# Thread: SOB 1000/MC-E 2S2P question

1. ## SOB 1000/MC-E 2S2P question

So I built an AW/Arcmania tower kit using a K-bin MC-E in 2S2P and an SOB 1000. Tested the light using 4xDuracell123A, 2xAW16340, 2xAW17670, and 2xIMR16340.

Ran great. No problems with heat even after a 17 min continuous run.

Tailcap draw for the 123A primaries was only ~0.6a, which seems surprisingly low. Tailcap draw for all of the 2xLi-ions was ~0.8A, again surprisingly low.

I would have expected 1000ma drive current, ~500ma per MC-E die, Vdrop~6.6V, and thus ~6.6W of power sent to the LED. If I assume 80% efficiency for the SOB as a WAG, then I get:

4x123A: 0.8 * 10V * Iin = 6.6W
2xLi-ion: 0.8 * 7.6V * Iin = 6.6W

Thus, I get calculated input currents of

Iin(4x123A) ~ 0.83A -> 800ma
Iin(2xLi-ion) ~ 1.09A -> 1000ma

But the measured tailcap draws fall short of these calculated values that would send ~6.6W of power to the MC-E. Instead, for the measured tailcap draws and the same assumed efficiency and battery voltages, I would estimate about 5W power sent to the MC-E and thus about 390ma drive current per MC-E die, not 500ma drive current.

What is the reason for this behavior? I checked the sense resistors to confirm that I really do have an SOB 1000. I even built two towers and both measure the same for tailcap draw and lux readings.

2. ## Re: SOB 1000/MC-E 2S2P question

The Iin should be around 700mA for 4 x primaries,depending on Vf.

2 x 3.1V x 1000mA / 85% = 7.3W

7.3W / 10.4V = 701mA Iin

But I don't think you have enough headroom for the 2 x Li-ion setup.

If the efficiency of the SOB is 85%, you would need at least 7.13V to remain in regulation.

6.2Vout x 15% = .93V

6.2V + .93V = 7.13V

-Michael

3. ## Re: SOB 1000/MC-E 2S2P question

Justin, which SF tailcap are you using? If it's a clicky, I'm guessing the drop in current is due to internal resistance. Try a Z41 and measure the current.

4. ## Re: SOB 1000/MC-E 2S2P question

Originally Posted by TexLite
The Iin should be around 700mA for 4 x primaries,depending on Vf.

2 x 3.1V x 1000mA / 85% = 7.3W

7.3W / 10.4V = 701mA Iin

But I don't think you have enough headroom for the 2 x Li-ion setup.

If the efficiency of the SOB is 85%, you would need at least 7.13V to remain in regulation.

6.2Vout x 15% = .93V

6.2V + .93V = 7.13V

-Michael
700ma, 800ma -- it all depends on the assumptions. You've assumed Vf ~ 3.1V. I assumed 3.3V at the expected drive current of 500ma per die. You assumed 85% efficiency. I assumed 80%. You assumed 10.4V for 4x123A. I assumed 10V.

Regardless, the calculated tailcap current is still higher than what I measured.

Not sure why you think I don't have enough voltage with 2xLi-ions. At a 1A draw, a 17670 should be able start at 4.0V and hold above 3.8V per cell for at least on the order of minutes. Granted, Wayne in another thread said that you need 15% voltage headroom above Vdrop. If Vdrop is 6.6V, then you need Vin~7.6V, and 2xLi-ions may somewhat borderline.

Regardless, 4xLi primaries certainly provides enough voltage headroom at ~10V input.

The question remains -- what is occurring here such that the SOB doesn't seem to be running at full output?

5. ## Re: SOB 1000/MC-E 2S2P question

maxspeeds,

Good question. Host was a SureFire 12ZM for the 2x17670 and 4xLi primaries cases and a SureFire 6P body for the 2x16340 (AW and IMR) case. Bezel was always a 12ZM TurboHead (T-62). Tailcap was a standard SureFire momentary-on tailcap (either the original 12ZM tailcap that is not a LOTC or a newer Z41 LOTC, though not the current generation Z41 -- all of my SF stuff is old).

I would think that it would have to be a heck of a parasitic resistance to negate a headroom of ~3.5V for the 4x123A case.

6. ## Re: SOB 1000/MC-E 2S2P question

Indirect measurement of efficiency or LED current is not the way to go. Is it 80% or is it 85% is all ?????. Who knows.

It's possible that your DMM sense resistor could be causing the regulator to fall out of regulation. A lot of DMM use a fairly high value like 1 ohm for the high current range.

It's possible that without the DMM in the tailcap it could be delivering 1A to the LED as expected. Since it is hard to derive from a tailcap measurement I would suspect the setup to be part of the measurement error.

It's always best to measure the current to the LED. Since you say you have two that measure the same I would suspect that both are doing exactly what they should be doing. I'd have a hard time imagining that two SOB1000 are both identically wrong.

My recommendation is to always use a bench setup and test the converter with the LED on the bench where the current to the LED is easier to measure or get to. After bench testing you can button it up and know for certain what it is doing.

I have a drawer full of different LEDs mounted to a heatsinks for this specific purpose. It allows me to hook them up to a converter on the bench and then measure with extreme accuracy what the current to the LED is.

Wayne

7. ## Re: SOB 1000/MC-E 2S2P question

One of the best ways to measure DC current is with a clamp on current meter. The vast majority of clamp on ampmeters are for AC current only. A small number of them will actual report DC current.

The make and model fo the one I use everyday is the Extech 380942.

http://www.extech.com/instrument/pro...99/380942.html

I know this one is expensive, but, it eliminates the possibility of adding the internal series resistance of the DMM.

Wayne

8. ## Re: SOB 1000/MC-E 2S2P question

Originally Posted by dat2zip
Indirect measurement of efficiency or LED current is not the way to go. Is it 80% or is it 85% is all ?????. Who knows.

It's possible that your DMM sense resistor could be causing the regulator to fall out of regulation. A lot of DMM use a fairly high value like 1 ohm for the high current range.

It's possible that without the DMM in the tailcap it could be delivering 1A to the LED as expected. Since it is hard to derive from a tailcap measurement I would suspect the setup to be part of the measurement error.

It's always best to measure the current to the LED. Since you say you have two that measure the same I would suspect that both are doing exactly what they should be doing. I'd have a hard time imagining that two SOB1000 are both identically wrong.

My recommendation is to always use a bench setup and test the converter with the LED on the bench where the current to the LED is easier to measure or get to. After bench testing you can button it up and know for certain what it is doing.

I have a drawer full of different LEDs mounted to a heatsinks for this specific purpose. It allows me to hook them up to a converter on the bench and then measure with extreme accuracy what the current to the LED is.

Wayne
Thank you again for your help.

If the SOB fell out of regulation at the start, wouldn't it run DD, and thus I would expect even higher tailcap current draw? I tested with 4xDuracell123A, and 2x(AW16340, IMR16340, and AW17670).

For clarification, I wasn't trying to back estimate SOB efficiency. I was just trying to get a seat of the pants estimate on what might be going on with the LED and SOB.

The specs for my meter claim 0.1V drop on the 10A DC scale. I suppose that could make a difference for the 2xLi-ion case, especially combined with other system resistances like the springs, the SureFire body contacts, etc. But the 4x123A case should be at at lest 10V input.

Note that in my tailcap current measurements, I remove the SF tailcap and complete the circuit using the DMM, touching one probe to the bottom battery contact and the other probe to the unanodized contact of the SF flashlight body at the tailcap threads. Thus, the tailcap is not in the circuit during my DMM measurements.

The DMM accuracy specs for DC current are fairly poor at +/-(2% + 10 digits), which can add another few hundredths of an amp error.

I believe that my estimates of battery voltage in are reasonable (~10V for the 4x123A and ~7.6V for the 2xLi-ion). At the measured tailcap currents, I get about 6W power from the batteries, ignoring any SOB efficiency issues. Unless the Vf of the MC-E ie exceptionally low, for 1000ma out of the SOB and 500ma to each MC-E die, I estimate Vf to be around 3.3V giving a total voltage drop of 6.6V from the LED. That gives 6.6W of power to the LED. Hard to do that if you deliver only 6W to the SOB unless efficiency is greater than 100%.

I specifically tested 4x123A to try to avoid "running in the margins" in terms of voltage headroom to run in regulation. I would think that 10V nominal for the input voltage should be more than enough. But I measured only 0.6A at the tailcap.

I suppose we can do a crude sensitivity analysis. Assume that Vbatt is closer to 10.4V-11V, not the 10V I've assumed for 4x123A. Then for 0.6A at the tailcap, power to the SOB could be 6.24W-6.6W. Suppose we factor in DMM measurement error (I'm going to ignore the 0.1V DMM voltage drop for the 4x123A case). That could add about +/-0.02A for the measurement error. That gives about 6.45W-6.8W power (assuming the error is +0.02A, not -0.02A).

Now if the MC-E Vf is very low at 500ma per die, say 3.1V instead of the 3.3V I assumed, then I need to deliver 6.2W, not 6.6W to the emitter. Then the tailcap draw has to be at least 0.62A, and now we are in the regime of the DMMs accuracy specs. If the SOB is running at 90% efficiency, then I need 6.89W from the batteries, which is still close to the 6.8W figure in the previous paragraph.

Also, my initial comparative lux measurements for the MC-E tower (before I did any DMM current measurements) vs some of my other lights, plus my visual perception of the light output from this MC-E tower in a SureFire TurboHead, had made me believe I was getting full light output. The tower really lights up a room.

I did measure the current to an emitter that I thermal epoxied to a heat sink -- a Cree P4 taken from an old Chinese drop-in for a SureFire 6P (intent was to use this Cree to test for function, not to test for accurate drive current -- I may have to set up another LED for that). I got something like 0.54A, which I thought was so far off that I dismissed it as setup error due to my skinny hookup wires (maybe 6"-8" long for each) used to connect the 2x16340 to the SOB 1000. The hookup wires have alligator clips on their ends, and I use small magnets to stick those clips to the battery + and -. The 16340s are also connected via a small magnet.

I also have about 2"-3" of 26 gauge stranded wire soldered to the Cree, each terminating in a spring-loaded test clip. Unfortunately, no specs on the test clips. They are the type that use a little curved metal tip to hook a wire or component. The spring action then pulls the metal tip back into the test clip to hold the wire or component in place.

I measured the current to the emitter by placing the DMM in series between the LED+ lead of the SOB and the LED+ lead of my test Cree.

So I guess after all of this long-winded discussion on my part, I am tending to believe that all is well and that small differences in the various values used in my calculations could add up to give an erroneous conclusion. I guess what still gives me some slight pause is my Cree P4 bench measurement.

Do you think I am off-base anywhere in the above discussion?

I may try one more build using an SOB 1200 and see what I measure.

9. ## Re: SOB 1000/MC-E 2S2P question

Originally Posted by Justin Case
Thank you again for your help.

If the SOB fell out of regulation at the start, wouldn't it run DD, and thus I would expect even higher tailcap current draw? I tested with 4xDuracell123A, and 2x(AW16340, IMR16340, and AW17670).

You are thinking of a boost converter where if the battery voltage rises to where the boost condition no longer exists (Vin > Vout) then it goes into DD and LED current then rises.

In a buck condition the regulator falls out of regulation when Vin < Vout and under these conditions current is always less than the regulated current. In a buck regulator LED current is never more than the regulated amount unless the converter is broken, shorted or defective or something else is wrong.

As for measurements and error from tailcap current. It's all speculative since you don't know the Vf at 1A nor do know the temperature of the LED when making the measurement. As the LED heats up the Vf drops. Meausuring the LED current it should be stable and independant of the LED Vf if the converter is working and thus many times more accurate to put the meter on the LED side.

The converter exact efficiency number is not know precisely with these input output conditions.

The way to tell if the converter is working or not is to measure the battery current for different voltages. Lowering battery voltage the battery current should be higher. Higher input voltage the battery current should go down.

If these conditions are true with lower input current as battery voltage rises then the converter is most likely in full regulation and most probably delivering the full current as specified by the sense resistor value. Whether this jives with your measurements or not is hard to determine.

I know you have done all the math, but, the test with the Cree P4 may have been valid and still be invalid with the tailcap. Why? If the DMM internal resistor is a high value and you had plenty of headroom on the power supply to the LED the measurement would be fine.

Inserting a resistor of any type in series with a converter can cause it to be unstable. Most of my converters have a compromise in input capacitance due the converter board size limitation. This means I should be using a larger value capacitor, but, didn't have the real estate for the large value capacitor. Instead, I used the "largest" value I could fit on the board. This along with the output capacitor value may make the converter board unstable under certain conditions. One condition that has always been known is to keep the bench supply leads as short as possible and to try and minimize the input resistance from the power source.

While I don't think the SOB is unstable with a DMM in series with the tailcap there are just too many variables to really trust this type of measurement and to allow this type of measurement to have any qualtative meaning.

Unless you are measuring the LED current where a small series resistance will not affect the converter board any other method has associated risks.

Ideally, your first proto should have had long LED wires and the LED should have been mounted externally to the light so that LED current would be easy to measure. This is the prefered method to very your first setup is a full bench setup if you have the equipment like a good power supply or on the bench with some form of battery holder. Once you know the LED current is correct you could measure the battery current and from there extract the efficiency of the converter and also measure the Vf of the LED. Once these numbers have been extracted and the light put together you could do a final check and verify that the tailcap current is aroudn the same number as when you measured it on the bench.

Take it with a grain if salt. It may be absolutely correct that the current to the LED (based on the tailcap current measurement) is less than expected or there is something that is not being accounted for. Whether that be the converter efficiency number, LED Vf, or converter stability to name a few, who knows...

As it stands now given only the battery voltage and tailcap current it would be hard to speculate whether the converter is regulating or not and whether the current to the LED is 1A or not.

One other point. We have sold a lot of SOBs and the vast majority work out of the box. Usually if the SOB doesn't work it doesn't work. Now it could be true that there are a lot of defective SOBs that the customers for some unknown reason fail to report it back to the shop for replacement, but, I find that statement hard to believe.

Wayne

10. ## Re: SOB 1000/MC-E 2S2P question

Originally Posted by dat2zip
The way to tell if the converter is working or not is to measure the battery current for different voltages. Lowering battery voltage the battery current should be higher. Higher input voltage the battery current should go down.

If these conditions are true with lower input current as battery voltage rises then the converter is most likely in full regulation and most probably delivering the full current as specified by the sense resistor value. Whether this jives with your measurements or not is hard to determine.
Yes, and this is another reason why the results seemed confusing before you've provided the clarifying information that explains things.

For the 4x123A, I measured 0.6A at the tailcap. For 2xLi-ion, I measured 0.8A. For the assumed battery voltages from before of 10V and 7.6V respectively, both battery sources appeared to be sending basically the same power to the SOB -- %efficiency*6W. So it looked like the board was running in full regulation.

Originally Posted by dat2zip
I know you have done all the math, but, the test with the Cree P4 may have been valid and still be invalid with the tailcap. Why? If the DMM internal resistor is a high value and you had plenty of headroom on the power supply to the LED the measurement would be fine.
In fact, the test with the Cree was with two IMR16340 cells, so that would give Vin ~ 7.6V, while Vdrop ~ 3.5V, so the voltage headroom was quite large.

Originally Posted by dat2zip
Inserting a resistor of any type in series with a converter can cause it to be unstable. Most of my converters have a compromise in input capacitance due the converter board size limitation. This means I should be using a larger value capacitor, but, didn't have the real estate for the large value capacitor. Instead, I used the "largest" value I could fit on the board. This along with the output capacitor value may make the converter board unstable under certain conditions. One condition that has always been known is to keep the bench supply leads as short as possible and to try and minimize the input resistance from the power source.

While I don't think the SOB is unstable with a DMM in series with the tailcap there are just too many variables to really trust this type of measurement and to allow this type of measurement to have any qualtative meaning.

Unless you are measuring the LED current where a small series resistance will not affect the converter board any other method has associated risks.

Ideally, your first proto should have had long LED wires and the LED should have been mounted externally to the light so that LED current would be easy to measure. This is the prefered method to very your first setup is a full bench setup if you have the equipment like a good power supply or on the bench with some form of battery holder. Once you know the LED current is correct you could measure the battery current and from there extract the efficiency of the converter and also measure the Vf of the LED. Once these numbers have been extracted and the light put together you could do a final check and verify that the tailcap current is aroudn the same number as when you measured it on the bench.
I will modify my procedures to incorporate these suggestions.

Originally Posted by dat2zip
Take it with a grain if salt. It may be absolutely correct that the current to the LED (based on the tailcap current measurement) is less than expected or there is something that is not being accounted for. Whether that be the converter efficiency number, LED Vf, or converter stability to name a few, who knows...

As it stands now given only the battery voltage and tailcap current it would be hard to speculate whether the converter is regulating or not and whether the current to the LED is 1A or not.

One other point. We have sold a lot of SOBs and the vast majority work out of the box. Usually if the SOB doesn't work it doesn't work. Now it could be true that there are a lot of defective SOBs that the customers for some unknown reason fail to report it back to the shop for replacement, but, I find that statement hard to believe.

Wayne
I would tend to believe that all is well. I suspect that, at a minimum, slight differences between actual vs assumed values for several of the relevant variables, DMM accuracy spec issues, and DMM voltage drop/internal resistor effects led me down the garden path. Thanks.

11. ## Re: SOB 1000/MC-E 2S2P question

Justin Case,

Thanks for taking the time (and thanks to dat2zip too) to share significant details with your findings and project. I too was wondering how an SOB driver would work with the MC-E wired in 2S2P and it is quite obvious you've spent much time experimenting and testing. After all of the information you've shared and the time you've put in I didn't see your opinion summarizing how well this driver solution works. Since you've got an SSC P4 U2 bin module to compare the MC-E too......what does the eyeball test indicate? Is the MC-E significantly brighter?

9x23

12. ## Re: SOB 1000/MC-E 2S2P question

Basically, the MC-E tower puts out a lot of light. The spill is very wide and bright in my T-62 TurboHead, over 3X brighter than for the P4 tower. The hot spot lux at 1 meter is also large and bright (about 50%-60% brighter than my Malkoff M30, for example), but not as bright as with the P4 tower. The MC-E hot spot is a large, fairly uniform disk of bright light. The P4 has more of the traditional hot spot plus corona. The overall size of the MC-E's hot spot is about the size of the P4's hot spot plus corona. The beam quality for the P4 tower is essentially perfect -- no noticeable anomalies even in white wall tests. In contrast, the MC-E hot spot has some very slight dark spots in an otherwise excellent beam. At very close range, you see the Cree MC-E "cross", but beyond a few feet it disappears. The SOB 1000 works well and heat does not seem to be a problem (I potted the SOB with thermal compound as a precaution). I also tried an SOB 1227 and that also seems to work well, although I don't really see any noticeable output difference vs the SOB 1000. I'd probably stick with the SOB 1000 for any future builds. Slightly less expensive and lower current draw from the batteries.

Edit: I think a better way to get some additional lumens is to stick with the SOB1000 driver but use an M-bin MC-E instead of the K-bin I've been using. Calculated lumens for a 2S2P M-bin using an SOB1000 is basically the same as for a 2S2P K-bin driven by an SOB1227.

Building a Seoul P4-based tower is a lot more relaxing than building an MC-E tower. With the P4 tower, you just build it cookbook-style. Follow the steps, and you should get a successful tower.

The three things that make the MC-E tower a hassle are the filing of the corners of the MC-E case, the bending down of the MC-E's leads, and the LED focus/centering. The first two issues deal with fitting the MC-E into the SureFire TurboHead's reflector hole. One little slip with the file and you can nick the MC-E's dome. Most likely, that won't have any noticeable effect on the light output or beam quality. But still, at \$20 or more per emitter, you'd like a pristine dome. Bending the leads gives similar concerns. It would really suck if you fatigued and snapped off a lead.

The third issue is LED focus/centering. These AW towers are already designed for Seoul P4 focus in an SF TH. But the Seoul focus tower height is not correct for a Cree MC-E. You need to raise up the MC-E an additional (nominal) 0.93mm, assuming that the P4 and MC-E dies are located right at the top face of their respective cases.

If you check the Seoul docs for the P4 and the Cree docs for the MC-E, you'll find that the P4 case (emitter) height is given as 2.38mm +/- 0.2mm and the MC-E case (emitter) height is 1.45mm +/- 0.1mm. That's where I get my 0.93mm figure above to raise up the MC-E. However, due to the measurement tolerances in the docs, the additional MC-E height for good focus could range from 0.63mm to 1.23mm. But at least from the statistics of small numbers based on the MC-E towers I've built, the 0.93mm target seems to work. Whether it's optimal, I can't say. It may be that the MC-E could use just a little more height, which would remove the very minor dark blobs that I can see in the hot spot in white wall tests.

I've not tried to use a mic or calipers to measure the bare emitter case heights.

Let me do that right now.

One P4 measures 0.097", or 2.46mm. Another measures 0.093", or 2.36mm. One MC-E measures 0.056", or 1.42mm. Height differences are 1.04mm and 0.94mm. So there appears to be some variation in emitter heights, resulting in variation in how much to raise up the MC-E.

Another unknown is how the AW tower height was optimized for Seoul P4 focus. Presumably, an actual SureFire Turbo lamp filament height was measured (or several were measured and averaged). Let's say that this height is H. The optimum P4 tower focus height then could be set to (H-2.38mm), if one assumed that the P4 emitter height is 2.38mm. But, if AW used a different value for the P4 emitter height, then that can also change the estimate on how high you need to raise the MC-E. Since I don't know how the optimized Seoul focus AW tower height was determined, all I can say is that 0.93mm of extra height is probably a good target value. I don't know how sensitive the beam quality is to focus height, but presumably 1.0mm vs 0.93mm won't be a big deal. And maybe it's beneficial to err slightly on the high side due to emitter height variations.

The hassle is that hitting that nominal 0.93mm target is a bit of a hit-or-miss affair. I don't have a shim that is 0.90mm thick (for example), with the remaining 0.03mm provided by a thin Arctic Alumina thermal epoxy layer between the shim and MC-E. All I have is one of Wayne's 0.030" (0.76mm) copper shims, with two layers of AA epoxy at the tower/shim and shim/MC-E interfaces to get me to ~0.93mm. I've run my MC-E towers for as long as 17 min continuous, with no tint shift or other indications of heat problems. So, it seems that even relatively thick ~0.1mm layers of AA epoxy are ok.

Associated with the shimming hassle is the MC-E centering hassle. All centering has to be done by eye. First, you have to center the shim on top of the tower. Any excess AA epoxy that squeezes out interferes with that visual centering. So, you try to wipe off the excess. Then you mount and center the MC-E. Same issue with excesss AA that squeezes out. Plus, the MC-E is squarish, not round, making visual centering that much more tricky.

I personally didn't find the soldering too difficult. IMO, if you can successfully solder to a bare SSC P4, you can deal with an MC-E.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•