Another Teensy 4.0 Breakout Board

defragster said:
Good work - fast. So a CAP isn't called for to support the SD power needs? Loglow breakout has 3 caps - one for USB - I assumed one was for SD.
Click to expand...

Modern uSD cards can have large current pulses. 512KB flash pages are common and write times must be fast to achieve 100 MB/sec rates. so large pulses can occur.

Here is the spec for for max current :
Code: 
6.6.3 Current Consumption
The current consumption is measured by averaging over 1 second.
• Before first command: Maximum 15 mA
• During initialization: Maximum 100 mA
• Operation in Default Speed Mode: Maximum 100 mA for SDSC and SDHC
  100mA (XPC=0) or 150mA (XPC=1) for SDXC
• Operation in High Speed Mode: Maximum 200 mA
• Operation in UHS-I Mode: Maximum 400mA (UHS50,DDR50) or 800mA (UHS104)
• Operation with other functions: Maximum 500 mA.

These are averages so there can be large pulses.

Here is a typical pulse and "mysterious house keeping" current.

SanDiskCard.jpg

The nearly 400 ma pulse is writing to the data area of the card. The later "flush data" is probably just a write to the FAT area. Modern cards treat these areas differently.

Teensy 4.0 won't get sustained UHS104 current since the 1.8V signal supply pin is hiding on ball J6, NVCC_SD0, which is not accessible. You can't switch to UHS-I mode.

A Teensy 4.x with a uSD socket and 1.8V signal capability will need to supply 800 ma of 3.3V in UHS104 mode.

I hope Paul produces a T 4.x capable of UHS104. I designed SdFat for Teensy 4.0 to do infinite size writes since the uSDHC controller can support infinite write sequences.
 
Bill Greiman said:
Modern uSD cards can have large current pulses. 512KB flash pages are common and write times must be fast to achieve 100 MB/sec rates. so large pulses can occur.
...
A Teensy 4.x with a uSD socket and 1.8V signal capability will need to supply 800 ma of 3.3V in UHS104 mode.

I hope Paul produces a T 4.x capable of UHS104. I designed SdFat for Teensy 4.0 to do infinite size writes since the uSDHC controller can support infinite write sequences.
Click to expand...

@Bill G: you should post about this on the pjrc.com/threads/57842-Future-Teensy-features-amp-pinout - "Next Teensy" thread
 
blackketter said:
Yeah, they are. With ample flux and a little cleanup from excess solder with a braid mine went pretty easy.
Click to expand...

If there are favorite sites that discuss things like flux, etc. it might be helpful for those of us whose knowledge ends with which part of the of the soldering iron to hold. :)
 
What size/type capacitors are people using for the USB host and SD cards? Are there other protections that we should be thinking about? Do we need diodes in the USB to prevent reverse current flow?. How about the ability to provide separate power and regulate it to 5v? Obviously with some of these things, you would need to go to daughter cards to add the extra components. I saw people designing the daughter cards, but I wasn't sure whether they had been used or not.
 
MichaelMeissner said:
What size/type capacitors are people using for the USB host and SD cards? Are there other protections that we should be thinking about? Do we need diodes in the USB to prevent reverse current flow?. How about the ability to provide separate power and regulate it to 5v? Obviously with some of these things, you would need to go to daughter cards to add the extra components. I saw people designing the daughter cards, but I wasn't sure whether they had been used or not.
Click to expand...

The USB Host daughter card I'm working includes the capacitors and the protection chip provides reverse current flow and ESD protection, but not any voltage regulation. This circuit was lifted from the Teensy 3.6 schematic.

I added a decoupling cap to the SD breakout board design, and I'm not sure what additional capacitance (beyond what's provided by the Teensy) is needed. It'll be good to look at the power line on a scope while writing and see the behavior.
 
Well, that's a total fail and a huge disappointment! Mine just arrived, and all three are headed for the trashcan.

1) The lower cut-out is completely missing, ie it's solid!
2) The plating on the the top cut-out through holes is mostly missing, and in a couple of cases was detached from the board and dangling in space! Not fixable - there is absolutely nothing to solder to! Looks like the plating was torn off by the mill bit.

I'm not about to waste a couple of T4's trying to fix this.

p,s. - these are the thin ones, which maybe why the plating tore off???
 
Last edited:
Yep - just like mine.

I'll look into oshPark's policies on returns/manufacturing errors. But I don't know about the hole plating on such small holes, it may be just not feasible...

@Blackiketter - Were your latest batch of boards ok?
 
Last edited:
Let me know what OSHPark policy looks like - the SD holes tearing out may be a design issue for the layout as ordered {thin board - heavy copper - not enough room to cut} - but missing the 2x5 routing when they did the large area seems like an error.

The large square looks similar between red/yellow p#59 - though the rout may be a tad higher based on the trace below? The USB pads look usable.

If the lower 2x5 were routed - the SD traces might pickup solder?

Maybe a redesign with a trace top and bottom - where one on each hole would not actually route anywhere - but provide a securing point for the through plating?
 
Seeing as I already have the board in KiCad I'll take a shot at playing with the layout, drill size, pad sizes, etc
 
DerekR said:
Seeing as I already have the board in KiCad I'll take a shot at playing with the layout, drill size, pad sizes, etc
Click to expand...

Cool. Looking at the right unit in p#59 pic above you can see the hole plating tore but is hanging on at the trace in two spots - so maybe a pad/trace on active and unused side might secure it from both ends for better results.
 
DerekR said:
Seeing as I already have the board in KiCad I'll take a shot at playing with the layout, drill size, pad sizes, etc
Click to expand...

If you could add labeling for the pins on the underneath side, it would be very helpful.

Also, I was noticing that pins 24-33 don't seem to be laid out logically.
 
I was just playing with rectangular pads. I think they would be more resistant to tearing from the board.
However, this IS blackketter's project. I think we should respect that, and work with him...
 
Sorry about that, guys!

I suspect that it's related to the thinner material. I haven't yet ordered the thinner ones, I'll do it now, along with another set of thicker ones.

@DerekR: The OSHPark folks are really helpful. Please let us know if you hear any feedback from them.

Thanks for your patience.
 
blackketter said:
Sorry about that, guys!

I suspect that it's related to the thinner material. I haven't yet ordered the thinner ones, I'll do it now, along with another set of thicker ones.

@DerekR: The OSHPark folks are really helpful. Please let us know if you hear any feedback from them.

Thanks for your patience.
Click to expand...

Are you to try some changes before ordering the same thing? Very odd the 2x5 routing got skipped.
 
I wrote to OSHPark and included the picture from above and got this response right away:

Hi Dean,
​
That's not unexpected, and an unfortunate example of why we currently aren't able to guarantee castellated areas on a design.
​
Each fabrication run is subject to interpretation by the operator at the time, so even if one returns as expected, a subsequent run on the very same service may not. It's possible the fab operator handling that 0.8 mm thick batch deemed those pads more likely a header, or simply decided that attempting a cutout there too risky. Via plating and pads, especially on smaller features may randomly be torn by the milling process.
​
So in the end, while we do allow the attempt and often it succeeds when following the guidelines, there's always a chance it may not work and why we don't currently guarantee them.

Chris
Click to expand...

They did tell me when I asked about castellations earlier that they couldn't guarantee it. I went ahead and ordered a set of 0.8mm thick boards as a test anyway.

I'm guessing that to get this done reliably we'd need to work with a fab house on a reproducible process.
 
An important factor for producing castellated holes is reversing the routing direction(climb cut instead of conventional cut) as to not rip out the plated hole.
A different end mill might be necessary as it has also to cut trough copper which is harder than glass fibre composite.
 
I guess I should have waited 2 more days. I was quietly watching this as wanted SD and USB for my Teensy 4.0's so I ordered a set of the PCBs last 2 nights ago in the .8mm thickness so I suspect in a week or so I'll end up with some boards that won't work as well :)

See how they are when they make their way to Canada
Jeff
 
I fear already that mine will look the same when they’ll arrive here in France in a few days... Why don’t they use a precise laser to do the cut-outs?
 
I've been looking at your breakout board for a week or so (along with the USB host companion board), waiting to finish a design of my own before ordering them all from OSH Park.

Should I order it in 1.6mm, or are these also hit-and-miss ?
 
Theremingenieur said:
I fear already that mine will look the same when they’ll arrive here in France in a few days... Why don’t they use a precise laser to do the cut-outs?
Click to expand...

The shiny gold double thick metal plating would be an issue for a laser too? I saw a water jet cut a cellphone and other stuff in half the other day, not sure that would have the needed accuracy.

Will be interesting if what you get matches the photo or the desired board.

Only KurtE noted on the thicker board - properly made it seemed but not as easy to solder over the thicker board?
 
trigger said:
I've been looking at your breakout board for a week or so (along with the USB host companion board), waiting to finish a design of my own before ordering them all from OSH Park.

Should I order it in 1.6mm, or are these also hit-and-miss ?
Click to expand...

I've gotten the SD card sidecar and the USB Host sidecar in .8mm and the ESP32 sidecar in 1.6mm. Assembled and all working fine. I haven't tested the 3.3v regulator on the ESP32 board yet, running off the built-in 3.3v from Teensy seems to be ok, even while doing some basic wifi.

I have a bit of software to write to get the firmware updated on the ESP32 module, but with the factory firmware I'm able to associate my phone with the module.

For the USB, SD and ESP32 sidecars, I don't think it matters what thickness. If you are ordering from OSHPark, the thicker versions are likely to ship faster.

For the Teensy 4.0 breakout board, folks have had poor success with the 0.8mm process, my single order of 1.6mm came back fine.

So, unless you have a specific need, I'd go with the 1.6mm for everything now.

IMG_5453.jpg
IMG_5454.jpg
 
You must log in or register to reply here.
Back
Top