JTAG connector part # for Beagle Play

I was not suggesting that the part be loaded. This would be an end-user installed item, as it was on previous Beagles being only a fraction of the cost of the Tag-Connect.

Maybe I am missing something but I believe the Wurth footprint is SMALLER than JTAG-AM62 on the image below based on the respective datasheets. Am I looking at the correct footprint? I get 22mm2 roughly for the Tag-Connect:

And for the Wurth I get 21.59mm2

Put it in place of this:

image

Thinking layer3 is the only layer that would need some changes. Traces would need to be shifted to left/right. If this design was in KiCad I could change this for you:-) Also, I am not requesting that the other JTAG connector be changed, just the AM62.

Sorry, but Iā€™m not letting you get away with that comparison @jkridner :slight_smile: .

My experience is that the Tag-Connect without clips is quite unreliable for development work without a custom jig to hold things in place. Itā€™s fine for one time programming tasks or a rare firmware update, but for debugging it just doesnā€™t work reliably even for much smaller SoCs. Even our holding jigs had to go througn several iterations to make it ā€œreliable enoughā€.

So, a ā€œrealā€ comparison is against the Tag-Connect with clips. A footprint for a ā€œfullā€ 10-pin through hole is 6.4mm long x 3.4mm wide. (21.7 mm^2) A footprint for a ā€œfullā€ 10-pin SMT connector is 6.4mm long by 6.3mm wide (40.3 mm^2)ā€“the advantage being no through holes to obstruct other layer routing.

See: https://media.digikey.com/pdf/Data%20Sheets/FCI%20PDFs/Minitek%20127_Catalog.pdf

A Tag-Connect without clips is 7.62mm long x 2.069mm wide (15.7 mm^2). A Tag-Connect with clips is 8.91mm long x 7.62mm wide (67.8 mm^2).

My smattering of dev boards concurs with this. Their one-shot interfaces are either 10-pin SMT .050", Tag-Connect, or custom solder pads. However, their debug probe interfaces are all 10-pin SMT .050" with one exception that is Tag-Connect with clips.

Iā€™m a big believer in .100" through hole. However, if you simply canā€™t stomach that, then the standard CoreSight 10 10-pin .050" ARM debug connector is really the way to go for debug ports.

I do understand that part of this is cost transfer as everybody continues to get stung by everybody whining that the RPis are ā€œcheaperā€ than a BeagleBone in spite of them being totally unavailable (I sympathize). So, youā€™d like to 1) not populate the footprint but 2) allow people to run a debug probe on these boards without having to pull out a soldering iron and 3) only charge those people who are actually using a debug probe. Only a Tag-Connect with clips hits that point.

Unfortunately, the Tag-Connect cables are sufficiently expensive (about $50) and fragile that I just see you irritating everybody. Those of us who are used to this will just roll our eyes and buy the cable and wish youā€™d used a standard .050" connector. Those who arenā€™t used to this will probably duck the cable and try to solder to the pads anyway to save the $50. And those who need reliability will pull out a soldering iron to avoid the problems with Tag-Connect.

So, the only people who would be well-served are those who arenā€™t sensitive to a $50 cable and donā€™t easily lean into soldering on a connector. In return you are inflicting a slightly flaky debug connection on them.

You have better feedback on this than I do, but Iā€™m just not seeing this.

I think 0.1" (2.54mm) is out of reach. 0.05" (1.27mm) is worth considering if it can fit.

Iā€™ll look into something like Documentation ā€“ Arm Developer.

1 Like

Iā€™m not particularly against the TagConnect and have used the TC2030 for various MSP430 boards. However it seems particularly awkward to use with the BeaglePlay.

I bought an XDS-110 debugger when experimenting with the PRU on the BeagleBone Black and that works just fine once the headerā€™s soldered on. The TC-2050 cable is fairly expensive for what it is, but then it seems thereā€™s no easy way to connect the TC-2050 to the XDS-110. Of course itā€™s possible to spin up an adapter board, but that seems unnecessary and a bit clunky - especially when youā€™re using the official TI debugger for the Sitara.

Iā€™ve looked at this connector, and I donā€™t see detents on the beagle play that actually let you attach the connector securely.

Thoughts on how this should be secured?

This is the 2050 - note the hooks are expected to plug into something?

I donā€™t see any way to solder on a header - at least with the tools I have.
Perhaps some solder paste and heat the entire post ?
For reference further down (another comment posting) showing the TC-2050 cable and beagleplay JTAG contacts.
Ideas - thanks.

These are spring loadedā€¦ I use: TC2050-IDC-NL Tag-Connect LLC | Development Boards, Kits, Programmers | DigiKey + TC2050-CLIP-3PACK Tag-Connect LLC | Development Boards, Kits, Programmers | DigiKey

Then they just stay in placeā€¦

Regards,

As Robert mentioned, the NL (no leg) version can be used. I donā€™t bother with the clips and just hold it in place when I need to.

When I was trying out the BeaglePlay I used an XDS110 and a custom adapter between the XDS-110 and the TC2050-IDC-NL TagConnect cable. I could successfully detect the BeaglePlaysā€™ CC1352 in Code Composer Studio. However, I didnā€™t have any luck with the Sitara on the BeaglePlay or the CC1352 on the BeagleConnect Freedom.

Thanks for the feedback. I see why the pins on the no clip connectors are angled a bit - thatā€™s what holds the TC2050 connectors in place. Newer way to do this, but 50 bucks .vs. 2x5 header is a bit much, but savings over the entire number of BeaglePlays purchased (consumer product) may be worth it.