Peak volt is 75.6V, I usually calculate with the average charge I have in real life.
But when it comes to what components can handle or actual design considerations, only actual battery voltage matters. The kg·m2·s−3·A−1 type of voltage.
An abstract “lead acid equivalency” or “nominal voltage” or “mid-charge approximation” is not useful for that, and seems to cause confusion.
Your antispark solution needs to be rated for and able to handle actual volts.
1 Like
Yeah i agree. Nominal ratings are fucking useless.
2 Likes
Well it hasn’t blow yet to this day 
Posted in the wrong thread but moved back over here (soz)
Righto this is a very preliminary look at the FlipSky 20S Anti Spark but I thought I’d share some pictures because I could take ages to give much else
Overview: (I like the bus bar for the negative)
Core power components:
MOSFETs
High Side Driver (bottom left)
Linear regulator for the circuit logic (top right-ish, square package below the bus bar and solder blob)
Disclaimer: This is only commenting on basic component choices and not the topology, and I’m not an expert. With the exception of that diode I didn’t check, they all seem to be 150V rated parts from big name manufacturers. The current ratings on the fets should be weeell within spec, if the traces are up to it they could theoretically do 676A continuous. 12AWG though lol.
If I get some inspiration I’ll check some less basic stuff like comparing it to application notes and see if that driver can match the gate charge of the 4 or whatever.
PCB:
I didn’t do a great job capturing this (and forgot to measure) but it is as advertised a big chonky aluminium PCB. This pic might be useful?
9 Likes
Thank you so much for taking a look at this switch and posting your findings!
A few (very technical) thoughts…
-
The continuous current rating of 169A in the datasheet is a theoretical number and can’t be used here. Everyone uses the front page current rating (it’s incredibly convenient) but it’s listed for a case temp of 25°C and once current flows in the FET the case temp is no longer 25°C, no matter what kind of heat sink you have. The datasheet also mentions that the current rating is limited by the max junction temp rating and that will be reached very quickly at high current levels.
-
Calculating the power each FET must dissipate at this switch’s continuous 150A rating (37.5A/FET) means that each FET creates more than 14W of heat! That would bring the FET temp to over 600°C, probably way over as I am being extremely generous with my calculations here.
-
The wire-to-board and bus-bar-to-board connections each create heat too. The four wire connections could add about 3W more heat at 150A. The bus bars would add at least another 1W. That means the board is creating about 60W of heat at 150A…it would not survive.
-
We need to use the on-state resistance and thermal resistance numbers to calculate how much current each FET can handle before reaching its limit of 175°C. If each FET was alone and on a square inch of 2oz copper that would be about 15A. But they are on tiny copper pads and each FET heats up its neighbor. That lowers the current rating considerably. Being on an aluminum PCB brings it back up a bit so let’s call it 10A per FET…about 40A total.
-
Adding up all the FET+connection+bus bar heating numbers for running at 40A continuous gives us about 7.5W of heat for that switch. In my opinion, based on my experience, that sounds about right for a normal board that size (with all the necessary compromises made to keep it that small). Using an aluminum PCB, thermal pad, and an aluminum case helps to raise the current rating some though. I can’t say by how much. A lot depends on the desired life span for the FETs. Some companies are willing to run them at their max temp since it doesn’t happen often during actual use. Other companies, for increased reliability, would lower the current rating so the FETs never exceed about 80% of their temp rating. For very high reliability designs the rating might be halved.
-
The board traces are the equivalent of a wire much smaller than 12AWG. This combined with the 12AWG wiring and all that component and connection heat means that the 150A rating is absolutely preposterous. But since esk8 doesn’t (typically) run continuously at high current levels this doesn’t have to be a big concern…unless, like me, you hate seeing these kind of bs ratings games being played.
-
Regarding the voltage ratings…the board layout looks to be pretty low inductance. As long as the FET driver was set up properly I don’t think there will be a lot of extra board-created voltage spiking (need to use a scope to know for sure) but since they say 80V max then I would stick to that or lower as there are spikes everywhere in an esk8 setup.
-
It appears that this switch uses the FETs themselves for the pre-charge instead of a separate resistor (with its own smaller on/off FET) to slowly let the current through to the ESC. This is perfectly acceptable for an AS switch but when used this way there is a certain FET operating characteristic that can cause big problems and can easily lead to FET overheating. This can lead to FET failure if the switch manufacturer does not take it into account. I have met power electronics product designers who were not aware of this operating characteristic so it’s certainly a possible failure mechanism.
-
If the aluminum PCB is not grounded then it just becomes a big antenna, possibly making any voltage spiking worse and/or interfering with operation of the switch because of electrical noise. I do not know if they ground the PCB aluminum or not.
Thanks again for posting that info and the photos! I always enjoy teardowns.
12 Likes
I have no idea about most of what you’re talking about but i appreciate it.
no need to go into detail etc, would be a wasted effort for my monkey brain 
9 Likes
TL;DR:
IMO the 150A continuous rating is ridiculous and if they did not take certain things into account in the design then the FETs can pop. 
8 Likes
Hahahaha thanks for all that, muuuuch more detail than I was going to be able to add. I’d like to add some basic calculations for thermal output in these things as practice, might try and update this over the weekend
2 Likes
omfg, thank u for the tldr
3 Likes
LOL…yea, I went deep on that one. Decided that posting the two pages of calculations might be a little too much though. 
Perishable skill, that type of analysis is. I needed to flex the neurons a bit.
7 Likes
we all appreciate the knowledge u bring, just that my dumb brain can’t process all these info at once 
1 Like
Here’s a “fun” fact…
At the 150A continuous rating of that switch just those four 12AWG wires alone will generate at least 70W of heat. If you can imagine placing your hand on a 60W light bulb you can get an idea of how silly that rating is.
5 Likes
i thought all the cont. rating is always “as long as u don’t exceed XX°C” kinda rating
1 Like
There can be a “do not use above xx°C” ambient temp rating.
If the continuous current rating is ambient or component temperature-limited then they need to say that. Otherwise we have to assume they are claiming a true continuous rating.
This device has a preposterous 300A “max” rating that would be the one we could possibly consider as being temp-limited. But this is a beyond silly rating since 12AWG wire melts…literally melts…after a few seconds at about “only” 220A.
Since we have no way of knowing the temps inside the case, and how close different components are to their max temp, we need to trust the manufacturer’s current rating. A temp-limited rating would place all the work on us the find out how hard a device can actually be used. Different setups would results in different current ratings.
A manufacturer’s current rating, IMO, needs to be set for the worst case which can be reasonably expected for the intended use. If the current rating is somehow limited then we need to be told. That way we don’t need to measure temps to know which current rating to use and we can trust that the manufacturer is actually selling a product we can use the way they say it can.
Otherwise the product can burn out…oh…wait…
3 Likes
I explained one of those phenomena (number 8) in a little more detail over here:
1 Like
That’s it. 
The particular characteristic I was talking about though is a (seemingly) little-known aspect of operating FETs in this linear region. I don’t want to mention it here as will be taken into account with my AS switch (when I dive back into it) and can help to differentiate it from other designs that don’t do this.
1 Like
Sorry, forgot to mention this…
It’s a really not good thing to have bits of metal bouncing around inside an AS switch case. 
I don’t know what kind of QC checks are done before putting the board in its case but, at least for this unit, the QC was…umm…inadequate. Where it is in this photo is no problem but that bit-o-metal will not sit there when the switch is being bounced around.
5 Likes
No one knows because there probably isn’t any 
3 Likes
ESPECIALLY when the bits of metal look like dicks.
8 Likes