Skip to content

Latest commit

 

History

History
135 lines (80 loc) · 8.16 KB

README.md

File metadata and controls

135 lines (80 loc) · 8.16 KB
Raw

(Please note that is a repository based upon a template (not even a fork as there's another repo which is the fork). Details of the design are described at the very end of this README.md. Go to https://tinytapeout.com for instructions!)

What is this whole thing about?

This repo is a template you can make a copy of for your own ASIC design using Wokwi.

When you edit the Makefile to choose a different ID, the GitHub Action will fetch the digital netlist of your design from Wokwi.

The design gets wrapped in some extra logic that builds a 'scan chain'. This is a way to put lots of designs onto one chip and still have access to them all. You can see all of the technical details here.

After that, the action uses the open source ASIC tool called OpenLane to build the files needed to fabricate an ASIC.

What files get made?

When the action is complete, you can click here to see the latest build of your design. You need to download the zip file and take a look at the contents:

  • gds_render.svg - picture of your ASIC design
  • gds.html - zoomable picture of your ASIC design
  • runs/wokwi/reports/final_summary_report.csv - CSV file with lots of details about the design
  • runs/wokwi/reports/synthesis/1-synthesis.stat.rpt.strategy4 - list of the standard cells used by your design
  • runs/wokwi/results/final/gds/user_module.gds - the final GDS file needed to make your design

What is this specific design about?

The design implements a 300 baud UART transmitter (8N1) with limited character set (0x40..0x5F) loading. Both limitations are described in a separate section below.

The 300 baud is a theoretical value as the limited clock rate as stated in the TinyTapeout FAQ:

What is the top clock speed? Not sure yet, I would like to get around 100kHz but with the current scan chain and 500 designs it’s looking more like 750Hz. TBD. We have a built in clock divider that can further reduce this speed down to 2Hz.

I am not sure if the actual baud rate will differ.

Input pins

The design uses all 8 input pins:

  • IN0: 300 Hz input clock signal (defining the baudrate)
  • IN1: bit b0 (the least significant bit) of the loaded and transmitted character
  • IN2: bit b1 of the loaded and transmitted character
  • IN3: bit b2 of the loaded and transmitted character
  • IN4: bit b3 of the loaded and transmitted character
  • IN5: bit b4 of the loaded and transmitted character
  • IN6: load word into shift register from parallel input (IN1..IN5) (1) or cycle the existing word with start/stop bits around it (0)
  • IN7: output enable (for gated output signals): 1 output is enabled, 0 output is disabled (permanently set to HIGH/1)

Output pins

The design uses all 8 output pins...

  • OUT0: UART (serial output pin, direct throughput)
  • OUT1: UART (serial output pin, gated by enable signal)
  • OUT2: UART (serial output pin, reverse polarity, direct throughput)
  • OUT3: UART (serial output pin, reverse polarity, gated by enable signal)
  • OUT4: UART (MSBit, direct throughput); typically set to 1 or can be used to sniffing the word cycling through the shift register)
  • OUT5: UART (MSBit, reverse polarity, direct throughput); same usage as above
  • OUT4: UART (MSBit, gated by enable signal); typically set to 1 or can be used to sniffing the word cycling through the shift register)
  • OUT5: UART (MSBit, reverse polarity, gated by enable signal); same usage as above

Design, simulation and usage

The heart of the design is a 13 bit shift register (built from D flip-flops).

wokwi-simulation

Why 13 bits? The UART transmitter is meant to use 8N1 configuration, where the 8N1 stands for ...

  • one start bit (with value 0),
  • eight (8) data bits,
  • no (N) parity bit, and one (1) stop bit (with value 1).

This sums up to 10 bits. To improve readability on the signal line I have added two idle bits (always 1 on the line) preceding and one subsequent idle bit (always 1 on the line) after the transmitted word. So strictly speaking, this is not pure 8N1 ... or rather 8N1 with additional pauses (or additional stop its). The additional pauses are meant, as mentioned previously, to improve reading and analyzing the serial data. This will reduce the net bandwidth again, but there's additional overhead in loading the shift register anyhow - and this is purely for experimental purposes.

Summing it up: the 10 bits from 8N1 configuration plus the 3 additional idle bits make up a total of 13 bits.

When being loaded the shift register will load the following values (from first bit being clocked out the last one; in the Wokwi schematic from bottom to top):

  • 1: MUX input is permanently tied to VCC: first preceding idle bit

  • 1: MUX input is permanently tied to VCC: second preceding idle bit

  • 0: MUX input is permanently tied to GND: start bit

  • X: bit b0 of the character to be transmitted (see description above)

  • X: bit b1 of the character to be transmitted (see description above)

  • X: bit b2 of the character to be transmitted (see description above)

  • X: bit b3 of the character to be transmitted (see description above)

  • X: bit b4 of the character to be transmitted (see description above)

  • 0: MUX input is permanently tied to GND: b5 of the character to be transmitted (cannot be configured)

  • 1: MUX input is permanently tied to VCC: b6 of the character to be transmitted (cannot be configured)

  • 0: MUX input is permanently tied to GND: b7 of the character to be transmitted (cannot be configured)

  • 1: MUX input is permanently tied to VCC: subsequent idle bit

The clock signal can be generated manually by clicking the Step push button or by switching the slider switch to the square wave generator. There's a bunch of LEDs that visualize the state of different control and data signals.

When a word has been transmitted, it will be transmitted again and again until a new word is loaded into the shift register or the output is disabled (the word will keep on cycling internally).

Limitiations of the implementation

  • The low baud rate of 300 symbols/second, i.e. 300 bits/second. (I think that this is an external constraint which cannot be fixed easily. Unless one might be able to implement a PLL or something...)
  • The limited character set of 0x40..0x5F (i.e. capital letters A..Z, and special printable characters @`[\]^_) have been chose arbitrarily to allow human-readable characters. This is because the message encoding has been limited to 5 bits (only 5 data input pins plus 3 control input pins). That means that the characters' three most significant bits are fixed to 0b010. This results in words being 0b010X'XXXX, i.e. between 0b0100'0000 (=0x40) and 0b0101'1111 (=0x5F).

Status/TODOs

I am a bit late to the party. I've started to think about the design on August, 31st - and submission deadline is already one day later on September, 1st.

☑️ Describe the design idea

☑️ Implement the design idea using Wokwi

☑️ Describe the implementation

☑️ Edit the Makefile and change the WOKWI_PROJECT_ID to match the project. → It's here: https://wokwi.com/projects/341631511790879314

☑️ Enable GitHub Actions

☑️ Describe the signal mapping to the ASIC hardware I/O pins/ Wokwi user interface in the simulation

☑️ Share your GDS on twitter, tag it #tinytapeout and @matthewvenn!

☑️ Submit it to be made

☑️ Join the community

🔲 Improve the implementation to work around the current limitations.

🔲 Verify implementation based upon the VCD recordings coming from Wokwi logic