GDuP progress thread.

[dat2zip]

The GDuP is a microprocessor controlled buck/boost converter for a single LED driver application. The primary use of this converter board is for the Aleph Light engines in any single cell flashlight (primary and rechargeable).

Here are some pictures of the converter board.

Top side view of the converter board.

Bottom view of the converter board.

Completed assembly. GDuP and anode board.

Converter board assembly view in the EScan.
NOTE: The taller inductor version does not fit in the ESCan. Only the shorter inductor.

Questions and links.

Does the GDuP have reverse battery protection? No

When will the converter boards be available? After the blackout period. Initially only light engines and completed lights will be available to recoup the majority of the NRE costs. After that the blackout period will be lifted and the converter boards will be put on the shoppe.

What will be the price of the GDuP converter board? TBD

What light engines will be available? There will be the GDuP Cree light engine and the GDuP Seoul light engine.

When will we see completed light engines? Status and updates will be posted as new posts on this thread and this portion will be udated as needed. Currently it's a rough guess that Seoul LEs will be available sometime this month (April 2008) and Cree LEs to follow.

How does the UI work? See 2nd post. (Updated 2010)






From the original post...
======================================================

Here is an early picture of a functional prototype

Wayne

 

[dat2zip]

A brief description of how the On state UI works. (Auto-calibration UI update at the bottom of this post)

Turning on the light on and then quickly off will cycle the light to the next level when it is turned on the next time. We will call this "tapping".

Tapping the light will cycle to the next level the next time it is turned on. Internally a counter is incremented. If the internal counter reaches 15 the next time the light is turned on it will calibrate the light.

The internal counter is retained even when the light is off. There are two ways to reset the counter.

1) Turn the light on for more than a second.

2) Calibrate the light.

CAUTION: It would at first appear that you would rarely get the internal counter to numbers higher than 10. But, a casual user can easily get over 10.

One way is to play with the light. Cycle it three times, grin with happiness the light works as stated. Set the light down. Now the internal counter is at three and has not been reset. Later that day you pick the light up and do it again. Happily the light works and you again grin with happiness and fiddle with the light off. Moments later you repeat this process. Each time you cycle three times the counter continues to increment. Eventually the light appears broken and starts flashing and no longer cycles. Leaving the light on for more than second doesn't seem to do anything.

You have now started the calibration sequence with the internal counter having a value of 15 or more.


Once you get into calibration you can do one of two things:

1) Let it calibrate. (takes a few minutes) Successful completion is with the light ramping up/down 5 times and then going into low.

2) after turning it on turn it off after it stops ramping up/down.

The light ramps up/down at the beginning of calibration and the end of calibration. The internal counter is reset right after the first set of ramp up/down and can be turned off without affecting the current calibrated levels.

A fresh battery is recommended if you want to perform calbration and there is no harm to run calbration. Calibration is done when the light or LE is assembled and there should never be a need for re-calibration.

The initial ramp up/down in brightness may appear as a blink or may to others be termed or phrased differently. If you look closely the light intensity varies. This is the ramp up/down and it should do this 5 times.

Wayne

Update:

New software was implemented in 2010. High output option available also.

How GDuP Auto-cal works:

When you turn the light on it will determine in 1 second if the current level is within specifications. If it is not within specifications it will blip quickly once a 1 second. This is the indicator it not calibrated. It will automatically start a re-cal procedure and if successful ( a few seconds to many seconds later) it will blip quickly twice. Once it blips twice it signifies it cal'd correctly and stored it away.

You can now cycle to the next level and wait to see if it auto-cals or not. You need to do this once and it will do this periodically with LED aging.

But, it's automatic. Just use the light. It should be correct after a few usages.

 

[Robban]

If I may give a suggestion I would reduce the timeout delay. Kai (from kaidomain) recently had a poll asking users how long they wanted the light to stay on before remembering the last state and the support for 5 seconds was pretty much non existant:
http://www.cpfmarketplace.com/mp/showthread.php?t=166388

Unless I'm mis-reading your explanation in which case I'll just shut up and stand in the corner 😛

 

[dat2zip]

If I may give a suggestion I would reduce the timeout delay. Kai (from kaidomain) recently had a poll asking users how long they wanted the light to stay on before remembering the last state and the support for 5 seconds was pretty much non existant:
http://www.cpfmarketplace.com/mp/showthread.php?t=166388

Unless I'm mis-reading your explanation in which case I'll just shut up and stand in the corner 😛

I'm not clear if that is the same thing or not.

Let's say you tap around a couple times and when you get to medium output level you decide that's the level you want for now. With a standard clicky like the McClicky switch you'd depress the switch all the way down and engage the switch. Now it's latched in the on state.

If you decide this is not the level you want, you have up to the 5 second window to turn it off to cycle to the next level. That means a quick pulse or a long click will both cycle the light. Setting it to 1 second will for many who are not adept or coordinated could have trouble toggling levels. Setting the timeout to 5 seconds allows even the slowest user to still cycle through the different levels.

Does that make sense? Do people want it less? If people use this with a reverse clickie does it need to be different?

If there is a need for it to be different I would be happy to make that change.

5 seconds is the timeout that is on the Shark/Remora. I've not heard of any issues with that UI.

Wayne

 

[jch79]

Wayne - this is awesome. The timing is perfect, IMHO. :thumbsup:

john

 

[dat2zip]

OK, I'm starting to get the picture.

The timeout period actually determines the minimum usage time. Anything less than the timeout period will cycle the light.

So, If you were wanting to use the light momentarily say to light the bookshelf up to find a book and then you turned the light off and the timeout period was 5 seconds, you would have cycled the counter and the next time you turn on the light it would be in the next state.

I'm in agreement that 5 seconds is probably too long for a hand held light. Wasn't a problem with the Makita lights I've built to date since when I turn them on they typically are on for several minutes minimum.

I will set the timeout to 1 second and do some testing.

Would there be interest in less than a second? I know most people could rapid fire and a timeout of a 1/2 second is probably preferred for the advanced users.

Wayne

 

[cmacclel]

Wayne if you want to send / sell me one I'd be happy to test one out 🙂


Mac

 

[dat2zip]

The code is all done and tested. There is only one minor annoyance that I've minimized, but, not completely eliminated.

The annoyance is the initial turn on has a flash of initial brightness and then the regulated level of light. I will continue to look into this, but, in the meantime I've got enough info to order the next revisions of PCBs and make minor corrections.

The total stack height right now is 7.2mm and I'll be reducing the stack to ~5mm which will then fit in the standard ESCAN and the XRCAN making upgrades to any Aleph light (single cell) possible. That would include the A1, A2, A3, A19 and any future new heads that might be brewing.

A 0.1 ohm sense resistor and the target values are 3mA, 80mA, and 400mA of which I haven't decided on the final values. For now it should be good enough to start getting some air time in an actual light and get some UI testing done.

Changing the sense resistor to 0.05 should double all the 0.1 ohm values and should cal the high at around 800mA.

One could consider the 0.05 version the standard and the other the mizer version. Of course one could use a larger sense resistor and cut the high down even further.

Other options are to modify the target values in the firmware to generate any custom levels desired.

The main converter board top view.

The main converter board bottom view.

Wayne

 

[kenster]

Wow! You are moving right along with this driver so can I place my order now? OK, next week. 😀 I`ve been wanting a driver like this for a long time and you couldn`t have nailed its features better. You didn`t mentioned changing the timeout setting at 1 second and giving that a test run. That should be much better than 5 seconds IMHO. Can`t see the sense resistor in the pics but I hope it is easy to get to for those that SUCK at the soldering deal. Like me. 😱 Could a person could piggy back solder on the 0.1 ohm with a 0.18 ohm to further customize the drive levels?

Looking forward to the GDuP debut on the Shoppe Website. :rock:

Ken

 

[McGizmo]

This converter is going to open some nice doors!! :thumbsup: There will be some fresh air for some lights presently collecting dust as well! Very cool!! 😀

 

[dat2zip]

Actually, there are a lot of enhancements I can look into. One of many will be to extend the range to approximately 1A and still retain low at around 3mA or so.

After many steps backwards today I finally got one unit into a XRcan with a Cree Q4 LED. While it still retains the initial flash the light is very useable and the UI is very easy and quite stable.

That's a good milestone... I'll make the PCB changes necessary and order another round of boards this weekend. Now I need to get back to the other jobs.

I couldn't resist. On this board I turbo'd it with a 0.05 ohm resistor. It's cal'd for the LED on the bench. Now let's see if I can enable cal when it's in the light. High is 800mA and low is around 20mA. Quite useful.

Wayne

 

[dat2zip]

Wow! You are moving right along with this driver so can I place my order now? OK, next week. 😀 I`ve been wanting a driver like this for a long time and you couldn`t have nailed its features better. You didn`t mentioned changing the timeout setting at 1 second and giving that a test run. That should be much better than 5 seconds IMHO. Can`t see the sense resistor in the pics but I hope it is easy to get to for those that SUCK at the soldering deal. Like me. 😱 Could a person could piggy back solder on the 0.1 ohm with a 0.18 ohm to further customize the drive levels?

Looking forward to the GDuP debut on the Shoppe Website. :rock:

Ken

I've been playing with the timeout and 1 second will more than likely be fine for most users. I've upped it to 1.5 seconds on this proto and it's actually somewhat annoying as I can accidentally cycle since I didn't keep it on long enough to keep the current setting.

5 seconds is definitely out on this driver.

Wayne

 

[jch79]

I love watching the progres & the process how you do things Wayne - it's great to see! :thumbsup:

Lookin' forward to it! This coincides nicely with my newfound LE assembling! 🙂

john

 

[dat2zip]

The two board proto had me bummed out big time. I was watching low regulate 3mA to the LED and 140mA from the power supply. That's not good efficiency. More like the two stage tailcap draw off the battery.

It bummed me out even more when I assembled the first converter board. It too did the same thing. That was extremely depressing.

Even though the converter is wired up in voltage regulation mode that initial startup flash I believe is overshooting (causing the flash) and when there is no load killing the converter board. It took me several days to identify what was wrong (internally shorted FET) I replaced the converter IC on both units and that fixed that issue.

The converter IC has good efficiency at low currents. Even regulating 1mA the converter is 70% efficient. Add another 1/2mA or so for the processor.

So, it's not safe to have no load and power up this unit.

Wayne

 

[dat2zip]

A rough schedule would be the following.

Order next revision converter board and anode board. ~1.5 weeks.

Assemble/test 1-3 days.

Those boards would be discrete boards. They would not be on panels.

At this point a small run of hand soldered boards could be made generating a preliminary preview or pilot run. Several factors will have to be considered to do this.

Assuming all is well the boards could then go into the production process with quoting, ordering and processing through a local CM. That takes 3-6 weeks and this is where I'll need a few $$$ to commit to.

After that the boards would be unprogrammed and they would have to go through programming. I'm leaning towards this instead of pre-programming them as it gives better flexibility to make different versions and to migrate to a newer revision with no scrappage of boards.

Looking at the calendar, That would put the next prototypes end of this month (sept) and possible production rollout Mid November assuming no hiccups or delays.

That's the ideal scenario and the chance of that happening... I'd give it a good confidence factor of 8 on a scale of 1 to 10 (1 bad, 10 good).


That's the plan of which it could change at any time.

Oh, I forgot. The stackup needs to shrink. The three things that will be shrinking will be the pcb for both the converter board and anode board from 1/16" to 1/32" thick FR4 and lastly the inductor needs to be reduced in height ~1mm. Right now I'm trying to source a thinner inductor and that could be a serious hiccup as lead times are 8-12 weeks for most inductors.

 

[kenster]

You can deffinately drop the timeout under a second and work everything fine. I would prefer it real short for the timeout but some might not be able to operate it and get frustrated? That`s no good. You should try it if it is easy to change. Also try using the tailcap as a twisty to operate it. Takes a little more time for most people to operate a driver like this using the twisting action. I know I have some twisty lights that could use a driver upgrade.

Mid November? Sounds great and right before my B-Day too! 😀


John, I am deffinately enjoying watching Waynes GDuP development and progress reports as well. :twothumbs

 

[Icebreak]

I couldn't resist. On this board I turbo'd it with a 0.05 ohm resistor. It's cal'd for the LED on the bench. Now let's see if I can enable cal when it's in the light. High is 800mA and low is around 20mA. Quite useful.

Wayne

Nice. I think a lot of guys will like to have a little GiDdy uP in thier flashlights.

- Jeff

 

[dat2zip]

It took quite a while to re-arrange the changes on the board. The board density is high, real estate minimal. It took me Sunday and all Wed to make all the changes.

I have been trying to complete another job and when I ship the these 20 boards I will have more time to focus on the GDuP and the next converter board (Hammerhead).

I'm trying to whittle down my parallel tasks to a smaller subset and hopefully maintain the smaller subset going forward. I do need breaks and I'm running myself into the ground at this run rate level.

Here is a picture of the 20 boards. They are almost 100% loaded. I still need to put the ICs on and run them through the functional test fixture and troubleshoot any that don't work. I expect another full week before I can ship these to the customer. (best case). It might take longer.

Today is Friday and late today I did fire off the main converter board to get made. It usually takes a week, sometimes less sometimes more. When the new converter board comes in I hope to have a new programming fixture for it and will be able to load and test the same day it arrives.

I don't have any pretty pictures to post.

Wayne

 

[jch79]

:wave: :candle:

My SS order is on hold until these suckers are available! I'm psyched!

:thumbsup: john

 

[dat2zip]

The 2nd revision raw board is now in transit and looks like it's in Kentucky right now. They might show up Monday, but, everything is kinda on hold with two other higher priority jobs.

I'm fixing a hole in my patio before the rain season comes upon us and that is on the top of my todo list. I also need to order more converter ICs or find some loose stock hidden somewhere. I can't seem to find any at the moment around the house and I won't be able to test them until I get more.

I bought an open case of 10CC syringes on Ebay and those might show up on Monday as well. I seem to have an affinity in collecting these. I do use them for lots of things and the 5CC is too small for many of needs I have.

I'll update here on the progress (forwards or backwards) when more is available.

Wayne