PICkit Headers

[T3sl4co1l]

Contents:
1. Rant
2. Question



1. Rant.

I hate PICkit3 programming headers.
They stick out.
They're sharp.
They're unshrouded.
There's no side-to-side alignment.
There's no upside-down alignment.
In short, they're made to be as generic and as useless as possible.

Microchip does not recommend ESD protection on these headers.  So they're vulnerable.  Pick up the board, touch the connector, zap, chip's dead.  98 boards on the shelf, 98 boards on the shelf...

The headers necessarily stick out, vertically, or horizontally outside the board outline, or horizontally with a large offset (because the PICkit3 itself has a super thick housing), in any case making it very likely that picking up a PCB will leave your with more holes in your hands than you started with.


2. Question.

Can we solve this?  Two solutions, preferably both at once:
1. Use a shrouded header, solving the alignment and finger-safety issues.
2. Add ESD protection as standard, to eliminate ESD handling issues.

Obviously, there is no shortage of 0.1" pitch male pin headers available from all over.  But it doesn't look like any of them have such a huge shroud as to allow the PICkit3 to mate (let alone unambiguously).  Any ideas?

2. I was just going to add BAT54S's to the three pins (MCLR/RST, PGED and PGEC).  Should be good enough.  Any un-obvious recommendations against that?  Could I add a series resistor as well?  (Not worried about EMC or signal quality here, given the short trace length, infrequent use, and "laboratory" environment where programming would be performed.)

Thank you for your time, 

Tim

[Wilksey]

I have a .1" header with a bit of correction fluid on the mating half of the PK3 where the arrow is so I know which pin is MCLR, this goes to a Molex keyed connector which the mating half goes on the board, Microclasp or Milligrid usually, I have cables for both.

As for ESD, I have never fried a chip programming it before, it's usually the slip of the probe on the TQFP's that do it when shorting two pins together whilst debugging with the ICD3!

[KL27x]

Doesn't need ESD protection, AFAIC. Why put schottky on the pins? Did you mean in series or clamping to VCC or what? I don't get what purpose BAT54 serves. I find out everyday how dumb I am... please Blueskull or T3 explain what this does? If I understand how you mean to connect them, one of us will undoubtedly learn something.

I highly prefer standard 0.1" pins. But if you prefer doing it backwards... take out the pin header and put in the socket/female header on your board? Problem solved? 90 degree pin header might also solve your problem?

I like generic... generic and accessible (pins > sockets for accessibility) = more useful, IMO. If I can't conveniently mark it, I just clip the MCLR pin every so slightly. My ICSP cable is clearly marked on the MCLR end.

[T3sl4co1l]

I have a .1" header with a bit of correction fluid on the mating half of the PK3 where the arrow is so I know which pin is MCLR, this goes to a Molex keyed connector which the mating half goes on the board, Microclasp or Milligrid usually, I have cables for both.

Hm, so, adapter cable?

As for ESD, I have never fried a chip programming it before, it's usually the slip of the probe on the TQFP's that do it when shorting two pins together whilst debugging with the ICD3!

Worried about handling.  Belt and braces, and all that.  It's a teensy risk, but it should be a zero risk!

I never put 0.1'' debug headers on my boards. I will either use tag-connect solutions or create my own pogo pin pattern.
So my solution: use tag-connect pads on your target board, then just buy a PicKit to tag-connect adapter cable from their website.

Ah...yeah.  I wonder how the customer would approve of them... 

We've used them before, and found them reasonable to use, the Tag-Connect thingys I mean.

It's another bit of kit, but it's nice having the zero-incremental-cost, no-profile footprint.

Doesn't fix the cable-to-PICkit3 ambiguity problem, but I suppose one could simply tape the two together...

ESD diodes are not a big issue as long as you use low capacitance variants. If ESD events may only happen in manufacturing process, not in user application, then small MOVs may also be used to save cost.

Are the programming pins really that sensitive to loading capacitance, or leakage current?  (Also, I'd be kind of surprised if MOVs helped much; the pins in question aren't 5V tolerant, so the internal clamp diodes should fry first.)

Doesn't need ESD protection, AFAIC. Why put schottky on the pins? Did you mean in series or clamping to VCC or what? I don't get what purpose BAT54 serves. I find out everyday how dumb I am... please explain what this does?

To clamp input surges (namely, ESD) to VCC/GND, diverting it away from the MCU pins.

BAT54S is a common part, very cheap, reasonably low Vf, and being two series diodes, can be conveniently strung between GND-pin-VCC to solve this kind of problem.

Clamp diodes work better than zener diodes, because the IC has internal clamp diodes already.  Most likely, when ESD is applied, it would be while the board is unpowered, so, if VCC is near 0V, the chip's clamp diodes will do their business in the 1-2V (above GND) range, long before a zener sinks much current (>7V?).  (That does work well for "5V tolerant" pins, though -- those are protected with on-chip zener or snap-back diodes, so the breakdown voltage is independent of supply.)  Using external clamp diodes, means a lower clamping voltage (maybe 1V above VCC), and diverting the energy away from the chip, preventing damage.

I highly prefer standard 0.1" pins. But if you prefer doing it backwards... take out the pin header and put in the socket/female header on your board? Problem solved? 90 degree pin header might also solve your problem?

Hmm, maybe.  There are double-ended, ah erm, headers, that fit regular socket strip in both directions.  Solves finger-touch and finger-poke problems.  Doesn't solve polarity or offset, and does need an additional part (though, it's at least the cheapest of possible adapters!).

Tim

[stj]

use right-angle headers and a short extension cable - like the chinese clones all come with for free. 

[KL27x]

Yeah, if by double-headed connecter, you mean just a section of long pin header, all you need. Stick it in the female end of your programmer/ICSP cable.

I think I'm still confused by ESD diodes. So you want to add external ESD clamping diodes is parallel with the ones already on the chip? To slightly lower VFD compared to the one's on the chip? I don't see why that is necessary. Anything below VFD of those diodes on the chip, already, is not ESD problem? I must be still too dumb to understand, lol. I thought anything 3.6V or less is not problem... and the clamp is to shunt 1000s of V back down to sub +- VFD to the rails.

Anyhow, I don't see how you can clamp MCLR to Vdd without creating an obvious problem if you plan to use debugger or HVP.

With PICKIT3/ICD3, I find these lines are fairly sensitive to capacitance. Very tolerant with PICKit2. But I wouldnt' worry about the capacitcace of a Bat54.

[T3sl4co1l]

I think I'm still confused by ESD diodes. So you want to add external ESD clamping diodes is parallel with the ones already on the chip? To slightly lower VFD compared to the one's on the chip? I don't see why that is necessary. Anything below VFD of those diodes on the chip, already, is not ESD problem? I must be still too dumb to understand, lol.

Yes.  Most importantly, it diverts the energy away.  During an ESD peak, the pin voltage might exceed "Vf" (that is, the "0.7V" that you think a diode should drop) by 10s of volts.  With the peak current being in the 10A range, that's 100s of W being dissipated by a teensy tiny microscopic trace and diode structure (on chip).  Whereas the BAT54 is a fat meaty junction, with huge bondwires (relatively speaking), and will only drop a few volts under the same condition.

Best is if you can afford to add a resistor following the diode.  So that you get a three-way voltage divider: ESD is shunted by the first diode, which leaves a few volts extra, which drops across the series resistor and internal clamp diode.  In this case, very little energy is dissipated by the chip's protection, and the circuit becomes nearly indestructible, even from malicious intent!

Anyhow, I don't see how you can clamp MCLR to Vdd without creating an obvious problem.

Erm... "most" pins I suppose I should say...  That one's 5V tolerant, so can use (needs?) a zener.  Or, I have a 5V rail in this circuit, I could use that too, I suppose.  (Unless it's occasionally given old-fashioned "high voltage" for programming, but that seems unlikely to me.)
 
Tim

[KL27x]

I never put 0.1'' debug headers on my boards. I will either use tag-connect solutions or create my own pogo pin pattern.

Me neither. Same. Except for a dev board... pin header for dev board, for sure. (Or if I think it is good to go with the first revision, I'll just hand solder a pin header to the ICSP pads for dev purpose.)

For customer board, I think you just give them w/e they want, no? I mean, if I'm the one that is going to flash them, and it's < 10 seconds per flash, for sure, pogo pin pads. (Or flash the chips before assembly).

T3: I think your ESD protection standard is probably overkill for most applications. But since you mentioned moving rubber belts and whatnot, I suppose you can't be to careful. I think chance of repetitive insult like with belt/friction or something is the main way you will defeat the internal clamping diodes.

[KL27x]

I initially hate Tag Connect, but I had to use it for 3rd party hardware for a client. I got over it by frankensteining a Tag connect cable to my own standard. I think I might even start using it.

The Blue one has the Tag Connect pin header assembly in the end. The steel guide pins are cut off and rounded so they can be used on a flat surface.
http://i18.photobucket.com/albums/b103/klee27x/DSC_3104_zpsqlm2mzut.jpg
Angled for ergos. (in-line arrangement of the pogo pins without the guide pins already allows you to tilt the interface... much easer to see what you're doing when the pads aren't in the shadow of your thumb, lol.)
http://i18.photobucket.com/albums/b103/klee27x/DSC_3105_zpsd3tl6mzv.jpg
Hot glue fixes everything.

[trophosphere]

I just put 0.1" male headers onto my programmer and adjust may layout so that the individual pins of the SIP footprint is slightly staggered like this. This way, I can just put the header in, program it, and then take it out without using any money getting special components just to program my board. It also saves on height and also reduces the chance of anything coming into accidental contact with those pins if you want to look at it from a possible ESD prospective.

[KL27x]

AcHmed99

Are those the kind, by any chance, that allow a long pin header to push all the way out the back? I found a 90 degree female header on a retail device like this. The spring clips wrap around the pin from the side, and they are rock solid. I have been meaning to source some of these for use with extra long pins.

[JPortici]

adapter cable all the way. what we use is a 10 pin IDC connector. pins are shared between programming and uart communication (to program data/parameters and communicate with the product)

[technix]

I use female headers instead of male ones for my project, and I have a short patch cable with pins on both sides for connecting them.

[deephaven]

+1 for adapter cable. Most of my target boards have 6 pin 1.25mm Molex headers for programming so they don't take up much PCB real estate. Never had an issue after 1000's of boards.

[Wilksey]

Yep, adapter cable.

[T3sl4co1l]

Hmm... ya know... there isn't enough space between boards in this particular project to use Tag Connect.    So, female headers it is!

Thanks for the ideas,

Tim

[KL27x]

And after tallying the answers.... this is why generic is best. Almost everyone just makes the thing work the way they want it to. Might as well have a generic pin header interface as a starting point. Last thing I want on dev/test equipment is a unique connector.

No they aren't. They are the kind Newark had on sale so I bought 20.
I'll be keeping my eye out for the ones you mention now, they sound good.

Thanks, anyways. I'll have to order a Mouser/Digi paper catalog someday. I must have thrown out my last one years ago. These guys hold very tight... it's the header on my Sure Electronics PK3 clone!

[calexanian]

And after tallying the answers.... this is why generic is best. Almost everyone just makes the thing work the way they want it to. Might as well have a generic pin header interface as a starting point. Last thing I want on dev/test equipment is a unique connector.

+1 for that. I have never found any difficulty with it. Way more convenient than fiddley double row connectors. as far as the protection diodes, I have never had an ESD failure in the many thousands of boards we have programmed that way here. It is also tolerant of being plugged in backwards as well.

People hate on the Pickit3 but I have never heard a genuine reason not to use it. My complaints lie in its general temperamental nature when switching from device to device and various versions of MPlab, but once you get it set up right they just kinda go along forever.

And seriously people, for the price nobody has any reason to complain!!! Use an adapter if you just hate .100 IDC headers that much!

[Buriedcode]

+1 for adapter cable. Most of my target boards have 6 pin 1.25mm Molex headers for programming so they don't take up much PCB real estate. Never had an issue after 1000's of boards.

This.  For my "quick 'n' dirty on stripboard, looks shocking" I'll of course use standard 0.1" unshrouded headers, but on later prototypes and actual commercial devices, the small 1.25mm molex's have it.  I made up 3 cables for it, with the PICkit end a standard 'male' crimp pin housing with a nook taken out of it, and a small plastic square on the PICkit's female connector.  I know it rarely damages boards/chips plugging it in the wrong way, but as I had to make a programming cable anyway.. might as well add polarisation to it.  Ever since PIC's had self-programmable flash, bootloaders on there for in-system updates.

I like the fact they have such a basic connector, its so generic, but as you noticed its also an arse unless one makes something custom.  Pogo pins are all well and good but what if you knock something up in an hour to test an idea?  Unless you've made up a batch of small adapter boards that convert these to 0.1" grid, you're stuck using custom PCB's.

As for protection, I've never killed a chip, and I don't wear an ESD wrist band.  But there will be a first time.

[KL27x]

Pogo pins are all well and good but what if you knock something up in an hour to test an idea?  Unless you've made up a batch of small adapter boards that convert these to 0.1" grid, you're stuck using custom PCB's.

When I am doing software dev, I always use a connector, of course. You don't need pogo pins except for batch flashing lots of boards after the firmware is done.

The purpose of pogo pads... they're practically free regarding board space/cost, no component cost, no assembly cost, for where you don't need to connect beyond a short <10 second flash. Just for initial firmware and/or updates.

The main reason I don't just do the wiggly plated thru hole thing is partly because I use toner transfer where applicable for prototyping. So connector, it is. Otherwise, that would seem like a great, cheap solution to batch flashing/testing/updates/dev. But pogo pads can be made a lot smaller in space than .1 PTH, too. There's a time and place for all these solutions. The pogo pin is cry once and save on hundreds/thousands of boards. Even if you preflash your chips, you might need to fix a bug or make an update, prior to shipping. Squeezing pogo pads in there even in a device not planned to be updated is like a "why not?" Manually flashing thousands of boards isn't fun, but it will save a lot of time/money if you made a fatal code error.