MARCH 05, 2013 09:11 PM Main PCB board with 3 CPU on it (including GPS).
Soldering kainda a problem - for indium alloy (In60Pb40)it is very time consuming, with indium(In60Pb40) soldering paste is less available and expensive, with indium (In100) paste is low strength (SMD components can not withstand good vibration), plus is 155C max, plus it is less sticky.
For ground station and CS prototype used Sn60Pn40. For relatively simple board process takes 2 hours for components list, 1 hour for manual paste distribution, 1 hour for placing SMD, 7 min in oven. For InPb soldering it will take around 2+6 hours.
And it will be less video from now on – next CubeSat flight software.
FEBRUARY 15, 2013 12:19 PM Ordinary competition's day
Memory - core type (real core) - 512 bytes - from old military computer, could survive under any solar weather, but small and heavy - uploading a video for a team's blog - the third party liability insurance - antennas (cool looking but a garbage) impedance matching - substitution for a lucky penny == 1 Lunaro Sterling coin - moving a GPS prototype into a schematics - main computer schematics - routing traces on PCB - sleep 8h (not recorded) - Clear sky over Vancouver == communication test 7h, 3.5 km (hard to catch == impossible to record) - PCB tracing ( GPS + memory storage module flash 2GB) - continue uploading video for a blog - sleep (recording skipped ). On a background: Alexander Galich's + Turetskogo choir + "June 31" movie's songs.
Nov 28, 2012,
Required power for communication on Lunar distances.Igor -
Assume MW7IC2425 25WT with narrow band PAE 40% will transmit 10WT (40dBm).Assume
nRF24L01 sensitivity -85dBm (at 1Mbps). Assume cascade 3 LNA amplifiers
(SKY67014-396LF) with noise 4dB, gain 36Db total. On distance 400000km path Loss
212 dB on 2450 MHz, all this will require on both side antennas with 28dBi gain
to get -80dBm RX power. Enough to work with nRF24L01’s sensitivity -85dBm and
margin for a noise 85-80 = 5dB. Using different LNA amplifiers (MC13850EPR2)
will give 6dB noise and 39 gain, outcome in this case -77dBm with margin for
noise 85-77 = 8dB. Pretty tight.
To test on Cubesat transmitter == 0.5WT transmit (27dBm), distance 400km path
loss 152 dB, symmetrical antennas with 8dBi gain will give in that case -70dBm
RX power. That is a goal for test flight.
For first test on ground, transmitter has to be stripped from power amplifier
with LNA cascade 26-29dB and direct 1mWt (0dBm) output to same antennas (8dBi)
as for Cubesat to get same -71dBm power on RX. Distance has to be 20 km (path
loss 126dB). Direct distance = Cyprus – SFU hill ==25, Grouse Mt. – SFU hill ==
15km, and Mt. Seymour – SFU == 9 km
For second test on ground, transmitter with 0.1WT (20dBm) and same antennas
require 2km (path loss 106dB) distance for same -70dBm RX. In this case LNA
amplifiers hast to be removed, or put into by-pass mode.
To calibrate 8dBi ‘s antennas, power amplifier and LNA has to be both removed
and -70dBm RX achieved on distance 200m. Adjusting antennas from original can be
done by reducing amount of turn on helix. Perfect spot for this - is my window
on 14th floor and a BC place near steam's plant
Reduced size helical shows 10cm length with 15 dBi, and on 40cm around 28dBi
(today it is only speculations). To confirm that needs to test on distance 200m
antennas with 1/3 of turns. And after that conformation has to be done on
regular helix 8cm (3 turns).
For testing 0.5-1WT PA pretty good is glass windows in my building – glass
reduces signal – measuring distance with calibrated 8bDi antenna helix allows to
calculate path loss over glass (LNA has to be bypassed) estimation for today is
15-25dB. Distance around 0.5km – 0.7km on spot of Olympic village can be used to
test full system. To get -69dBm RX power needs to path loss 112 dB (4km) on
0.5WT or 115dB (5.6km) loss for 1WT.
Useful link for this (instead all formulas searches, implementation inside calc
and etc.) -
Nov 21, 2012.
PAE and efficiency. Boris – Q: – if PAE of transmitter
amplifier == 40% then around 60% (definitely less, rest is noise) goes to the
heat – assuming that a thermal radiation is only available method to heat
transfer, then some of the heat will be possible to recycle by reverse Peltier
thermoelectric heat pump. That is easy to check on our next tests with 1WT
transmitter. Even more - with 25WT eating by amplifier and PAE=40%, 15WT goes to
heat, each watt recycled will improve Ecclesiastes' ratio – “time to harvest
power” / “time to transmit”, that will give for each one watt recycled = 2%
savings in total time transmission == 0.5 low resolution frame per 10 minutes
transmission session , or on 16hours (for required 15min HD video) will be 25
minutes savings. Efficiency of Pertier’s elements is kainda a question
(especially reverse one) – in some sources it is stated == 5-10% in some up to
50% - needs to make experiments, but even with 5% it will be 1% savings in
Nov 21, 2012.
Calculations, calculation, and nothing more than
calculations. Thanks Gregory! – FR-4 PCB’s dielectric constant ranges
Er= 4.2 – 5.0 (assumed Er=4.8). Height of dielectric =62mils. Thickness of the
board = 1.25 oz/ft^2 (1.7mils). Useful link to calculate Z0 =http://www.ekswai.com/en_microstrip.htm
. Smith’s calculator resources
http://sss-mag.com/smith.html (very good Smith.exe by Prof. Fritz
Dellsperger from Switzerland). As a result for
on reverse calculation impedance on
input pin = (58.535 –j9.834) on
output pad (5.180+j59.876) –
error 20% based on Er error. Looks like will be require for MOSFET to narrow
amount of channels from 128 to 32. That can give PAE around 40% - for clamed in
spec consuming 25WT it will be 10WT of
transmitted power (with harvesting on 3-rd week max = 4- 5WT it will
require 6 minutes to harvest for transmitter, 4 minutes for control and 1 minute
for transmit totaling == 60sx10K = 600K/min, or 10 minutes == 1 HR picture ,or
24 low resolutions frames, or 15min HR video will be transmitted during 16
http://hamwaves.com/antennas/inductance.html for new design of antenna
it is Zc=128om. For switches good
HMC784MS8GE(10WT, isolation -32dB). Needs to re-trace PCB, order and
test. That is best for today.
Nov 15, 2012.
Class D and class E,F amplifiers.Alex- correction for previous
post about class D: “. . .PAE (Power Added Efficiency) can be around 100%. Some
matching for impedance should be done, but that is a minor question.. . .”
Last statement is not true. According:
impedance eats 60% of efficiency. Suggestion in article to use Class E with
proper impedance matching in topology. But for class E and Bluetooth this also
that with 22.7 dBm (186mW) it is 36%PAE and for 26.4 dBm (437 mW) PAE = 30%.
Needs to ask – does they actually made such device or it is just theoretical
calculations on simulators.
Tubes/wave. To make experiments needs to have ability to get high vacuum
inexpensive way – glass can be printed, everything can be printed and packed
into glass. It is funny – to get the vacuum (free on the orbit) needs to have a
vacuum (expensive on earth) – Isn’t it?
Antenna will arrive tomorrow. It will be possible to make measurements with 0dBm
to compare helical and modified helical designs. I discussed today with Boris
crazy idea about make modified antenna with active antenna’s sections. He looked
at me like professor of physics looks at perpetuum mobile’s inventor. Active
And you have no idea what I bought looking for wave tubes! PDP-11 processor! –
Old! golden pins shining brightly! (at least on picture). It will be embedded
into Cube-sat – may be we can connect a power pin to a ground!
Yours truly Donno
Nov 14, 2012.
Class D amplifier. (translation). Alex- Finally
got what is class D amplifier (thanks for Sasha ‘s tip) – two MOSFET with
similar characteristics (one is N another is P type) working as switches. Gates
and drains connected, one source (P)on a power another source (N) on a ground.
In simulation shows when on gates voltage low then 1.1V, then on drains is
ground, and when on gates >1.1v, on drains VDD. Even gain around 8-10Db can be
achieved, but it does not matter. If to skip the gain then switches will be in
open or closed state, as a result PAE (Power Added Efficiency) can be around
100%. Some matching for impedance should be done, but that is a minor question.
Switching voltage depend on MOSFETs, the same can be said about impedance
matching design. On exits needs to put filter to make sin-cos-things.
Simple like a Byte! Isn’t it?
Some researchers (in scientific patents) stated that it can be important
solution for low cost tooth-blue’s power amplifier because GFSK (Gaussian
Frequency Shifting Keying == shift 180 degree) less picky then π/4DQPSK (45
degree shifting) and 8DPSK(8 phases). On Wi-Fi that idea does not work.
But like in that funny story about the hunter lost direction in a forest, does
not matter what you find, matter how can use it.
a) Variety MOSFETs for 2.4Ghz available only in one polarity. Maximum what can
be constructed from such devices is class A, A-B (theoretical PAE 60%)
b) Check for available switches bring “Ops”. Switches working on huge
frequencies with power 1-5 watts can switch from channel to channel with maximum
frequency up to 1GHz (1 nanosec), and for 2.4GHz needs to have 0.41(6) nano –
Oops in full scale.
Well, what is about “space” – yep – looks like it working but topology for such
chip needs to be ordered as customer’s order. Prices in that case will be around
big green numbers.
With switches looks like more complicated case – restrictions on good toys is in
place – you know –
What was left – (a) tubes, but needs to account power for heating, all those
passive and active induct-capas-res-tors on the way into the tube and a signal's
path out. (b) Traveling Wave tubes. (c) go to school to learn how transistors
That’s it. Nicht gut. Today best is PAE 50% maximum, may be for Cubesat it is Ok
but definitely not for the moon, solar panels on shoelaces required.
Yours truly, Dunno.
Nov 13, 2012.
Class D Amplifier. (translation will be later). Шура!
Наконец-то понял, че такое класс Д усилитель (туп был - сознаюсь! спасибо Саше
за подсказку) - два полевых транзистора (один N другой P полярности) одинаковых
характристик работают как ключи - входы (затворы) соеденены вместе, истоки
запаяны тем же кренделем - сток одного на питании, сток другого на земле. Когда
на входе одно значение (0) то на истоках земля == 0 вольтов - а при значении 1
на входе == на истоках получается питание.
За счет того что схема сидит либо в открытом, либо закрытом состоянии КПД
Все просто как байт.
После выхода ставят фильтр для обрезания в синусоиду - некоторые утверждают (в
высокоученых публикакциях), что зело быть решение для усилка мощьности
голубозубого == т.к. максимум надо чтоб пролез сдвиг по фазе на 180 градусов.
(для Вай-Фай не подходит = там что-то замучено со сдвижкой).
Но, как в том анекдоте, с заблудшим в лесу мужиком, от отчаяния кричащим - А-УУ!
- Услышь-те ! Памаги-те! Кто-нибу-ть! - кричал -кричал - вдруг, кто-то ему лапу
на плечо - оборачивается - медведь - Ну хорошо мужик - вот я тебя и услышал - а
поможет ли тебе это?
так и у меня - от того что что-то нашел - не значит что решение найдено -
Пунктик номер один - на нужную частоту полевые транзисторы только одной
плярности. Максимум что можно собрать из таких транзисторов, это класса А-Б
усилитель (тетеретическийКПД максимум 60% )
Пункт номер два - прикинул что можно использовать обычные свитчи - ан -нет! -
свитчи пашущие на сумашедших частотах, с мощностями 1-5 ватт, (!) переключаются
с канала на канал (!) с максимальной частотой только до одного гигагерца (1
наносекунда) - а надо 2.4гига (или 0.41 и шесть в периоде наносукунды).
Начал рыть, а как же для космоса работают - да работают - только топологию для
такого устройства разрабатывают индивидуально - после разработки профессора
пишут статью - а цена такой цацки минимум 100 штук зеленых.
А для свичев пронблема еще веселей - судя по всему лапу наложили ребята с
шевронами - и хотя есть надежда как в случае с гироскопом - в спецфикации одно,
а на деле переключаются быстрее - но это вряд ли! - должны бдить!
Остаются радио лампы - но у них дикие габариты - нужно учитывать все емкости и
индуктивности по пути в лампу и из оной. Или лампы бегущей волны. Или изучать
принцыпы работы транзисторов.
Вот такая карусель получается - максммум КПДна сегодня 50% - для кубсата может
быть ОК - а для луны явно не хватит батарей - на шнурок придется дополнительную
привязывать - что не есть зер-гут!
5, 2012. Sania, Thanks for info. That is exactly what I
need now. I download exact software last week. It allows matching impedance of
PA and antenna, LNA and BT. Demo version of it take in account of a strip of
cupper trace and impedance can be easily matched. With that software finally I
matched PA and helical antenna it is simple as ABC – impedance of antenna 180
and exit of PA 50 just need to place tiny 403 resistor = 180-50 = 130om. Thanks
again for info. Gregory also give me nice website (in Russian) with detailed
explanation what to do -
http://www.avr123.nm.ru/soglasie.htm. Gregory also shake head about my
approach – he asked what oscilloscope I am using to check signals – My answer-
“build in 1976” - and he replied - “Саня ! на коленке ничего не сделаешь!” ==
"Alex! Good oscilloscope for 3GHz is mandatory."
Status is on critical part today (communication amplifiers) and options
- Amplifiers needs careful PCB’s
re-design – will takea time.
- I finally ordered antenna,
will be printed on 14 November, only after tests I think will be possible to
draw patent – nobody knows may be I am wrong with my “reduced size design”. As
far as it is impedance will be 70-80om, for cubesat it will be smaller size
(more space in cube will be available) – for ground station also smaller – for
example amount of material to print is 1/3 of original helical. But until tests
done it is premature to talk that it will be working.
- Victor suggest to use LAMPA
BEGISCHEI VOLNI (traveling wave amplifier) – looks promising. 70Db. Spread
spectrum. Prices 25-2000. Up to 1kwt. Questions: how to heat electrons
emission’s element, how to place in Cubesat (it is long – also available shorter
version – but nobody knows parameters – kainda Russian military leftover), what
to do with series of magnets (from another prospective magnets is not bad it
will stabilize Cubesat by itself). How to cool down internal element inside
helix. Used previously successfully in space. Not sure about – just
investigation and experiments will take time. May be require experiments in
- Also found modem for cubesat
on 2.4Ghz frequency (price=1000-2000CND) in Calgary
http://www.microhardcorp.com/n2420.php . But not sure about error correction
and compatibility with my hooping frequency protocol. I can give-up protocol in
favor to use that modem, but I am not sure – for cudesat it will work ok but
what will be on main path – the design of scalable 10wt TX and protocol still
will be required. That modem uses 1wt transmit with 5V – 500ma = 2.5wt power
needs to be accumulated before TX. For a modem recommended just a regular
antenna. Our approach sound better – if antenna (again if- after tests – today
it is a speculation) will be polarized then 10 cubesats flying together will be
not created disturbance to each other in communication; in favor to hooping
protocol the same - interference avoidance can be done by protocol also.
- On tip from Alex - still
working on it – need class C amplifier – found yesterday one – PAE=45% but input
require 40mwt – needs to have “middle” amplifier from 0Dbi to 15-18Dbi. MOSFET
is a probably choice for class C – MOSFET is heavy for me – but proper design
can give PAE=90%.
- No information from Serg – he
promised to check with his friends - they did in Donetsk region modems on 2.4GHz
– modem supported IP connection btw mines on distances 10-20 km.
Interorbital did tests – needs to force development. Communication is a critical
one – that is a main – Power plant also was not started yet – but I hope that
solar panels can be simpler – I planning to control it by a PIC and empty space
from antenna can be used to deploy some extra panels.
Jun 6, 2012.
Well, communication protocol done. Now core BT module (cubesat/lunar
modem) is capable to work in modes:
(a) 3 frequencies listening; 3 frequencies for transmit; 2
from 3 majority bits correction, bits shifting correction fix for preamble lost;
synchronization of a frequency switch; lost switch frequency correction; adjust
frequency on temperature drift.
(b) 1 frequency listening; 1 transmitting; the same functionality as
on (a) but switching is disabled.
(c) 1 frequency listening; 3 transmitting; again same
functionality; that will be main mode for a cubesat and lunar probe.
(d) 3 frequency listening; 1 transmitting; difference from (a) that
it will be 3 receivers monitoring 3 frequency but transmitting goes over 1
frequency. That mode will be on ground station – to maximize probability for
cubesat/lunar probe to get less noisy data;
In mode (d) instead of processing and fixing packets inside micro processor
data will be tunneled to a PC and fix of data can be done locally on a more
powerful desktop, or distributed to network of a computers for simultaneously
error correction (for such process can be used screen savers with instant win
notification for a Team Plan B supporters).
Mode (c) allow to have transfer data from cubesat/lunar probe without delay for
frequency switch synchronization. That can increase data transfer from 50
kbit/sec to 70-80. Transfer date for upload is not critical. Another reserve is
tweaking core module from 250kbs to 1mbs – that can give increase of a speed up
to 160kbs for download data from cubesat/lunar probe, for sure such increase
will require more computer power after ground station processing, and will make
a sense only with good network of a “fixing” packets screen savers.
Why to invent a communication protocol? Usually 99% of a time of software
development spend (lost) on following standards. Needs to do something fast ==
ignore standards and write from scratch (see code at
Range test, two helical antennas, one for Cubesat prototype 3 turns
another for a ground station with 1/3 of a turns on 3D printed antenna
transmitting power 1mWt, LNA 12dB. Normally that is a Bluetooth with a range of
10m. Testing on 25m == OK, testing on 100m == OK, but it shows that best
reception will be with antennas pointing 10-15 degrees up instead of direct
pointing (do not judge seriously = truly speaking designers was zero experience
in antenna’s design), test over water at False Creek at Vancouver did not show
good reception – looks like ground or water disturb signal, lifting tripod with
transmitter and receiver manually just 1m up improve communication. Problem was
corrected on a 5 block near Queen Elisabeth Park – road is strait, not much
cars, hills from both side == manual holding and pointing allows to confirm
communication over 450m.
That actually brings limits of a test for a ground station, the best place will
be a Squamish Chief == face of cliff is around 700m and ground station has to be
pointed with 45-60 degree to horizon.
Next step – amplifiers to increase power transmitting to 1Wt (in Canada allowed
4Wt), and mobile ground station assembly/functionality.
Communication system requare to communicate with probe on a distances ranged
low earth orbit (200-300km), high orbit (>500km), earth-moon flight (< 400,000km).
Communication with a probe will be on 2.4GHz frequency. Core of a transmitter and
receiver will be done with a regular Bluetooth chips. Instead of an antenna transmitter
there will be a connect to power amplifier capable to 100Wt pick transmit power.
Receiver pin on chip will be connected to exit from low noise cascade amplifier
capable of 92dB gain. Both amplifiers and Bluetooth chip will be controlled by additional
Protocol will support:
speed, medium speed 5bit/3bit with majority error correction, compatibility, and
hash protection. Mode transfer: half-duplex with hardware flow-control and each
packet with sequence number.
Portable ground stations will include 4 helical double size antennas, receiver’s
and transmitter’s amplifiers, same chip as a probe and microprocessor unit. On portable
ground station microprocessor unit will be connected to a personal computer using
a serial port. On PC special software will collect packets and send its over IP
to a central ground control station. Because of restrictions to communication on
2.4GHz band in different regions of the Earth some stations will be working in “receiver-only”
mode. 6 stations will be located around the globe to cover 24 hours communication
with probe, these stations will require a permit to operate in transmit mode. All
stations will be equipped with horizon – azimuth orientation system controlling
from the same communication software. Backup (manual) orientation assumed.
Each microprocessor unit (on probe and on ground stations) is
capable of remote software download. Each microprocessor unit can work as a regular
AT modem and as a standalone device controlling network for the probe’s internal
communications system. In normal mode of operation the probe will be accessible
as a web server with a designated IP address. Web server will be based on a main
computer module controllable by GET/PUT requests. Output of such requites will be
in XML format with telemetries readings /statuses of a probe.
Frequencies of a probe's transmitter can drift because of temperature. Testing for
temperature characteristics of a transmitter must be done and internal temperature
stabilization, temperature measurement of probe's transmitter can be used for reception's
turning. The same must be done for the on board receiver, it internal temperature
are to be recorded and used to turn frequencies of earth located transmitters.
Actually there is no guaranty that from first launch will be achieved long range
communication, In this case all failures in communication system has to be analyzed,
system can be redesigned and a second launch should be used to achieve mission goal.
for a probe/craft and for ground stations
Ats9 command supports:
from ATS9=LunBMPF1F2F3 i.e.
- check sum
- no big preambula, not transfer to processor
02 - channel 1
frequency == 02
0J - channel 2 frequency = 0x0A
3k - channel 3 frequency = 0x3B = 59
unsigned char Config01;
baud rate for uart communication
00 - 9600
01 - 19200
10 - 38400
11 - 57600
M MM - mode
000 - working 250Kbit check sum 32 bytes payload + 2 bytes chek sum at the end
of data 0xff
001 - 250kBit no check sum transmit 30 bytes payload (at the end 0xff padding)
check sum calculated by rf24l01 on receive 32 bytes - to get chk sum for
010 - 250kbit 3 bit data no check sum each bit converted to 3 bits (payload per
packet 10 bytes)
011 - 250kbit 5 bit data no check sum
100 - 250kbit carrier freq 1 = 1 freq 2 = 0 starting bit on freq1 followed by
freq2 stopping bit ferq1+freq2 for even (0,2,3,6,8) and freq1+freq3 for odd for
odd (1,3,5,7) P - addr == 3 byte preambula + 3 bytes preambula (01010101 6
bytes) X - transfer full received packet to processor