Feedback/impressions on the PhD-M6 custom battery pack

[Justin Case]

Regardless of the max limit of AW's protection circuit, asking his 17670 cells to deliver 4A-5A is asking a lot. The usual recommended max discharge rate for a standard LiCo chemistry Li-ion is 2C.

AW's 14500 cells also have a protection circuit with an upper limit of 5A +/-0.5A. I seriously doubt that pulling 5A from a 14500 is a good idea, regardless of that 5A limit for the protection circuit. 14500 C is advertised as 750mAh. 17670 C is given as 1600mAh.

 

[Justin Case]

deleted

 

[mdocod]

Hi donn,

Sorry, I mixed up the voltage/bulb arrangements in the response, but the concerns are still valid...

The high current limit on the PCB is only set that high in order to allow for the cold-start of high power bulbs. There is a combination of effects taking place. It's not unusual for current to spike between 10-20+ amps when a bulb is cold and being fired up. At the same time, there is a delay before the protection circuit can react to the high current condition. By moving the current limit higher on the PCB, it's possible to "trick" the delay into allowing that high current bulb to fire up and move on to steady state operation before the protection trips. The 5A limit on the PCB has nothing to do with the maximum continuous safe discharge of the cells.

The maximum recommended continuous discharge rate on most LiCo cells is between 1.5C and 2C rate. The PhD-M6 and any other PWM regulation presents further complications. Since the load is not continuous, but instead, in the form of higher current pulses, more energy from the cells is translated into heat as a result of cell resistance than would have been in a continuous discharge. This results in more heat buildup in the cells, which requires further de-rating for continuous discharges.

In our application, the MN21 is running at ~9V X ~7A X~50% duty cycle. The peak current on each pulse is actually in excess of the 5A limit of the PCB, but the fast pulsing creates a scenario where the PCB can not react to those high current peaks. The RMS current would have to rise above 5A to trip it more than likely.

I wasn't originally planning on sharing this, but when Will was working on the optimization of the soft-start to allow the MN21 to fire up (it wouldn't originally), I decided to test it to see how much "over-head" we had in the design for the PCB tolerances that we might run into. The 64275 will fire up at 6.8V, which is about 44W. (I didn't test this a ton, just cranked up the bulb numerous times, ran it a few minutes, and that's it) This convinced me that we have the over-head required to run the MN21 without problems from the PCB with the modified soft-start program.

Eric

 

[wquiles]

Regardless of the max limit of AW's protection circuit, asking his 17670 cells to deliver 4A-5A is asking a lot. The usual recommended max discharge rate for a standard LiCo chemistry Li-ion is 2C.

AW's 14500 cells also have a protection circuit with an upper limit of 5A +/-0.5A. I seriously doubt that pulling 5A from a 14500 is a good idea, regardless of that 5A limit for the protection circuit. 14500 C is advertised as 750mAh. 17670 C is given as 1600mAh.

+1

Well said Justin.

I think Will and I are going to have to discuss putting some kind of warning or disclaimer or something in place that says we are not going to configure the PhD-M6 in scenarios where the intention is to exceed 35W power consumption.

Exactly. We have tested and re-tested the PhD-M6 at the default values with as many bulbs as we could, and have a great feel of the "envelope" that we can support. We can stand firm in stating that the driver will work well within those limits and bulb recommendations.

However, whether the end customer is using default voltage settings or custom voltage settings, the end user can exceed the cells/pack beyond the safe area that we tested - there is nothing we can do about it. But every action has a consequence. If the pack/driver is damaged when going beyond the safe area we tested and recommended, then I don't feel it would be fair to call that fair/normal use, and repair/replacement of parts & labor should be covered at the owner's expense.

 

[Justin Case]

What do you think of the following:

Level 1: 6.8V (MN21)
Level 2: 7.3V (MN15, MN20, N62, 1111, 64250, 5761 eventually)
Level 3: 8.0V (1164, 1274, Hikari 5607)
Level 4: 10.8A (1185, 1331, 1318)

I like the original Level 1 at 5.0V, but I have only one N1 lamp, while I have lots of 1164s and 1274s. So I'm sacrificing the low-low for another upper-mid voltage setting. I figure that the MN15 at 7.3V can be my long-running lamp. I reduced the 7.5V level for the MN15 and MN20 down to 7.3V to try to accommodate the 1111, 64250, and 5761. I listed the N62 in my notional Level 2, but it's not clear to me if I'd actually use it. The notional Level 3 of 8.0V tries to balance safely driving the 1274 vs. getting some reasonable overdrive for the 1164 and especially the 5607. If I didn't have so many 1274s, I'd probably go for 9.0V for Level 3.

 

[wquiles]

What do you think of the following:

Level 1: 6.8V (MN21)
Level 2: 7.3V (MN15, MN20, N62, 1111, 64250, 5761 eventually)
Level 3: 8.0V (1164, 1274, Hikari 5607)
Level 4: 10.8A (1185, 1331, 1318)

I like the original Level 1 at 5.0V, but I have only one N1 lamp, while I have lots of 1164s and 1274s. So I'm sacrificing the low-low for another upper-mid voltage setting. I figure that the MN15 at 7.3V can be my long-running lamp. I reduced the 7.5V level for the MN15 and MN20 down to 7.3V to try to accommodate the 1111, 64250, and 5761. I listed the N62 in my notional Level 2, but it's not clear to me if I'd actually use it. The notional Level 3 of 8.0V tries to balance safely driving the 1274 vs. getting some reasonable overdrive for the 1164 and especially the 5607. If I didn't have so many 1274s, I'd probably go for 9.0V for Level 3.

I look forward to hearing Eric's comments but these look good to me. Once you finalize these, please update your order the sales thread ASAP, as your pack is likely to be one of the ones in the first batch

 

[Justin Case]

Will do ASAP. Thanks.

 

[LuxLuthor]

I just updated the document - available from the main sales thread. Comments/suggestions welcomed

I emailed you just a few!

 

[LuxLuthor]

I liked some of the suggested custom settings. After going over my charts and Ictorana's graphs, I'm thinking of getting this setup with my 2nd PhD-M6. I'm using general guidelines of not exceeding 5A draw, as between the heat, limited 17670 mAh capacity, and desired run time, I don't feel it would be practical (including the venerable 5761). I'll use AlanB, JimmyM, or direct drive in Mag's for those.

Level 1: 6.8V (MN21 & 1160)
Level 2: 7.4V (MN15, MN20, N62, 1111)
Level 3: 9.3V (1164, Hikari 5607)
Level 4: 10.8A (1185, 1331)

 

[wquiles]

Will do ASAP. Thanks.

No problem Justin

I emailed you just a few!

I guess I got lucky there only a few :naughty:

I will work on the updated document - thanks

 

[Justin Case]

In terms of Wh and thus run time, there is little difference between the 5761 driven at 7.3V/5.55A (40.5 Wh) vs the Hikari 5607 at 9.3V/4.3A (40 Wh).

In terms of current draw, both far exceed the recommended 2C discharge limit (3.2A for the AW17670). Double bacon cheeseburger vs triple bacon cheeseburger.

 

[wquiles]

In terms of Wh and thus run time, there is little difference between the 5761 driven at 7.3V/5.55A (40.5 Wh) vs the Hikari 5607 at 9.3V/4.3A (40 Wh).

In terms of current draw, both far exceed the recommended 2C discharge limit (3.2A for the AW17670). Double bacon cheeseburger vs triple bacon cheeseburger.

I ordered the boards with twice the oz of copper (2oz vs the standard 1oz), and made the high current traces short and very wide, and selected (with help from Jimmy) a low Rds FET (5mOhm) with 25Amp drive rating, but at the end of the day we are just very limited by the 17670 cells we have available today. The good thing is that the driver itself should work really well if/when the IMR cells ever become available.

That being said, if you are deciding between those two, I would go for the lowest current draw.

 

[Justin Case]

The Hikari is the lower current draw bulb and it is also really inexpensive and probably more readily available than the 5761. I wish I had my bulbs with me to inspect them and to run some beam quality tests. But I vaguely recall that the Hikari filament is relatively large and didn't focus that great with a VLOP FM Ver2 Deep Mag Reflector. Can't recall the 5761's filament. Not sure what kind of beam the heavier OP KT4 would deliver. The WA bulbs all use fairly small, fine filaments. As I recall, they all seem to produce nice beams in a KT4.

The deal in my case is that I have more 1111s, 5761s, and 1274s than Hikari 5607s and 1164s. That skews my choices more toward optimizing for the 1111/5761 and 1274, i.e., Level 2 at 7.3V and Level 3 at 8.0V.

I think if I had a boat load of only 1111s and 5607s, I'd set Level 2 to 7.4V and Level 3 to 9.4V (or to Lux's 9.3V).

For high output, the 5761 is comparable to the 5607 in estimated lumens. The downsides, besides bulb price/availability, are even higher current draw and probably the need to wait for IMR17670s to hit the market.

 

[LuxLuthor]

The Hikari is the lower current draw bulb and it is also really inexpensive and probably more readily available than the 5761. I wish I had my bulbs with me to inspect them and to run some beam quality tests. But I vaguely recall that the Hikari filament is relatively large and didn't focus that great with a VLOP FM Ver2 Deep Mag Reflector. Can't recall the 5761's filament. Not sure what kind of beam the heavier OP KT4 would deliver. The WA bulbs all use fairly small, fine filaments. As I recall, they all seem to produce nice beams in a KT4.

The deal in my case is that I have more 1111s, 5761s, and 1274s than Hikari 5607s and 1164s. That skews my choices more toward optimizing for the 1111/5761 and 1274, i.e., Level 2 at 7.3V and Level 3 at 8.0V.

I think if I had a boat load of only 1111s and 5607s, I'd set Level 2 to 7.4V and Level 3 to 9.4V (or to Lux's 9.3V).

For high output, the 5761 is comparable to the 5607 in estimated lumens. The downsides, besides bulb price/availability, are even higher current draw and probably the need to wait for IMR17670s to hit the market.

Basically I agree with your thinking on this from a total Watts standpoint, so no wrong choices. I'm more of a Bacon Double Cheeseburger guy, but I have a few points worth making for others considering your 5761 IMR idea.

The 5761 is not even close to fitting any of the three KT4 reflector openings I have, and I'm not boring them or sanding the bulb envelope just for this purpose. The Hikari fits easily.

With the 5.55A (@ 7.3V) current draw, the 5761 is a problem with 17670 LiCo cells & protection circuit. The Hikari as you say is cheap, smaller glass envelope, although a somewhat larger transverse filament. Either will turn out fine in a textured MOP reflector like in the M6.

Keep in mind that while the IMR will tolerate higher amp draw, it will have less storage. If you extrapolate from AW's 18650

LiMn 1600mAh / LiCo 2200mAh = 0.72 Let's round up and say that LiMn has 75% the capacity of LiCo.

So if 17670 LiCo has 1600mAh, then LiMn version should be reduced to 1200mAh, which isn't much when you are sucking out 5.55 A for the 5761. I would guess you probably have less than 10 mins by the time the realistic IMR capacity and voltage curve is used + unknown trigger point of moon mode cutting the balls off the mighty 5761.
​

Of course these are just the estimated extrapolations of an incandescent madman, but that's what went into my thinking in avoiding the 5761 in this application.

 

[mdocod]

The 5607 is delegated to either level 1 (7.2V) or 2 (8.5V).

Realize what I did, in referencing LuxLuthors charts, I mixed up the 1164 and 5607 in translation. (too many tabs!).

I expect to see IMR, LiFePO4 or some other safe high amp chemistry in this size eventually.

If LiFePO4 cells come out in this size, then any setting above ~7-8V driving a relatively high power bulb would likely not be in regulation. We have to cross our fingers for LiMn chemistry cells to be able to drive more bulbs in regulation.

Hint Hint > AW wink wink

I wouldn't say bulb life is the primary factor in determining drive levels. It's one of them, but it's no more important than unit price, output level and beam quality.

I fully appreciate the glory of driving cheap bulbs hard for the thrill of that brilliant white efficient output. However, the PhD-M6 project is an effort to turn the M6 into a better tool. Any recommendations I make within the scope of this project have to be on the conservative side wherever possible. Cranking through a bunch of cheap bulbs is not an option for folks who depend on that tool. Bulb life is extremely important when I pick drive levels that could be life and death for someone busting down a door.

*wait, hang on guys, let me swap in another bulb -bullets whizzing by-, anyone seen that that really small hex driver? ah crap....*

I guess the hex driver would come in handy later though. Maybe they could use it to pry bullets out...

I have to treat this project as if we are working on a serious tool. With that in mind, my recommended drive levels for bulbs will probably stop short of where many people may choose to drive them.

Eric

 

[mdocod]

Hi Everyone,

I feel that there is still confusion regarding normal operating current of bulbs in steady state, compared to how the PhD-M6 drives bulbs (or any PWM driver for incans). To say that we shouldn't exceed a particular RMS bulb current (3A, or 5A, or whatever), is not necessarily correct. In practice, setting some personal limits on bulb power is never a bad idea when using lithium cobalt or lithium primary cells.

The high current limit on the cells is tripped based somewhat on RMS current lasting long enough to trip it, and it is somewhat tricked by the PWM anyways. Remember, the MN21 is running at 7A peaks in our application and it's not tripping the circuit. I've tested the 64275 briefly, which would actually operate at something closer to 8-9 amp peaks under this PWM.

A 2C discharge in the case of a 3x17670 pack consisting of cells that are new and in great shape means that roughly 40 watts is being depleted across the entire circuit, including the power wasted as heat to resistance before anything even gets to the intended device to be powered up (in our case, a bulb). Under a normal discharge condition (non pulsed load), a 2C rate for a 3x17670 pack would generate heat in the cells at a rate of 4.5W, and about 2W would be wasted in other contacts and such (in a flashlight application). The result is that the power consumption of the device to be powered should not exceed ~33.5W for a normal direct drive 2C load, again, assuming brand new cells in great condition. At a 2C rate, one could say that typical LiCo powered flashlights are operating at ~84% efficiency after considering losses at the batteries and contacts and such. Dealing with these losses is best performed by reducing resistance wherever possible. Including finding lower resistance cells (this is why we need IMR cells!)

When we use a higher voltage source to drive lower voltage bulbs with PWM regulation, we get get higher current peaks, so a higher percentage of the total energy flowing from the cells gets converted to heat within the cells.

Rough Example: WA1111 @ 7.2V, 3.8A, 1.9 Ohm.

1.9 Ohm bulb resistance + ~0.5 Ohms for everything else = 2.4 Ohm load.

With cells at ~80% state of charge: 12V pack / 2.4 Ohm = 5A

12V x 5A = 60W total power consumption during an "on" pulse of the regulator.

0.5Ohm / 2.4Ohm x 12V = 2.5V lost to resistance
1.9Ohm / 2.4Ohm x 12V = 9.5V hits the bulb

9.5V x 5A = 47.5W hits the bulb during an "on" pulse of the regulator.

PWM @ 57% = 27W RMS to the bulb. The drive level is power based. The bulb never sees 7.2V or 3.8A, but it runs just the same.

2.5V x 5A x 57% = ~7W converted to heat, ~5W of that heat is right in the cells.

The total power dissipation (including ALL losses, not just the bulb operating power) occurring in this snapshot of running a WA1111 is 34W. Or ~1.7C discharge rate.

The efficiency loss to resistance gets slightly worse compared to direct drive. The example I gave above for a direct drive application shows ~83% efficiency into a 2C load, here, at 1.7C, it's more like ~79% efficiency.

I think that the most important thing to keep in mind, is that, because of the way that we are achieving regulation here, it's going to be normal to see more heat buildup in the cells for a given power output. If a 2C rate would normally cause ~4.5W of heat at the cells, and a 1.7C rate under one example of PWM regulation causes slightly more heat at the cells, then one must realize that at a 2C rate, we are going to exceed the heat build-up rate that the cells were originally rated to tolerate in a continuous discharge at 2C. This means that the user needs to take care in preventing continuous discharges of high power lamps. Cool down periods are important for safety.

So.... One could say, that a WA1111 IS a 5A bulb. You just have to run it at 9.5V instead of 7.2V.... Oh, and don't forget to turn it off 106 times per second

Eric

 

[mdocod]

Hi Everyone,

I spent pretty much all night last night calculating a bulb chart for the PhD-M6...

bulb_chart_v1.pdf

The information in the chart can be used to make decisions about driving various bulbs with the PhD-M6. Not every bulb in the chart is appropriate for use with the PhD-M6 on lithium cobalt cells, but have been included to maximize the amount of available information. For each bulb, I have listed some information about the theoretical range of drive levels that should work pretty well for the bulb, approximated torch lumen and runtime for each bulb, as well as some information about how taxing the bulb is on the cells themselves. Many of the bulbs that "could" be used should not be used with LiCo cells but each user will have to decide where their comfort zone on this is. If you want to play it safe, don't use any bulbs that have a "C" rate above 2, if you want to play it really really safe, do not use any bulbs with a heat factor rating above 1.

In studying the chart, I think many practical enthusiasts will see that the MN61 @9.0V may prove to be the most logical balance of runtime, output, and safety with no continuous run limit.

I've come up with an interesting specification that I am simply going to call "heat factor." The heat factor is a rough calculated estimate of the ratio of energy being converted to heat in the cells while running a particular bulb compared to the maximum rate of heat buildup that would normally occur under a direct drive 2C rate discharge of the cells. This information helps give the user a better understanding of how "hard" on the cells a particular bulb is, independent of just looking at the runtime and calculating a "C" rate. A heat factor of 1 means than the heat produced in the cells during the discharge is expected to be similar to what would have occurred during a normal direct drive 2C discharge. It is advised to never run a bulb with a heat factor above 1 continuously through the discharge, as it will likely result in cells over-heating.

The runtime ratings are on the conservative side in many cases but aren't guaranteed. Cell condition and operating temperature will effect actual runtime.

I actually learned a lot myself while running all of the numbers. The "heat factor" can also be looked at as a general efficiency of the configuration. Higher heat factors mean more energy is wasted as heat in the cells, lower heat factors mean that less energy is wasted in the cells. High heat factors are caused by very intense low resistance high current loads.

Most of the bulbs listed have their heat factor, run-time, and "C" rate calculated based on the recommended drive level in the far right column. For a few bulbs, I made 2 separate listings to show the difference in runtime, heat factor, and "C" rate based on running at different specific drive levels to show how different drive levels effect the results. For those specific listings, the voltage is listed next to the bulb name in the far left column.

Actually interesting to learn, that the 5607 and 1164 aren't really much harder on the cells than the MN21. The actual total energy drain from the cells is only slightly higher than the MN21, since they operate at higher voltage, and higher resistance, the heat factor is actually a bit lower.

Eric

 

[wquiles]

Thanks much Eric :twothumbs

 

[Justin Case]

Basically I agree with your thinking on this from a total Watts standpoint, so no wrong choices. I'm more of a Bacon Double Cheeseburger guy, but I have a few points worth making for others considering your 5761 IMR idea.

The 5761 is not even close to fitting any of the three KT4 reflector openings I have, and I'm not boring them or sanding the bulb envelope just for this purpose. The Hikari fits easily.

With the 5.55A (@ 7.3V) current draw, the 5761 is a problem with 17670 LiCo cells & protection circuit. The Hikari as you say is cheap, smaller glass envelope, although a somewhat larger transverse filament. Either will turn out fine in a textured MOP reflector like in the M6.

Keep in mind that while the IMR will tolerate higher amp draw, it will have less storage. If you extrapolate from AW's 18650

LiMn 1600mAh / LiCo 2200mAh = 0.72 Let's round up and say that LiMn has 75% the capacity of LiCo.

So if 17670 LiCo has 1600mAh, then LiMn version should be reduced to 1200mAh, which isn't much when you are sucking out 5.55 A for the 5761. I would guess you probably have less than 10 mins by the time the realistic IMR capacity and voltage curve is used + unknown trigger point of moon mode cutting the balls off the mighty 5761.
​

Of course these are just the estimated extrapolations of an incandescent madman, but that's what went into my thinking in avoiding the 5761 in this application.

All quite sensible. Fortunately, all of my stock 5761s fit through my KT4s.

IMO, when you get down to the bottom line, really none of these HO bulbs -- e.g., 5761, 1185, 5607 -- are all that practical. They all require fairly high current and 3x17670 in whatever chemistry is just going to be limited in feeding that appetite. It is true that the 5761 is probably by far the hottest running bulb, which further limits its already limited practicality.

I'm leaning toward running a 1274 as my workhorse bulb. The MN20 and MN61 also are candidates, but the SF lamps are way more than the 1274's $5 price (ignoring the sunk cost of the FM MN bi-pin adapter).

 

[LuxLuthor]

Hi Everyone,

I spent pretty much all night last night calculating a bulb chart for the PhD-M6...

bulb_chart_v1.pdf

Eric

Thanks for doing this. I wonder if there is a way to get this into a format for printing landscape on 8.5 x 11 paper, and/or put on Excel to add lines to track entire line easier? It takes a while to compare some of this to other tested results. That being said, I'm not sure I agree with all your numbers (i.e. recommended voltage range & lumens--for example the 1111), but it is still a useful comparative reference, and represents a good bit of work. Where are you getting your lumens readings from? Do we have CBA-II test results running the AW 17670 at various Amp loads? If not, I can do that quickly.