Is there a third Teensy 4.0 breakout board?

Status
Not open for further replies.
Some members would like to use the FRDM4236 breakout board to test USB host function. Since blackketter made a nice small USB card I will make the FRDM4236-V2. It will bring out a 6-pin female header to be used with blackketter's USB card. So this is a good time to make some changes.

https://forum.pjrc.com/threads/57672-Another-Teensy-4-0-Breakout-Board

https://oshpark.com/shared_projects/4v0NmpMj

I can bring one contact of the ON/OFF switch to a pad near the DAC1 pin, if you would like to make the DAC1 pin as an external ON/OFF switch just short that pad to the DAC1 pin. The other contact of the switch is connected to GND. The DAC0 pin is still not used.

Not be able to bring D24/D25 to those two pins. Need too much changes in layout.
Not be able to bring D+ and D- to those two pins. Need too much changes in layout and too risky.
 
@FRDM4236 :: Got the FRDM4236 board in today's mail. Looks good - amazing so many wires already pulled across the board!

Not sure there is room for them already - mixing USB anywhere past the rest of the T4 - the 2x5 pad lines and the SDIO lines wouldn't have anywhere to land without enlarged board.

A forward piece to pick up the USB D+/- could get VUSB and GND - and bring those pins out as a group of 6 as done and used by the BlackKetter USB board from that breakout.
 
Yesterday I received two packages in the mail, (a) more T4s to destroy, and (b) samples of FRDM4236's board. So I got to work and assembled two of my slightly modified (stretched) versions of Blackketter's board, and one of FRDM4236's board. The good news is that all three worked immediately with no need for rework, and all have been tested for read and write on all 40 GPIO pins. Fantastic!

After just one evening of working with both these breakouts here are some impressions:
1) They are both useful and neither one is better than the other. they are different and serve different functions. I want to have both in my toolkit.
2) I am now a convert to the use of castellated through holes for attaching to underside pads. In the past I've insisted on installing female headers (on loglow's T3.x breakouts) so that the Teensy could be simply unplugged, thus avoiding a major rebuild if something went wrong (which it too often did). With these new breakouts construction is fast - less than 20 mins - and I have no qualms about committing a T4 permanently to the breakout.
3) Soldering to the castellations is easy and nothing no be afraid of. A tiny bit of non-corrosive flux, a narrow tipped iron, plenty of good-quality soldering braid (I use MG brand - there is a lot of junk out there), and you're good to go. Don' panic if you do get a few solder bridges, that's where the solder braid comes in! Just be quick, and don't let the iron dwell on the pads more than is absolutely necessary. Practice with a cold iron to get the tip angle right to contact the the castellation and the pad.

The Blackketter Board:
3top.jpg
a) This might be called a bare-bones breakout because it has no additional components or sockets, and simply exposes the hidden T4 pins in two extra 14 pin columns beside the T4 pins, giving a total of 56 pins. The order of the pins around the board seems somewhat random, (chosen for ease of PC board layout), and you may need to keep a list handy.
b) Because of the adjacent columns of pins this board cannot be used in a breadboard, and that might be a problem for many users.
c) I've built four of these now. Assembly was easy, with no soldering problems. Only the very first one showed any functional problem, where a single pin is held high, and measures about 6.7 kohm to +3.3v. The rest of that device seems to work fine, and the other 3 units are fully functional on all GPIO pins.
d) Just my personal taste, but I like the form factor of the "squat" 14 pin height for fitting to my prototyping boards, and for now I think I will continue to use these breakouts.
-------------------
Aside: I've mentioned previously that for prototyping I use Velleman Eurocards, which are basically FRP 0.1" perf boards with Cu plated horizontal "islands" on the back, and they come with either 1, 2, or 3 holes per island. They are good quality, and I've never had a problem with a pad lifting from heat. The standard Blackketter board only works with one hole islands, whereas I only work with 2 or 3 hole islands. My solution was to use KiCad PCB software to stretch the board in width and include a gap of 0.1" between the two pin columns on one or both edges. The photo above shows the single gap version. However, I found after I built this board (which only works on two hole islands), the the double gap version actually works on both 2 and 3 hole islands, so that's all I be purchasing in the future!. I've had these built successfully in 1.6mm FRP at OSHpark (but not 0.8mm - which was a disaster, as several on this forum found out :mad:).

The FRDM4236 Board:
2grn.jpg
This is a nicely designed board that creates a breakout in the form factor of a T3.5/T3.6, ie a DIP package with 24 pins/column - totaling 48 pins. The 20 additional pins are allocated for GND, +3.3V, 16 GPIO pins, and two unused pins. The additional GPIO pins (24 - 39) are laid out in order around the board) Across the center of the board are two rows of 5 through holes, one for an I2C OLED display, the other for access to the Teensy pins PRGM, ON/Off, VBAT, GND.
The is a mounting area for a microSD socket. an ON/OFF tactile switch and some SMD components (for the SD card only).

a) This is a very useful general purpose breakout, especially for use with a breadboard, and there is no reason why it cannot be permanently built into a project.
b) The current boards are fabricated as 0.8mm PCBs. Assembly was easy (but read the instructions carefully). FRDM4236 has taken care to make soldering the castellations easier by creating physical barriers between adjacent solder points. No problems once you realize that you have to hold the iron at a steep angle. in the slots. FRDM4236 has been quite clever about the design of the castellations for the 1mm pads. He has made it such that there are small "peninsulars" between the pads to minimize solder flow.
c) The unit worked flawlessly from the initial power-up.
d) There is a problem however. There is no way to switch the power source between VUSB and Vin. As with any T4, Vin is wired to VUSB unless the jumper is cut. On this board there is no provision (or any available pin) to put an SPDT switch or three pin jumper header. (Loglow's boards allow for a switch or header to be soldered on). If you apply voltage to USB while an external source is connected there is a serious risk of damage. I cannot see a clean way to fix it here.

My recommendation for any redesign would be to modify the power connections, use the two empty pins for something useful - my vote would be for VBAT and ON/OFF, and perhaps get rid of the OLED pins and that second row of through-holes.

Will I buy more in the future? Absolutely.
 
Last edited:
@defragster: Thanks, as you suggested I extended the PCB and picked up the D+ and D-. I added an external 5V pin for the VUSB of blackketter's board. The new PCB is available at Oshpark. See more info. of the new PCB at http://www.trainer4edu.com/edubase_teensy/frdm_4236.html

@DerekR: Thanks for your nice write-up. Glad you can use it. I will think about using a SPST slide switch for selecting power source.
 
FRDM4236 sent me a board - now assembled to an OSHPark Purple T4.

20191017_005637r.jpg20191017_005614r.jpg

I used tall pins so the bottom female header PJRC Audio Rev D board could be attached.

I did the DMM test and found no shorted/bridged SDIO pins!
Board was not washed - only flux used was from solder core - but after doing each group, before flux got cold, blotted with a paper towel

> Powered up to Blink - opps I had not even powered the Purple T4 before soldering.

> Ran the CardInfo test for T4's BUILTIN_SDCARD, then moved same card to Audio board and it also worked with chipSelect = 10

> Ran Audio sketch ToneSweep and that worked to headphones.

> Tested the FRDM board's POWER button to work OFF and ON

>> Ran test code from loglow board thread (see below) to set all pins PullDown and show when change from LOW ( hit 3.3V on each pin ) , and then PullUp and show when change from HIGH ( hit GND on each pin ):
- No shorts between pins, all pins read Zero or One as expected!

<edit> Here is the Hi Low test sketch. Run it with SerMon active and read screen notes.
View attachment HiLowTest.ino

> Start with a 3.3V wire and touch in turn to each digital pin, watch for single pin change to 1 and 0 on release.
> hit SerMon ENTER to change mode
> Change test wire to GND wire and touch in turn to each digital pin and watch for single pin change to 0 and 1 on release.

** Pins out of sync on test start or after Enter - stuck in wrong state will be shown.
>> Pay attention to GND and 3.3V pins and never touch VIN 5V pin!


More to test - but those are the hard parts?

Soldering was straightforward and error free. I did a quick wipe or two with Solder Braid to make sure the SDIO pins were safe - and to get some off the tall pin tops so the female header could seat.

Going to solder a short female header on the i2c pins and got an SSD1306 that will fit there.

It looks like there is clearance to audio board for the 0.5" Heat Sink in use

>> Looking forward to ordering the USB_HOST capable version of this board … will wait a couple of days for the new design to settle.
 
Last edited:
Soldered on the short female header for i2c. Put on new style SSD1306 with the same pin order ( older/common ones had VCC/GND swapped )

Using Adafruit test code with Wire it is running well on this FRDM4236 board - using those provided mid board pins!

This cute trick easily gets rid of the long delays to make the code less boring to watch - also would be handy to wipe out debug delay()'s quickly:
Code:
#define delay(a) delayMicroseconds( a*10 )

And this one bumps the 400 KHz i2c speed to 1 MHz - where last two params are provided to specifiy optional CLK SPEED 'USE, EXIT':
Code:
Adafruit_SSD1306 display(SCREEN_WIDTH, SCREEN_HEIGHT, &Wire, OLED_RESET,[B][U] 1000000, 1000000[/U][/B]);
>> This takes a subset of Adafruit SSD1306_128x64_i2c.ino draw tests from 8.8 secs to 3.9 secs
 
@FRDM4236 - SOMETHING ODD with the FRDM4236 breakout and i2c.

I can run the i2c SSD1306 display on a PJRC breakout at 912 or 960 MHz with heat sink like on the FRDM breakout. It also runs at 960 MHz on a bare pinned T_4 with no heat sink.

However for some reason at either 912 or 960 MHz the FRDM4236 breakout using the i2c PCB pins noted in p#31 - or wired direct to pins GND/3.3V/SCL/SDA like the other T4's it LOCKS UP!

It does run at 816 MHz.

This board does have a Heat Sink as used on other units where OC speeds work - and this FRDM breakout can run other sketches at 960 MHz - but something about the 1 MHz or 400 KHz i2c and the i2c SSD1306 locks it up?
 
@defragster: Thanks, as you suggested I extended the PCB and picked up the D+ and D-. I added an external 5V pin for the VUSB of blackketter's board. The new PCB is available at Oshpark. See more info. of the new PCB at http://www.trainer4edu.com/edubase_teensy/frdm_4236.html

@DerekR: Thanks for your nice write-up. Glad you can use it. I will think about using a SPST slide switch for selecting power source.

Why not a link to either your OSHPark shared boards or directly to this one. It was a bit hard to find. Or at least one on WWW.trainer4edu.com. 3 of the one without the bump out for USB host on the way to me now.
 
Last edited:
@DaQue: Sorry for confusing anyone. The one you should visit is http://www.trainer4edu.com/edubase_teensy/frdm_4236.html for assembling instructions. On the top of that page you can jump to Oshpark to get the V3 board. I don't remember if I ever posted the http://www.trainer4edu.com. This is harder to find the FRDM4236. I already deleted the V2 board at Oshpark. I will add a note at Oshpark page to remind people to order a correct board.

@defragster: Thanks for your nice write-up. The I2C wiring is straight forward. It should work, I don't know why, may be caused by contamination. Hopefully your next board is OK. I don't like non-clean flux. I have not used it for 20 years. I use water-soluble flux. It's much much easier to solder. Easy to clean, no flux residue, no contamination, but you need to dry it thoroughly. Hold it with a clamping scissors while you use a heat gun (don't over heat it) or blow it with a small USB fan all night. The result is a very professionally cleaned board. You will like it.

For environment protection, the non-clean is the way to go, so it may be a bit guilty for not using non-clean, but not much, because it's so little in prototyping. Less guilty than drinking coke with a plastic straw.
 
I have not used non-clean flux regularly for 20 years, except a very few special cases like this to demo and on a board cannot be cleaned by water.
 
@DaQue: Sorry for confusing anyone. The one you should visit is http://www.trainer4edu.com/edubase_teensy/frdm_4236.html for assembling instructions. On the top of that page you can jump to Oshpark to get the V3 board. I don't remember if I ever posted the http://www.trainer4edu.com. This is harder to find the FRDM4236. I already deleted the V2 board at Oshpark. I will add a note at Oshpark page to remind people to order a correct board.

...

Thanks for a quick answer. I was looking for a V2 link is what my self inflected issue was so I ignored the V3 linked page. The bump out on the side for extra USB support was not something I was interested in and at V2 total Price: $7.15 vs Total Price: $10.10 for V3, and the fact that was the one defragster used that's what I ordered. I highly respect defragster and if its good enough for him its good enough for me.
 
I have not used non-clean flux regularly for 20 years, except a very few special cases like this to demo and on a board cannot be cleaned by water.

I took a soldering class years ago as part of ongoing training related to ISO9000 for a former employer and the trainer didn't fully trust no/non-clean flux. His issue was that its only promised to be inert once its fully activated at soldering temperatures. Hand soldering it's bound to get on spots that don't get that hot.
 
@FRDM4236 I’m curios and if you don’t mind what issues or concerns caused you to break the headers into multiple pieces?
 
@DaQue: I did not intentionally to break it into pieces. I started it with V1 to make a Teensy V3.6 compatible before seeing blackketter's USB host. I was going to stop there, but some forum members would like to have USB host feather available, so I thought about making a small add-on board for blackketter's USB host to be available for members who bought V1. It would be a lot cheaper.

I never used USB host and don't know how other use USB host, so I came up with the V2 which is not the way people will use. defragster suggested to extend the V1 pcb to cover the interface for bk's USB host board. He also kindly let me know how he uses USB Host. That's why I made the V3. I think that the V3 and bk's USB host board will work for USB host applications. For anyone who does not need USB host ( I saw quite a few of them) the V1 is OK. I deleted V2 on Oshpark.
Another reason not to post an Oshpark link on the forum is that you could not change it. Someone may click an old link and order an old product. If you have a few updates on your PCB, it's time consuming to find out the correct link to click, especially the thread is long.* If you have a web page to dedicate a PCB, you can update info. and provide assembling instructions.

Regarding your finding on non-clean flux, I will add one more problem, in wave-soldering or DIP soldering production the pre-heating must be controlled precisely. If it's too cold, many through hole parts will pop out. If it's too hot, the flux will dry out. Also it may cause solder bridges need to be removed, that's a second operation. That's was 20 years ago, now the flux may be improved. I only go to an assembly house that uses water soluble flux and Insist that they must guarantee to wash thoroughly.

Nothing is perfect, since I bought out the water soluble flux, I need to tell bad things about water soluble. I forgot to mention yesterday.

1. It smells bad, may be harmful, use a USB fan to blow air away from you when soldering. For soldering a small prototype it's not a problem, but if you do a lot in a short time, you can do it in your garage or other place, not in bedroom.

2. After soldering you need to wash it ASAP. The acid in flux residues can cause corrosion if you don't rinse it thoroughly. Wash it with warm water and soup, Brush it with a toothbrush. Shake water off before drying it. Refer my previous message for how to dry it. NEVER LEAVE IT OVER NIGHT WITHOUT WASHING IT, or your board will be all corroded next morning! Can never be cleaned again. You can test it on an obsoleted PCB with a few parts. I usually would wash it in an hour.

Is it worth the troubles ? It depends on your preference. The water soluble flux will give you very shining solder joints and a professional look on your PCB to impress your customer, for that reason alone, some people will go for it.
 
@FRDM4236 - SOMETHING ODD with the FRDM4236 breakout and i2c.
...

Unfortunate follow up to this "ODD" post - this OSH Purple T4 soldered FRDM4236 board has gone offline - LDO or something in Pin #1 corner getting warm and no 3.3V is showing up - so not even able to get Bootloader. Didn't do anything I'm aware of to cause this and nothing smoked or burnt - just powered today and nothing after it was on last night working with SSD1306 speed anomaly.

At this time I have that same display running on a non heat sink PJRC T4 at 912 MHz and temp is 63C currently it was 65 a bit ago when the room was warmer. The display is running FAST and well using the updated i2c clock work by @mjs513 at 3.3 MHz. That takes the 3.9 secs reduced from 8.8 secs down to 1.62 seconds per loop of code posted in T4 Beta thread - where the i2c alternate clock code is also posted. It ran about 6 hours at 816 MHz - then about 2 hours ago I upped the CPU to 912 MHz and all is well.
 
Quick follow up on Support from FRDM4236:

Received prepaid 1 day USPS return label and sending that assembled unit back today. He indicated an assembled unit will be on its way to me when ready.

One other diagnostic I did was this as requested and indeed not good results for reason TBD:
Checking all the pins against 3V These pins ‘RING’ continuity on my DMM.
• 25,17,18,20, GND

In the mean time that same SSD1306 display is running fine on another PJRC T4 - a long time on Wire i2c and now an hour or so on Wire1 with no trouble at T4 912 MHz.

** Note - this was one of his $7 shipped premade boards where he had the extras for test.
 
@defragster: From what you explained I am sure that the T4 is overheated and burnt. The V3 and GND and a few other pins are all short together internally. The 3V regulator is in the thermal shutdown mode. The heat does not only dissipates through heat sink, it also dissipates through PCB. The FRDM4236 board basically added an insulation on the T4 for heat dissipation. So it will make the T4 controller hotter. That's a fact. it will not dissipate heat as good as a bare T4 board. If the FRDM4236 works at 820MHz, that's perfect OK with me, even 600MHz meets the spec. Every time you overclock it, it will generate more heat and degrade the chip. Sooner or later it will stop working earlier than its life expectancy. That's maybe what happened to your test. It was OK for a while, then stop working.

I understand that your testing is for evaulation purpose. If you want to make FRDM4236 work with 920MHz, you may find a way to dissipate extra heat, like to use a USB fan to blow it while testing. It could work.

My testing units are all working at 600MHz today whole day. Tomorrow will try to run at 820MHz. If it's all OK I will mail you one on Wed. or Thursday.
I am not going to test it at 920MHz.
 
@defragster: From what you explained I am sure that the T4 is overheated and burnt. The V3 and GND and a few other pins are all short together internally. The 3V regulator is in the thermal shutdown mode. The heat does not only dissipates through heat sink, it also dissipates through PCB. The FRDM4236 board basically added an insulation on the T4 for heat dissipation. So it will make the T4 controller hotter. That's a fact. it will not dissipate heat as good as a bare T4 board. If the FRDM4236 works at 820MHz, that's perfect OK with me, even 600MHz meets the spec. Every time you overclock it, it will generate more heat and degrade the chip. Sooner or later it will stop working earlier than its life expectancy. That's maybe what happened to your test. It was OK for a while, then stop working.

I understand that your testing is for evaulation purpose. If you want to make FRDM4236 work with 920MHz, you may find a way to dissipate extra heat, like to use a USB fan to blow it while testing. It could work.

My testing units are all working at 600MHz today whole day. Tomorrow will try to run at 820MHz. If it's all OK I will mail you one on Wed. or Thursday.
I am not going to test it at 920MHz.

Have to assume I was careless or unaware of IDE speed setting when first moving to SSD1306 before adding the heat sink that generally works on the other boards to regulate temp controllably and have been through a lot more.

@FRDM4236 - your call on the replacement unit and timing.
 
Updated above - smoothed the filler PCB board to T4 could better shift. PinTest shows digital function on all pins including the SDIO #'s 34-39

CardIfno still giving failure? No idea why - two cards both work with same upload to PJRC T4 Beta which should match the wiring and setup for "BUILTIN_SDCARD";



NOTES ON {pt 2}: Teensy 4.0 Ultimate Breakout Board:
Found a 12V supply to external USB HDD bay with a fitting connector (not sure of what size is common - but I have 3 non fitting}:
> USB to PC was powering - showing Yellow VUSB power
> Sometimes - plugging external power maintains the VUSB Yellow?
> Unplug both and plug external and get BLUE VIN power
> Plug in USB and it stays BLUE VIN - expected - perhaps it depends on which is higher 5V?
> Unplug External 12V and Blue VIN stays lit with VUSB? Changed when toggled USB Host switch?
> MidBoard chip 'regulator' Hot to touch with 12V external power - it reads 124F or 51 C - nothing connected or soldered to PCB

External power is 12V 3A supply - USBHost Lan Test:
> Upload Ethernet LAN USBHost dongle code - at 600 or 816 MHz - which should work without heat sink.
>> With USB power and External connected with upload it does start - or USB alone as expected
{ Have USB connect dongle with 5v line switch to cut power but leave data and GND to PC }
>> Remove 5V USB power and powered from Ext VIN - T4 restarts/goes offline - will not run from onboard LDO with 12V input? Button will not take to Bootloader?
When running from USB power - switching OFF HOST power the LAN USB dongle shuts down and then when switched on it reattaches to the network. So that says the PCB and USB on USB power are all good, something about running from External VIN with this 12V supply, which as noted is for a long working external USB bare SATA drive connector box .

>> Found another 12V 1.5A supply with fitting plug that has the same behavior. Upload and run with VUSB and VIN it works. Cut the 5V VUSB line and the unit restarts or shuts down in some fashion. At this CPU speed the Host supplied VUSB it always enough.
- reading the Tindie doc it says 300 ma max at 12V - it is running at 600 MHz - just dropped to 528 Mhz and same result. And when VUSB is cut the LDO quickly goes too hot to hold a finger on. And I see the NOTE - Yes the Jumper is 'PREREG' - I wondered what that was for. I was hoping the external supply would be enough to drive even heavier USB Host loads - like a HDD which never work from Host Usb power through Teensy, where the LAN can.



I'm assuming that you cut the VIN-VUSB trace on the Teensy, as not separating it could cause some power issues. As for the USB Host power, it isn't designed to receive power through the port, but there is an inline diode, so you should be fine.

I'm not exactly sure what you mean by cutting the VUSB trace. Could you explain in more detail? The LDO shouldn't be overheating just powering the teensy - it's possible that there is a short, but I test all the boards before I ship them.
 
copied these Ultimate breakout posts to other thread : forum.pjrc.com/threads/57953-Possible-new-shields

I'm assuming that you cut the VIN-VUSB trace on the Teensy, as not separating it could cause some power issues. As for the USB Host power, it isn't designed to receive power through the port, but there is an inline diode, so you should be fine.

I'm not exactly sure what you mean by cutting the VUSB trace. Could you explain in more detail? The LDO shouldn't be overheating just powering the teensy - it's possible that there is a short, but I test all the boards before I ship them.

Yes, the VIN<>VUSB trace was cut. Tested before installing on breakout with DMM on pads and then plugged it in to USB and it did not power.

RE: "Cut the 5V VUSB line " and "when VUSB is cut" :: if you mean that, I have a USB Dongle with a switch on the 5V line from the USB connector. When switched it lights an LED and disconnects the 5V line feeding the Teensy USB connector - but leaves GND and data lines connected.

The Teensy runs when the LDO gets hot, so not going to shutdown - but was surprisingly hot when fed from 12V external - and also disappointing in supplied power being less than onboard T4 from 5V with regard to powering a USBHost connected device. Would hope for at least 700mA.
 
Last edited:
@defragster: Thanks for your testing, it's a lot of time spent. I am sorry for the problem. I will test here. When it's done and if nothing is wrong, I will send you one for your testing again.

I stopped sharing until defragster new testing is done and with a good result. For anyone who needs a breakout board urgently can get other two breakout boards.

https://forum.pjrc.com/threads/57122-Teensy-4-0-Breakout-Kit

https://forum.pjrc.com/threads/57672-Another-Teensy-4-0-Breakout-Board

FOLLOW UP ON THIS ::
FRDM4236 has done more testing on his end with SSD1306 and a T4 on the FRDM4236 board and it not seeing issues from board design! It works at high speed when properly cooled for those over spec states.

As for the board returned to FRDM4236 and the problem was found to be in the 1062 MCU chip:
Code:
*******************************
Your T4 is burnt, V3, GND and pin 25 are shorted out.  I first removed 3V regulator, then bootloader chip, they are still shorted out.  After removing the BGA controller, no more short circuits.
Short circuits are in the BGA chip.

Apparently when I started the SSD1306 i2c setup the IDE was left in a HIGH SPEED OC state and the heatsink and TempMon and F_CPU_ACTUAL printing was not in place yet and I weakened the T4 and then it gradually failed with further testing. When first setting up the SSD1306 I kept rerunning first to get the i2c speed maximized and I spent time in Adafruit and PJRC code leaving the unit running while discovering the Adafruit params for setting i2c speed and the 1 MHz T4 limit on i2c in current release. During that time the display would stop and I was seeing it as i2c speed issue - when it must have been left in a high voltage OC state with no cooling! … or that 1062 MCU was just not ready for 900+ speeds longer term.
 
Status
Not open for further replies.
Back
Top