Fullscreen HTML5Med Res (??MB)HTML4

The Pico Digital Video Terminal allows you to communicate with and control a Micromite, PicoMite or WebMite with modern, easily obtainable equipment such as USB keyboards and HDMI displays. Raspberry Pi Pico Digital Video Terminal Part 1: by Tim Blythman B oot-to-BASIC computers like the Micromite and PicoMite are an easy way to learn about programming, but they can still leave you tied to a fully-fledged computer as a way to communicate with them. Alternatives like the VGA PicoMite can stand alone but require legacy gear such as a VGA monitor. With this project, you no longer need a PS/2 keyboard or VGA monitor! Put simply, the Pico Digital Video Terminal is an updated version of the ASCII Video Terminal from July 2014 (siliconchip.au/Article/7925). The ASCII Video Terminal allows a microcontroller with a serial interface to connect to a PS/2 keyboard and VGA monitor or composite video display. The Pico Digital Video Terminal does much the same, although it provides more modern interfaces. It also provides a virtual USB-serial port to allow such a microcontroller to easily connect to a serial terminal program such as TeraTerm or minicom. It is modular, allowing some functions to be left off or customised. Instead of a PS/2 keyboard, it uses a USB keyboard. Even wireless USB keyboards are cheap and plentiful these days, so you can easily go wireless. The Terminal also delivers digital siliconchip.com.au video from its HDMI connector, allowing it to be connected to most modern TVs and monitors. There are even compact HDMI displays designed for computers like the Raspberry Pi, which should also work well. The signal is not strictly HDMI-­ compliant but uses a backwards-­ compatible Digital Visual Interface (DVI) set of supported resolutions that are supposed to work with all HDMI-­ compatible displays. We can’t strictly call it HDMI as such; licensing restrictions exist on using the ‘closed’ HDMI standards and trademarks. Regardless, the Terminal has worked successfully with all HDMI-equipped displays we have tried. We considered calling it “I can’t believe it’s not HDMI!” We have also implemented a USB host interface to communicate with the virtual USB-serial device on boards like the Micromite and PicoMite. The Digital Video Terminal features & specifications » » » » » » » » » » » 640×240 pixel monochrome (80×30 character) display option 320×240 pixel colour (53×20 character) display option HDMI socket with DVI-compatible digital video USB-A socket for keyboard (works with wireless keyboards) VT100 terminal compatibility USB-C socket for 5V USB power Three status LEDs Fits in a compact enclosure (105 × 80 × 25mm) Tested with the Micromite, PicoMite and WebMite Turns a development board into a standalone computer Works with other USB-serial capable boards, including: Raspberry Pi Pico/Pico W (including CircuitPython & MicroPython); Arduino Leonardo; CP2102 USBserial converters; and Micromite/Microbridge » Baseline DVI output over HDMI connector » USB host for keyboard » Flexible and modular design Australia's electronics magazine March 2024  45 Parts List – Digital Video Terminal 1 double-sided PCB coded 07112231, 98 × 68mm 1 black double-sided PCB coded 07112232, 99 × 22mm (front panel for H0190 enclosure) OR 1 black double-sided PCB coded 07112233, 99 × 27mm (front panel for H0191 enclosure) 1 ABS instrument case 105 × 80 × 25mm [Hammond RM2005STBK; Multicomp MP004813; Altronics H0190] OR 1 ABS instrument case 105 × 80 × 30mm [Hammond RM2005MTBK; Multicomp MP004811; Altronics H0191] 1 Raspberry Pi Pico programmed with 0711223A.UF2 (MOD1) 1 Raspberry Pi Pico programmed with 0711223B.UF2 (MOD2) 1 Raspberry Pi Pico programmed with 0711223C.UF2 (MOD3) 1 HDMI-compatible socket (CON1) [Stewart SS-53000-001] 2 USB-A through-hole right-angle sockets (CON2, CON3) 1 USB-C power-only SMD socket (CON4) [GCT USB4135 or similar] 3 6mm through-hole tactile switches (S1-S3) 4 2-pin headers (JP1-JP4) 1 4-pin header (LK1) 6 jumper shunts (JP1-JP4, LK1) 6 20-way header pins (optional; for MOD1-MOD3) 6 20-way header sockets (optional; for MOD1-MOD3; will not fit in H0190 enclosure unless low-profile types are used) 4 self-adhesive feet to suit the enclosure (eg, 8mm round) Semiconductors 2 2N7002 SMD N-channel Mosfets, SOT-23 (Q1, Q2) 3 green 3mm through-hole LEDs (LED1-LED3) Resistors (all M2012/0805 size SMD, 1/8W, 1%) 6 10kW 2 5.1kW 3 1kW 8 270W 4 22W The Terminal is a compact device that fits into a small ABS enclosure. The three Pico microcontroller boards communicate with a USB keyboard, provide a virtual USB-serial host interface and deliver a digital video signal. The Pico or Pico H version will work (both are shown here). Short-form kit (SC6917, $65): includes everything except the case. Choose which front panel PCB you want (for Altronics H0190 or H0191). Picos are not supplied pre-programmed. 46 Silicon Chip Australia's electronics magazine Terminal also provides an ‘upstream’ USB-serial device interface that is transparently passed through to the downstream USB-serial device at CON2. This allows these boards to behave as though they were connected directly to a computer, even though the Terminal sits in between. This arrangement enables data from the attached keyboard to be sent to the Micromite or PicoMite. Similarly, data from the Micromite or PicoMite can be displayed on the digital video output, while still being monitored by the computer. The Terminal will also work with many low-cost USB-serial adaptors, allowing it to communicate with a computer. We have verified that those based on the CP2102 chip work. VGA PicoMite comparison You might also be wondering what the Pico Digital Video Terminal offers that the VGA PicoMite does not. The VGA PicoMite (July 2022 issue, siliconchip.au/Article/15382) supports some advanced graphics options that the Terminal does not. Otherwise, using a regular PicoMite or WebMite with this Terminal is similar to working with a VGA PicoMite. The main differences are that you need a PS/2 keyboard and VGA monitor or adaptor to use the VGA Pico­ Mite, while the Digital Video Terminal lets you use a USB keyboard and HDMI monitor. We did recently publish a USB to PS/2 Keyboard Adaptor (January 2024; siliconchip.au/Article/16090) that lets you use a USB keyboard with the VGA PicoMite, but you’re still stuck with needing a VGA monitor or a VGA-toHDMI adaptor. It is possible to combine a VGA PicoMite with the Digital Video Terminal to get a dual-screen PicoMite setup, also allowing you to use a USB keyboard. It could even be a triple screen if you connect an LCD panel to the PicoMite! The Terminal will work with many types of development boards and not just the ‘Mites. Some Arduino boards and even boards that can run versions of the Python language (like MicroPython and CircuitPython) will work with the Terminal. It does not even have to be a separate development board. You can actually build the Terminal with a PicoMite siliconchip.com.au or WebMite embedded onto the main PCB and enclosed in the same compact case! We wouldn’t be surprised if some readers modified our software to create a complete, standalone device that doesn’t need an external board to be connected. We’ll detail some compatible devices later. A word of warning The Pico Digital Video Terminal uses an open-source software library to generate the digital video signal, and this library ‘overclocks’ the RP2040 microcontroller on the Pico to achieve the necessary timings. The original library by Luke Wren can be found at https://github.com/Wren6991/PicoDVI We are using a fork (derivative work) from Adafruit that interfaces the library with the Arduino IDE, available from https://github.com/adafruit/ PicoDVI We are using much the same hardware as Luke used in his prototypes; many other people have also tried this library. It’s impressive that it can generate digital video from a cheap and readily available microcontroller board. Luke notes that the signalling generated by our circuit is probably not wholly compliant with all the DVI and HDMI specifications. Nonetheless, it appears to pass all the critical tests. In the year or so since this library was released, many projects have used this software and custom RP2040 boards have been created for generating digital video. We have yet to hear of anyone who has had problems due to the overclocking or the signalling variances. The RP2040 on our prototype runs at nearly double its specified 133MHz and is barely warm. Every HDMI monitor we have tried has displayed the video correctly. We haven’t tried extreme cases like very long HDMI cables, but we see no reason for that to cause problems. So, this project does some things that are not ‘in spec’, but we and many other people have found it to work well. Connections Fig.1 shows how and where external devices connect to the Terminal and how it is arranged at a block level. Assume that we are using the Terminal with a device like a Micromite, it will connect through USB connector CON2. A complete, standalone system can be made by plugging a USB keyboard into CON3 and connecting a suitable display into CON1. 5V power is provided to USB Type-C connector CON4, feeding all the connected devices. Keys typed on the keyboard are converted to serial sequences by MOD3 and are transferred over the internal serial link to MOD2, which passes them over the CON2 USB connection to the Micromite (or PicoMite etc). The Micromite sends data back to MOD2, from where it is sent to MOD1, which behaves as a terminal display device and delivers video via CON1. With a Micromite, the serial data (from MOD3 to MOD2 and MOD2 to MOD1) takes the form of ASCII characters and VT100-compatible Escape codes. Thus, you can type on the keyboard, and the Micromite will respond as per its programming and display its output on the monitor connected to CON1. The output on CON1 is intended to mimic a terminal program such as TeraTerm. If you connect a computer to the micro-USB socket of MOD2, you can communicate with the Micromite as though it were directly connected to the computer; MOD2 will also transparently pass data between these interfaces. The grey lines show the path of data to and from the Micromite. A USB-serial interface is also provided on MOD3 to allow for the configuration of the keyboard interface properties. Each of MOD1-MOD3 has one LED that can be used to show the status of their respective connected device. Of course, a Micromite is not the only device that can be connected to the Terminal. We will look closer at what devices are compatible with the USB-serial host interface of MOD2, which includes several other development boards. Fig.1: this block diagram shows how the three Picos (MOD1-MOD3) interact and the external interfaces they provide. MOD1 delivers a DVI digital video signal to CON1, MOD2 communicates with the target computer via CON2 and MOD3 interfaces with the USB keyboard at CON3. siliconchip.com.au Australia's electronics magazine March 2024  47 Circuit details Fig.2 shows the detailed circuit of the Pico Digital Video Terminal. MOD1, MOD2 and MOD3 are Raspberry Pi Pico microcontroller boards loaded with different firmware. They each perform one of the main functions of the Pico Digital Video Terminal. We will explain in detail why there are three separate microcontroller boards in the Software section and what the firmware does there, too. The VBUS pins of MOD1-MOD3 are connected together, along with the VBUS pins of CON4, a USB-C socket. The CC1 and CC2 (configuration channel) pins of CON4 are each connected to ground via 5.1kW resistors, which presents the Terminal as a power sink requesting 5V. This is necessary for compatibility with USB-C. Any of MOD1-MOD3 and CON4 can supply 5V to the circuit. Power would typically come from CON4 if the Terminal is not connected to a computer, or MOD2 if it is connected to a computer. MOD1-MOD3 might also be connected to a computer for configuration purposes. Since each Pico provides a distinct 48 Silicon Chip function, we will discuss each in turn before describing how they work together. Note that each Pico has a corresponding connector, switch and LED numbered the same. MOD1 connects to CON1, S1, LED1 and so forth. MOD1 MOD1 is responsible for generating a digital video signal to send to CON1. Its GP12-GP19 (general purpose I/O) pins are each connected to a pin of CON1 via 270W resistors. With the cable impedance of around 50W, each of these pins will sink 10mA when pulled low by the microcontroller or not sink any current when high. The pins are arranged in pairs driven in a complementary fashion, providing the current-mode differential signalling needed. More detail on the TMDS signalling scheme that encodes the video data can be found in the panel titled “DVI-D, HDMI and TMDS standards”. MOD1’s GP9 pin (pin 12) also connects to green LED1 via a 1kW resistor, which is used as a status indicator. S1 is wired between MOD1’s 3V_ EN pin and GND. When pulled low, it shuts off the 3.3V regulator that powers Australia's electronics magazine the RP2040 microcontroller on MOD1, effectively resetting the micro. JP3 and JP4 are wired to GP8 and GP7 (pins 11 & 10), respectively, with the other side grounded. The software applies an internal pullup and can thus check the jumpers’ states by checking the voltage pin levels. The software uses these for configuration. GP10 and GP11 (pins 14 & 15), configured as I2C SDA (data) and SCL (clock), are wired to a 3.3V-to-5V level shifter consisting of Mosfets Q1, Q2 and the associated 10kW resistors. We used a similar scheme in the USB to PS/2 Keyboard Adaptor project mentioned earlier, to interface the 3.3V microcontroller to 5V PS/2 keyboard levels. The operation of this part of the circuit is explained in that article; the design is well-suited to open-drain busses like I2C and PS/2. A low level on either side is propagated to the other side; without a low logic level on either side, the respective pullup resistors maintain the lines at high logic levels. The 5V side of the circuit is connected to the DDC (Display Data Channel) SDA and SCL lines of the CON1 siliconchip.com.au Fig.2: the full circuit shows the eight 270W resistors that limit the current for the TMDS links of the DVI connection to the correct level; around 10mA. The two USB host Picos (MOD2 and MOD3) require only a USB-A socket and a pair of resistors to perform their roles. Most of the remaining circuitry provides communication between the Picos. HDMI connector. This interface can be used to interrogate an I2C EPROM on the display device to check its capabilities. CON1 also provides 5V to the display via JP2. This can power the EPROM on the display device, even when the display is switched off. The display can also connect the 5V back to the HPD (hot plug detect) pin on CON1. The 10kW/10kW divider allows the Pico to safely detect the presence of a connected display at its analog-­ capable GP27 input pin (pin 32). MOD2 and MOD3 The circuitry around MOD2 and MOD3 has been intentionally kept similar to MOD1 to simplify development. We imagine readers will have different ideas for things that can be done with this hardware, in conjunction with different software. Like MOD1, MOD2 and MOD3 have their switches (S2 and S3) wired between the 3V_EN pin and GND. Unlike MOD1, MOD2 and MOD3 have their respective LEDs connected to GP14 (pin 19) via 1kW resistors. These LEDs are otherwise identical status indicators. siliconchip.com.au The remaining circuitry connects USB-A sockets CON2 and CON3 to their respective boards via 22W series resistors. Readers might recognise this from the USB to PS/2 Keyboard Adaptor, which used a similar arrangement to interface the regular GPIO pins GP15 and GP16 (pins 20 & 21) to a USB connector. Jumper JP1 connects MOD3’s GP14 to MOD2’s GP17 pin. This gives MOD2 a second status LED in case MOD3 is not used. Our original design planned to use a single Pico for both USB interfaces; we will revisit that in the Software section. Interconnections Apart from the VBUS and GND lines that are interconnected around the circuit, there are serial data pairs (RX and TX) between each of MOD1, MOD2 and MOD3. The RP2040 has two UART peripherals, so each module has two incoming and two outgoing connections, one to each of the other modules. The following explanation expects two jumper shunts to be fitted to LK1. One is fitted between pins 1 and 2; the second is between pins 3 and 4. Australia's electronics magazine Both MOD2 and MOD3 communicate with MOD1 using their UART0 peripheral on GP0 (pin 1, TX) and GP1 (pin 2, RX). These are crossed over and connected to MOD1 at its UART0 for MOD2 and UART1 for MOD3 (GP4, pin 6, TX and GP5, pin 7, RX). MOD2 and MOD3 connect to each other via their crossed-over UART1 pins; this means that pin 6 (TX) of MOD2 connects to pin 7 (RX) of MOD3 and vice versa. Note how this continues the theme that MOD2 and MOD3 have much the same external connections. An alternative configuration of LK1 involves fitting a single jumper between pins 2 and 3. In this case, data from MOD3 comes into MOD2’s UART0 instead of UART1. You can probably see the spirit of how the Terminal achieves its aims, but of course, the detail is in the software. Software The firmware on all three Pico boards makes good use of the RP2040 PIO (programmable input/output) peripheral. We discussed the PIO peripheral in our review of the Pico March 2024  49 DVI-D, HDMI and TMDS standards DVI (Digital Visual Interface) was a standard developed in the late 1990s as a progression beyond the analog VGA (Video Graphics Array) interface. Part of the motivation was to switch to a digital communication format due to the increasing prevalence of digital displays like LCDs, removing the need to convert to and from an analog signal as required by VGA. Cathode ray tubes are analog in nature, requiring, for example, a ramped analog voltage to perform the horizontal and vertical scanning of the display area. Plasma panels, LCDs and OLEDs are more digital, having discrete pixels rather than a continuous phosphor surface, hence the preference for a digital interface. Our series on Display Technologies (September and October 2022, siliconchip. au/Series/387) has more background on those different technologies. Despite the name, DVI can carry an analog VGA signal, which made it wellsuited to the transition away from VGA. In practice, there were DVI-D (digital), DVI-A (analog) and DVI-I (integrated [digital and analog]) variants of the cables and connectors. DVI can only work if the display adaptor, cable and display all support the same digital or analog variant. For example, a DVI-D Fig.a: HDMI connectors carry many of the same signals as a DVI connectors, although they omit the analog (VGA-compatible) signals; HDMI is purely digital. Pins 1-12 carry the TMDS lines and are sufficient for a working video signal. 50 Silicon Chip output will only work with a DVI-D or DVI-I cable and a DVI-D or DVI-I input socket on the monitor. It is the digital variant of DVI signalling that the Terminal implements. However, the connector itself is HDMI due to the ubiquity of displays equipped with HDMI sockets. It is possible to connect a DVI display to the Terminal using nothing more than a passive HDMI-DVI cable or adaptor. HDMI high-definition multimedia interface Similarly to how DVI was backwards-­ compatible with VGA signals, HDMI is also compatible (by design) with a subset of DVI-D. One of the main advantages of HDMI is that the connector can carry digital audio, video, control signals and even data network traffic. The overlapping parts of DVI and HDMI that we are implementing in the Terminal use TMDS (transition minimised differential signalling). We will explain that below. Fig.a shows the pinout of the signals carried by HDMI cables; pins 1-12 have the TMDS signals and their shields. Newer versions of HDMI use higher data rates, compression and encoding schemes. For example, they can also implement different colour spaces, including HDR (high dynamic range), while DVI-compatible signalling uses a straightforward 24-bit RGB colour representation. HDMI adds other data channels between the video source and the sink (or display), and we have added a provision to interface to one of these, although nothing apart from TMDS is necessary for the Terminal to drive a video display. The DDC (display data channel) allows the HDMI source to determine what video formats a sink can accept. Since all HDMI devices must comply with the baseline DVI specification, implementing DDC is unnecessary, as we are not producing a signal beyond the baseline. The DDC used on HDMI is electrically the same as I2C. It is implemented (on the display or sink) as an I2C EPROM with a 7-bit I2C address of 0x50; a host can read this to find the display’s capabilities. We have connected these pins to a pair of I2C pins on MOD1 via a level converter, allowing it to read the sink’s DDC chip. A 5V supply is provided by the HDMI source to power the EPROM, so it can be interrogated even if the device itself is switched off. Internally, the source also connects that 5V back to the HDMI HPD (hot plug detect) Australia's electronics magazine pin, allowing the source to detect when a sink is connected. The sink may also be able to disconnect the HPD pin, for example, when switched on or off. HDMI implements other communication protocols that we have not provided connections for. Some of these protocols are not specified in all versions of the standard, but we’ll note them here for completeness. Consumer Electronics Control (CEC) is a one-wire bidirectional serial bus. It allows connected CEC-capable devices to control other devices. This means, for example, that a single remote control can operate many devices. Audio Return Channel (ARC) allows audio to be sent ‘upstream’. A typical use for this would be when a TV is showing a source that does not come from the receiver (eg, a tuner built into the TV). In that case, the ARC channel can send the audio to the AV receiver to route to its speakers. HDMI Ethernet Channel (HEC) allows Ethernet communications over HDMI cables, but it has been deprecated in the most recent HDMI versions. WiFi has mostly taken over its role of providing internet connectivity. TMDS transition-minimised differential signaling The critical part of both DVI and the baseline HDMI standard is the transition-­ minimised differential signalling that sends the video signal from the source to the sink. It consists of four shielded, twisted pairs of wires, each carrying a differential signal. Differential signals over twisted pairs make the receiver somewhat immune to common-mode noise since a similar signal will be induced in both wires in the pair from an external source. When the difference of the signals is calculated at the sink, the noise effectively cancels out. Also, the available voltage swing is double what it would be with a single line, adding 6dB of headroom to the signals. Of course, HDMI is not the only technology that uses differential signalling. Serial standards such as RS-485 also use differential signalling, as do USB and Ethernet over twisted pairs (eg, 10BASE-T and 100BASE-T). Electrically, the signalling scheme requires that one wire of each pair alternately sinks 10mA (from a 3.3V rail in the video sink by the video source). Which wire siliconchip.com.au sinks the current determines if it is a ‘0’ or ‘1’ being sent. In practice, the Terminal hardware drives its pins high (to 3.3V) or low, achieving the same effect. The cable impedance in series with the 270W resistors allows the right amount of current to flow. This is quite a high-speed signal for a microcontroller to send off-board, and quite a bit more engineering is involved than this simple explanation implies. Still, it should give you an idea of what the Pico needs to do at the hardware level. 8b/10b encoding TMDS uses a coding scheme that reduces the number of transitions that need to occur over the twisted pair, which reduces electromagnetic emissions. The clock differential pair has a 50% duty cycle and operates at a frequency equal to the pixel clock, thus giving two transitions per pixel. PLL (phase-locked loop) hardware at the sink allows the fullspeed bit clock to be recovered without having to be transmitted over the HDMI link. The three other pairs each carry a series of eight-bit data bytes encoded into 10 bit-times (hence 8b/10b encoding), meaning that there is one 24-bit pixel transmitted per pixel clock (see Fig.3 overleaf). Note that other different 8b/10b encoding schemes also exist. Of the 1024 combinations possible with 10 bits, 460 are selected to encode eightbit colour values; this means that some values have more than one encoding. Four further combinations are used to encode control data and are chosen to have a relatively high number of transitions to assist with clock recovery. The four control combinations encode two bits of data. On one of the colour channels (channel 0, blue), these encode the horizontal sync and vertical sync signals. The sync signals are naturally sent outside the times that colour information is transmitted. Channel 1 is allocated to the green component of the video signal, while channel 2 carries the red component. The timing of a baseline (640×480 pixel) digital DVI signal is practically identical to the corresponding VGA signal, but with digital signals instead of analog, so only the digital encoding and decoding steps need to be added (to analog VGA), with no changes in timing. A frame of 640×480 video actually consists of 800 horizontal and 525 vertical pixels. 800×525×60Hz (60 FPS) gives the 25.2MHz pixel clock rate. Fig.b shows how such an analog VGA signal would be encoded and then recovered. Another desirable property is for the running average DC level to be low. The coding scheme helps with this, as it allows 10-bit symbols with a high DC offset to be avoided. In this case, a zero DC offset means a long-term equal number of ‘0’ and ‘1’ bits. The eight-bit values with more than one encoding have 10-bit values with opposite DC offsets, so even long runs of the same eight-bit value can be transmitted with a combination of symbols that result in a low combined DC offset. The remaining codes with only one encoding and the control codes mostly have an equal number of ‘0’ and ‘1’ bits, so they can be transmitted without affecting the DC offset appreciably. The Terminal does not transmit audio, but HDMI allows that to be carried (along with other data) during the sync periods by using control codes to signal the presence of audio data. Fig.b: this shows how the signals found in a VGA analog video signal (on the left) can be encoded to be sent over a fourchannel TMDS link and then back. So, the four-channel TMDS link can be seen as a digitally encoded version of a VGA signal. siliconchip.com.au Australia's electronics magazine March 2024  51 Fig.3: the four differential pairs of a DVI signal look similar to this if seen on an oscilloscope. Note the symmetry and DC balance of all the signals. in December 2021 (siliconchip.au/ Article/15125). It is a specialised processor that implements a few instructions focused on I/O pins. It can easily emulate serial communication peripherals like UART and SPI, or produce PWM signals, including those needed to drive a servo motor. There are two PIO peripherals and each has four state machines. The state machines are the ‘processors’, so up to eight separate emulated peripherals can be created on the RP2040 chip. Each PIO has a memory that can hold 32 instruction words. Each state machine has a four-word deep input FIFO (first-in, first-out) buffer and a similar output FIFO buffer. If only input or output is needed, the two FIFO buffers can be combined into a single eight-word buffer. Being a 32-bit architecture, the words are 32 bits wide. Each state machine also has an input shift register and an output shift register. These take in chunks of data and shift them in or out one or more bits at a time, as is needed for the serial protocols noted earlier. The DVI implementation is a good example of a simple use of the PIO. Each pixel in a DVI video stream consists of 10 bits clocked out very quickly. The nominal pixel clock of a typical baseline 640×480 at 60Hz signal is 25.175MHz; the bit clock is thus 10 times that: 251.75MHz. The PIO operation for DVI video involves passing 10 bits of data at a Making an all-in-one computer Replacing MOD2 with a Pico (or Pico W) programmed with different firmware is possible. This is the reason for link LK1. A shorting block can be fitted between pins 2 and 3 of LK1, and the serial data streams from MOD3 and towards MOD1 are now both connected to UART0 on GP0 and GP1. The following PicoMite BASIC OPTION can be set to use UART0 (on GP0 and GP1) as the console: OPTION SERIAL CONSOLE 0,1,B The default baud rate used by MOD2 and MOD3 is 115,200, which should match the PicoMite’s default. If necessary, it can be changed with another option. OPTION BAUDRATE 115200 The PicoMite is now integrated into a computer with a USB keyboard and digital display interfaces. It’s a bit trickier than usual to access the spare I/O pins. Still, if you were looking to experiment with the BASIC language or perhaps the WiFi interface of the WebMite, it would make a compact machine for those purposes. You could also write your own code to run in place of MOD2, creating a custom computer. There are several projects around that emulate older computer platforms using the Pico. 52 Silicon Chip Australia's electronics magazine time into the PIO’s FIFO, which it then simply clocks out serially at the bit clock rate. The RP2040 also has a direct memory access (DMA) peripheral, which we use to ensure that the PIO is consistently fed data from RAM without requiring the main ARM processor’s attention. The processor simply needs to set data in RAM and arrange for the DMA peripheral to move that into the PIO as needed. The nominal 251.75MHz figure noted earlier dictates the overclocking needed; the Pico is clocked at 252MHz, resulting in a signal that is within VESA tolerances. The DMA and FIFO actually handle data in 20-bit blocks, sending two pixels at a time. In theory, the 32-bit ARM processor could work with blocks of three pixels (or 30 bits), but three does not divide into the 800 pixels that constitute each horizontal scan line (including sync periods). A good amount of data still needs to be generated to feed the PIO. Therefore, one of the two processor cores on the Pico is dedicated to encoding data from an RGB or monochrome frame buffer into the TMDS form that the PIO requires. The other core of the MOD1 Pico listens in on its serial ports and behaves as a VT100 terminal. The VT100 was a ‘dumb terminal’ introduced in 1978. They were standalone hardware devices that allowed numerous users to connect to a large mainframe computer over a simple serial interface. Nowadays, they mainly exist as software emulations by serial terminal programs such as TeraTerm or minicom that can run on a desktop or laptop computer. The VT100 standard allows ‘Escape codes’ to perform functions like moving the cursor around, changing the text colour and clearing the screen. They are called that because they start with the ASCII Escape character (0x1B). Since the various Micromites and PicoMites use the VT100 protocol, this is the most straightforward way to interface with them. This also means that MOD1 (and its associated components) behaves effectively like a dumb display terminal. MOD1 maintains a buffer of characters and their attributes (such as colour or underlining) and manipulates the buffer according to the data that arrives on its serial port. The buffer is then siliconchip.com.au rendered for display and sent to CON1 as a DVI video signal. LED1 is illuminated whenever MOD1 detects a voltage from the HPD pin of CON1, which indicates that a display device is connected. MOD2 MOD2 uses the Adafruit TinyUSB software library to allow it to behave as a USB host to a USB-serial device. The library incorporates the Pico-PIO-USB library, which allows the USB interface to operate on a pair of GPIO pins. We use the PIO peripheral to implement a USB port instead of the RP2040’s dedicated USB peripheral for two main reasons. Firstly, connecting to the GPIO pins is much easier since the internal USB data pins are only available at the Pico’s onboard USB socket or two test pads on the Pico’s underside. Secondly, it gives us two USB ports, allowing the Terminal to transparently connect the Micromite (or whatever is connected via CON2) to a computer, both relaying and intercepting data between the Micromite and the computer. That is most of what MOD2 does. It bridges the link between the device connected at CON2 and a computer connected at MOD2’s micro-USB socket. It can also inject data received from MOD3 as well as exfiltrate data to MOD1, as shown in Fig.1. The LED connected to GP14 is illuminated whenever a compatible device is connected to CON2 and flickers when data is received from MOD3. The Pico-PIO-USB library uses nearly all of the PIO resources, which is the main reason why we need a dedicated Pico for this role and can’t integrate it with either of the others. MOD3 Like MOD2, MOD3 also uses the Adafruit TinyUSB software library to allow it to behave as a USB host, except in this case, it expects a keyboard to be connected. Thus, the circuitry around MOD3 can be much the same as MOD2, but the software is different. The software has much in common with the USB to PS/2 Keyboard Adaptor; that project translates strokes from a USB keyboard into PS/2 scancodes, while MOD3 on the Terminal translates them to serial data and VT100 Escape codes. It also monitors the Number Lock, siliconchip.com.au Silicon Chip kcBBack a Issues $10.00 + post January 1995 to October 2021 $11.50 + post November 2021 to September 2023 $12.50 + post October 2023 onwards All back issues after February 2015 are in stock, while most from January 1995 to December 2014 are available. For a full list of all available issues, visit: siliconchip.com. au/Shop/2 PDF versions are available for all issues at siliconchip.com.au/Shop/12 We also sell photocopies of individual articles for those who don’t have a computer Scroll Lock and Caps Lock keys and changes the state of the keyboard LEDs to suit. The serial data it delivers depends on the state of the Number Lock and Caps Lock keys. MOD3 lights up LED3 whenever it detects a keyboard is connected and flickers it briefly whenever a key is pressed or released. You will see different patterns on LED2 and LED3 during regular operation. Our early prototypes combined MOD2 and MOD3. The Pico-PIO-USB library can support a USB hub so multiple devices can be connected, but it only supports a single hub. Many USB keyboards incorporate a hub, at least internally, so we found that many keyboards did not work when connected through a hub. Since a Pico is much the same price as even the cheapest hub we could find, we opted to simply add another Pico and USB socket. It is a more elegant solution as everything fits neatly into a single enclosure. While MOD2 and MOD3 can only perform the specific role they have been programmed for, the symmetrical arrangement of their external connections means that they can be physically transposed on the PCB with only minor software changes. Hardware notes That is all we can fit into this article, so next month’s second part will describe assembly, testing, configuration and use. We have provided the parts list this month to allow you to collect various components needed for assembly. The design uses a few SMDs, but they are primarily passive parts in M2012 size, so they are not too difficult to solder. The HDMI socket has pins at a 0.5mm pitch, so it is probably the most challenging part to solder. Still, it is not too difficult if you have good flux paste on hand, decent lighting and perhaps a magnifier. There are two case options. One option is to use the 25mm-high Altronics H0190 (or equivalent). This has a corresponding front panel PCB that is coded 07112232. This case is ideal if you plan to permanently solder the Picos to the PCB via pin header strips. An alternative is the 30mm tall H0191, which uses a front panel PCB coded 07112233. This case allows you to fit MOD1-MOD3 using sockets, so they are removable. That might be handy if you are considering using an alternative firmware for MOD2, as described in the panel (“Making an SC all-in-one computer”). Other configurations for advanced users While the Terminal we are describing here is intended to parallel the ASCII Video Terminal in function, the modular nature of this project means that it can be altered to work in various ways. For example, you could build the Terminal with only the parts surrounding MOD1 and use it to deliver a custom video signal to a modern digital display. Fitting just MOD2, its USB socket and surrounding components will give you a USB-serial interceptor device like the one we plan to describe in an upcoming Circuit Notebook. Fitting just MOD3 and its associated components will give you a device with similar capabilities to the USB Keyboard Adaptor for Micros (February 2019; siliconchip.au/Article/11414). However, all of these options require the Pico to be wired up to another device via a serial link to be useful. Australia's electronics magazine March 2024  53