Obviously, everybody has a grab bag of favorite features. But I tend to think there are limits about what can be done.
I tend to think some of the constraints might be:
- PJRC.COM is a small company, This tends to limit the number of designs to 1-2 per year.
- Obviously the higher the price, the fewer people will buy new Teensies. Right now, the sweet spot is the $20-30 range for the consumer. Perhaps it can go a little higher, but I would suspect $40 or $50 may be the limit. I doubt the retail price of the Portenta H7 ($104) would be sustainable for the long term.
- I wonder how important is backwards compatibility? Paul has done a heroic effort to make the Teensy 4.x compatible with the Teensy 3.x if you stick to the main devices (first serial port, first I2C bus, first SPI bus, position of the first 10 analog pins, position of the power/ground/VUSB pins), but of course once you get past the main devices, things are not as compatible between Teensies.
- For many people (including me), having the main Teensy pins be breadboard/prototype board accessible is very important. However, having pins at 0.1" means much bigger chips to bring out more pins, which can drive up the cost.
- I think one of the useful new features in the Teensy 4.1 is the solder pads to add 1-2 more memory chips, including 1 chip that might be persistent.
- How important is USB host, ethernet, and hopefully wifi/bluetooth to people on the board directly?
So one thought that I've had is the basic Teensy 5.0 should be the size of the Teensy 4.1, and if possible the pinout should be the same as the 4.1 for those pins. This way, we won't need a revision E audio shield, etc.
No SD card on the Teensy 5.0. Instead use the area to bring out as many pins as we can, using a standard connector. Perhaps a ribbon connector, perhaps something else.
Come out with daughter boards that can mount either under the Teensy (or over it) to attach to the high density pin output and bring out the various functionality people want (JTAG debugging, display/camera, flexio, micro-SD card, etc.).
It would be nice if we had pads like the Teensy 4.1 for external memory. From one standpoint, it would be nice if there were two versions, where one had the chips attached, and the other didn't have chip support (given how few places in the world there are to order psram chips). But I can understand having multiple models is a cost to PJRC.
Personally, I would like to see a lipo battery JST-2 pin connector and a charger chip to charge lipo batteries from USB, but I can certainly live without it.
An alternative might be to grow the Teensy 5.0 by 0.2" wide. Have the standard pins from the Teensy 4.1 on the inner row, and all of the extra pins on the outer row. That would give us 48 additional pins. I don't know if we want the USB host, ethernet, and the inner pins to be in the same location as the Teensy 4.1. It might make sense to move these to the outer row pins, but I can also understand keeping them in the same position.
The main way I would probably use this is to mount stacking headers on the main pins, and then only mount headers above the board for the outer row pins. That way people wanting to use a traditional breadboard for the main pins can do so, but you could use jumper wires to connect the other pins. People who design PCBs, could design it to take both rows of pins.