Close

Radio upgrade

A project log for 3D printed truck

3D printed running aid

lion-mclionheadlion mclionhead 03/11/2021 at 18:550 Comments

So far, lions only used unamplified radios of 5mW on their own boards. Extended use of these radios has proven disastrous. Their range is only 15ft. The mane problem is near phone towers. The GSM bands overlap the 900Mhz ISM band so they die completely. Frequency hopping did nothing to improve it.

5mW was good enough for 13 years of indoor flying & stationary weather stations. When they're well above ground & using well aimed directional antennas, they can go 500ft. Thus, it was hoped they would be enough for outdoor driving.

There are only 2 low cost chip amplifiers on the digi key: the 1W SE2435 & the 500mW RFX1010. The lion kingdom went with the 500mW because it's much easier to hook up. As usual, it was a 2 week snail mail time. The limitation on power is the battery life. Some reduction of the output power or the duty cycle will be required to manage the battery life.

The XBee Pro 900 made 50mW & never had any dropouts when driving. It was worthless in the longer ranges required for flying, but it was flawless for driving. It was just too big to fit in the modern controller. It burned 210mA all out. The duty cycle got it down to 100mA.


The gold standard in the commercial world is the DJI Mavic, which only transmits 400mW. The top end Spektrums only transmit 7mW at 2.4Ghz. They obviously rely on directionality.

The mane problem with getting any more power is the small form factor, small battery, & not being able to use a directional antenna.

The upgraded board was a bodgeorama, but it was the lion kingdom's 1st home made transmitter above 5mW.

The trick with the RFX1010 is the antenna input is shorted to ground at DC.  It needs a 33pF blocking capacitor.  Then, when the transmit enable pin is on, it burns over 150mA with no input.  When the transmit enable pin is off, it burns nothing.  This pin needs to be actively turned off when not transmitting.  Receive enable & mode have to be grounded to get it to completely shut down.  The control pins are floating.  Mercifully, the switching time is only 1us.

Then of course, the high power requirement of the amplifier caused problems.  The ADC became unreliable when the amplifier ran.  Making it use the dedicated RC oscillator got it to work for some reason.  The LP2989 voltage regulator got stuck on 2.5V until it got a 100uF input cap.

When it finally worked, it only burned 50mA for the full .5W.  The duty cycle was low enough to use full power.  Frequency shift keying is clearly visible on the spectrogram.

The 9 discrete channels used in the frequency hopping mode are clearly visible on the spectrograms.


The fully populated board.

Testing the battery charger, LED, & speaker.  Testing is essential, since it's not reworkable in the enclosure.

Bench testing was ugly, with much worse results than the unamplified radio. It seemed to be saturating the receiver. Moving it farther away or covering it improved reception. Hot glued the bodge wires down instead of making a new board. Buttoned it up & it worked much better in the field. It managed to fend off the phone tower & range was no longer a problem. There were still dropped packets near the phone tower but not enough to kill it.

The impedance matching is obviously horrendous. Having the receiver right next to the ground is horrendous. The lion kingdom was just hoping for enough improvement to do the job. The battery has to be charged after every drive now. It took 30 minutes to hit constant voltage on the inductive charger. Something would have to be rigged out of a multimeter to get the number of coulombs.

The big question is exactly how much power it's radiating, compared to a Spektrum or a DJI. It spreads the 500mW across 100kbits. The Spektrums concentrate 7mW in a very narrow bandwidth so they have long range. The DJI video transmitter radiates 1W across a 20Mhz band of frequencies, so it has shorter range. No-one knows what the DJI uses for remote control. It may be the same as the Spektrum.

This leaves the gyro glitches & the erratic speed.  I2C needs to be done in software.  That was how lions made it more reliable, 10 years ago.  

Discussions