The FrankenBebi Project

[CaptTom]

Originally Posted by capt-couillon

1. As stated by "Goboatingnow" the requirements are not necessarily valid outside the US Colregs lines.
2. The cost of testing by an outside laboratory and submission for approval of certification by the USCG is well outside the budget of this project. In addition, FrankenBebi lights constructed by kit, or open source parts list would not be covered even if certification were obtained for the "ready to use" version.
3. Using the values specified in CFR33 84.15 Table 84.15(b) and the absolute minimum bin output value at the 21.5mA drive level we are using will be in excess of 4 miles (27 candelas) Typical output at 21.5ma is closer to 32000mC per led. Beam-spread overlap at distances exceeding 200 meters should raise perceived intensity to the 45000mC range. In any case, this is twice the required range for recreational vessels up to 50 Meters. It is expected that recreational vessels in excess of 50 Meters will not be using the FrankenBebi

That's pretty much what I meant; just make sure you meet the requirements. I wasn't suggesting going for USCG certification at this point.

I believe you'll find that the US CFRs are taken from specifications found in international treaties, and the US lighting requirements (outside the demarcation lines) are pretty much the same as the rest of the world.

As for multi-sector lights, I agree that might be best to put off for later. As a power boat I'm not really interested in tri-color or dual lights. Just an all-around white, but with the ability to light the forward (steaming light) sector (225 degrees) separately from the aft (stern light) sector (135 degrees). That seems like it would be fairly simple.

[Celestialsailor]

Since these seem to be made from off the shelf, hardware store, plumbing items, Why not have a "free" set of plans with part numbers for the LED's and plumbing parts. Kind of a cruiser helping cruiser. I would imagine parts no more that $20.

[capt-couillon]

Originally Posted by Celestialsailor

Since these seem to be made from off the shelf, hardware store, plumbing items, Why not have a "free" set of plans with part numbers for the LED's and plumbing parts. Kind of a cruiser helping cruiser. I would imagine parts no more that $20.

We are trying to go that way as much as financially possible for the DIY cruiser (of which I count myself as one) There are costs however in a development project of this type, (parts, assembly equipment and materials, time we could use doing our own projects, etc). Rather than asking for donations to further current and future improvements in design we are hoping to recoup those costs (and maybe a little extra for the grouch-bag) from the "give me a well designed product at a reasonable price" crowd. The "DIY Kit" will include all the electronic components (LEDs Resistors, IC chips) you need to build a FrankenBebi. It will also include the printed circuit board we have developed, to make assembly much easier than a wire to wire "rats-nest". You get the hardware store parts, wire and potting compound locally. We will supply the parts at a cost less than what you would pay at the local Radio Shack, and the quality is the best we can find (No Shenzhen Market knock-offs). The final option is we will provide a full parts list with no schematics. If you are ambitious enough to go the full route, I'm sure with a parts list you can figure out how to wire it up. TANSTAAFL

As the design for the FrankenBebi has not been finalized, I can't give a final price on all the options. I would however expect the DIY Kit to be in the price range you mentioned ($20-$25).

Thanks for your interest and input.

[capt-couillon]

Originally Posted by capt-couillon

(Edit: It would appear Igor has fallen asleep at the switch, and the project webpage server is currently [Tues Morning] unavailable. We are trying to shake him awake to resolve the issue. Thanks for your patience)

At some point today while I was in the laboratory building a jig for the drill press, Igor woke up and got the server back on line. The FrakenBebi Project Page is back online.

[Celestialsailor]

It sounds great and if you can make a few bucks at it the same time...fantastic. I look forward to your kit. I know they use to use 15 LED's. I liked the one for a cigarette lighter. I always thought the Bebi was a great design.

[capt-couillon]

Michael and Kendra had a great (and reliable) design. We are fortunate to have it to use as a basis, with the latest technologies. The 15 LED design works very well on a number of levels. Viewing angle overlap, as well as an optimal array layout for nominal 12 V systems. We will be including the 12V "lighter plug" as an option on both the DIY KIT and the ready to use FrankenBebi. Personally, I prefer a little more watertight connector above decks, but the option is get a bare-wire version and add your connector of choice.

Speaking of design (off-topic) today would have been the 120th birthday of my Hero

TKs again for your interest and comments.

[capt-couillon]

News Update for the FrankenBebi Project

[GordMay]

Are the “... results of test at this point (are) certainly exceptable ..."
“acceptable” or are they “exceptional” ...?

[capt-couillon]

Originally Posted by GordMay

Are the “... results of test at this point (are) certainly exceptable ..."
“acceptable” or are they “exceptional” ...?

That would be acceptable.
Thanks for the catch Gord... Igor don't spell so good, but is fluent in idiot.
We took away his breakfast, and he has made the correction.

[capt-couillon]

New news at the FrankenBebi Project website.

The site has been to moved to a new server which should eliminate the earlier problems we had during periods of high traffic.
We now have an official "webmaster" who has improved the look and navigation as well.

The project continues to make progress. Igor has posted the latest news and updates online, as well as sending out the new newsletter later tonight. We have also posted a schematic of the latest (ver: 0.10.3) prototype under test including the photoswitch circuit. Other additions include a white-paper on the goals and design philosophy of the project. At this point, the project is still (for the most part) on schedule to meet our goal of distributing the beta-test versions in early December and production versions in early January 2014.

Applications for Beta-Testers have now been closed. Thanks to all who applied for your time and interest.

As always, we appreciate any and all input and suggestions either here or via email. Contact info is available on the site. We are hoping with the assistance of the new webmaster to be able to integrate some type of comments section on the site to allow interested individuals to kick around additional ideas based on the FrankenBebi design. Right now, all our energy (and funds) are focused on the original goal of a replacement for the original Owl, but we have had some great suggestions for future efforts.

Thanks again for your interest and support.

[B26Trimaran]

I showed a friend TD with years of knowledge with LEDs and electronic design and he had this comment:

"Interesting approach, but I would have used 3 resistors to limit the LED current, not one.* If for any reason one LED craps out the remaining strings of three will be driven at more than their rated current.* 2 more resistors cost a couple of cents.* The photo switch should be running on the regulated power, as variations in supply voltage will cause shifts in the* brightness level at which it switches.* I suspect that a CMOS Shmidt Trigger IC would be able to be used with a much lower standby power draw typically millionths of an ampere."

Cheers
B26

[capt-couillon]

I am not the electronics guru on this project, but as the point guy I will answer to the best of my knowledge. I will pass your friends comments on to the "wizard" for him to chew on....

Interesting approach, but I would have used 3 resistors to limit the LED current, not one.* If for any reason one LED craps out the remaining strings of three will be driven at more than their rated current.* 2 more resistors cost a couple of cents.

In order to use resistors as current control devices, it would require 5 (one for each series leg of the parallel array. Also not too efficient with the voltage swings seen in a marine battery bank. The single 12ohm resistor is being used to control the output of the LM317 which is a (simple) linear regulator. Output of the 317 is set to 104mA or 20.8mA per series leg. If we were to loose one led and hence its associated leg, the current would now rise to 26mA per leg. Since we are underdriving the leds to extend their life, this is still only 86% of max allowable current. Fact is, if you loose 20% of the leds you should return it for replacement anyway as your 360 visibility sector is now impaired. The latest design has the led legs connected in side by side configuration with each block of three covering ~48 degrees, not counting overlap. The next iteration of the PCB design will (if we are succesful) stagger the led connections so the loss of 3 leds would be overlapped by the adjoining good leds. No two adjoining leds would be on the same series leg. Reduced range in case of a failure, but no dark sectors.

The photo switch should be running on the regulated power, as variations in supply voltage will cause shifts in the brightness level at which it switches

According to our wizard.. this is not the case. However, the photoswitch is now under testing, which includes variations in battery voltage from 12V (25% SOC) to 15.5V (equalization cycles) During these tests we will watch for serious deviations in the light level vs switch point as suggested.

I suspect that a CMOS Shmidt Trigger IC would be able to be used with a much lower standby power draw typically millionths of an ampere.

I would love to be able to pass a schematic on to the wizard for such a circuit. A microamp draw at idle would be great. We will be happy at this point with 10mA as that is about as low as we can go with the LTR and even with the op-amp, still get a reliable gate voltage at the mosfet. Have your buddy sketch something up? Always need a good plan B, sometimes it works out better than plan A. Keep in mind the PCB is only 1.4 inches in diameter and designed for assembly with average DIY hand soldering skills. Thru-hole only, no surface mount components.

Thanks for your interest and comments. We will pass them along, and pay attention to the possible issues with the current design during the coming weeks tests.

[capt-couillon]

New news at the FrankenBebi Project

Prototype PCBs arrived Friday afternoon, and the first try at population and assembly into the case was successful, A few issues, but it would seem the basic design premise is sound.

Schematic and PCB rendering for the next iteration (ver: 0.11.03) have been posted. New version includes a blocking diode to prevent damage from reversed power leads, and some additional significant design changes.

Coming soon... A documents repository will be added to the webpages, containing all relevant files, including the Eagle
.sch and .brd files, .pdf copies of schematic and board layouts as well as licensing information. The project is being released as Open Source Hardware, under the CERN OHL v.1.2 Open Source Hardware License.

Thanks for your interest and support.

More info on the latest updates may be found at the project webpages

[Paul Elliott]

Sorry if this has already been discussed, but do you believe that the forward voltages of the LEDS will always be matched enough to provide adequate current-sharing among the five series-strings? I would typically add a small "ballast" resistor to each string, not for current limiting, but to ensure that the forward drops in each string are balanced. This will reduce efficiency and voltage overhead a small amount, but the resistor only has to provide a tenth of a volt or so drop. The value will depend on the specified Vf variation for the LEDs.

I am just leery of paralleling diodes or diode strings. One string having a diode with a slightly lower Vf will hog all the current. This also depends on the V/I curve of the LEDs. If it has gentle enough slope to it you may not need the series resistors.

[capt-couillon]

Evening Paul, thanks for your comments and interest.

While ballast resistors have been brought up several times, here are some of the factors that were considered prior to the decision not to use them.

The LEDs being used in the project are being supplied by a US based manufacturer (Cree) with a good reputation for maintaining the tolerances stated in their spec sheet. In addition, the product has been binned to reduce the variation of Vf. A random sample from our first batch order of LEDs (100) showed Vf variation was within specs, and fairly evenly distributed above and below the mean average of the sample. It is hoped that the additional component expense of high quality binned LEDs will be offset by the long term increase in reliability and reduced component failure rates. Further testing of the various prototype units has showed no serious variation in Vf (or If) amongst the 3 LED series strings in a given unit.

In our eyes a straight series circuit of 15 LEDs would have several disadvantages. The largest problem that presents itself is the failure of a single LED, will fail the entire unit. In a series/parallel array, while a single LED failure will also extinguish the other 2 LEDs in the leg, the rest of the array will continue to function. As we are only driving the LEDs at around 60% of max, a string failure will only increase the current to the remaining strings to 86% of max without over driving them. Another disadvantage is the fact a series of 15 LEDs would have a voltage drop well in excess of practical limits for a small, simple (and hence reliable) device.

The final consideration is the problem of stuffing 5 more components on an already crowded 1.4" diameter PCB. At this point we feel we are reaching the maximum density that will still allow assembly by an individual with average soldering skills and equipment. The density of a board containing only the basic driver circuit is not a problem, but the addition of the photoswitch circuit increases density to what we feel is its upper limit. The economics of PCB fabrication does not allow us to produce individual boards for each possible option so we have to go with a single board design that will cover all possible variations.

I hope this explanation helps... No matter how many times it is discussed, a potential problem will generate potential solutions. Many of our current and future design considerations have and will be based on comments from individuals not directly involved in the Project. This is the great advantage of Open Source Hardware. We certainly don't know it all, and surely have not considered every possible issue (yet). If one feels the design can be improved, and we are not willing to make the changes, you have all the tools and basic work to design your own variation freely available.

One quick example of a question that spurred a major design change concerns the failure of a single 3 LED leg. The question was specifically about overdriving the remaining legs (which we had already considered) but brought the insight that with the current PCB layout all 3 LEDs would be side by side with a failure causing a dark sector of visibility over 48 degrees. The latest iteration of the PCB design distributes each of the 3 LEDs in a leg in such a manner that no LEDs in the same leg adjoin each other. Beam spread overlap from the remaining LEDs will cover the dark areas with no loss in 360 degree coverage. A further discussion of this design change is covered on the PCB page, but a good example of how a previously discussed problem generated a solution to a previously unexplored issue.

Please feel free to contact us/me at any time with questions and suggestions. Either via e-mail, this forum, or PM...