How to Access Back-side Teensy Pins?

Status
Not open for further replies.

gfvalvo

Well-known member
Hi all.

Wondering how one goes about making connections to the I/O pins that only appear as pads (with no thru holes) on the back of Teensy PCBs. Is there an adapter board of some sort available? The only other thing I can think of is to surface mount it like an LGA package. Seems beyond the capability of most hobbyist, myself included.

Thanks.
 
There are several adapters, depending on your needs.

For the Teensy 3.2, most of the ones that I'm aware of include:

For the 3.5/3.6 there are:

Be sure to check the licensing conditions on the boards that aren't sold commercially before using them (i.e. KurtE, FrankB's boards).

And of course if you need just one or two connections, you can carefully solder a wire to the pad.

Unless you need the Teensy to be 1.4" long, if you get the Teensy 3.5/3.6, they have more holes available for use with breadboards/prototboards.
 
Last edited:
Again it may depend on your goals and which Teensy.

Michael has a nice list of many of the available options above.

You have the option of just soldering wires to the pads. However you need to be careful as to not tear the pad off the board...

You can get a set of pins to solder onto the bottom of the Teensy. For example for T3.2, I use something like:
http://www.digikey.com/product-detail/en/samtec-inc/TSM-107-02-S-DV-A-007/SAM8883-ND/2685683

TSM-107-02-S-DV-A-007_sml.JPG


Or you can use some of the adapter boards. Note: some of the adapter boards use the above like connector. Others use a castellated approach where you basically create a board that you cut out a location in the bottom of it, that cuts these pads more or less in half, and then you solder this board onto the Teensy including from these half hole pads to the pads on bottom of Teensy. Example of one I made for T3.5/6
T3.5-castellated-bottom.jpgT3.5-castellated-top.jpg
 
When it's time to buy more I'm going with https://www.oshpark.com/shared_projects/Gnvbt7io again. I like having a reset button when I play around with low power (shifting to <24 MHz clock after or at the end of setup). It's not perfect especially if you like using stacking headers but it's cheap and I found it quick and easy to assemble. A issue with any of these is with that many pins it gets to be a pain working it loose from a solderless breadboard. A plus related to that is these all make the Teensy section at least quite a bit stuffer. I suggest popping in the 32khz crystal option when you build whichever one you choose. Doing it later could be a pain.
 
Teensy 3.2 adapter board without need for SMT header on bottom

@DaQue, glad you have a good experience with the Teensy 3.2 Breakout Board R3 at OSHPark. We just put one together and are testing it this weekend. We need to use Teensy 3.2 in a production product of a few hundred per year and I am told the SMT header will be a real hassle in production. We need access to some of those signals. The https://www.oshpark.com/shared_projects/Gnvbt7io adapter board is attractive because it has no SMT header and can use standard socket strips to plug into. My tech tells me the castellations are much easier to align and solder.
 
I used one of the BreakOut boards to put a T_3.x into the shape of the beta T_3.6 PROTO with top headers and that worked well.

I've also used the one FrankB did shown above on PCBS.io - I see PCBS.io is now in the new owners hands - shipping from FL rather than west coast - pricing looks the same and ship times are suggested to be as good or better than before.
 
@DaQue, glad you have a good experience with the Teensy 3.2 Breakout Board R3 at OSHPark. We just put one together and are testing it this weekend. We need to use Teensy 3.2 in a production product of a few hundred per year and I am told the SMT header will be a real hassle in production. We need access to some of those signals. The https://www.oshpark.com/shared_projects/Gnvbt7io adapter board is attractive because it has no SMT header and can use standard socket strips to plug into. My tech tells me the castellations are much easier to align and solder.

It might be worth your while to see if the Teensy 3.5 would work instead of soldering boards to the 3.2. You would get:
  • 16 more digital pins;
  • 13 more analog input pins (11 overlaid of top of the digital pins, 2 pins that are both analog input and analog output, and 2 pins that are only analog input -- note the 3.6 doesn't have these last 2 analog input pins);
  • 6 more PWM pins;
  • Alternative pins for RX1/TX1;
  • Alternative pins for CAN0TX/CAN0RX;
  • 2 more serial UARTs (RX4/TX4 and RX5/TX5);
  • Alternative pins for SPI0;
  • 2nd i2c port;
  • CS1/SCK1 for the second SPI port;
  • More memory, higher clock speed.

Sure, it is an inch longer, the pins in the back row are relocated, and $5 (US) more than the 3.2, but it would eliminate the extra board, and it sounds like it might simplify your build. The 3.5 like the 3.2 is tolerant of 5v on the digital pins (the 3.6 is not 5v tolerant).
 
Last edited:
Good points MichaelMeissner ... that spare inch includes a faster SD card as well - not sharing other SPI bus pins.
 
Good points MichaelMeissner ... that spare inch includes a faster SD card as well - not sharing other SPI bus pins.

I fully agree it would be a much easier build and probably the better solution would be to use the 3.5 over the 3.2 with breakout board but be careful when comparing all the IOs on the 3.2 with breakout to all the IOs on a 3.5 through hole and back side. IMHO a better comparison for your build would be all the IOs on the 3.2 with break out board to the IOs with through holes on the 3.5. Teensy 3.2 f/bb A0-A20 (21) and 0-33 (34) digital vs Teensy 3.5 through holes A0-A9 and A12-A22 (21), digital 0-39(40). The lengths are basically identical to each other when you compare that way too.
 
It might be worth your while to see if the Teensy 3.5 would work instead of soldering boards to the 3.2. You would get:
  • 16 more digital pins;
  • 13 more analog input pins (11 overlaid of top of the digital pins, 2 pins that are both analog input and analog output, and 2 pins that are only analog input -- note the 3.6 doesn't have these last 2 analog input pins);
  • 6 more PWM pins;
  • Alternative pins for RX1/TX1;
  • Alternative pins for CAN0TX/CAN0RX;
  • 2 more serial UARTs (RX4/TX4 and RX5/TX5);
  • Alternative pins for SPI0;
  • 2nd i2c port;
  • CS1/SCK1 for the second SPI port;
  • More memory, higher clock speed.

Sure, it is an inch longer, the pins in the back row are relocated, and $5 (US) more than the 3.2, but it would eliminate the extra board, and it sounds like it might simplify your build. The 3.5 like the 3.2 is tolerant of 5v on the digital pins (the 3.6 is not 5v tolerant).

You just sold me on using a 3.5 over 3.2 and breakout board for future builds that won't on a 3.2 without breakout board. And don't forget the 3.5's floating point unit too.
 
Last edited:
It might be worth your while to see if the Teensy 3.5 would work instead of soldering boards to the 3.2

Thanks for the suggestion. If we can get the I/O we need without using the pads on the back, which it seems we might, I would be sold on 3.5! Will have to look more closely at that. Plus we can use the onboard uSD instead of our own socket. It has a dedicated interface too and doesn't share one of the SPIs? If 3.5 will work, I agree it would be a simpler solution. Assuming the 3.5 is stable when executing the libraries we use. I don't have any sense of how different the execution path is between 3.2 and 3.5/6

What's the state of wired Ethernet support for these parts? We can stick with the WIZ850io module for now and do a revision later to use the onboard RMII. The ship date for this project is nearing so we want to change as little as possible at this point while making the system easier to manufacture. Thanks...
 
Last edited:
What's the state of wired Ethernet support for these parts? We can stick with the WIZ850io module for now and do a revision later to use the onboard RMII.

Stick with WIZ850io for now. RMII is unlikely to be really stable until early 2018. I'm pretty sure there will be progress made on it this year, as I'm doing now for the USB Host port on Teensy 3.6. But I wouldn't recommend depending to use it for an important project in 2017. It's going to take time to really get things highly stable.
 
Anyone know who CPT is on OSH Park? T3.2 BB R3 needs tweaking. Don't order it!

I'll try to find out from OSH Park but there may be privacy concerns.

CPT designed the very useful Teensy 3.2 breakout board R3 (thank you CPT whoever you are) and we have just discovered a design flaw which can lead to shorts from pads to the ground copper pour. We have already built over 250 of these, and about 4% of the first 120 failed post assembly testing... We are pretesting bare boards now and so far in the next 100 we have found one bad board... We are fixing the production files as best we can by editing the Gerber text files directly but it would be nice to change the design source files. Ironic, isn't it: most of the "open source" hardware at OSH Park do not include the source files. Those that do must link to an external repository. So it is really "open Gerber" hardware. Anyway...

Kudos to OSH Park who called this issue to our attention when I placed another order last week.

I will describe the issues in detail with photos and give a link to the improved Gerber files in a couple of days when they are posted to OSH Park. We will be ordering 100 of the tweaked boards and will have quick results to report. We will bump the rev to 3.1 so you can tell them apart.
 
Great, thanks! Will update as promised once we have a revised board which passes OSH Park scrutiny too.
 
Teensy 3.2 Breakout Board R4

Update: we just ordered 9 boards from OSH Park. If they test out and look OK we will publish the design files and make a pull request to HWGuy. BTW the files imported nicely into Altium 16. Here are the changes we made:

All the Teensy pins except Vin and ground don't have signal connections on the board, they just pass through via the soldered pins on Teensy, so making those PTH just makes them harder to solder. So we have made those NPTH with no pads. There were some floating ground pour islands we have fixed. We changed the SW to a washable SMT part and one which is readily available, updated the board rev to R4. Fixed soldermask clearance from pads/traces and increased copper pour clearance (those two caused exposed slivers of the copper pour adjacent to pads/traces if mask was not perfectly registered, which could lead to a short when soldering). Diode and bottom paste jumpers, traces, left unchanged.

Here is a screen capture of the top:
Top_2018-02-07_16-34-12.jpg
Bottom:
Bottom_2018-02-07_16-38-18.png

@HWGuy if you want one of the prototypes I will mail one to you when they arrive. No cost.
 
Status
Not open for further replies.
Back
Top