From: drowe67 Date: Wed, 9 Dec 2015 05:49:37 +0000 (+0000) Subject: updated task list and block diagram X-Git-Url: http://git.whiteaudio.com/gitweb/?a=commitdiff_plain;h=54a7acab9e10d84d747f09fb01baf43fcdd6b0f6;p=freetel-svn-tracking.git updated task list and block diagram git-svn-id: https://svn.code.sf.net/p/freetel/code@2500 01035d8c-6547-0410-b346-abe4f91aad63 --- diff --git a/sm2000/doc/sm201- Block Diagram v0.1.ppt b/sm2000/doc/sm201- Block Diagram v0.1.ppt index 91c46feb..80f40c28 100644 Binary files a/sm2000/doc/sm201- Block Diagram v0.1.ppt and b/sm2000/doc/sm201- Block Diagram v0.1.ppt differ diff --git a/sm2000/doc/sm202 Task List.txt b/sm2000/doc/sm202 Task List.txt index 37736db1..421b8332 100644 --- a/sm2000/doc/sm202 Task List.txt +++ b/sm2000/doc/sm202 Task List.txt @@ -1,84 +1,124 @@ -[ ] Goals - [ ] Demonstrate TDMA repeater - [ ] superior weak signal performance (-135dBm) - [ ] diversity to handle multipath - + TDMA diversity, switch rx LO quickly? - + demo overcoming large fades? - [ ] wideband audio - [ ] high bit rate data - [ ] low cost - [ ] Full SDR so we can define waveform - [ ] SSB up and down converters to allow full waveform control - [ ] no SoCs - -[ ] Doc - [ ] Block diagram - [ ] Uses cases in paragraph form - [ ] Gain, NF, IP3 budget spreadsheet - [ ] Power budget spreadsheet - [ ] Current per active device - [ ] Power dissiapation per active device - [ ] Resistor power dissiapation per active device - + RF bias currents can be high - -[ ] RF Specs - + these are targets, main goals are functional, leave - others to get sparkling RF performance - [ ] spurious tx emissions -45dBC @ 1W - [ ] IP3 10dBm - [ ] Blocking 95dB - [ ] System NF 2dB - [ ] ACS 50dB, - [ ] MDS (1200Hz) -135dBm, - [ ] Tx phase noise -124dBc/Hz ?? - - -[ ] Design - [ ] LNA selected with RF switch - [ ] switchable power as well - + as power consumption is so high - [ ] AGC in rx to facilitate SSB operation - + a control line from uC DAV/PWM? - + pump up/down GPIO - [ ] LO buffer that can be bypassed by DNL - [ ] put extra 0805/603 compons in parallel for Z matchhing and dnl - + e.g. on collectors and emitters of RF amps - [ ] protection for LNA etc if RF switches get set the wrong way - [ ] xtal filter power - + limit, how many uW? - [ ] linear reg for vadc ADC and vref - + jumper select to test +Goals +===== + +To demonstrate advanced VHF/UHF Digital Voice features on an open +software and hardware platform. + +[ ] feasibility of "$100" TDMA repeater + + show our radio working as a TDMA repeater without a diplexor + + temporary mount on hill top, city wide coverage + + no split so saves spectrum, potentially low cost (to fill coverage gaps easily) +[ ] 10dB weak signal performance gain over analog FM and 1st gen DV + + "smackdown" video of current state of the art against our solution +[ ] diversity to handle multipath + + TDMA diversity, switch rx LO quickly + + demo overcoming large fades +[ ] wideband audio + + for example use 10dB gain to send 10 kbit/s Skype type audio over channel +[ ] high bit rate data + + 10 - 100 kbit/s +[ ] low cost + + several hundred $ sales price + + approx $100 BOM + +Project Workflow +================ + +Manhatten Prototype -> Sch Entry -> PCB Design -> Load and Test -> Demos + /\ /\ + || || + Proto Modem SW Modem SW + UI SW + Driver SW + +Themes +====== + ++ Full SDR so we can define waveform ++ SSB up and down converters to allow full waveform control ++ no SoCs ++ Demos of advanced DV features is "enough". We don't need + sparkling performance in other areas such as all RF specs. + +Task List +========= + + +[ ] Documentation + [X] Block diagram + [X] Uses cases in paragraph form + [X] RF design spreadsheet + [X] cascaded Gain + [X] cascaded NF + [ ] cascaded IP3 + [ ] Power budget spreadsheet + [ ] Current per active device + [ ] Power dissipation per active device + [ ] Resistor power dissiapation + + RF bias currents can be high + + don't want to smoke and SM parts + [ ] Summary for each rail + [ ] Specs + + We don't care too much about IP3, Blocking, ACS, tx phase noise, + power consumption, they can be modest, equivalent to middle of + range VHF radios. + + We do care about the following. Check off when demonstrated on + prototype of Rev A + [X] 12VDC operation + [X] NF < 2dB + [X] MDS (1200 bit/s 2FSK at BER < 0.03) of < -133dBm + [ ] MDS (1200 bit/s 2FSK at BER < 0.03) of < -130dBm (3dB de-sensitisation) + with -20dBm blocking signal outside of 144 - 148 MHz + [ ] spot check at 88, 108, 150, 200 MHz + [ ] 1W tx power with spurious emissions -45dBC + + +[ ] Remaining Manhatten Prototype Tasks + + Manhatten Style prototype used to build and test as much as possible + [ ] 1W PA + [ ] PIN diode TR switch + [ ] PA filter + [ ] Input (1st) BPF + [ ] TDMA switching + [ ] how to switch PS rails + [ ] tx and rx meet specs + [ ] Parts Selection - [ ] what constitutes a RFC? - + recc general part - + current consumption - + RF characterisistics - + e.g. for PGA 103 and RF amps + + cross off below when part selected and tested in prototype [ ] New audio amp chip + bit better, higher power than LM386 - [ ] should we add a LCD? - + how would this be used? - + Could use cheap Nokia 5110, at least in first rev. Well documented and libraried. [ ] General purpose VHF transistors + [ ] General purpose IF transistors [X] 1st mixer: ADE-1 - [ ] RF switches + [X] RF switches + Brady has suggest AS17992LF [ ] 1W PA transistor [ ] LCD + + Could use cheap Nokia 5110, at least in first rev. Well documented and libraried. [ ] encoder - [ ] first LO: Si5351 + [X] first LO: Si5351 + also 2nd LO - [ ] optional Si570 + [X] optional Si570 + low phase noise soldered in option, provide footprint + Si5351 still used as 2nd LO - [ ] how to select Si570? - + jumper 0hm resistor? + [ ] Prototype parts list to Edwin for review + + just major parts, semiconductors + + make sure they are obtainable + + look for low cost equivalents + [ ] Use Cases + + need software/test procedure to support the following use cases + [ ] test modes + + useful for devevelopment and debugging + [ ] rx BER + + prints BER on LCD when continuous 2FSK/4FSK test frames sent + [ ] tx BER + + sends continuous test frames using 2FSK/4FSK [ ] VHF FM radio - [ ] Open Source VHF DV radio - [ ] Low MDS + [ ] Open Source PTT VHF FreeDV radio + + continuous or TDMA? Can they be the same? [ ] TDMA Repeater [ ] Diversity rx + use 2nd output from 5351 @@ -91,11 +131,32 @@ to use, e.g. set a frequency [ ] Changing waveform + [ ] Software Modules + + see sm201 Block Diagram page 2 [ ] VHF Modem + + cont mode first, we can adpat to burst/TDMA mode next + [ ] ideal non-coherent 2FSK/4FSK BER versus Eb/No perf + [ ] tracks +/- 500ppm sample clock offsets with no impl loss + [ ] +/- 5kHz frequency offset acquisition in 1000ms on continuous signal + [ ] C/Octave versions + [ ] user defined symbol rate (nominal 1200, 2400), carrier freq, sample rate + [ ] can it cope with say 3000 symb/s to allow time for TDMA guard times and sync? + + e.g. a gd fit with integerat Fs/Rs ratios + + might need TDMA design at same time to answer this question + [ ] variable nin style demod C function call, approx 20ms packets + [ ] DR to review function headers + [ ] automated test script to copare C and Octave units results + + to ensure identical implementation + [ ] command line ./mfsk_tx and ./mfsk_rx + [ ] demo OTA (1200/2400/2FSK/4FSK) using SSB radios or SDR [ ] VHF protocol + + do we need two modes, cont & TDMA? + + or is there a way to combine? So one rx does both? Autobaud? + + can we optionally use FM friendly machester encoded waveform? [ ] TDMA framework around modem, burst modem - [ ] diversity + [ ] rx TDMA diversity + [ ] test frame tx/rx modes [ ] New Drivers [ ] LCD [ ] Encoder @@ -103,35 +164,80 @@ [ ] Si5351 [ ] UI for setting a frequency [ ] USB console + [ ] USB tx/rx signal streaming + [ ] Code to support running Modes + [ ] Analog FM + [ ] TDMA repeater + [ ] Diversity RX + [ ] FreeDV PTT TDMA radio + [ ] Risks - + and experiments/tasks/prototypes to mitigate - [ ] DAC and ADC tied together - + demo tx/rx switching using this approach - [ ] Can we use ADE-1 bir-directional? - + what about diplexor? - [ ] Investigate use of other mixers - + but doesnt have to be perfect? - [ ] Celestra mixer - [ ] Glens - [ ] coupled res in shielded box or cots part? - [ ] how to swithc mixer/xtal filter - [ ] what else wcould er use instead of NE602? - [ ] ADC perf + + check off when risk has been nailed, e.g. by experiment/test + [X] Can we use ADE-1 bir-directional? + + yes, use TIA and PIN diode switching + [X] coupled res in shielded box or cots part? + + make little shields out of blacnk PCB + + just make box, doesnt need lids + [X] how to switch mixer/xtal filter + + tx modulator transistor is part of rx load + + alo use of TIA + [ ] Is NE602/612 still easy to get? + [X] ADC perf + SFDR etc, which is why we run it as 100kHz + if not good enough we'll need an external ADC + test against two tone, MDS and ACR + + Result: met 135dBm spec on prototype so OK [ ] Power dissipation of active devices and resistors + easy to exceed limits in SM resistors and transistors + RF drivers draw lots of current - [ ] termination of mixer + + mitigation: power spereadhseet in doc section + [ ] IF termination of 1st mixer + + how to test if this is OK? [ ] do we need a way to bypass the LNA? + or is it OK even with strong signals + + mitigation: cascaded IP3 calculations in doc section + if this compares to current middle range radios then OK + [ ] xtal filter power + + what is the limit, how many dBm are we using at the moment? + [X] prototype RX works + + 1200 bit/s 2FSK at < -133dBm with BER < 0.03 + + theory with NFis -134dBm at 0.02 + [ ] prototype TX works + + 1200 bit/s 2FSK at 1W spurious -45dBC + + demod by SDR at high SNR with 0 BER + [ ] TDMA switching + + demonstrate design supports tx/rx switching fast enough for TDMA + -[ ] PCB layout +[ ] Sch Entry + [ ] linear reg option for vadc ADC and vref + + jumper select to test + [ ] put extra 0805 or 1206 components in parallel for Z matching and DNL + + e.g. on collectors and emitters of RF amps + + between building blocks + [ ] inductors + [ ] coil winding details if air core + [ ] tolerance + [ ] statement on Q so we know if we can subs SM or need Hi-Q air-core + [ ] alignment of coils WRT each other to prevent coupling + [ ] location of shields if rqd + [ ] how to adjust/if adjustment rqd + + e.g. adjust filter by variation in coild spacing against filter spec + [ ] 4 by 0.1 inch pin headers in a square pattern to disconnect building blocks + [ ] select Si570 with a jumper + [ ] review of foot prints for discretes + + ease of debug, e.g. if we need to add a LC filter + + power handling + +[ ] Rev A PCB Layout + [ ] footprints for all devices + [ ] block diagram showing map of building blocks + + separate analog and digital [ ] shields [ ] where do we put sheilds [ ] how to fabricate? + + try small pieces of black PCB [ ] foot prints [ ] probe points + VHF freindly with nearby grounds @@ -140,8 +246,9 @@ + can we get a 50-ohm probe for cro or spec-an? So mechanically it can sit on board? [ ] 50 ohm points for spec an? - [ ] ways to disconnect modules so we can insert tracking gen and spec an - + 0805 0 ohm jumpers + [ ] 4 pin 0.1 inch headers to disconnect building blocks so we can insert + tracking gen and spec an + [ ] Manufacture [ ] is it feasible to have 6 turn coils with one turn taps in production? @@ -149,32 +256,3 @@ [ ] can people wind toroids? [ ] How will we test and align? -Unorg notes ------------ - -What we care about: - -+ Software defined waveform (even linear ones like PSK), so must be - linear up and down conversion, eg. direct FM is out -+ TDMA so needs a uC with bare metal, no USB and Host PC type devices due - to latenecies -+ Functional demos of advanced features -+ Your idea is you are willing to roll up you sleaves and make it happen. - -What we don't care about: - -Other people are good the aspects below, we are not. So we focus -where we can make a difference, like innovateive coding, good modems, -a new codec, and open source. We are doing enough! Don't have to -a tick all boxes first time around. - -+ minimal cost -+ your favourite radio architecture - there are many ways to up and down convert a signal. -+ High RF performance, such as phase noise, blocking, ACR, high tx - power, multi-band operation, low spurious, component selection, low - power consumption. -+ Your favourite chipset -+ Your idea unless you are willing to roll up you sleeves and make it - happen. -+ gold plating - is the feature you are proposing going to add to our - schedule? Will it introduce risk? diff --git a/sm2000/doc/sm203 gain_noise_calcs.ods b/sm2000/doc/sm203 gain_noise_calcs.ods index daf010d8..04c7cde2 100644 Binary files a/sm2000/doc/sm203 gain_noise_calcs.ods and b/sm2000/doc/sm203 gain_noise_calcs.ods differ