Path: csiph.com!eternal-september.org!feeder.eternal-september.org!nntp.eternal-september.org!.POSTED!not-for-mail From: "Don" Newsgroups: sci.electronics.design Subject: Re: CoB LED filament analysis Date: Wed, 15 Apr 2026 11:36:14 -0000 (UTC) Organization: A noiseless patient Spider Lines: 78 Message-ID: <20260415a@crcomp.net> References: <20260411a@crcomp.net> <10ri6ll$34rpp$1@dont-email.me> <76bptkt0snst4e4ui156gh76ij4mlu4vdm@4ax.com> <10rikn2$3923u$1@dont-email.me> <1urptkdi5gicce55ke373nf5u3eofef9tb@4ax.com> <20260414a@crcomp.net> <64mstkhuv9792mr44ac6fo6ks9355v53uq@4ax.com> <10rm82e$dqln$1@dont-email.me> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Injection-Date: Wed, 15 Apr 2026 11:36:15 +0000 (UTC) Injection-Info: dont-email.me; posting-host="de7e6991c0916a4eed74131a08798a3a"; logging-data="915853"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX1+4JdOGoSU6cxUUqpUEJdx1" Cancel-Lock: sha1:Pa92xe62U/YHceVK0JRTMcixANQ= Xref: csiph.com sci.electronics.design:743037 piglet wrote: > On 14/04/2026 8:57 pm, john larkin wrote: >> On Tue, 14 Apr 2026 19:42:17 +0100, JM wrote: >>> On Tue, 14 Apr 2026 08:21:18 -0700, john larkin wrote: >>>> On Tue, 14 Apr 2026 12:25:57 -0000 (UTC), "Don" wrote: >>>>> john larkin wrote: >>>>>> JM wrote: >>>>>> >>>>>>>> The diodes are visible next the resistors even the bond wires are faintly >>>>>>>> visible in the photo. >>>>>>> >>>>>>> You are probably correct, but easy enough for Don to check by using a >>>>>>> DC supply to check conduction with both polarities. >>>>>> >>>>>> Yes. Graph and don't guess. >>>>> >>>>> OK you guys, lots of excellent ideas! The DC idea's easiest, so it's >>>>> first. Big Clive's youtube link now appears as a Footnote on the >>>>> pertinent page: >>>>> >>>>> >>>>> >>>>> The webpage also includes a couple of new images where 68 VDC is applied >>>>> to barely illuminate the filament's CoBs, first in one direction, and >>>>> then the opposite. How do you graph such empirical data? >>>>> Anyhow, as expected, seven filament CoBs illuminate regardless of >>>>> DC polarity. Perhaps each of the seven CoBs contains a couple of LEDs, >>>>> cross connected to conduct current through the anode of one LED, while >>>>> the cathode of its coupled LED blocks current? >>>> >>>> The xray sure doesn't look like there is a bridge rectifier. >>>> >>>> https://crcomp.net/ledfilament/top.png >>>> >>>> If 68v is the turnon threshold, and a blue LED needs, say 2.8 to light >>>> up a bit, there are roughly 24 LEDs in series. >>>> >>>> What someone should do is graph current vs voltage, both polarities. >>> >>> https://mega.nz/file/h98BgYSS#xF8fYYcMfE7qY0TWBPLaC3QjEQ0eeRiZnUr_qJsqdx4 >>> >>> (Note that connections to a diode die are top/bottom). >> >> What use would diodes be in series with the assumed end resistors? >> >> Draw a circuit please. >> >>> All he needs to do is measure the current at a couple of dc voltages >>> (both enough to illuminate the LEDs) and calculate the series R from >>> delta V/ delta I. What else is there to know? > > You ask, you get ... > > Congratulations, you did it again! Your adroit analysis is anticlimactic. Because it eliminates my expectations of exotic electronic elements such as miniaturized MOSFETs. Yet, in retrospect, it's simple to see how power elements demand a relatively large footprint in order to handle a high voltage. It's only simple now because you saw it first. Anyhow, your radiographic interpretation eloquently puts everything into perspective. Is it possible for you to give me permission to include it on the webpage with a credit to you? You also confirm my suspicion of a half-wave rectifier at each end. One half-wave connected to Line and another attached to Neutral. This gives me enough insight to include an illustrative kicad schematic on the page. Then the only things left to do are a detailed write-up and a pesky V-I curve. Even if the curve only shows a resistive load it still gives me an excuse to teach myself how to handle a Fluke current probe. -- 73, Don, WD7Q veritas _|_ liberabit | https://www.qsl.net/wd7q vos |