Andrew Starr's STM Project Journal

Adding a Serial Flash Device to the RCM5700

As mentioned previously, the STM will need a file store from which to serve user interface web pages to a requesting client. As the Dynamic C development system does not currently support partitioning the program flash for the use of the FAT file system library, I have had to add an external flash chip. An examination of the RCM5700 schematic, and of the module itself, revealed that Rabbit have made provision for adding a serial flash chip to the module, but have simply not loaded it for the RCM5700. Presumably, a future Minicore module based on the 5000 series will feature a serial flash.The AT45DB081D is a 1Mbyte flash with an SPI interface. The device footprint is an SO-8, so it is fairly easy to solder. The 2 associated resistors and the decoupling cap are 0402, but 0603 components will just fit if you are careful. I used a 30x binocular microscope and fine tweezers, which greatly assist when hand-soldering SMD.

These lines were added to SFLASH.LIB to configure the system to use the flash device:

   #elif RCM5700_SERIES
        // RCM5700 with added ATDB45081D
        // define Serial Flash on serial port B, chip select on PD6
        #define SF_SPI_CSPORT                 PDDR
        #define SF_SPI_CSPIN                    6
        #define SF_SERPORT                    SBDR
        // serial port RXB and TXB are both hosted on parallel port D
        #define SF_SERPORT_ALTRXTX    0x11

The flash was tested with the sample utility sflash_inspect.c (found in samples/RCM5700/serial_flash). I had to comment out the lines at the top of the source that check the board type, to prevent the compiler throwing an "unsupported board type" error. This utility allowed me to read and write ASCII data to the flash.

A word of warning: if you want to try this yourself, please bear in mind it will most likely void your warranty, and if you break your module, Rabbit Semiconductor most likely won't want to know. In my opinion though, the modification is a low-risk one.

Block Diagram, Rev A

Here's a block diagram of the STM, showing the major functional parts of the system.

 

Outline of Proposed STM Design

The following is a summary of the major design decisions I have made so far. Each section will be
covered in detail in future posts as the project progresses.

Mechanical construction

• Scanner

I've chosen to use a 'tube' piezo with quadrant electrodes; this seems a mechanically compact and simple solution (compared to tripods, bimorphs etc.). It also seems to be by far the most popular choice in other STM designs I've seen.

• Coarse approach

The problem of bringing the sample into tunnelling range of the tip without adding excessive vibration or mass to the system seems ideally suited to a lightweight precision linear motor. PiezoMotor AB (www.piezomotor.se) make such a product, the Piezo LEGS motor. I have selected this motor for the STM coarse approach mechanism.

• Vibration isolation

By keeping size and mass down to an minimum, I'll try go get away with a small stack of steel plates or ceramic tiles separated by rubber o-rings. If I need to go to a suspension-type system, hopefully this can be minimised by cunning use of eddy current damping with Halbach arrays of rare-earth magnets or something.

• Tunnelling tip

I have some narrow-gauge Pt-Ir wire and tweezer cutters purchased from SPI Supplies.
Hopefully the 'angled cut' technique will work to yield tips that are suitable for atomic-
resolution imaging, otherwise I'll need to investigate the various other techniques out there (such as electrochemical etching).

• Carrier

The carrier on which the scanner, sample stage and coarse approach will be mounted will be an approximately 'U'-shaped piece of Macor machinable ceramic. With a ceramic tube scanner, aluminium oxide rod (in the linear motor) supporting the sample stage, and a ceramic carrier, it is expected that the effects of thermal drift will be kept to a minimum.

Electronics

• Microcontroller

The microcontroller will implement all signal processing, waveform generation and comms with the host computer. I have selected the RCM5700 from Rabbit Semiconductor for several reasons:

1. the Dynamic C development system is free and comes with extensive application libraries.
2. The price of a module and evalution board is quite reasonable (mine cost $US49).
3. The module specifications are quite good - lots of I/O, 50MHz clock, 16 bit internal bus, 1Mb of flash and 128kb of SRAM, onboard Ethernet.
4. A free port of the real-time kernel uC/OS-II comes with the development system.

• Tunnelling current amplifier

There are many ways to implement an amplifier for converting the 10nA or so of tunnelling current to a useful voltage. At the moment I'm favouring the design featured in the article, "A low noise and wide-band ac boosting current-to-voltage amplifier for scanning tunneling microscopy" by Dae-Jeong Kim and Ja-Yong Koo, published in the Review of Scientific Instruments 76, 023703 (2005). This design attempts to overcome the bandwidth limitations inherent in amplifiers having large feedback resistances with parasitic capacitances, while retaining good noise performance.

• Piezo drive

I have opted for a piezo voltage range of +/-35V, a compromise which should yield a reasonable scan range,but is still low enough to be safe, and not require expensive high-voltage opamps.

• Coarse approach control

A Piezo LEGS motor driven in 'nano-stepping' mode is capable of 10-20nm resolution. However, nano-stepping requires several complex DAC-generated waveforms, so this task may well be farmed off to an auxiliary processor, responding to digital commands from the RCM5700.

• Sample bias

Sample bias voltage will be user-selectable, +/- 0-10V, generated by a DAC. 10 bits resolution should be sufficient. In practice, the bias voltage will usually be less than 1V, but surface modification experiments will require a larger voltage range.

Software

• User interface

I have decided that the STM will be controlled by the user via a Java applet in a web browser. This allows the control software to be integrated into the STM, with no software needing to be installed on the host PC (aside from the Java VM, which will most likely be installed already). The web browser will download the Java applet from the STM, and execute it. The applet will then open a TCP connection back to the STM to exchange control and image data.

I will probably model the data protocol on something like SCPI.      

(http://en.wikipedia.org/wiki/Standard_Commands_for_Programmable_Instrumentation)

• System control

Since the STM will have a number of processes running concurrently, such as ethernet comms, the tip-sample servo loop and the X-Y scanning waveforms, I will be using a real-time kernel - uC/OS-II. The Java applet and support HTML files will be stored on a FAT file system located on a serial flash chip. A web server process will serve these files on demand from the host PC, and an FTP server process will accept updated files for development purposes. Other uC/OS-II processes will implement the various STM functions.