FlightAware Discussions

Piaware using x86, aarch or arm64 architectures

History, but interesting to brows, if you have free time :slight_smile:

Anyone working with the new x86 Raspbian

Run Piaware from Win/Mac Laptop/Desktop - Absolutely No Software Installed on Hard Drive

Piaware in Car while at Airport?

1 Like


Here some stats before and after switch between pi and laptop. As we say at Russia: “Feel my finger inside? But my hands are here!”
Dunno, really, I don’t understand why the same RF-hardware (except computing device) shows such difference. Ehm… I know the real difference between ARM and IBMPC arch, I know pretty much about radioelectronics… But I still don’t give a… reason why. Power consumption is exact the same at usb-ampermeter (but! I’ve seen few moments when power consumption! hopped up while laptop installation in use). I know that ALL USB ports at RPI have the SAME power bus so you can’t charge even old (not smart-) phone. I assume that is the reason. Let’s just calculate some simple Ohm units =)
OK, rtlsdr dongle (forget about mods, just for example) with onboard LNA powered by 5V USB. If we just calculate LNA consumption (powering rtl chip has no matter, allways the same to be clear), so 20 db of LNA should be transferred to “times” by logarithmic law. Just to be clear - 10 db are 10 times, but 20 db are 100! In that case, if we operate just LNA power consumption only, a 20 db gain (which is pretty common for average user) takes not 150 mW but 1.5W! See? Even for a 3A copyrighted reserved trademarked PSU the difference already is more than 30%! And now just take a look at rpi usb specifications…
Now I think my point is clear enogh.

Now, just to be clear, lets see a real antennas installation.


Here we see two antennas at one mast. Pretty large dipole lower and classic spider ADSB higher.
Both antennas are calculated and tuned to their frequencies - large made for FM broadcast (3 meters diapason, exact size of each wire wire is 76 cm, what gives us classic half-wave dipole), and that little pimp at he top is widely known ADSB “spider”, real measures are 6.7 cm for each wire (again, tuned classic half-wave dipole for 1090 MHz - you may calculate youreself).
So the key point of LNA powering is SNR, sound to noice ratio. If SNR is low, rtlsdr dongle ruled by its driver tuning its LNA gain to get more db of snr. Simple logic, nothing hard. So, everyone may ensure by own experiment - make two antennas for different diapasons (one should be ADSB 27 cm band, of course) and try to receive something.
So what I expected, what I got and what is completely ensures Ohms, Hertz and Faradeighs (sorry for my bad Eng, dunno real his name spelling but you got it) laws.

  1. If you catch the wave length complicated to dipole - you don’t need to use amplifier, because of resonance;
  2. If you try to catch ADSB using FM broadcast band dipole - you’ll catch it, but you WILL ENFORCED to filter strong wideband signals, which overlaps smaller bands (GSM overpowering filter problem - heard about here?), so you still should use amplifier and filter. Both are going to be power loss reasons and RPI cannot handle it;
  3. and mirror-sided test - catch the FM station using ADSB antenna. Of course, a school geometry shows a SIN(a) law power loss that causes same signal loss, that as bigger as more frequency difference is. LNA needs power again to compensate, dongle getting hot (yep, that’s physics!), RPI USB cannot handle again.
    So we need kill two rabbits by one shot:
  4. Less power consumption at dongle side;
  5. Less power loss at antenna side.
    Second killed by making good antenna, first killed by one two ways:
  6. Screw economy, just gimme a power!
  7. Well-tuned antenna and external LNA (a part of 1st way).
    So any way you go - screw the economy. You’ll get problems with throttling CPU (USB at max current causes undervoltage > less CPU powering > less calculating power > more gain loss > LNA needs more power > go to beginning), overheating CPU (calc power dissipates thermal power > need cooler > using 5V GPIO at 200 mA or much higher beyond typical > go to previous reason) and so on, so on…
    Yes, pretty much factory installations working fine, BUT - coax cable shouldn’t be more than included 10m or so, because LNA will need more power caused of signal power loss… See?
    Just to feed my antennas I need about 15 meters cables!
    So, the final point is… Buy a penthouse =)

If everything else is identical and you’re confident it isn’t a power or software problem, then I’d go for the “except computing device” bit - the receiver is in two different RF noise environments. Maybe your laptop is unusually noisy.

That’s not how it works; amplifiers generally don’t consume power proportional to gain.

I’m not clear what your point was.

Ehm… decibells and “times” are linked exactly proportional decimal logarithmic law, and real power consumption including the another basic law… damn, dunno appropriate Eng word… Ideal operating coefficient? Perpetum mobile staff, you got it. More power you consume, more power you dissipate to nothing (LNA at dongle heats just like any other electronic. More gain - more pain).
That is how it work exactly. And there are no any other phisycs in our universe =)
So real RPI installation cannot show more than 25 db of onboard LNA gain because its datasheet limits. Otherwise - normal USB 2.0 may handle up to 40, but dongle gonna be melted.

Here’s a stackexchange answer that may help explain: https://electronics.stackexchange.com/questions/322023/is-amplifier-power-consumption-relative-to-signal-gain-or-output-volume

The TL;DR is that for many amplifiers (and specifically class A is probably what we’re interested in here for a LNA), the power consumption is dominated by the static power consumption and does not depend on output power much at all. Also, it depends on output power, not gain, and there is a design limit on output power where the amplifier stops being linear and you can’t get more gain out of it.

FWIW the R820T has something like 50dB of configurable gain (as used by librtlsdr, anyway) and I can assure you that there is not a factor-of-100000 difference in power consumption between the two ends of the gain range.

1 Like

And again - I don’t see any reason to struggle against statistics =) Just linked to USB 2.0 (real, not such as op RPIs) of old laptop and got about 30-50% data traffic encreased and twice larger “line of sight”. Just look at my profile.

My bad English… I mean exactly the power consumed and and power went out. Don’t try to catch a grammar mess. 10 times are 1 db only, 100 times are 10 db.
You say up to 50 db LNA can give. Yes, it can. But do you really sure that RPI USB may give 100 000 times more power? =) Even if LNA takes 1microAmps for zero ratio, than 50db are 0.1 A wich is already at the top border of RPI, but real signals a bit (actually, 50-100 times) more, and only a real USBus may handle more than 1A.
See? Oh, whatever, I got much more planes and distance, so that experimental and normal theoretical-based data is competely correct so there are no reasons to discuss =)

No, because it doesn’t need to.

The R820T datasheet is sorely lacking in useful data, but here is an example from another variable-gain LNA (Analog Devices AD8330):

Vdbs is the gain control voltage, nominally 30mV/dB (so this graph shows a gain control range of approx 0dB - 50dB). There is not much change in supply current (maybe 10%) across the entire gain control range - certainly nothing logarithmic there.

The “LNA wants more power” argument is a complete red herring. I’m not disagreeing with your observed numbers, but your explanation for why they changed makes no sense.

1 Like

I get way more variation than that without making any changes - this is the past few days
image

The gain is on the antenna input (analog signal), not on the power input from USB. As in nano W levels of antenna power.
Example: http://otadtv.com/digital_tv/signals.html

The operating power used by any computer is generated by the digital transistors switching on/off and it’s wasted as heat.

1 Like