My LimeSDR is dropping frames

No, that’s great thanks :smile:

I’m having a heck of a time getting my Lime (Lemon) to do anything and I was using FM bcst as a reference. So this that waaay interesting to me. See My Lime seems completely deaf

Two things come to mind for me
1> “Change advanced power settings” to disable “USB selective suspend setting” (at least while troubleshooting, it could be MS Windows trying to turn off or reduce power to USB devices after 15 minutes).
2> Disconnect any other USB devices that are using the same controller, to eliminate them as a possible source of the problem. I use a dedicated PCIe USB 3.0 controller card with “NEC/Renesas µPD720202” chipset for SDR (I have not used this configuration with the LimeSDR [not shipped yet], but I have with other SDR devices problem free).

Side question: Is there any recommendation for radiators and/or fan on LimeSDR? I am using @luftek designed cases (great job btw) but is causes my Lime to overheat :(. I dismounted SMA pigtails from second channel on the case hoping additional holes would increase air flow which helped a little but not enough.

Could it not be that your computer is dropping the frames and not the limeSDR? Maybe it’s just not fast enough?

1 Like

Yes, that can be the case…

I am considering to make measurement setup and plot whole available spectrum response
on modified Rx1 W input. For at last 500 kHz to 2,7 GHz with my Aglilent E7495 on
full span (500kHz-2.7GHz)

My Rx1 L input is left as is, because I have one project in relation to EU GSM bands
RF levels within urban areas.

73 Djani

doubtfully… In my recent test when I’m not dropping frames I have 30%-ish GPU usage and 15% CPU usage (4 cores i7). When I start dropping frames CPU and GPU usage decrease significantly.
I pretty sure this is due to overheating… running lime without the case and/or putting fan blowing into the board helps.

25mmx25mm heat sinks are ubiquitous as are others for the remaining hot ships on the board.
I use glue ons for my RTL-SDR boards and they work.
So far I haven’t heard much about the use of real heart sinks in the right sizes so I figured I’d throw out the obvious reminder.
Heat sink first, then fans.
That is usually the way it works. The other way around produces far less of a result and wastes energy on a fan that is not doing much to actually cool the affected chip.

Could also be USB issues on your computer. In a couple of discussions here the topic of USB3 has come up and for some early USB3 implementations there are inherent problems. Sometimes the computer vendor releases firmware updates.

Are you using USB2 or USB3?

I kind of ruled out USB issues (I’m using USB 3.0 btw) and I’m pretty sure this is overheating. When I run 10Msps it runs for hours without any problems and chips on the board are warm but can be touched with bare finger. When I do 60Msps after 15-20 minutes chips on the board are hot and barely touchable. I am going to order heatsinks for each chip and maybe mount a fan inside the case.

Hi leoha,

Did you already decide which heatsinks part number you will use?
Are you going to glue those?

I don’t know yet but I can let you know once I decide. I just probably try to fit dimension of chips to dimensions of heatsink (low profile preferably) and glue them with either heat conductive glue or some thermal pad.
I don’t have time to do this now but i’ll try to find something on the weekend.

This post is about fans, but maybe will be useful:

My first thoughts are to use the aluminium case that goes with the @luftek end plates.
Then to buy some aluminium square bar cut to the right length and add thermal foam pads to
one end of each (the end that will press against the chip). The other end of the bar will be bolted
to the case.

When the case is closed the bars/pads will gently push against the main chips taking the heat out to the case.
I can then add a standard large heatsink to the outside of the case and a thermally controlled fan like those
mentioned by @Zack .

I really don’t like the idea of blowing air over the board as what ever you do it will accumulate a layer
of sticky dust which will act like a thermal blanket.

  • Charles

[quote=“g4guo, post:21, topic:881, full:true”]
When the case is closed the bars/pads will gently push against the main chips taking the heat out to the case.
I can then add a standard large heatsink to the outside of the case and a thermally controlled fan like those [/quote]

The pad idea is interesting. I don’t know how much pressure these parts can take and alignment might be tricky.

This is true and with so many small parts the potential for problems goes up as dust, etc. accumulates.

Sounds right that it’s heat. Do you have a way to read the temps besides finger test? I ask only because it would be interesting to generate a heat profile for these. I have to find my infrared thermometer and do that.

Therefore my solution blowing only internal air / disipating the heat on box walls
and second heatsink with fan is outside on the box top.

This is the method Red Pitaya use in their diecast case.
I found some 2 mm thick thermal pad on eBay (usually used for GPUs etc).
The centres of the main chips on the Lime are inline so it shouldn’t be too difficult
to drill the holes. You can get 10 mm and 20 mm square Aluminium bar on eBay.

The biggest problem will be cutting the square bar flat and to the right length.
I can always put a washer in there made out of copper sheet if I over do the cut.

You can get copper sheet of various thickness’s (again on eBay).
I even found some heatsinks the same dimensions as the case.

Worth a try anyway.

  • Charles

Interesting that one of those parts is listed as “Active” and they have 0 stock and the other is listed as “Last Time Buy” for 2017-12-31 and they have ~12,000 with EOL announced on July 2016.

FWIW I am going to use a slow fan that just recirculates air internally and one like these mounted externally.

@leoha if you’re curious about the power usage here’s a good synopsis:
I wonder if anyone has a USB3 power doctor to check the real numbers.

I too have concluded this is the simplest approach. Additionally I plan to use a large case with heatsinks adhered/mounted on the inside of the case walls as well as heatsinks on each of the chips.

FYI, for the external fan/heatsink, I read a lab report that found that drawing/pulling air over the heatsink performs better than pushing air into the heatsink. IIRC most CPU coolers draw/pull air.