706 FSK Modem +HDLC +UART (PoC) :: Quicker, easier and cheaper to make your own chip!

How it works

This is a proof-of-concept design to sketch out the TT_UM digital interface for a later project design that will attempt to incorporate both analogue and digital aspects of the basic skeleton shown in this project.

The design is based on the classic circa 1988 model design used in Amateur Radio Packet systems by G3RUH. The initial specification is looking to achieve data rates of between 4800 and 64000 baud, but the design maybe able to service audio 1200 baud packet radio as well.

The design is 1-data-bit per symbol.

The original TNC (Terminal Node Controller) was a Z80 CPU and 8530 Serial Communications Controller. So inline with this I expect to provide an 8-bit CPU (as a future TT project) as a companion to this so the two items taken together should be able to form a complete communications solution of a capable TNC. This is an area I spent a significant amount of my teenage youth understanding and experimenting with that gave me a good grounding in all the digital electonic, radio and computer/CPU theory/practice that is still in use today.

The original PCB board design used:

a x16 master TX CLOCK line of the data rate.
was based on 12v audio interface/opamps, and 74HC TTL logic
was capable of the range of baud rates with minor modifications, the most used speed in my experience is 9600 baud
the TX DAC was 4 x 8-bit samples per bit, with the waveform lookup using a 12bit address that can see previous bit information sent
EPROM were used directly to provide waveforms, these have a number of jumper set modes to allow compensation for non-linear responses at the TX-AUDIO and RX-AUDIO

Due to the need to perform ROM lookups, this is operating in 4 phases sharing 6-bit output from module, and 4-bit input to module. The 4 phases cover a sequence of:

TX nibble low (6bit address)
TX nibble high (6bit address)
RX nibble low (6bit address)
RX nibble high (6bit address) It is not clear if this arrangement a good choice. There is also a programmable latency on the reply, of zero-cycles or one-cycle, the shifts the expectation of the result.

I also need to validate the DAC 8bit loading scheme prevents any chirping (visibily to DAC of partially loaded data, due to multiplex timing differences) of the data because it is loaded in 2 halves.

The master clock (CLK pin) due all the above, it is ncessary to run the clock pin at x4 the x16 of the original design.

data rate baud	master clock (CLK)	tx clock	tx sample clock
4,800	307,200	76,800	19,200
9,600	614,400	153,600	38,400
19,200	1,228,800	307,200	76,800
38,400	2,457,600	614,400	153,600
64,000	4,096,000	1,024,000	256,000
76,800	4,915,200	1,228,800	307,200

Table is in Hz or Baud

The master clock (pin CLK) is driven at x64 the synchrnous data rate. The tx clock rate is derrived from this 'CLK divide-by-4'

The UART clocking is also derived from CLK, and each side (uart RX and uart tx) can be individually configured to be 1:1 or 2:1 the synchronous data rate:

Uart TX x1 = data rate x1
Uart TX x2 = data rate x2
Uart RX x1 = data rate x8 (due to majority voter, 8 sample buffer)
Uart RX x2 = data rate x16 (due to majority voter, 8 sample buffer)

As you can see maybe there is some headroom for faster transmission speeds within a TT project, before needing to increase DAC resolution and explore 4FSK/6FSK/QAM etc...

There are 3 main functional areas with the design:

The type of FSK modem is 2FSK (dual tone) outputting continious wave.

Upper Digital (included here)

This incorpotates a full-duplex HDLC frame processor attached to a UART (ttl interface), the UART process encodes the frame in format similar to KISS format used by TNCs, with a few modifications.

Lower Digital (included here)

This manages the receiver clock recovery PLL circuit and interface, the original designs used EPROM lookup tables with 12bit address (which has visibility on at least the previous encoded bit) and provides an 8bit data output.

The data outputs are then fed into a respective 8bit DAC

The receiver has a PLL lock detector which is used to provide DCD (Data Carrier Detect) signal. While the hardware design is capable of full-duplex operation it is often used in Amateur Radio situations in a half-duplex situation with a carrier sense channel sharing algorithym.

Lower Analogue (not includes in this PoC design, see next iteration)

The parts that are missing from the design:

8bit DAC for transmit waveform shaping, using 4 samples per bit
opamp for transmit audio anti-aliasing (low-pass filter?) circuit to remove harmonic noise from the output audio
8bit DAC for receiver clock recovery feedback, using 16 samples per bit.
opamp for receive audio signal interface, this maybe moved to an external board due to needing to protect the TT IC from over voltage from being attached to usuall 12v equipment or maybe 36v when using some ex-commerial radio trancievers. This may have been a comparator circuit (unsure at this time), fed into a DFF to synchronise the incoming data to the x16 (of datarate) clock recovery timing
2 x opamp to provide PLL lock detection (unsure how this works atm), I would guess it can detect when the signal is being centered and has been centered for some number of samples, maybe via slow capacitance charge up when the UP/DOWN line is managing to meet an approximate 50%/50% duty cycle per x16 clock recovery tick.
2 x opamp to provide zero-crossing detection, this is used to provide the PLL its feedback mechanism (the UP/DOWN line) to advance or retard the edge alightment.

It is hoped all items can be incorporated into the same design using the analogue GDS facility with TT and connected to the respective lower digital signal.

At this time we bring out the interconnection points (between analogue and lower digital) to the external interface of TT and we provide a configuration mechnism to be externally or internally driven/internally sourced. This should allow for a significant level of simulation and experimentation by users of the project to understand and explore FSK/PLL theory by picking a testing configuration combintation, being full-duplex it should be able to loop-back at various levels to understand each part better. While also providing those with a Ham Radio license to try out on air communicating with their local users or AMSAT.

Have fun... 73s de G7LED

How to test

When the final design is completed, there should be a number of visible and testable aspects available to observe the working of various functions.

I am not expecting this PoV project to yeild good result due to the limited time spent on it just before submission deadlines for TT06.

Check back with the repo for a testing regime.

External hardware

At this PoC stage, testing with RP2040 and FPGA external boards to validate the electrical interface acrhetecture makes sense and provided the most options.

#	Input	Output	Bidirectional
0	Rx Data	UART TX	Rx Clock (bidi)
1	Tx Data	UART CTS	Up/Down (bidi)
2	UART RTS	UART DCD	TableAddr[0]
3	TableData[0]	Rx Error	TableAddr[1]
4	TableData[1]	Tx Error	TableAddr[2]
5	TableData[2]	Sending	TableAddr[3]
6	TableData[3]		TableAddr[4]
7	UART RX	Tx Clock Stobe	TableAddr[5]

706 FSK Modem +HDLC +UART (PoC)