Let's say you work for FTDI Chip...

[KE5FX]

How does something like this happen?  Is it poor recruiting practices?  Drug use in the workplace?  Somebody's idea of a joke?  Or just institutional indifference?

[CM800]

All a matter of perspective




And it fits well with the general style of USB slim products (e.g. on a USB cable)

[T3sl4co1l]

The added 2mm of trace length is utterly immaterial, FYI.

Tim

[KE5FX]

Yeah, obviously it's immaterial... but, again, why do it this way?  Cypress did the same thing on the QFN version of their FX2 chip, good luck turning that one 90 degrees.  I don't buy the cable argument but I agree it's the only halfway-plausible excuse.

The best way to do it would have been to bond out one of the lines to two pins, one on each side.  The resulting stub would also have been immaterial, but I guess they couldn't afford to sacrifice a pin elsewhere.

[chris_leyson]

Somebody has got their panties in a twist, it's not like it's microstrip at GHz only USB 2.0. Their job is to layout and package Silicon not to layout your PCB. If you put the chip on the bottom side of the PCB it wouldn't be a problem. Put your toys back in the pram and deal with it.

[CM800]

Easy. Just buy bottom mount connectors and mount it on top side. You will have D+/D- reversed.
Since I always use bottom mount connectors on top side, I actually like this pin out layout.

Out of curiousty, what is your reason for always using bottom mount connectors on top side?

[KE5FX]

Easy. Just buy bottom mount connectors and mount it on top side. You will have D+/D- reversed.
Since I always use bottom mount connectors on top side, I actually like this pin out layout.

I'll confess I had no idea there was such a thing as a "bottom mount" USB connector.  The connector I used here is this one, perfectly standard part with no mention of what side of the board it's meant to be mounted on.

I've used the QFN version of the CY7C68013 (FX2) before, but in that case I actually did mount this connector on the bottom of the PCB.  It was only in a recent project with the FT245 that it occurred to me to wonder why they chose this particular pin order.

Somebody has got their panties in a twist, it's not like it's microstrip at GHz only USB 2.0. Their job is to layout and package Silicon not to layout your PCB. If you put the chip on the bottom side of the PCB it wouldn't be a problem. Put your toys back in the pram and deal with it.

Found the FTDI employee... 

[DaJMasta]

The best way to do it would have been to bond out one of the lines to two pins, one on each side.  The resulting stub would also have been immaterial, but I guess they couldn't afford to sacrifice a pin elsewhere.

But USB is a differential pair, you don't want them to be split up on the chip they're going into...


Maybe they think routing towards the interior of the chip instead of out from the pins is less likely to get in the way of using its other IO and bypassing and such in single sided load designs.  Then it's just a matter of whether you want the chip slightly on the left or the right of the connector, and they chose left. 

[Monkeh]

I'm pretty sure the use of a mini connector is far more suspect than their pinout.

[KE5FX]

I'm pretty sure the use of a mini connector is far more suspect than their pinout.

But I like mini connectors. 

I hate micro connectors, and there's not enough room for the ruggedized Amphenol full-size 'B' connector that I usually use.  That leaves mini-B.

[CM800]

I'm pretty sure the use of a mini connector is far more suspect than their pinout.

But I like mini connectors. 

I hate micro connectors, and there's not enough room for the ruggedized Amphenol full-size 'B' connector that I usually use.  That leaves mini-B.

What about the glorious new USB-C masterrace?

[KE5FX]

I'm pretty sure the use of a mini connector is far more suspect than their pinout.

But I like mini connectors. 

I hate micro connectors, and there's not enough room for the ruggedized Amphenol full-size 'B' connector that I usually use.  That leaves mini-B.

What about the glorious new USB-C masterrace?

No experience with those, even as a user.    They seem to cause a lot of confusion, though, and I don't have any of those cables around here.

[CM800]

I'm pretty sure the use of a mini connector is far more suspect than their pinout.

But I like mini connectors. 

I hate micro connectors, and there's not enough room for the ruggedized Amphenol full-size 'B' connector that I usually use.  That leaves mini-B.

What about the glorious new USB-C masterrace?

No experience with those, even as a user.    They seem to cause a lot of confusion, though, and I don't have any of those cables around here.

I have them on my phone and laptop.

Great connector, sturdy and reliable. you can also plug it either way around. (no flipping it back and forth until the superposition collapses) You don't need to worry about all the excess wires. You can use it legacy on a 2.0 device easily enough.

[T3sl4co1l]

The best way to do it would have been to bond out one of the lines to two pins, one on each side.  The resulting stub would also have been immaterial, but I guess they couldn't afford to sacrifice a pin elsewhere.

But USB is a differential pair, you don't want them to be split up on the chip they're going into...

Just how much splitting up do you think is important here?  No, seriously -- not being pejorative, it's exactly the question we should be asking ourselves all the time.  How much does it matter?

Do we really need that bypass cap?  Do we really need (or even want) differential-routed traces?  Do we really need length-matched traces, or controlled-impedance traces?  How much?

'Tis a question left sorely unasked, resulting in people often making very bizarre priorities, that only hurt the project overall.  Test!  Measure!  Ask!

Tim

[Mr. Scram]

What about the glorious new USB-C masterrace?

The USB C connector seems to be quite a bit more complex.

[amyk]

Yes, this pinout was probably designed for putting things on the other side of the board.

I remember reading some datasheet from a different manufacturer, for a USB device which had automatic D-/D+ detection.

On the other hand, MCUs with the GPIO pins for one port scattered on all 4 sides of the QFN is a bit more of a why? situation...

[cyberfish]

Here is MC33883, a H-bridge gate driver chip.

C1 and C2 go to an external capacitor for the integrated charge pump. They have no other function.

[CM800]

Here is MC33883, a H-bridge gate driver chip.

C1 and C2 go to an external capacitor for the integrated charge pump. They have no other function.

A little bit confused as to your point?

I'm reading it like:

A car has a steering wheel to steer it, it has no other function.

[cyberfish]

Here is MC33883, a H-bridge gate driver chip.

C1 and C2 go to an external capacitor for the integrated charge pump. They have no other function.

A little bit confused as to your point?

I'm reading it like:

A car has a steering wheel to steer it, it has no other function.

It's the fact that they are on opposite corners of the chip. A more sensible design would put them next to each other. If those pins had other functions, those other functions might have justified putting those pins the way they are.

[SiliconWizard]

This C1/C2 pin assignment looks funny at first sight, and is certainly an annoyance for routing purposes.
I'm guessing it may have something to do with the way the power rails are routed inside the chip die though, imposing a pad placement. Then you can't really make bonding wires cross one another to make the final package pins easier to route. Or you would need an interposer, which would make the package costly.
Just a thought.