In ‘>’ order the KiCad footprint is 1 3 2. The KEC footprint, shown in posts by tormyvancool, is 2 3 1. In Fairchild it’s 1 3 2.
EDIT: in both datasheets the functional order (read in the same way ‘>’) is B C E. But the number/name connections are different because the numbering order is different.
Yep indeed. By checking all the microdetails.
Yes this was the tricky part.
And (however we name footprint pads) it is the only one what should be understood as ‘pin order’ for bipolar transistor.
If KiCad SOT23 pads were named BOTTOM_LEFT, BOTTOM_RIGHT and TOP there would be less risk of such error, I think.
But I don’t postulate such change 
@Aris_Kimi in a long discussion I had privately an don GitLab with him, he made me read the IPC 7351 page 77 if I remember correctly.
On that document the pin 1 is upper left he 2 lower left.
Right?
Ok, if you get the KEC datasheet you can see that IT IS the SOT-23. But they inverted it.
It’s NOT question of the S or else.It’s question they didn’t respected the SOT-23.
So who’s right who’s wrong is hard to say.
I went back to the project and I rebuilt the history (it was schematised in Jan 2022).
Today I also recuperate the email with the chat with JLCPCB
Yes I chosen for the MMBT version.
When I went on JLCPCB THAT MMBT was not available (shortage) at THAT time.
I asked to JLCPCB in chat and they answered that the component was not available due semiconductor shortage.
INDEED at THAT time, even the 555 were not available at all. 0 in stock. I needed the LCM: no way. There was only one type in VVIP package …
This confirmed the shortage issue (bias confirmation?). Right till here?
I chosen the only one (at THAT time) available (so I even not opened the MMBT datasheet itself).
Yes it was marquee with S but I opened it’s Datasheet to verify the data.
Up there is written and reported as you see on my screenshot.
Having chosen the MMBT in KiCAD, it reported the 1 and 2 inverted compared to the datasheet.
HENCE:
Of course once delivered, the circuits were not working.
As result: today we’re here.
Unfortunately the reason they had stock of this part was probably because it is not pin compatible
???????
According to everything that was shown here 2N3904S is pin compatible with MMBT3904.
So far I see the MMBT Emitter = 2 while the S has Emitter = 1
I agree, assuming you are talking about the device (why would you mean anything else ?)
I think you are talking about the symbol ? if the devices have the same function for the pin in the same position (regardless of arbitrary numbers) then they are “pin for pin” compatible.
Pin numbering for the MMBT3904 is different . . . pin function is the same.
You must have different definition of devices being pin compatible then me.
For me if device can be replaced by the second one by simply assembling it at the same footprint than they are pin compatible. And you can replace MMTB3904 with 2N3904S that way.
Electrons flowing through transistor don’t care how we named pin/pads.
So the KiCAD symbol for MMBT3904 is actually correct . . .
SOT23 is a generic footprint and in this case it doesn’t match, sorry but in this case I think caveat emptor applies.
Piotr check here please
Here below check the SOT-23 on the right side.
The pin 2 is the upper left
While the MMBT is the lower left
Always SOT-23
So it’s not compatible at all
Otherwise they couldn’t be delivered wrongly mounted.
Or you should give me a hint to understand what’s occurred then …
Perhaps the S didn’t respect something?
What’s your thought?
SOT-23 from KEC = version S
SOT-23 from OnSemi (Farichild) = version MMBT
SOT-23 from IPC 7351
Both are SOT-23 but Emitter and Base are flipped each other, since they have the same pin number (in the S and in the MMBT)
2N3904S
MMBT3904
They are the same pin out, SOT 23 cannot be mounted wrongly, there are 3 pins, you can’t mount them upside down.
Checking with V7.0.2, the MMBT3904 automatically gets the BCE to actual device package as shown in the isometric drawing above correct.
This is what matters. Datasheet pin numbers mean nothing as various manufacturers cannot agree on pin 1
The man’s hands fall…
Do you read what people write to you?
And your red arrow shows that upper left = 2 = Base.
For MMBT3904 upper left = 1 = Base.
And for electrons it is not important if it is 1 or 2. Important is that it is Base.
It was told so many times that I simply don’t believe that you didn’t catch it yet.
Have you been touched in ‘fast reading’? If yes than (according to my experience) it can be the source of problems (during this thread and when you ‘corrected’ something when designing that PCB).
Edit:
Get back to your first post. Pictures you gave just also shows that.
B from MMBT3904 datasheet corresponds to pin 2 on the 2N3904S datasheet and pin assignment list tells that 2=B. So both are the same.
Ok so can you explain (however I have made everything with 6.0.2) why it was mounted differently? With the base at the place of the emitter?
Can you? I want to understand the culprit at this point.
YES I do read what ppl says but I see it seems no one is watching the elephant in the room.
The devices mounted (the S version) had the Emitter placed in place of the Base and of course: all the made devices where not working.
EDIT: do you know (I told also but it’ was lost into the discussion) how I have made them work? Just unwelding the transistors and upside them down … it was easier for the moment
For example, we change the outputs on the same contacts of the microcircuit… You order a reel by the name of the model and you get non-working devices… It turned out that the manufacturer released a new revision and updated the documentation but the developer did not check because the general name remained the same… This happened within one year… Kicad libraries are relevant only on the day of writing, everything else can change and everything needs to be checked… Imagine that there is no digital designation at all on sot 23 and you need to put it yourself)
From 6.0.10 . . .
I suspect you either had a custom symbol and/or custom footprint. You are in the best position to understand what happened though, you may have the original KiCAD 6.0.2 files and could investigate.
I’d like to understand what happened also, I might learn from your experience. 
It is not possible to mount SOT23 not correct so there are only two possibilities:
First we should be sure if we have 1 or 2. From all this thread I can only say that 1 has higher probability then 2. And sorry if you say it is 1 or 2 I will not take it as axiom now.
Show design or at least screen shot of PCB fragment being enough to find if we have 1 or 2.
If left side of your picture from first post is how it was during you designed it than it looks that you have not changed pad numbers in KiCad SOT23 footprint but you have symbol with Emitter being pin 1.
I switched KiCad to default libraries only for the while. Now I’m back with my libraries.
As most probable I suppose that you have taken 2N3904 symbol and changed the footprint from THT to SOT23. So the situation 1.
I use only my libraries and each symbol has assigned footprint and I never ever change it.
