Why put a type-A USB connector on a peripheral device?

These connectors are *very* inexpensive and sort of the defacto standard. I believe the idea is they plug directly into the PC port or the port on the hub. Why would you need two cables? You already have a cable on the pin header... I suppose you are plugging the header into a prototyping socket? I find using the stiff USB cables to the prototype socket board is like using a motorcycle chain to secure a laptop. The socket board gets dragged around by the cable.

But yes, it easy to find USB extension cables just for this.

snip

and keeping the usb side of thing short also seems to be more reliable in noise environments

-Lasse

Perhaps mechanical considerations. You can have it both ways: The common type-A USB plug and Micro-USB receptablce are sufficiently strong to survive as PCB mounts. The uncommon Micro-USB plug, or the cantilevered abomination in the above URL, are not very strong. I suppose one could attach a pigtail cable to the PCB, which ends in a Micro USB plug.

I'm not sure if one could protect devices and computers from both sourcing +5V DC power. While this probably would not blow anything up, the power circuitry would not enjoy being backfed.

Extra credit for miswiring the cable:

If the next generation of SBC (single board computers) which might use a Micro USB receptacle on the PCB for USB instead of just as a power connector, it might be useful on a USB to RS-232 adapter board. However, I would expect to see abominations such as this: which methinks should best be avoided by not succumbing to temptation.

usage case:

if you want to add a TTL level UART interface to your PC or laptop.

A-male to A-female cables can be had fairly easily although that may not be a type sanctioned by USB-IF

Can it be either a host or device? When USB was new, type-A connectors at both ends wasn't uncommon.

Duh. Turned out my question was a dumb question.

For what I'm doing, I was planning to embed the adapter in my peripheral device for flashing firmware, and use a cable because I didn't want the adapter's connector sticking out.

I'm just having a problem finding an adapter with a mini or micro USB port that a cable would plug into which also breaks out to through-holes the CP2102 outputs that aren't wired to the header. I need TX, RX, DTR, RTS, Vcc and GND, and most adapters provide both 3.3V and 5V at the header, but not RTS. The two adapters I referenced give me RTS at a non-header through-hole, which is fine, but they have the wrong (for me) USB connector. And I can't find one with the right connector in the $1 range. But I'll keep looking.

How about one of these?

It turns out that mini-USB sockets are 10 cents each in either through-hole or surface mount. So the answer appears to be to get the USB-to-UART adapter, with the wrong connector, for about a dollar, remove and discard the type-A USB connector from it, and mount the adapter board to the main PB board of the project as a daughter board.

Then mount the mini-USB socket to the main PC board positioned at a hole in the case. Then the routing of the PC board will connect what used to be the USB lines on the adapter to the new USB mini socket, and the adapter outputs to the right processor pins.

I still need to check for the potential problem that the adapter's outputs might interfere with the normal functions of the processor pins they're connected to - when nothing is connected to the USB socket. I don't think there will be a problem, but I just need to get an example and check for protection diodes and other things. And of course study the CP2102 datasheet.

Anyway, if this works, the user will just connect a USB cable from his computer to the mini socket on the project, and flash the new firmware. No need to buy a Launchpad, or connect the project board to it.

I guess I'm a bit confused. If you are building a board from scratch, why not just add the CP2102 to your board? It looks like you might need to program the various USB settings if the default settings aren't good enough, but I can't imagine they aren't.

So why bother to buy this module when you can get just the chip?

Because:

1. So far as I can tell, the module would cost me less than the chip and associated parts would.

1. The chip is only available in surface mount. So if I use the module, I don't have to deal with soldering it, and, if I want, I can still make the main board through-hole.

Ok, the SMT issue hadn't really occurred to me. It has been so long since I worked with through hole other than to wire up one of those prototyping plugin boards.

The chip is the same price as the board and the passives are virtually free, around a penny a piece if you buy a hundred or so. If you aren't buying a hundred, why do you care about a few pennies?

May I ask what you are building? What sort of device can you build these days that is all through hole?

I'm just an amateur, and everything I do is through-hole. I'm just one of a number of people involved in this project, which at this point is all through-hole. If it goes any further, there will probably be one more through-hole version, and it was my idea that it would nice not to have to use a Launchpad to flash new firmware. But in the unlikely event that it goes big-time, of course it will done properly in surface mount, and at that point the chip can be included on the main board.

The cost isn't significant, but I meant that at least it's not more expensive to use the module. But I wonder where you would buy the chip. The module is $1.22 on Ebay. Where would the chip be available to me in that price range?

There's nothing special about it. It's just the processor, a rotary encoder, LEDs, a buzzer, an output jack, and a bunch of transistors, resistors and capacitors. But the processor is available in DIP, so it's still possible to do the whole board through-hole.

I'm an older guy, with a bit of a tremor in the hands, and for me the big problem with doing surface mount at home is parts placement. It's just an exercise in frustration. And while boards in small quantity and stencils are available at reasonable cost, I don't think manufacturing is yet. What we need is OSHBuild.

I think these type of adapters are intended to be directly plugged into the USB port of a pc. The 6 pin male header with the TTL signals can be connected to a female header on a small microcontroller board. If a mini or micro USB connector were used it would require a cable.

eBay.

I am similar. I have had a tremor for over a decade and it really gets in the way of doing fine work. I am a professional and I typically have a contract assembler do the work as I am building 100's, but prototypes are a bother.

This part is 0.5 mm pitch QFN which is not too bad. The pad underneath can be soldered without too much trouble by having a via in the center of the pad and applying solder while heating from the bottom through the via. The pins can be soldered the same way as QFPs, tack two corner pins, add flux copiously and then apply solder to all the pins by dragging a (small) ball of solder across the pads. Inspect and clean up any solder bridges with solder wick or just touching with a fine iron tip. Do the pins before the center pad.

I give these instructions that I have read others use successfully, while never having tried it myself on QFNs.

Over a decade ago when my tremor was much less, I successfully built a small power supply not much larger than a penny. It used 0603 passives which I soldered possibly 10 times each to get just right. I would solder just one end first to tack it down, but even that was *very* hard to do. I have some 88 pin QFN chips waiting for me to solder. We'll see when I get to it.

The only parts that actually require an oven reflow are BGAs. No way to reach the inner pins with an iron. The positioning of the part is typically less than with some QFNs (although there are some *very* fine pitch BGAs too) but it requires using a metal stencil to squeegee solder paste onto the pads with careful alignment. All far too much for me.

I believe there are Asian assembly houses that work pretty cheaply. Many will kit the parts for you. I haven't used any of them myself, but they may be in my future. Maybe I need to put something together like OSHBuild. Hmmmm....

Well, I have one of these modules that I got for programming my ocilloscope kit, and it has the CP2102 on it. And just looking at it, there's no way I would try to solder it with an iron. The "pins" are wrap-under.

As the availability of through-hole parts has dwindled, I've looked at the SM processes on Youtube, and with one exception, it looks like they've got everything covered, and at a reasonable cost. Kicad is free. You can get boards and stencils at good prices, and with the right toaster oven from Walmart, and a k-type thermocouple, you even have a reflow oven at very low cost.

But for us shaky types, what's still not quite there is pick and place. But if you look on Youtube for pick and place DIY, you'll find some interesting progress along those lines. Actually, if you don't have a tremor, a simple vacuum placement tool can be had for the cost of a cheap aquarium air pump, some tubing, and a syringe. And that would certainly make life easier vs trying to use tweezers.

Or, what I probably should do is make contact with a junior high or highschool nerd who would come over and do parts placement for $20 when I need that done. Steady hands. :-)

Don't let that dissuade you. Solder wicks. Use plenty of flux and all will be good.

If you want.

And don't use socialist solder. Use the real thing.

I did that when I was in high school: Employee #1. ;-)