FTDI 8-bit Microcontroller - FT51A

[benSTmax]

It seems FTDI jumped in the 8-bit microcontroller bandwagon with their FT51A http://www.ftdichip.com/Products/ICs/FT51.html

Did anybody used this MCU yet?

A brief presentation:
1. CPU core is 8051-type running at 48MHz with a claimed 48MIPS@48MHz
2. USB, UART, PWM, I2C, SPI, 8-bit parallel FIFO, 8-bit ADC
3. UART/SPI/FIFO with DMA functions
4. Up to 16 digital I/Os
3. Lowest cost variant (when buying directly from FTDI. I couldn't get this parts on Digikey) is for FT51CS and its price is: £205.00 (or $292) for 100pcs

More information about the FT51A:
FT51A Datasheet:
http://www.ftdichip.com/Support/Documents/DataSheets/ICs/DS_FT51A.pdf

Development board, Evaluation board and Debugger:
http://www.ftdichip.com/Products/Modules/FT51AModules.htm

FTDI official press release (dated Jan 6th 2016):
http://www.ftdichip.com/Corporate/Press/FTDIPR63.pdf

and a relatively new (Jan 7th 2016) press article on this MCU:
http://www.electronicsweekly.com/news/products/micros/ftdi-chip-adds-support-boards-for-ft51a-mcu-2016-01/

I tried to look for more reasons or advantages of this MCU over the similar existing ones but I couldn't find something compelling.
Similar parts from SiliconLabs (EFM8UB10F16G), Cypress (CYPD2103) or Microchip (PIC16F1455) cost ~$1/piece or less for 100pcs quantity (on Digikey).

I am trying to understand why would someone use these FTDI FT51A parts which are more expensive than other similar micros.

Would you use them(FT51A) instead of the other similar parts enumerated above (and others which I did not) and if so what would be the reasons for doing so?

[amyk]

I guess they're just betting on people recognising the FTDI name, but after things like this (ironic that this thread is right next to it)...

https://www.eevblog.com/forum/microcontrollers/ftdi-gate-2-0/

...I don't think I'd want to use them.

[dannyf]

The datasheet has a lot more information. Seems to be a fairly advanced / feature rich 8051.

Maybe they are targeting those applications with 8051/USB?

[benSTmax]

The datasheet has a lot more information. Seems to be a fairly advanced / feature rich 8051.

Except for the integrated hub, I cannot see much advantage over the SiLabs EFM8UB1 series https://www.silabs.com/Support%20Documents/TechnicalDocs/EFM8UB1_DataSheet.pdf
The price difference between the FT51A and the EFM8UB1 of ~$2 is too much to pay for the integrated USB hub. If I really want to hook-up 4 USB-MCUs on a single USB connection to the USB Host, I would rather use 4 x EFM8UB1 & an USB Hub chip, rather than chaining 4 x FT51A.

Variant 1 - EFM8UB1 and USB Hub chip
4 x EFM8UB10F16G + 1 x USB Hub Chip (4 ports chip from Cypress, TI or Microchip) = 4 x $1 + 1 x $1.50 = $5.50

vs

Variant 2 - FT51A only
4 x FT51A = 4 x $2.92 = $11.68

hard to beat the price point ...

[andersm]

They've also introduced the FT90X series of 32-bit MCUs, which uses their own, proprietary core. The architecture documentation hasn't been released yet, but it's a RISC with 32 GPRs.

When you can get an M3 with USB for <$2 and a M0+ for <$0.5, why bother?

From their perspective, the ubiquitous support for USB means they have to find new business somewhere, hence eg. their display controller ICs. For customers, I suppose if you've been using their products for USB connectivity they might be worth a look, especially if the software support is good.

[westfw]

I found the FTDI documentation/manuals for their VNC2 chip to be particularly awful (well, closer to non-existent, I guess. ("Just use the library functions that we provide!"))  And no "community", either.  Sigh.
If the FT51x series is using a standard 8051 core, I guess that would be a big help...

[Howardlong]

They've also introduced the FT90X series of 32-bit MCUs, which uses their own, proprietary core. The architecture documentation hasn't been released yet, but it's a RISC with 32 GPRs.

They're going to have to work very hard to sell that into anything. New skills, new toolchains, new architecture, and from what I could see, no USP.

[benSTmax]

I still cannot understand why trying to come up with an MCU these days. The profits in this area are thinner than ASPs and in order to get some market-share fast, they would need some disruptive product or a similar to existing solution MCUs but with an outstanding support.
I cannot see any of these with FT51A nor FT90X. Price is higher than a Cortex M0 w USB, SW and firmware support is not anywhere near their competitors in this market.
I wonder why they didn't focused on the things they do best. Like bridging solutions with good driver support.
One other idea that comes to my mind is maybe they're trying to create a bigger portfolio expecting to be bought by a larger semi company (see all the acquisitions throughout 2015 till now) at a better price. More dough for the shareholders in the case of an acquisition. Maybe Microchip will buy them out with the "change" they got from their recent Atmel acquisition ... 

[andersm]

They're going to have to work very hard to sell that into anything. New skills, new toolchains, new architecture, and from what I could see, no USP.

At least they're using a GCC toolchain, so your knowledge will transfer. They've also pushed their port upstream, so at least it looks like they're trying to do the right thing. If you really cared, I suppose you could trawl through the binutils sources and figure out the instruction set that way.

[Howardlong]

They're going to have to work very hard to sell that into anything. New skills, new toolchains, new architecture, and from what I could see, no USP.

At least they're using a GCC toolchain, so your knowledge will transfer. They've also pushed their port upstream, so at least it looks like they're trying to do the right thing. If you really cared, I suppose you could trawl through the binutils sources and figure out the instruction set that way.

It's also Eclipse based... but then so is TI's CCS and NXP's LPCXpresso, and despite using the same IDE and even core, they work quite differently from an end user perspective, they almost might as well be two different IDEs. I'm just shuddering thinking how bleeding edge it's going to be, I wonder if they bought the core in, or it was the result of someone's postgrad project?

[andersm]

It's also Eclipse based...

It's just an IDE, you can use whatever you like. There's even a separate programming tool and a GDB bridge. However, I don't know if these components are open-sourced, or the protocols documented, so development might be Windows-only for now.

[f4eru]

I wonder why they didn't focused on the things they do best.

Coz it's a sinking ship. USB<>UART is a temporary solution in the long term.

I wonder why they didn't focused on the things they do best. Like bridging solutions with good driver support.

I would say they have "driver  includes malware" reputation now, so Eeeeek

[up8051]

Hello,

I'm trying to use FT900 chip from FTDI, I bought a starter kit and debbuger.
I use Eclipse based toolchain version 1.2.0.

But at this moment:
1. Technical documentation of the chip is poor.
2. Debbuger is very slow and hangs very often.
3. Under debbuger there no acess to peripherial registers, it is unimaginable for me  .
4. Degugger have some bugs and its rather not useable.
4. Only simple examples are avaliable but they do not show the main advantages of the FT900 - streaming
5 USB stack (from MCCI) is not avaliable yet.
6. No technical forum or commuinity

I tried FT900 because of build in HighSpeed USB PHY and I2S port.
I'm waitng for new version of toolchain (1.2.1) but I do not expect a big improvement.

Overall rating

Regards,
up8051

[mikeselectricstuff]

I can't see how anyone in their right mind would risk designing in a part like this from an unknown in the CPU market, unless it offered some massive benefit over exiting products.
Who knows what availability will be like? Toolchain quality?

If nothing else, the investment in learning a toolchain that is no use for any other potential alternative products is a major cost.

I had a brief play with the VNC2 ages ago and it was pretty terrible - poor docs and crashy, cumbersome tools.

There are plenty of other parts out there with USB functionality so I really can't see why they've done this