FT232 versus FT231

Overview

In his Adafruit interview Fred Dart, the CEO of FTDI, talked about the differences between the UART FT232 and FT231.

For the design of the Nusbio, I started with the FT232R and later switched to the FT231X.
I spent a good amount of time studying the 2 chips.

Nusbio G1 - Early version based on the FT232R

I was surprise to hear that:

"it can only drive out at 3.3V max though it has 5V tolerant inputs"

and also in the interview it was not mentioned that the FT231X has a 1024 bytes (512+512) transfer buffer, versus the FT232 which only has a 384 bytes transfer buffer (128 + 256). The buffer size in synchronous bit banging mode makes a big difference in terms of byte/second transferred.

Nusbio G1 - Based on the FT231X

The FT231X has multiple advantages over the FT232R

1. Price
   $2.12 over $4.50 for one chip.
2. Buffer size 1024 over 384. A larger buffer increase the transfer performance when using the synchronous or asynchronous bit banging mode. I do not know if this make a difference in UART mode (I supposed it should) but with Windows the serial port are limited 128 000 bauds.
3. Package
   SSOP-20 over SSOP-28, easier to solder. For both I would not use a regular solder iron. I recommend to use an "electric skillet"or toaster. Obviously the FT231X take less space on the board.
4. Voltage
   Based on the FT231X datasheet the 3.3 Volt out should feed VCCIO, therefore all 8 GPIO would output 3.3V. Making it a 3.3V devices.
   For now in Nusbio I feed VCCIO with the USB 5V or 3.3V configurable with a jumper or a soldered connector. By default it is set to 5V, making Nusbio a 5V device.
   
   I need to check with FTDI if that is ok, though it seems to work and Fred Dart confirmed the input are 5V tolerant. The question is the output and VCCIO are 5V tolerant.
   
   At the URL http://www.ftdichip.com/Products/ICs/FT231X.html, I found this
   

         True 3.3V CMOS drive output and TTL input.
         (operates down to 1V8 with external pull ups and is also 5V tolerant)

   From Twitter @FTDICHIP sent me back the following:
   @MadeInTheUSB the VCCIO is really meant to be 3.3V Absolute max ratings as quoted are this. It will withstand a 5V input but output at 3.3V.
   
   I am going to update Nusbio board to send 3.3V to VCCIO, this mean the 8 GPIOs will send 3.3V.  Nusbio VCC will remains 5V with up to 500 mA.
   
   
5. Clock
   With the FT232R, there is an issue with the internal clock in bit banging mode. It can only be viewed with an Oscilloscope. The clock is not stable. If you bit bang 1 0 1 0 1 0, you would expect a stable 50% duty cycle what ever the baud rate. But what I saw on the Oscilloscope was all over the place. I found one person on the FTDI forum that remarked the same problem. Adding an external clock fixed the problem (Adding an external clock does not increase performance).

The FT231X has multiple disadvantages over the FT232R.

1. The USB data lines require each a 27Ohm resistor and a 47pF capacitor.
   More work to build a device
2. FT232R is definitely 5 or 3.3 volts chip, where the FT231X is a 3.3V chip (though the gpio input are 5V tolerant. And gpio can be configured to 5V, but not "recommended" by FTDI)