6.0 V, if the LED forward voltage + dropout voltage is less than the battery, it will drop out of regulation? To state this differently, it is strictly a linear voltage dropout regulator and essentially "wastes" excess battery voltage as heat and is incapable of boosting the supply voltage.

If I have an LED with nominal forward voltage of 3.5V, am I not better of direct driving it on single LiIon battery rather than using the AMC7135 regulator? There is a chance of over driving the LED when the battery is fresh off the charger but as far as brightness is concerned, DD will always be brighter (assuming you avoid getting the LED in angry blue state).

Why are these kind of regulators so popular? Am I missing something?

- Vikas

 

[jeffosborne]

Hi Vikas!
Your assumption is correct, but your conclusion is not.

If you direct-drive, you still need a resistor to keep from frying the LED. That resistor is NOT a better solution than the AMC7135 IC.
This IC is popular because it is inexpensive, easy to use, and delivers 350ma regulated. Two in parallel are common to deliver 700ma to an LED.

When the battery voltage falls below the LED forward voltage, regulation does stop (not needed) and the LED remains brighter than it would have been if a resistor was used.

Cheers,
Jeff O.

 

[Justin Case]

Is it true that even though AMC7135 has operating voltage of 2.7 ~ 6.0 V, if the LED forward voltage + dropout voltage is less than the battery, it will drop out of regulation? To state this differently, it is strictly a linear voltage dropout regulator and essentially "wastes" excess battery voltage as heat and is incapable of boosting the supply voltage.

If I have an LED with nominal forward voltage of 3.5V, am I not better of direct driving it on single LiIon battery rather than using the AMC7135 regulator? There is a chance of over driving the LED when the battery is fresh off the charger but as far as brightness is concerned, DD will always be brighter (assuming you avoid getting the LED in angry blue state).

Why are these kind of regulators so popular? Am I missing something?

- Vikas

You have it backwards. To run in full regulation, Vbatt > Vf + AMC7135 dropout voltage (which is only about 0.12V). Thus, if you have a relatively low Vf LED of say 3.3V or 3.4V, then 1xLi-ion can drive the LED in regulation for essentially the full capacity of the Li-ion.

For DD, the output depends on the discharge curve characteristics of the power source you are using. As your batteries run down and voltage sags, output will decrease. Not so, if you use the AMC7135 based driver.

Sure, DD efficiency is 100%. But the AMC7135 efficiency is quite good also. Efficiency is essentially Vf/Vbatt. If we assume Vf of 3.3V, then efficiency is probably no worse than about 80% initially, quickly improving to perhaps about 90% as the Li-ion sags to its quasi steady state voltage, and finishes at around 95%.

Also, the AMC7135 driver allows you to avoid that concern of overdriving the LED. If you have to go with a dropping resistor, then it seems you may as well use an AMC7135 board instead.

 

[Greg G]

How rugged are the 7135 boards? I have ruined several drivers by dropping the light onto the concrete floor at work.

I have a couple of 18650 Cree XR-E lights in DD. The emitters were chosen to not draw too much on a fresh battery, and they are well heatsinked. Those are the two lights I have the most confidence in.

 

[Fichtenelch]

Is there a possibility to get 2 or 3-modes with a 7135-based board?
I need some drivers, from 0.5 to max. 2.8A...for a p7, xr-e r2, xpe- r5 and golden dragon..and an mc-e

 

[bstrickler]

Is there a possibility to get 2 or 3-modes with a 7135-based board?
I need some drivers, from 0.5 to max. 2.8A...for a p7, xr-e r2, xpe- r5 and golden dragon..and an mc-e

ShiningBeam.com carries a 3-mode P7/MC-E driver that runs at 140ma, 1,000 ma, and 2500 ma. It appears that Bryan is out of them at the moment, but you can e-mail him, and ask when he'll have them back in stock (he sells out of them fairly quickly).

~Brian

 

[chiption]

Why not try OCP2152?It's max current is 1.2A;Input 2.5V to 5.5V;output can be set at 3.7 or 3.8V,It's the best Voltage for CREE Q5.
Even can use LiFEPO4 battery and LiCoO2 battery.

 

[Justin Case]

If Vf is around 3.8V at 1000mA drive current, then an LDO regulator like the AMC7135 is the wrong way to go anyway since the driver will fall out of full regulation at Vbatt ~ 3.8V + dropout voltage (about 3.9V for the AMC7135). Factoring in voltage sag under load, and a 1xLi-ion source probably won't run in full regulation for very long.

If Vf is lower, say 3.3V at 1000mA, then an AMC7135-based driver is a very good choice for a 1xLi-ion battery configuration. Vbatt has to stay above only about 3.4V to run in full regulation, which should be easy for just about any brand of Li-ion, 16340 or larger.

The problem with a traditional buck IC converter based driver board is that they typically have a relatively high voltage headroom or overhead. So the buck driver board won't work very well either if you have a relatively high Vf LED. Even the good buck drivers are around 0.5V headroom. Many are at 1V or more. For the higher Vf LED like your Cree XR-E Q5 example at 3.8V, 1xLi-ion input voltage is already too low. For a low Vf ~3.3V, you would need to maintain Vbatt at least above 3.8V (and possibly higher) to run in full regulation, which potentially wastes a fair amount of battery capacity.

LiFePO4 at 3.2V in won't work well at all driving an LED with Vf~3.8V using a buck driver. For a low Vf LED around 3.3V or so, you may as well direct drive the LED with LiFePO4.

If you are running 2xLi-ion or 2x123A or more, then the voltage headroom is no longer an issue. It's the single Li-ion case that is the demanding one.

 

[Vikas Sontakke]

What is the typical Vf for P7 or MC-E at about 1.5-2.0A current? Is it more towards 3.8V or 3.3 ? I am asking about C-bin or K-bin, respectively.

Thanks,
- Vikas

P.S. I am trying to find if 2-mode driver-less SSC based light makes more sense than the 5-mode one.

 

[Justin Case]

The Vf bin figures for the P7 are for full power (700mA per core, 2800mA total). Since the I Vf bin is 3.25V to 3.50V, while the J Vf bin is 3.50V to 3.75V, the estimated Vf you might get at 1500mA or 2000mA could be something like 3.1V to 3.5V. Depends on the bin and where in the bin (i.e., toward the low end, in the middle, or toward the high end). I've measured 3.0V at 1000mA (250mA per core).

 

[Benson]

How rugged are the 7135 boards? I have ruined several drivers by dropping the light onto the concrete floor at work.

If this concerns you, pot them completely in epoxy.

But IMHO they're pretty rugged, aside from the potential for melting the solder and the chips sliding off in extreme overheating (e.g. if you run them from ~6V or so with poor heat-sinking). There's no coil hanging off or anything.

 

[TorchBoy]

If you direct-drive, you still need a resistor to keep from frying the LED. That resistor is NOT a better solution than the AMC7135 IC.

"Better" depends on how you define good. In most respects AMC7135 solution is better than a resistor, but a resistor is better in price, versatility, (marginally) for simplicity. :thinking: That's probably about it.

Sure, DD efficiency is 100%. ... Efficiency is essentially Vf/Vbatt.

And with a resistor? If that's also DD I wouldn't call it 100% efficient.

Why not try OCP2152?It's max current is 1.2A;Input 2.5V to 5.5V;output can be set at 3.7 or 3.8V,It's the best Voltage for CREE Q5.
Even can use LiFEPO4 battery and LiCoO2 battery.

Are you saying it's a constant voltage board? I don't have Cree any XR-E Q5s with a Vf at 1 A as much as 3.7 V.

 

[Justin Case]

Quote:
Originally Posted by Justin Case
Sure, DD efficiency is 100%. ... Efficiency is essentially Vf/Vbatt.

And with a resistor? If that's also DD I wouldn't call it 100% efficient.

Ok, if you want to quibble even DD is not exactly 100% efficient since there are undoubtedly real world contact resistances, hookup wire resistance, internal battery resistance, and other parasitic losses. IMO, adding a dropping resistor is just another source of resistive loss.

In most respects AMC7135 solution is better than a resistor, but a resistor is better in price, versatility, (marginally) for simplicity.

Regarding the claim that a resistor is better than an AMC7135-based driver for versatility, I think that's very debatable. The resistor approach identifies a specific resistor value for a specific LED Vf/If and battery Vbatt. The same model (e.g., SSC P7, I Vf bin) of LED can have a very different Vf (after all, the Vf bin is 0.25V wide, giving a range of 3.25V to 3.50V), giving you a different resistor value. And Vf actually changes (decreases) as the LED heats up. The spec for an MC-E (I can't find the spec for a P7 off-hand), for example, is -4mV/C. Since you use a resistor to drop the input voltage to something closer to Vf, the LED's NTC behavior will take it farther from the input voltage. At 50C, for instance, Vf can decrease by .004V/C * 50C = 0.2V. If you change from 3xNiMH to 4xNiMH, you will also get a different value for the resistor you need. If you change the LED from a P7 to say a P4 and happen to get the same Vf, you still will calculate a different resistor value because the current is different. With an AMC7135 driver, the main limit is the operating voltage range (6V max, preferably as close to Vf as possible, while leaving enough headroom for long running in regulation). Within that limitation, all you have to to is hook up the driver and go. It doesn't matter if you run with a Seoul P7, Seoul P4, Cree XR-E, 3xNiMH, 4xNiMH, or 1xLi-ion. For the resistor approach, you can always round the resistor value to some approximate figure like 1 ohm and call it good. That might make the resistor approach equal in versatility to the AMC approach. Now, you might argue that the resistor method is more versatile because you can apply the method to any DD setup where you need to match Vf and Vbatt more closely. Sure. But that seems like a dodge to me. I could make a general claim that using a driver is a very versatile method because you can select the right driver for the configuration of interest. But that driver might not be very useful for any other configuration. Is that really versatile?

Quote:
Originally Posted by chiption
Why not try OCP2152?It's max current is 1.2A;Input 2.5V to 5.5V;output can be set at 3.7 or 3.8V,It's the best Voltage for CREE Q5.
Even can use LiFEPO4 battery and LiCoO2 battery.

Are you saying it's a constant voltage board?

The OCP2152 is an IC, not a board. The linked datasheet shows that you select some resistor values to set the output voltage when you implement the IC in a driver board application circuit.

 

[TorchBoy]

Ok, if you want to quibble even DD is not exactly 100% efficient since there are undoubtedly real world contact resistances, hookup wire resistance, internal battery resistance, and other parasitic losses. IMO, adding a dropping resistor is just another source of resistive loss.

Yes, and sooner or later that resistive loss is large enough that it can easily be quantified, which immediately gives less than 100% efficiency. Unless you're making heat, not light.

Regarding the claim that a resistor is better than an AMC7135-based driver for versatility, I think that's very debatable.

I'd say it depends at what you're looking at. In terms of output current the AMC7135 doesn't give you a choice for lower currents unless you use PWM (and sacrifice even more simplicity) and capacitor-smooth the output.

Don't get me wrong, I think the AMC7135 is great.

The OCP2152 is an IC, not a board. The linked datasheet ...

I must have missed the link.

 

[Justin Case]

Sorry about that, here's the OCP2152 link.

Mag internal resistances are non-trivial already. CPF is filled with low resistance mods to the switch and tail spring, most notably.

Even if you hard wire the Mag switch in the typical way, you still have two internal switch contacts. Most of the Mag mod guides don't address bypassing these two contacts (I didn't bypass them in my Mag mods either).

The tail spring is relatively high resistance, although I forget what it is off-hand. The classic resistance mod is to strip the anodizing inside of the tailcap and solder in some copper solder wick to connect spring top and bottom.

Then there are the contacts between the bottom battery and the tail spring, the contacts between each battery, the contact between the top battery and the Mag switch, and the contact between the Mag switch grub screw and the Mag body tube.

Even if this is only 100 milli ohms total, which I think is a low guess (just picking a nice round number), at roughly 3A current to a P7, you are looking at 0.3V drop.

Re: AMC drivers running at lower output currents. There are plenty of multimode AMC7135-based drivers that use PWM. What would you do in a resistored approach to get multimodes? Dump the Mag switch, find some new switch, and use it to switch between various resistor values? You've just lost what little advantage in price and simplicity that you had using resistors vs AMCs. Same with using the more elegant d2Flex solution, plus you need to mod the Mag switch for momentary operation.

 

[TorchBoy]

You're now talking about simplicity in one part of your system over complexity in another. I've done a few switch swaps, although not in Mags - I'm just not into Mag mods so have no idea what would be simplest overall. The overall simplicity varies for each situation. In my particular cases replacing the switch was easier than getting multiple modes with the momentary switch already there. For a couple of them, since I now had three on positions from the switch to play with, running the two lowest modes with resistors turned out to be easiest, with an AMC7135 board for high and turbo (making use of the original momentary switch again to switch on the extra chips on the board).

FWIW in one of those mods I did find horrible high resistance. It was a while ago but I think I managed to chew out over half an ohm by soldering several poor contact joints (eg, springs to conductors). It made a difference for the turbo mode because of the voltage that was being lost dropped the AMC7135s' output under the Vf of the LED at 1 amp.

 

[TorchBoy]

Sorry about that, here's the OCP2152 link.

Thanks. I see it's a step-down converter chip, apparently with a 0.5 V overhead. So it can be used with a LiFePO4 battery only if the LED's Vf is ~2.5 V.