Sold/Expired Feeler - Programmable Hotwire Regulated Driver drop-in for D-M@g

:sigh: :thinking: :shakehead

Today seems to be one of those days.

Tonite I installed the first article regulator into a 3D M@g and loaded a FM 9xAA NiMH holder in one end. In the other end went a WA1164 bulb.

I checked the bulb on my power supply and it put out a nice orange glow at 1.5 volts, quite a reasonable amount of light at 2.0 volts and it max'd out the power supply at about 3 amps and 6 volts. Nice white output even there.

When I installed the bulb into the socket and tested the assembled flashlight I noted two things. One is the socket's grip on the bulb. It has less tension than the Kiu, and it doesn't hold the bulb as straight - the bulb can tip. So I think the Osram socket is a bust, and Starlight deserves a gold star for his socket grip test suggestion. I didn't expect that much variation in the socket, the Osram appears to be very high quality.

The other problem is the bulb doesn't light up. That has to be my problem, as the meter measures output when the bulb is removed. I suppose it could be the socket, but I will have to diagnose that later, I'm out of time and energy today.

So I think we need to find another socket. I have a couple of different ones coming to test, so we'll have to see if any of those turn up good. Folks who want a Kiu socket should make sure they have one available that we can use. Hopefully we will find something better, but just in case..

On other fronts, my chosen FET is fading from availability so I'll have to choose another. I found one that is a little smaller physically and has adequate specs. I'm going to have to get some and test them and then buy quantity as these parts keep changing.

One other side thing I've been evaluating is converting the design to (mostly) surface mount. I'm not sure I will use it, but I have converted much of it to surface mount (really a hybrid). It is not quite finished, and not all of the parts will fit in surface mount, but it might help reduce costs a little, and it makes things fit a little better. Not sure if it is worthwhile since it will add another test cycle which will consume much of the savings.

:candle:

Alan B said:

When I installed the bulb into the socket and tested the assembled flashlight I noted two things. One is the socket's grip on the bulb. It has less tension than the Kiu, and it doesn't hold the bulb as straight - the bulb can tip. So I think the Osram socket is a bust, and Starlight deserves a gold star for his socket grip test suggestion. I didn't expect that much variation in the socket, the Osram appears to be very high quality.

Click to expand...

Final setup is a thing of beauty!!!

Alan, I got a dozen of those sockets, and besides the fact that they are genuine Osram brand, they have a "U" shaped holder inside the hole that clamp the pin from both sides, unlike KIU's which is a scissors type style that once you push it over with larger bulb, it does not spring back to hold smaller bulbs.

Thumbnails



I stuck a WA 1164 and it gripped nicely in both holes. Then I inserted larger Osram bipin in one of my samples a dozen times, trying to loosen it up like what happens with KIU. It did not loosen up. I went back to the 1164, and it again gripped the same as originally.

I suspect that you may have one that started a bit looser, but it is possible to move the wire underneath slightly to one side, and expose one side of "U" arm. I was able to push a heavy pin or jewelers small blade screwdriver blade outside of arm and bend it slightly more inward, then repeat with other "U" arm, so it gripped even tighter.

Again, it's up to you and we don't want to get hung up on something as silly as a bulb holder, but I never found a way to get the KIU slot back to holding a WA bulb once enlarged.

LOL....well before I go to bed, figured what the hell, I'll take these apart and see what's up between them, so now you can see that KIU actually put a small segment of spring to help keep tension on leafs, and it is also a "U" style. So, the same could be done with the Osram. I ground down the bottom flanged a bit with dremel pointed grinder, and it then pushed up easily....and could be shoved back down and anchored again. Well now we know.

Excellent work!

This shows why the bulb can 'tip' with the Osram. It isn't significantly constrained against 'rocking'. With the Kiu the pin is held along its length.

The contact area is also quite different. With the Osram it is two tiny spots, with the Kiu it is two lines. The spring in the Kiu is also quite important to maintaining good contact. It is definitely going to produce lower resistance. This is not important at 3 amps but at 12 amps it is very important.

Fascinating! Looks to me as though the Kiu is custom made. Thanks for taking them apart and making good photos.

One other comment I forgot to add last night is the rubber button cover fits really nicely on the new regulator sled. It is almost hard to get out, it fits so well. And the clicky pushbutton has a nice feel to it. Much more positive and short-throw than a M@g switch. This is one area the first prototype did not test since it did not support the rubber cover.

Great work Lux on the socket research, thank you.

During my modeling phase I also took apart a Kiu socket and noticed this spring feature. I thought the spring tensioned clips on Kiu's socket would be a standard feature for any bulb socket because of the design intent. Any chance Osram sells them this way?

Alan, PM sent.

At the end of the day if it adds a few dollars to the cost to get Kiu sockets then I don't think that would be the end of the world. I have 2 of the AW soft starters and I really like them, the Kiu socket works well and I just stick to various Osram bulbs with the same size pins.

I have my brightest light working now - 5 x IMR 26500 cells running a 64458 bulb. I can play with that until my programmable hotwire regulated driver is ready and then I can run my Elephant II with 8 x IMR 18650 cells, letting me program in how hard I push the bulb.

What is needed to be able to program the driver? Do we get everything we need (except the PC)? If not then can Alan supply everything (USB connection, software, etc) and at what cost?

People who want Kiu sockets should purchase them direct from Kiu and supply them to me when we are ready. Kiu won't sell them to me directly.

Programming requires:

1) programming interface dongle, USB to ISP, available from Digikey, Mouser, etc, about $35

2) DVM to calibrate the readings, required when changing voltage ranges

3) programming adapter to reach into the regulator board in-situ, will be available from me, about $10

4) free software, from net

5) sourcecode and procedures, from net

I could possibly put together a cd of the stuff, if folks want to avoid downloading from the net. Half of it comes on a cd with the programmer, so I could do just the other half. Not sure I can supply the programmer software anyway due to license restrictions.

I don't plan to sell the programmers, though some special arrangement might be possible.

Looks like a programming setup would be about $50. It can program any number of regulators, and many other things that use Atmel AVR ISP programming.

What would folks prefer?

Alan B said:

...
The aluminum sled cost goes down by about a third at 200 vs 100 quantity. It is the most expensive part in the unit by far. However it is a large capital investment with notable risk to purchase 200 sleds....

Click to expand...

I can't imagine you won't eventually sell 200 of these. (Or maybe just 200 total sleds, by offering them unstuffed too.) But I see it might take a longer than you want to tie up your funds.

I bet you could line up "investors" to share the risk of an extended pay-back period. On the other hand, you'd still be in the middle and the extra accounting might be more hassle than help for you.

Let us know if you'd like to explore the idea.

Alan B said:

...What would folks prefer?

Click to expand...

I'm sure there are a number of solutions that would be fine for me. But, I'd suggest you pick a programmer, vendor, etc. and generate a reference shopping list. That would keep you out of the loop and I can't imagine anyone capable of reprogramming the driver can't download some software and place an online order.

If we order a KIU kit, does that contain enough parts for you to build two drivers, or just one?

A Kiu kit includes one socket, but he will sell a spare socket with a kit so you can get a second socket at the same time.

You guys can get the programming software, USB<==>Chip Dingle-Dongle from Digikey at this link. (or if you prefer, it is on this Digikey page as part #ATAVRISP2) Here is what you get for $34, although the latest software on disk is also at their website (link is in letter below).



Note: All of this information and questions are already addressed on Alan's first post, or one of the links therein, including the FAQ link.

Alan B said:

I don't plan to sell the programmers, though some special arrangement might be possible.

Looks like a programming setup would be about $50. It can program any number of regulators, and many other things that use Atmel AVR ISP programming.

What would folks prefer?

Click to expand...

If there were an option to get all the required hardware in one package from you then that would be my preferred option, I don't mind downloading some software if it is available for free from the internet.

If I were to order the USB-chip dongle thingy myself then it would cost $34 + over $20 for the international freight. Getting everything in one package from you would save me a significant amount of international P&P cost.

KiwiMark said:

If there were an option to get all the required hardware in one package from you then that would be my preferred option, I don't mind downloading some software if it is available for free from the internet.

If I were to order the USB-chip dongle thingy myself then it would cost $34 + over $20 for the international freight. Getting everything in one package from you would save me a significant amount of international P&P cost.

Click to expand...

I will have to check into this. Also, the programmer may be available from a distributor in your country. I may be able to help out if not.

Status Update - The software version that I tried first in the new article regulator was the problem, I went to an earlier version and it worked. I need to finish the newer version as it has the support for the six voltage ranges.

During this process I was also reminded that testing with a current limited power supply and a light bulb load is problematic. The rapid on-off pulses can confuse the current limiting circuit in the power supply. It works fine with a meter, but connect the load and the setup appears to have problems or quit working. This will have to be documented in the manual. Adding a large capacitor across the supply to absorb these current pulses can solve the problem.

The multivoltage range version of the software will need to be completed and debugged before the regulator will operate as designed. At this point I believe the regulator hardware is working normally but I cannot fully test it until the software is complete.

I made some progress on the software and on the new hybrid surface mount PCB layout. The new board is looking very good and I'm planning to get a few of each on the next prototype run. I also updated the programming adapter and the switch board. There is a little more to do on the boards before the next prototype run.

I'm going RV camping this weekend, and I don't know if I will have internet, but I am taking enough stuff to work on debugging the software so hopefully the First Article regulator will be operational by the end of the weekend.

Success - we have ignition, er, light. The problem was the FET drive protection code was a bit too agressive, it was shutting down during the bulb startup on the WA1164 for some reason. I turned the gain down on this loop and it operates normally.

This allowed me to take the next step and test the bulb position with the standard spacers and FM reflector. This produced a great focus at 1 turn out. So again, we have the excellent modelling effort by Nanomiser paying off - things line up as designed. I wanted best focus to come a little out from fully tight, and it does. Excellent! :thumbsup::thumbsup:

Incidentally, we're RV camping, so this is a nice diversion.

I also have been working on the circuit boards and have redesigned the programming board with a single switch since we aren't using the others.

So it looks like we can ship this test unit to LuxLuthor in the next day or so. Shucks I'm going to have to solder up another one so I can have one too...

I'll have to test the regulator a bit more before shipping it off. Make sure it works in the dark. :naughty:

Have a great afternoon, I can't wait for ... dark.

Great job Alan! :twothumbs

Can hardly wait to see some pics / beam shots. :naughty:

Beamshots of a regulator. Hmmm.

Tonite I re-ranged it for 2S LiMN. Worked fine. So I've checked it on 2S and 4S ranges so far.

I made up a newer longer programming adapter. The original one was a bit short but did work. This one has room to spare.

I took some new photos but have not prepared them for upload yet. The Eye-Fi is transferring them from the camera to the laptop as I type this, via WiFi. Very nifty SD card with integrated WiFi.

Sweet!

Found one more bug. Darn.

Double-click to maximum voltage quit working. Strange.

Debugging led me to a subtle integer math problem that crept in when fast-start was added. It only showed up when operating very close to the max with the ADC and VBULB which happens when you use 2S settings. Slept on the problem, realized what was going on and fixed it this morning. Too late to ship it today as I've got to get ready for the carpool, and I should test it a bit more this evening and make sure nothing else has crept in with the new changes. As I type this the output is 7.300 plus or minus a few millivolts (no load). The 2K uP is 57% full with pretty close to all features turned on.

Take your time, better to find bugs like this since you are so familiar with software.