FlightAware Discussions

FlightAware ProStick Plus + heatsink = bad performance

Now being so so close to putting the antenna outside, I was installing cooling on the FA ProStick Plus. Hoping to get a little extra performance with that.

I applied 1mm of thermal pad. See photo below.

The capacitors are cut out of course and some other components too. Reason: Those components did not feel hot so I thought that attaching them to the thermal pad and heatsink might actually attract heat towards them.
The area in the bottom center of the stick had been hollowed out from the heatsink. No contact possible between heatsink and those 2 components. The chip on the left bottom has been filled up with thermal pad to allow contact with heatsink. The component above is still cut out. The component in the right bottom is cut out because it is tall, it has direct contact with the heatsink sticker.

Result: signal -75%.

Then I covered all of the cut outs with paper tape, making direct contact more than impossible. Result: Little bit improvement. But still -50% from original.

The further the heatsink is moved away from the pcb, the better the performance.

So what is up here? There are users running the FlightAware ProStick Plus with cooling. Did they actually not test before and after and run degraded now? Did they apply a big thermal pad to move the heatsink up?

Remove the compound in the analog section where the signal comes in.

PCB traces for high frequency change their impedance with anything close to them.
Anything but air around them will degrade the signal.

2 Likes

Hi Wiedehopf,

Thanks.

I guess removing the thermal pad from there also means that the heatsink should not be hovering above there?
What area is it exactly? Can I still cover the R820T2?

I which components carry frequency high enough to be suscepible.
The heat sink touching the top of a chip is probably not a problem.
It just can’t touch the traces.

See also the pad for the rtl-sdr v3: https://www.rtl-sdr.com/buy-rtl-sdr-dvb-t-dongles/

My guess is the biggest problem is the exposed solder right at the connector.
Just experiment.

That pad by itself is useless anyway without a metal surface to be pressed against, do you have one with holes for the capacitors?
Just removing the plastic casing should remove enough heat.

Apart from that i’d suggest using a fan lowing on low speed.

Like wiedehopf said, conductive heatsinks should not be close to the RF traces, the parasite capacity affects their performance.
Removing the plastic case helps removing the trapped hot air. Then you can have active cooling of the chips if you want.

Bigger question is… why people insist in putting the sensible electronics (receiver, Pi) outside. It’s so hard to pull a coax cable to inside the house?
You will anyway pull some cable to power that box outside (CAT5 UTP usually).

With my setup the stick is high up in a fully closed antenna pole, almost eliminating coax except for a high quality pigtail. Active cooling is not possible. From there a 3m USB cable to inside the top of the house.

With only the thermal pad applied performance seemed ok, but I just peeled down 1cm from the SMA side. Performance +20%. So your theory got confirmed…the thing is just to find out what exactly can be covered and what not. I mean the bigger the heatsink the better for cooling. Those little ones will only buffer up and after that the chips will still be as hot as they are without heatsinks.

Heatsink looks pretty shitty but it worked very well concerning heat:


So yes, capacitors are cut out. Stuff below the middle cap too, to not bring heat to them, because they were not hot. On the SMA side (left in pic) I cut out some, but apparently not enough.

I think it is save to assume that even without thermal pad, and the heatsink hovering above, so with air in between, there will be a performance penalty. I have to cut off more from the left.

You are doing this for 3 meters of coax cable versus USB cable?
You will still have solar heat gain, no matter what.

Heatsink I just do, because I can. Tuning, tweaking, modding.

Concerning heat from the sun, I have no idea how that will work.
A 2 meter aluminium pole is on a stainless steel wall mount.
About 20 cm below the pole top is the FA USB stick.

That’s just asking for low voltage at the dongle which will seriously degrade the performance.

Just get 3m of coax.
If you really want to minimize the loss, get LMR400, but even LMR200 isn’t terrible for such a short length.

If you are that into min-maxing, i’d go with an external LNA any day (inside after the 3m of cable).

1 Like

Sad to hear. I wish there was a big best practices topic with things like this and the heatsink signal distortion. Would have saved money and a lot of time.

I will measure the voltage at the cable end (under load) and also try a comparison of 0,5 m versus 3 meter.

If you really want to make it work you could run the +5V using an extra cable from the RPi to the dongle. (grab the 5V from the pin header on the RPi)
Then you just need a good power supply that delivers 5.1 V and it should work :wink:

Problem is also you can’t really test if an additional filter would help, because in quite a few places the blue dongle internal LNA gets overwhelmed by mobile phone signals.

Still your setup will likely deliver good performance.
Go ahead and try it, just because others would do it differently doesn’t mean it can’t work :slight_smile:

I guess you’re going to put everything in a sealed box? What if you would use a small and cheap 1.5W 5V solar panel to power a small 5V 750mW one inch fan to cool the electronics?

If is a sealed box, the fan it would just produce more heat inside that space.

1 Like

@wiedehopf
Pi is powered by the official psu, which is said to be 5,1V. In case I run the extra wire…should I only use that extra wire or connect it in addition to the regular USB +5V?

Interference in this area seems ‘low.’ I am no expert thus no numbers but it’s at the edge of an already flat village with less than 50.000 people.

@alleyoopie
Setup is/will be as follows:

Pi inside the house, in the attic.
3 meters of USB cable to the FA stick.
FA stick with or without cooling, depending on the results from testing.
20cm of high quality pigtail SMA-N.
Stanislav 5dB antenna.

Mounting:

  • Wallmount stainless steel 4 bolts and 1 data hole straight through the wall. 30cm circular pipe pointing upwards. Outside diameter 38mm, inside 34mm.
  • Converter piece stainless steel ~20cm. Sliding inside the wallmount 10cm to achieve an inside diameter of 35mm for the piece on top of the remaining 10cm.
  • And that is a 2 meter aluminium pole, 38mm outside diameter, 35mm inside diameter.
  • In the top is a 35mm to 25mm pipe.
  • Inside the 35mm pipe is a 25mm pipe that is the antenna holder.
    Pipes’ insides thus become sealed off from the outside.

So basically an elegant horn (with everything inside) sticks out and upwards from the house attic with antenna on top of that.

I disagree, it would keep the electronics much closer to the air temperature in the box.
Thus the chip temperature is dramatically reduced.
(10 C above box temp is much better than 35 C above box temp even if it’s still rather hot)
The additional heat produced by a low power fan will be negligible.

The air temperature in the box will mostly be determined by outside temperature and the sun.
Couple of watts for a large box are also better dissipated to the outside then then air moves around.

1 Like

You are going to save yourself so much trouble by just running 3m of cable.
Especially considering the stick failing.

LMR400 coax: https://www.amazon.de/Coaxial-Cable-Male-SMA-LMR-400/dp/B07MJ34W2S/
https://www.amazon.de/Coaxial-Cable-Male-2-5m-LMR-400/dp/B07MJ35QPT

Attenuation for 5m is less than 1 dB.

I think that switching to an Airspy mini would make a bigger difference efficiency wise…

Also found out this:
When holding the 35mm top piece in the hand or having contact with it while it is sticking out of the pole, increases messages per second from Dump1090 by 20% most of the times. Sometimes performance drops a bit. I cannot really glue myself to the wall outside, so how to achieve this by default? Grounding the stuff is not an option as it would attract lightning.

It is very annoying to run into all these weird RF behaviours that severely affect performance, yet are unclear to understand.

That about sums up RF engineering from my point of view. I’m convinced it’s mostly voodoo, especially the microwave stuff.

Yeah, a friend basically says the same. Maybe I should just get over it. :lying_face: Difficult with a Kaizen attitude.