GDuP progress thread.

[THE_dAY]

thanks Wayne,

looking forward to the new driver... and the **CENSORED**!

 

[dat2zip]

Unofficial low runtime. At ~23 hours (midnight) the light stopped working. I took it to the emergency room and all the elves were not to be found. I had to do my own surgury. After xrays, CT scan the doctor found the anode board had popped away from the XRCAN. It was touch and go with beads of sweat over a hot plate, but, it's alive again!

I dug into what I thought was an old pile of proto raw boards and I found the container of anode boards. Yippie. The other two boards died yet again for other reason. I replaced one for the third time and it appears to visually look good and I can program it and calibrate it. It doesn't hold the regulated values for some reason. It works fine for while and then it cycles to new and completely bogus values after that. I might have to wait for the batch to come back from the CM before I can make any more LEs. For now I'll put this one through the paces in medium and high.

Wayne

 

[mosport]

Hi Wayne,

Wondering if you've had any progress fixing the regulated value problem?

Hope all is going well and looking forward to leftover halloween candies with my next order

 

[dat2zip]

Hi Wayne,

Wondering if you've had any progress fixing the regulated value problem?

Hope all is going well and looking forward to leftover halloween candies with my next order

After my last post I spent a long time investigating the low and how it behaves. I tweeked the computed low control value and it turns out that it has no effect. I isolated the problem to the high side sense ic and the way it works is what is limiting the low end. At low sense voltages the sense ic generates no output voltage. It just stops. This limits the control range to approximately 100:1. With the default 0.1 ohm sense resistor the bottom end is around 10mA (approximately).

That puts a kabosh on the 500:1 or so I was shooting for. Another thing I spent a day and several restless nights was the high end. The LED current feedback to the uP is again limited by the sense IC in the fact that it runs off the LED voltage and thus is limited on how high it's output voltage can swing by the ic supply voltage as an upper bounds. We run into the upper bound by the gain of the ic which is 20. If we look at the LED voltage at the high drive current and pick say 3.8V for Vf or supply to the sense IC and assume the output can swing to the rail - 1.0 or 2.8V we now know the upper bound.

2.8V output divided by 20 is 0.14V and 0.14 across a 0.1 ohm resistor is 1.4A.

So far soo good for a 20mA low and 1A high. But, for a 3mA low we need to increase the sense resistor almost a factor of 3 to scale the output current range.

Scaling down the 1.4A max down by three limits the top end to 466mA max and of course we need headroom so 400mA or so would be tops.

That might be ok. I need to do more testing to determine and refine these values.

===== Two working proto status update.

The first one I made had the initial flash startup and I don't use it much as the initial flash tends to destroy any night vision and makes low annoying to use in the very dark environments. I still play with this light once in a while to see if it's still working. I still love the three levels even if it does flash. Anyone want to put a bid on it? The LE or LE in a mule head I'll sell if someone is interested in it.

The second proto is in my pockets as a A19 or mule head depending on the mood of the day. The newer boards have inverted in the hardware design and has access to enabling the LED controlling the initial flash to have initial darkness. The second proto from a dead cold start (first time use) shows signs of a initial flash which is not near as bright as in the first proto. I believe I can adjust the uP startup delay to get rid of this. After the first turn on the light has no noticeable flash or startup issues.

Let me re-iterate. For you speed freaks who can type faster than a speeding bullet and can click a clickie faster than the switch is capable. Comcast is looking for people like you for their commercials... Seriously, this means that the light is initially dark or off for some 80-100mS from when you apply power. Some of you will notice this. It seems to have some lag feeling relative to other lights. The good news is you real fasties will be able to cycle to the next level if you can catch it during this period. It's theoretically possible to cycle without the light coming on. Can I do it? No.

Since this feature is configurable in software CPF should consider the ramification. Ponder this thought for a while. Maybe, we should make the initial off period a tad longer to take advantage of this. Or, making it longer detracts or adds to the lag irritating you more than it's worth. Please keep you comments shorter than my posts.:sick2:

===== Additional status

I ummm, what happened.... Uh, O yeah. Two weeks is a long time. I was working on another prototype on the bench and was burning it in leaving the light on in low overnight abt two weeks ago. In the morning I noticed it was full brightness and drawing some 1.7A from the supply. It was happily making a nice heater of the LED.

I did some investigation quickly since we were getting ready to go somewhere for the day and found the feedback resistor had a cold solder joint. I re-soldered the resistor and turned it back on low. When we got home late evening I noticed it was again on full brightness. I troubleshot this all the next day and determined the IC had wigged out for some reason. This particular board had the ic changed 3+ times and it was getting too frail for any additional surgery so it was scrapped.

I had plans on working on another fully loaded board that the ic had fried and have not done anything since. I had hoped to get back some 50 boards from the CM and was going to resume debugging/refining and isolating issues using boards from this batch.

I want to confirm the first turn on initial flash can be fixed in software and once that is confirmed do some testing on the useable range. I am currently thinking a mizer and regular version each having three levels of brightness.

Of course this is Halloween evening and I could just be posting a big lantern of story. Maybe, I should have added more spook, mystery and ghosts to the post.:green:

The CM has quoted the assembly for production which means I have the pipeline primed once we have determined all is well.

This means the 50+ converter boards will be standard thickness and I have mating anode boards of standard thickness and a large batch of the thinner anode boards. The production converter boards will be the thinner version and both thinner boards will be required to put this into the ESCan/ESScrew assembly. The only LE I plan on making with the 50 are XRCans and that pretty much limits the useage to Mule and A19.

Unless there is other feedback the current thought is to make a few LEs and get them out to a few Beta testers and then get feedback in a few days or so of useage. If nothing is found that needs changing the remaining 50 will be made into production LEs and these will be made available. I think they will be offered as mule heads since this is (IMHO) and awesome configuration. It also makes it sound like I had planned this offering only SOB mules and leaving the 1X123 configuration open for the GDuP as the primary offering for the 1X123 configuration.

Is there a post text size limit? LOL

OK, now I'm pooped. I need to get out to the garage and get some work done.

Wayne

 

[dat2zip]

Put in ~10 hours trying to resolve the startup issue. Tried many different things. Have not exhausted all options. Until I can come to terms on this issue I won't consider it good enough. I have received 1/2 of the 50 converter boards back from the CM and they are ready to go once I get the firmware completed.

Multitasking between building light engines for the Mule lights, working with Vendors on new parts, tooling. The Makita Dock is moving forward. I've ordered production parts.

Wayne

 

[dat2zip]

Crap... Just deleted my post. Gotta retype it all now.

I spent most of last night in simulations to balance the tradeoffs in the design and to come up with a reasonable solution. Once I got to that point that I was happy with I tested that configuration this morning and have accepted it as a viable solution.

The flash is still there. It's minimal now and no where near blinding as it was before.

I will be making preliminary light engines with this version of firmware and hardware and will be contacting the beta testers to get them the test units.

Wayne

 

[jch79]

:thumbsup: Great news! It's cool to read about all that goes into these tiny boards! Thanks Wayne.

john

 

[chimo]

Sounds great Wayne! Looks like a great product there!

Paul

 

[bluecrow76]

Can't wait! And where do beta testers sign up? :twothumbs PM me if there is room for more.

 

[dat2zip]

OK,

I have delayed assembling the tested converter boards only for the simple fact of the process in assembly is difficult and I had to come up with a fixture, jig or process that would be not only painless, but, make it as efficient as possible minimizing the effort to assemble the assembly.

The light engine assembly is already the most complex one to date and the microprocessor has a special fixture required just to program the board. This all leads to more steps to make one light engine. Not only have I be mulling over options and ideas I've been working on other fixtures for other projects and investigating solutions and thermal management for the Shark converter board.

I'm looking for three beta testers. They can either PM me their shipping addresses or email me direct. I'll select up to three people. Being a beta tester means reporting back here in this thread or starting a new one and giving their honest views and opinions.

Since the converter board is more or less cast in concrete in the light engine it will not be possible to re-program these with different options or variables base on feedback presented.

I encourage all members to participate and engage their opinions and views during the beta process as I'm hoping to make final tweeks into the production release code and get this program on the way.

Today is Tuesday evening. I'd like to try to ship these out this week if possible.

Wayne

 

[dat2zip]

Did I mention the Beta testers will keep the lights as part of being a beta tester! :thumbsup:

Wayne

 

[cmacclel]

I'm in for a beta tester. PM sent.

Mac

 

[jch79]

+1! :twothumbs

Email sent to the yahoo account! :thumbsup:

Thanks Wayne!

john

 

[PhotonFanatic]

Still would like to be a beta tester, so +3!

 

[griff]

I also would like to
be a beta tester, so +4

 

[dat2zip]

Just to clarify the Beta unit will be a complete Light Engine installed in a bare mule head, 1X123 bare aluminum, McTC bare tailcap with a McClickie installed. I would like the Beta testers have the capability to swap out the LE into other heads so I can get a feel for the LED drive levels the board is currently configured for so that we can get the low, med, high levels set.

If a set of drive currents are already known I'd like to see them posted as well.

There will be too many variables if the bare board was sent out.

I know the GDuP will work in many other things. This phase is only the complete light.

I need to think about this and open the discussion of the option of having just the board as part of the Beta testing. Those of you that were looking for this option might want to chime in with your thoughts to help the thought process out. I do have more boards and the next step was to get all the components changed to bring them current physically speaking. I wasn't planning on programming more with the current code until after the Beta feedback and make any code changes.

The board has some limitations as it stands and I have not decided on the exact route I will be taking as we move forward. Keep it like it is or re-design or what. The current design I could extrapolate the 2mA drive levels from empirical table values as one possible option. In order to do this will take approximately 1-2 weeks of characterization and developing a formula for computing and extrapolating levels I can't calibrate to.

Wayne

 

[McGizmo]

I haven't seen one of these yet either and I am excited at the potential!! :thumbsup:

 

[cmacclel]

I haven't seen one of these yet either and I am excited at the potential!! :thumbsup:

These new boards look very promising I'm also excited that there release is getting close!

Mac

 

[datiLED]

I would love to be a Beta tester for the GDuP!

 

[LED Zeppelin]

I would be honored to be a beta tester as well.
PM inbound.