Does this DIP socket exist?

Might be easier to do something that's sideways, at 90 degrees, but you'd need a bit of room.

RL

NO. Think about it...

sounds like it's time to dig in the junkbox for a ribbon cable and some sockets.

You could use a normal DIP socket and short wires to connect it to the pcb but I doubt if this is suitable for a mcu running at more than a few MHz. Better get a kit with an in curcuit programmable mcu.

A company called Ares is in the business of making mounting adapters... plug-and-socket (or pin-and-SMT-pad) miniboards which can swap around pinouts. They might possibly have something of this sort.

If you want to homebrew it you could make your own pin-inversion board... a socket on one side, pins to solder to the PCB on the other side, and a whole lot of traces running around to invert the placements. You'll certainly need a double-sided (at least) daughterboard, probably with a lot of vias. Maintaining proper signal integrity may be tricky, depending on the speed at which the MCU is operating.

Usually, modern MCUs can be reprogrammed in-circuit. If it's really a DIP, with through-hole leads for a socket, you MIGHT consider putting a socket on the wrong side, and get a DIP header and hand-solder the header to the upside-down chip.

That presumes the 'wrong' side has clearance to mount a socket, and that the (relatively tricky) hand-soldering of the chip is within your capabilities. You will want to LEAVE THE LEADS UN-BENT, because the MCU presumably has to fit into your programmer socket right-side-up as well as the target socket upside-down.

Purchase, or make up a 40 pin to 40 pin flat ribbon cable with 40 pin dip crimp connectors on each end. Manufacture using either a small proto board, or etch a small board that allows you to mount a standard socket, plus whatever socket you want to use, such as a zif socket.

Make the cable as short as possible, and mount on the other side of the board with double sided tape.

That is how I overcame these problems in 1979

Don...

I think maybe I have seen them.. will take a look on an old dev board tonight. Very deep with tall sides so it contacts on the fat part of the DIP pins. Even if I do discover one, good luck finding them in

I saw some 28 pin DIP sockets like that about 25 years ago. I didn't see a logo or any markings on them. I think they were for the EPROMS in early car computers.

get something like this

solder you cpu in upside down, careful not getting solder on the part of the pins that normally go into the socket

you now have the sockets pins for when you need it upside down, and the original pins when you need it right side up

-Lasse

If the OP is still monitoring this thread, what chip are you using ?

Seems like a hard way to go if there is a modern way to "fix" this.

I found a deep one, but it still won't hold an upside-down DIP.

Might have been remembering the old Moto programmers that required the PLCC to be inserted upside-down; the deep socket a 16-pin job in a MC68HC705K1CS programmer.

Best regards, Spehro Pefhany

How many pins? What sort of frequencies are you dealing with?

There are two boards that are joined with headers; these have to be soldered for reliability. The chip is mounted between them. There might e enough space to mount a 90 degree socket, and be able to reach in and pull out the chip (and put it back afterwards).

Maybe two 90's plugged into each other, one soldered to the wrong side? Is there enough stand-off height that a 90+chip would fit into another 90 on the board?

So you un-plug the 16f84 so you can re-program it ?

Correct ??

I then take it that the ICSP pins are also not available on the backside of the board ?

I would also take it that this is a development project. That's why you need to re-program the part.

I am trying to understand why you just do not program the 16F84 in-circuit ? That's what ICSP is for.

Am I missing something about your setup that prevents this ?

hamilton

easily

pcb but

Not for any dip, single sided is plenty. For PQFP 44 to 144 leads; double sided is easily sufficient for lower clock speeds (20 MHz or less). For higher densities or clock speeds it gets pretty ugly pretty fast, including significant on adapter board bypass caps (mostly small stuff "C" though).