UPDATE on 20160106: 21 months to the day after the writeup below, FTDI finally announced that they have started to sell evaluation boards and chip samples for the FT51.

FTDI recently announced two USB-enabled microcontrollers. As is (unfortunately) the trend these days, the announcement appears well in advance of silicon being available.

One is a custom-instruction-set 32-bit micro with features that have been found in various ARMs for many years. The other is an 8051-based design. Limited details are available here.

I find the 8051-based chip intriguing, but am very wary, as I’ve been burned by FTDI in the past.

I’ve used FTDI’s USB-to-serial chips in many designs and have been pleased with those chips, but their Vinculum products have been nothing but an exercise in frustration.

The original Vinculum chip could be programmed with one of a handful of firmware images. As such, it was limited to applications that just happened to need exactly what functionality the firmware happened to provide. Since the firmware was completely a black box, there were all sorts of nasty surprises. I once had a prototype using the Vinculum that logged data to a USB thumb drive, and it killed the drive in under a month because (unbeknownst to me) the Vinculum firmware was repeatedly writing the FAT table and using up all the flash write cycles.

Its successor, the Vinculum-II, sounded promising. It offered user-programmability of their custom-instruction-set 8-bit micro. However, FTDI sabotaged the product from the onset with lackluster documentation. Everything was cloaked as “proprietary”. The instruction set, the register map of the USB peripheral, and the object-code-only driver libraries were all opaque. If their driver libraries worked, all of this secrecy would have been tolerable, but the drivers didn’t work respectably. In evaluation of the chip, I found that the throughput of data (after going through their contrived driver API) passed from one port to another was abysmally slow (a few thousand bits per second).

Perhaps the FT51 will right the wrongs of the previous Vinculums.

Unlike the Vinculum-II, the instruction set is known, and it is the venerable 8051. I cut my embedded teeth on ceramic UV window package NMOS Intel 8051 devices a few decades ago. Back then, C compilers for the 8051 cost many thousands of dollars, but nowadays, there are free ones in the form of SDCC. I abhor Eclipse (the IDE that FTDI will apparently be using), so I can say for certain that I will not be using that.

Unlike the Vinculum-II, it appears the USB peripheral will be documented. It is described as “comparable to popular D12”. It is so popular that a Google search for “usb d12” returns almost nothing of consequence. :( However, I presume that they are referring to the Philips (now NXP) PDIUSBD12 and perhaps the RTL is a derivative of what was used for the FTDI FT120.

A preliminary datasheet for the FT51 is available from FTDI by registering to receive their marketing emails. One quirk that I noticed in that datasheet is that they have two distinct UART peripherals: one is integrated with the 8051 core and is good for only tens of kilobits, and the other is a full-featured UART with performance consistent with a conventional FTDI USB-to-serial chip.

It is pure speculation on my part, but I have to wonder if the 8051 core in the FT51 is the open source implementation by Oregano.

.

back to main page contact