Post by thread:[PIC:] ICSP of SOT23 10Fs.

Hi. I've been thinking of the issue of ICSP of the 10Fs in SOT23 packages. The "product" will be very small (one 10F/SOT23, one SMD LED and a few SMD passives, probably just 10x10 to 15x15 mm's). The two methods I've thought about are : 1. Using a bed-of-nails using "pogo" pins. Pro: easier to lay out the PCB. The ICSP-pads could be anyware. Con : a bit more complicated hardware setup. Have to make a fixture with the pogo pins and probably a couple of indexing pins with matching holes in the PCB. 2. Using contact fingers on one side and a PCD edge connector. Pro : Reasonable easy harware setup using a standard connector. Con : Have to route the signals to the edge of the PCB and possible have the extra space for the depth of the connector. I don't want any additional hardware (connector) directly on the PCB. I'm hoping to be able to make a few 100's of these... Any thoughts ? Regards, Jan-Erik. ____________________________________________

Hi, Method #1 is probably not reasonable for just a few hundred boards. Method #2 takes up too much space on the finished PWB. How about creating a design that can just be repeated across a panel with waste areas between each "module"? The waste area would carry the required programming traces out to the edge of the panel with appropriate connector(s). Program each module, then snap the board apart (requires v-groves or very closely spaced holes). At 10x10mm, you could do a whole lot of boards on one panel. Also easier to assemble than many individual small boards. Ken spam_OUTklumiaTakeThisOuTadelphia.net {Original Message removed}

On Sun, 28 Nov 2004 10:45:42 -0500, you wrote: >Hi, > >Method #1 is probably not reasonable for just a few hundred boards. >Method #2 takes up too much space on the finished PWB. How about >creating a design that can just be repeated across a panel with waste >areas between each "module"? The waste area would carry the >required programming traces out to the edge of the panel with >appropriate connector(s). Program each module, then snap the board >apart (requires v-groves or very closely spaced holes). At 10x10mm, >you could do a whole lot of boards on one panel. Also easier to >assemble than many individual small boards. Also, you could bus all the power together either for the whole panel, or in rows, to reduce the number of pads per board needed. What I did on a couple of 12f67x designs is put a group of 5 small vias ( 2 staggered rows of 3 and 2), and use a hand-held probe made using bed-of-nails type testpoints - the holes hold it in position while programming. Works fine. > >.....klumiaKILLspam@spam@adelphia.net > >{Original Message removed}

Jan-Erik Soderholm wrote: >Hi. >I've been thinking of the issue of ICSP of the 10Fs in SOT23 packages. > >The "product" will be very small (one 10F/SOT23, one SMD LED and >a few SMD passives, probably just 10x10 to 15x15 mm's). > >The two methods I've thought about are : > >1. Using a bed-of-nails using "pogo" pins. >Pro: easier to lay out the PCB. The ICSP-pads could be anyware. >Con : a bit more complicated hardware setup. Have to make a fixture >with the pogo pins and probably a couple of indexing pins with matching >holes in the PCB. > > > I've done the pogo pins bit before. No extra alignment posts are needed if you make the pads thru-holes and use the chiseled probes. The probes are almost self-aligning. You can probe from the top (just holding the probe array), or probe from the bottom in the usual way. I like arranging 5 pads on 2mm or 2.54mm centers in a line. >2. Using contact fingers on one side and a PCD edge connector. >Pro : Reasonable easy harware setup using a standard connector. >Con : Have to route the signals to the edge of the PCB and possible >have the extra space for the depth of the connector. > > > You will have poor results because the contact point pressure is too low; if the PCB is not very clean, or if the surface is slightly oxidized, you can't cut through. A good way to do this (what I did for Imagic in 198X) was to design a test station with the important contacts taken to the edge of the PCB, like conductive ribbons of tinned copper; when the PCB array was properly inserted, photo-sensors would cause the probes to be forced against an almost arbitrary point on each conductive ribbon. VERY fast testing ; 12 PCB's was tested at the same time, with LED's indicating which PCBs were OK & which were bad. We tested and sorted Good/Bad for 12 PCB's in less than 1sec each. NOTE: Pogo pins have a finite lifetime, usually 10K actuations. Usual failure is the internal springs breaking. This has to be monitored. >I don't want any additional hardware (connector) directly on the PCB. >I'm hoping to be able to make a few 100's of these... > > --Bob -- Note: Attachments must be sent to attachKILLspamengineer.cotse.net, and MAY delay replies to this message. 520-219-2363 ____________________________________________

Kenneth Lumia wrote : > Hi, > > Method #1 is probably not reasonable for just a few hundred boards. > Method #2 takes up too much space on the finished PWB. How about > creating a design that can just be repeated across a panel with waste > areas between each "module"? The waste area would carry the > required programming traces out to the edge of the panel with > appropriate connector(s). Program each module, then snap the board > apart (requires v-groves or very closely spaced holes). At 10x10mm, > you could do a whole lot of boards on one panel. Also easier to > assemble than many individual small boards. Yes, that's an idea. I'll probably get the modules panelized (if that's the word) from some boardhouse anyway. Now, what I forgot to say, was that the actual firmware on each module will not be the same from one module to the next. Well, a couple could have the same code, but it will vary a lot. If things work out as planned, they will be sold in "sets" of, let's say, 10-20 modules (together with a separate power supply module), each with separate firmware. If I prepare 10 sets at a time, I could of course program 10 similar modules at the same time also... This is going to be a simple "office-desk" solution. Thanks for all ideas ! Jan-Erik. ____________________________________________

Mike Harrison wrote : > What I did on a couple of 12f67x designs is put a group of 5 > small vias ( 2 staggered rows of 3 and 2), and use a > hand-held probe made using bed-of-nails type testpoints > - the holes hold it in position while programming. Works fine. OK. Good idea. That was the single-point type of pins, right ? And Bob Axtell wrote : > NOTE: Pogo pins have a finite lifetime, usually 10K actuations. > Usual failure is the internal prings breaking. This has to be > monitored. Well, if I ever reach 10K units in this project, I would definitly consider a broken spring as a minor issue... :-) :-) I would condider it a success if I reach a couple of 100 units. Jan-Erik. > ____________________________________________