Hotwire PWM Regulator for Mag D body, JM-PhD-D1

[JimmyM]

I'm starting this thread to see about interest and introduce my version of the Hotwire regulator. This is a spin-off of the design work that Alan_b, wquiles, myself, and others have been working on. The basic theory, design, software, etc have been developing for a while over in this thread. HERE

My version will be similar to the JM-SST in that, it will reside under a KIU base and use PWM. Other than that, this is a true PWM regulator.
I haven't begun power testing at high levels yet.

I'd like to see what thoughts you folks have regarding this.
SInce this design will be incorporated into an existing Mag switch. You have 2 options.
1) The internal circuit is completely dead when off and it will only leak as much current as the MOSFET itself will allow. That's in the area of a few nano or micro amps. However, you're limited to just ON or OFF operation.
2) The internal circuitry is live (but will sleep). This will leak more current due to the voltage sampling divider and the circuitry using some small amount of current. In the area of a few hundred micro amps. But this will allow things like multiple levels.

Also, there is the subject of adjustability. The circuit is reporgrammable and the source code is available. So you can have me program it for certain voltages before sending it out or you can reprogram it yourselves if you're feeling adventurous.
There is another possible option. I'm going to work on a way to add an adjustment pot to the regulator. This way, you can adjust your own voltage.

To Do List:
Run high power tests with 30V FET
Run high power tests with 40V FET

OK, so that's it. I'd like to open the floor for discussion.

UPDATE 1/6/09:
An onboard pot to adjust voltage will be used. This will maintain it's setting even if you change packs.
The 30V version looks good. I'm waiting on a 40V FET (that would be used in the actual design) to run high power tests. I have another 40V FET but that one is too large to use in the end design.
I have run tests using the 64623 and 64625 bulbs with the 30V FET with 29.4V input and 12VRMS output. Peak current on the FET is about 21 Amps. Tests have run as long as 3+ hours with no issues. I have more parts on the way to test 40V FET version.
We (Alan, Will, and I) have noticed an inductive "kick" that applies greater then supply voltage to the FET when it turns off. I'm investigating possible ill effects caused by this. Thus far I have seen none.

 

[drew2001]

Hi Jimmy,

I'm absolutely interested. Wanted to bump your post up and look forward to others expressing participation as well. I'd say for my app it would be # 1 as I only use the mag623 (so far) and have yet to go into higher watt outputs. Looks like the chances for upping the outputs in a D cell size mag is going forward. I also like the idea of owner operator adjustable pot option for the v regulator... options options fun stuff.

Will this next gen board have a slow start option as well?

For those reading about this for the first time, I'll add that Jim has been great at working his magic and timely fulfilling the orders. I've been in contact with his design revisions on my own JM-SST board and been happy with the outcomes.

Below is a link to my first attempt at installing the circuit board in my mag623

https://www.candlepowerforums.com/threads/186715

Just wanted to post success stories in confidence. Looking forward to yours, Alan's, Will's and others efforts.

Thanks again. :thumbsup:

 

[JimmyM]

It WILL have a softstart feature. Sorry I forgot to mention that part.

In the next few days I'll be doing some high power testing with the FET I plan to use. I'll start with the 5761 and move up to the 64623 and 64458.
I'd like to see if the smaller FET can handle the high current of the big bulbs.

 

[JimmyM]

Another question for you folks.
What is the highest voltage Mag bodied mag? Would it exceed 35V?

 

[donn_]

Better ask Lux that question. :devil:

 

[Metatron]

Another question for you folks.
What is the highest voltage Mag bodied mag? Would it exceed 35V?

i have an elle II with a 400watt globe just waiting for an appropriate switch, so 40v would be right on.

 

[JimmyM]

Better ask Lux that question. :devil:

True. I think I remember someone using a 36V bulb. Hmmmm. That "may" be outside the realm of possibility for this unit. But we'll see. The ing is... If I can use a different regulator internally, that frees up more room, making more room for a bigger FET.
Of course, I have to see how popular the programming ability is for folks. If most people just want them set and be able to tune the voltage via a pot, then I can gain some room for a bigger FET.

 

[Metatron]

True. I think I remember someone using a 36V bulb. Hmmmm. That "may" be outside the realm of possibility for this unit. But we'll see. The ing is... If I can use a different regulator internally, that frees up more room, making more room for a bigger FET.
Of course, I have to see how popular the programming ability is for folks. If most people just want them set and be able to tune the voltage via a pot, then I can gain some room for a bigger FET.

now ur talking great idea a pot!

 

[Raoul_Duke]

I'm interested.

I'd like to think I could re-program thins, but I would be happier on a fresh pack, and measuring V bulb, and using a pot.

Where would the pot be? would we have to drill a hole in the kiu, or could it be done by removing the rubber boot of the switch.

Also would it be possible for a couple of wires to be hidden under the boot for measuring V bulb.

The tailcap could be unscrewed for the leakge issue, and the multi lever is good for usefull vs insane output.

I'd like to be able to get a full run without heat concernes, but have the option of going back to full power when required.

Would be nice if it switched on at full blast ( ok maybe a few mseconds SS) and then ramp down.

Whatever way you do it I'm interested. ( edit = Sold :twothumbs )

 

[JimmyM]

I'm interested.

I'd like to think I could re-program thins, but I would be happier on a fresh pack, and measuring V bulb, and using a pot.

Where would the pot be? would we have to drill a hole in the kiu, or could it be done by removing the rubber boot of the switch.

Also would it be possible for a couple of wires to be hidden under the boot for measuring V bulb.

The tailcap could be unscrewed for the leakge issue, and the multi lever is good for usefull vs insane output.

I'd like to be able to get a full run without heat concernes, but have the option of going back to full power when required.

Would be nice if it switched on at full blast ( ok maybe a few mseconds SS) and then ramp down.

Whatever way you do it I'm interested. ( edit = Sold :twothumbs )

For the ON/OFF version, no modding of the guts of the switch are required. You would just solder a wire from the top if the switch to the pad on the regulator. Just like the JM-SST. So a pot under the rubber switch boot would require a lot more modding. The pot for this version would be accessed via a hole in the kiu base. To measure voltage, however, you will need a True RMS voltmeter since the voltage to the bulb will be a PWM/Square wave.
The multi mode version would require that the switch me made momentary. This is pretty easily done.

 

[Alan B]

For the ON/OFF version, no modding of the guts of the switch are required. You would just solder a wire from the top if the switch to the pad on the regulator. Just like the JM-SST. So a pot under the rubber switch boot would require a lot more modding. The pot for this version would be accessed via a hole in the kiu base. To measure voltage, however, you will need a True RMS voltmeter since the voltage to the bulb will be a PWM/Square wave.
The multi mode version would require that the switch me made momentary. This is pretty easily done.

NOTE - DC RMS meters with the proper characteristics are not common, and are not required.

Measurements can be done with an averaging DVM and a calculation. It is probably better to recommend this technique than getting or using a DC RMS meter, since most inexpensive RMS meters won't do it properly, and folks who have or buy RMS meters won't be happy if they buy one and it does not work for them on DC RMS.

Procedure:

Measure Vbatt and Vbulb mean and calculate Vrms, or measure Vbatt and assume a desired Vrms and compute the Vbulb mean to adjust to. This is documented in the design thread and in the program source (copied here):

* RMS Calculations
*
* VbulbRMS = sqrt(VbattDC * VbulbMEAN) # to see what you have
*
* VbulbMEAN = VbulbRMS^2 / VbattDC # to set a desired value

One tricky thing is to adjust the output voltage without losing the low battery voltage protection. The calibration needs to be correct for the protection to work properly, so changing the calibration is not really what one wants to do with the output voltage pot. It might make more sense to have a voltage adjustment pot that merely put a voltage into a chip pin that was a "request" input, and use this to control the desired output voltage. The calibration would be separate, and the low battery voltage would be unaffected by the adjustment.

Alternately, develop a "one button interface" program rather than a pot for the interface.

Lots of choices.

-- Alan

 

[JimmyM]

You're absolutely right, Alan. I forgot about that.

Also, the IRLR/IRLU7843 runs the 84458 at 19.9V RMS on a 25V DC supply with barely any warmth. I had to touch the tab with the skin just back from my fingernail to detect it at all. An infrared thermometer would be nice. I'll be testing the 64623 on 16V later.
Regarding heat. The original JM-SST didn't have any heat sink at all. This board design has a much larger copper area to dissipate heat if it accumulates. 2 oz copper and 16 thermal vias to a copper pour on the bottom of the board. Similar but smaller type of thing on the FET driver.
If I can manage to get rid of the programming header I might be able to make room for the mighty IRF2804(40V) or IRF3206(60V). Then 500+ watt regulated Mag mods might be possible. Of course I would have to leave that up to LuxLuthor capable hands to test. He he he.

 

[JimmyM]

One tricky thing is to adjust the output voltage without losing the low battery voltage protection. The calibration needs to be correct for the protection to work properly, so changing the calibration is not really what one wants to do with the output voltage pot. It might make more sense to have a voltage adjustment pot that merely put a voltage into a chip pin that was a "request" input, and use this to control the desired output voltage. The calibration would be separate, and the low battery voltage would be unaffected by the adjustment.

Alternately, develop a "one button interface" program rather than a pot for the interface.

Lots of choices.

-- Alan

All true. It's a fine balance between operator simplicity and functionality. That's kind of what I'm trying to feel out here. I like the "request pin" input idea. I've mostly been messing around with FP7. simple ON/OFF with regulation and softstart. The low battery protection would be cool if I can fit all that stuff on the board. Because the programming header would still be required to set the low voltage shutdown. So many options, so little space. I'd hate to have to fit an additional pot. We'll see.

 

[SafetyBob]

OK, Jimmy, here it goes:

I would like a simple on/off regulated, softstart thingy for a 623 setup first. Now, I want you so set it up, I will tell you what battery setup I will use.....niMh or emoli. All I need to do is install the thing and done....ready to go.

I think if we keep it simple, limited to three or four bulbs for now, it would be best for you for inventory and runs of forty or fifty would be feasible for you to keep up with you would be stuck with stuff you can't sell.

Now, if it would be possible for the low voltage shutoff, then you would be even more of the man that you already are in most of our opinions. But if it complicates the design so that it becomes a big problem to fit into the switch or assembly, then lets forget it........it would be nice though!

Now, the future. I know this is for M@gs, but you know there is a small crowd of us crazy ones contemplating how to fit aircraft landing lights into a cheap lantern assembly.....we would appreciate something for that 400, 600, 800 or 1000 watt light too. But we can talk about that later after the M@g situation is solved or into production.

I appreciate all your work on this. It's really up to you if you want to manufacture a complete switch assembly or just a board that most of us can install ourselves.

Bob E.

 

[JimmyM]

OK, Jimmy, here it goes:

I would like a simple on/off regulated, softstart thingy for a 623 setup first. Now, I want you so set it up, I will tell you what battery setup I will use.....niMh or emoli. All I need to do is install the thing and done....ready to go.

I think if we keep it simple, limited to three or four bulbs for now, it would be best for you for inventory and runs of forty or fifty would be feasible for you to keep up with you would be stuck with stuff you can't sell.

Now, if it would be possible for the low voltage shutoff, then you would be even more of the man that you already are in most of our opinions. But if it complicates the design so that it becomes a big problem to fit into the switch or assembly, then lets forget it........it would be nice though!

Now, the future. I know this is for M@gs, but you know there is a small crowd of us crazy ones contemplating how to fit aircraft landing lights into a cheap lantern assembly.....we would appreciate something for that 400, 600, 800 or 1000 watt light too. But we can talk about that later after the M@g situation is solved or into production.

I appreciate all your work on this. It's really up to you if you want to manufacture a complete switch assembly or just a board that most of us can install ourselves.

Bob E.

It's not a matter of bulb set-up per se. It's just a voltage issue. Set it to what ever voltage you want, then it will hold that voltage as long as the battery voltage is at least that high. So it should be able to run any bulb between 6V and 30V provided it doesn't draw more than ~12A (I'm guessing there, but testing so far looks promising). I can't make a whole bunch of different versions. So everyone is going to be the baddest thing I can package into a 30mm round board. The WA1185 bulb will be driven by the same unit that drives the 64623. Just set differently.
The low voltage cut off is possible, but it would have to be implemented via software programming. There just isn't enough room on the board for 2 pots AND a programming header.
All but one of the components in the new design, thus far, is included in the PhD Proto board. It's just the FET driver that's different. I'm going to pick a couple up and teset it out. The driver on the proto board can pass 9A, the one I've selected for the D Mag, can only do 2A. So it may affect switching speeds, but not much. The switching transients are already like a wall on the scope. Testing will tell the tale.

Regarding the future. I'm already cursing the 30mm board for its size. The ones I build for the Lantern bodied lights will be chock full of stuff. Low voltage cut-off, multiple settings, adjustable via a pot (Set max regulated voltage via a pot, and the 3 settings will be something like 75%, 88%, and 100% of set voltage or whatever). I even found a chip that can do per cell voltage monitoring for up to 10 li-ion cells. And it will have a nice beefy driver than can handle dual or quad or more FETs and still drive it properly. 100 Amps would be manageable. They'll be much smaller runs and thus more expensive, but that's the way things go.

But keep the ideas coming, guys.

 

[LuxLuthor]

Jimmy,

I think there are only a few that would be willing to take all the steps to make >250W work. There are issues of reflector size, host, heat, battery source/charging, practical illumination, run time, etc. that limit what is worth doing.

Beyond the 150W you are using over 12A; on high current--NiMH or safe chemistry 18650, run time sucks. Actually, above 50W run time sucks. I guess what I am saying is for the developments/testing to satisfy regulation >30V is hard to justify.

For some the POT solution ala AWR's HotDriver example is the only thing that would be workable. When I spoke with him, most people ONLY wanted a set, programmed driver that was setup for a particular bulb, and didn't want to even understand his simple POT adjustments.

For others having the program header would allow much more adaptability & flexibility. You don't need a POT (or 2 POTs) if you are willing to set program instructions...as long as the instructions on what/how to change is made simplistic "paint by numbers." I think many would be happier with program header.

Perhaps having both choices available (POTs or Header) is the wisest?

 

[donn_]

I'm gonna pipe in on this thread as well, although I already listed this "wish" on Will's M6 thread.

I'd like to see this regulator built into a battery sleeve, rather than a light or a battery pack.

The reason is my choice of cells. I use A123s as much as possible, in both sizes.

I'd like to see a programmable battery sleeve system, which can be made to fit a variety of M@G tube-lengths, and can regulate the lower voltage A123 cell format. I'd like it to take 26650 A123 cells or Emoli 26670s.

In D-M@G tubes, it would be a sleeve system.

In C-M@G tubes it would be positioned at either the front or rear of a stack of cells.

In fatter M@G tubes, it would be an insert which would interface with cell packs/holders of 18650 A123s.

 

[JimmyM]

I'm gonna pipe in on this thread as well, although I already listed this "wish" on Will's M6 thread.

I'd like to see this regulator built into a battery sleeve, rather than a light or a battery pack.

The reason is my choice of cells. I use A123s as much as possible, in both sizes.

I'd like to see a programmable battery sleeve system, which can be made to fit a variety of M@G tube-lengths, and can regulate the lower voltage A123 cell format. I'd like it to take 26650 A123 cells or Emoli 26670s.

In D-M@G tubes, it would be a sleeve system.

In C-M@G tubes it would be positioned at either the front or rear of a stack of cells.

In fatter M@G tubes, it would be an insert which would interface with cell packs/holders of 18650 A123s.

A regulated battery holder. That would require a lot of fab work that I just don't have the equipment or knowhow to do. I've mentioned, in passing, to Lux, that he and I might do a kind of "project". He builds packs, I build the regulators. He lives very closeby. The downside of building it into a pack, is that either it's ALWAYS on or you have to remove the pack to slide a tiny switch to shut it down. And you would still have to contend with the stock switch components.

 

[JimmyM]

I just ran some tests with a 64623. 15.7V RMS output from an 18.1 V source. The FET barely warmed up at all. No heat sink. 9.8 amps RMS measured.
No problem. I'll have to build a parallel bulb rig to more heavy load the FET.
So it looks like I've found my FET for the driver.
Now if we can work out the details.
My first instinct would be to leave out the low bat shudown, keep the programming header.
I'm still on the fence regarding the pot.

 

[donn_]

The ultimate would not be a pack, but a holder.

Imagine a delrin sleeve in segments; A front endcap to interface with the light source, and a tail endcap to interface with the tail-spring.

Build the regulation into either the front or rear endcap, whichever works better.

Stick as many segments as needed in between the two, to hold the cells.