Yeah Phil, exactly
Nice words Dale
Say TA is Oshpark Chinese or do you have another source? Could be interesting!
Oshpark is US based and great for the ~22mm and smaller drivers but they get pricey for the larger drivers.
For drivers over 30mm it is cheaper not only individually but also total cost to order from china. You get 10 PCB’s for less then the price for 3 from oshpark.
I have tried 2 places in china, they both turned out good.
Yeah, I missed that. It does turn on too late. I got too used to looking at fets for bucks (it's probably not great for that either, but for other reasons).
deleted... some comments confusing the 800dp with 808, never mind.
I have to wonder DEL if you did the same thing as me and mentally mashed up 404 and 800 into 808. The (obsolete) 808 is 20A max. The 800 is 50A.
By the way, I notice that all of these have very similar power dissipation ratings, not surprisingly since they're the same package. What this means is the real difference is still just the Rdson and the corresponding difference in I^2R heat production.
I would love to be missing something here, and there are a few inconsistencies on the datasheet, but this is how I got it:
The 50 A is a ‘package limit’ - and it is kind of deceiving. This is only the rating of the bonding wires or similar inside the package (and keeping its temperature below what would damage the package). The actual transistor junction will have more restrictive current limitations. For a FET this is mostly related to keeping the junction temperature in check.
If you look at the ‘absolute max. ratings’ table on page 1 (note that the transistor junction will be at 150 C in these conditions!):
- Continuous drain current: 50 A…and if you follow footnote (a) this is the ‘package limit’.
- Next line: 35.4 A for 10 s (25 degC ambient and mounted on 1 sqin of PCB).
- Next line: 28.2 A for 10 s (70 degC ambient and mounted on 1 sqin of PCB).
For more than 10 s of operation we have to look at the SOA graph, bottom of page 4. The bottom curve is for DC and would apply to continuous operation at some current and some voltage drop over the FET. We are switching the FET hard (minimal voltage drop) and therefore care only about the peak on this curve - which is around 20 A for the SiR800. (You can also see the 35 A point on the 10 s curve, as listed in the max rating table on page 1.)
Staying below the SOA lines ensures that the transistor junction is being operated at safe temperatures. The FET’s lifetime will be reduced if operated above the SOA curves. How much is a complicated topic, but good designs do not do this.
I for one didn't mean 50 is realistic continuous, so ok no disagreement on that. but comparing 10s numbers (for their described mounting), the 404 is listed at 36.6 amps. The 800 is 28.2. Ok. On the graph I can't read off the SOA for DC. There seems to be something wrong with that graphic actually. Since the mounting is somewhat arbitrary I find the difference between 10s and DC also somewhat arbitrary. Either are good for comparison and neither describes the real Q8 usage.
Yes, the 404 is clearly better. I guess I thought that was a given. Rdson is lower and for whatever reason they claim a little better power handling for the 404, probably indicating a little better junction to case resistance. Based on rdson and power handling it's actually a bit weird that the current ratings aren't more different than what they are.
I don't know if this difference is just on a scale of better and worse or if it crosses some threshold for meeting the light's "needs". 20A seems like a bunch for "DC" (continuous) for a "good" design anyway, but for tripod mounted it could make sense and and I don't know maybe one FET makes it sustainable and one doesn't, for the 5 minutes while there's enough battery voltage to actually draw 20A anyway.
Anyway, I'm not sure your point about package limited. The package limited current is 50A, not 28.2. If you could actually keep the case cool, presumably you could run 50A. However the graph on top of page 5 makes no sense. They seem to basically be refusing to rate it at 50C case temp above 50A on the premise that you can't actually keep the case temp that low at 51A anyway. Of course you can though, with the right cooling hardware, and of course if you did, the max current should continue to go up as case temp goes down, much like the dashed line, not like their flat line in the sand.
I do not know why they do it, but it seems to be common. The package limit covers only the mechanical assembly and not the actual transistor junction. Internal conductors simply need to be large enough so they do not fuse or burn the casing material.
For the junction power dissipation dominates and things get more complicated.
I suspect that their "package limited" statement means this: You shouldn't ever run it above 50A because if you do, a thermal protection set at 125C (or whatever it is we said the max case operating temp is) won't actually protect the chip. (however, one set at 50C would, even at 80A). This is square-box operating parameters. Stay in the box, and all is well. But there are other shaped boxes that could be drawn instead.
Well, like Cree emitter datasheets this one is misleading and not real world correct.
I’ve pushed the 800 well over 30A and I know Richard has gone considerably higher. Could be that our sub 4V usage is in conflict with the 10V-20V ratings, I don’t know. I DO KNOW that they rock and make a LOT more than the sheet says.
I use the 4x 9V MT-G2 as a room filling light for taking photographs, have run it in excess of 30 minutes and, while I’m sure it’s not as bright then as when I started, it’s still illuminating a 2 car garage size room like daylight. (that’s some 400 sq.ft. and a little) This is running 3 of the Trustfire 32650 cells in series.
The Meteor is 12 XP-L V3 1A HI in parallel, on 4 18650’s in parallel, measured at 33.7A at the emitters and some 11,700 lumens.
A BTU Shocker with 3x 9V MT-G2 making some 10,000 lumens on 3 18650’s in series is also using the SIR800DP.
FWIW, I’m one of the first ones to use this MOSFET, Wight told me what to look for and I dug through data sheets for hours to find it, maybe 2 years ago. I have never had one fail to date. (as far as that goes, Comfychair pointed me at the SIR404DP and I bought a dozen of em to try, making me also one of the first here to use this FET as well, to the best of my knowledge. None of these have failed me to date either and I have one running Narsil to 4 Nichia 219C in my Ti/Cu home made EDC that I use daily)
They work, in other words. Well.
Old Vishay document, but relevant:
http://www.vishay.com/docs/74670/74670.pdf
A 50% safety factor will explain a lot.
I’d take one. Just seeing this thread now!
Wow just wow
I wrote it in a pm and say it again, never expected small electronic parts could be more exciting then a good movie 
Will update the list later.
Please add me to the list. Thanks.
That’s some incredible work guys!
Please add me to the list as well.
Cheers.
Well, I'm not sure why ThorFire can't use these, accept maybe cost, and if they really have 1,000 pieces on hand. 88 cents is good for us, maybe too much for them, dunno.
well, it actually says they have 999 lots available, so a half a million pieces.
I see those stock #'s on Ali all the time - I find them hard to believe for flashlights and parts, etc. You see the buy history of a few sold in 6 months, then 1,000's on hand - doesn't make sense. Do you think Simon has 29,950 L2's in stock with 53 orders in the last 6 months? Or 247,806 Convoy C8 XML2's with 1,151 orders?
Dunno, it sounds utterly insane. I think they list potential inventory just in case some big buyer comes along.