I have been experimenting with 40M, which looks good. Others have been running with 60M:
So why is the sample rate limited? Are we risking board damage, or is something(what?) out of spec?
My guess would be heat dissipation. The PCB in the mini has quite a low surface area which can’t dissipate as much heat, as fast as a physically larger surface area. A general rule would be the hotter the temperature of a PCB the lower the mean time between failure will be.
Another issue would be crosstalk. the LimeSDR-USB/LimeSDR-PCIe has a 12 copper layer PCB.(extra ground planes to isolate signals) , where as the LimeSDR mini is a 8 copper layer PCB. When you increase the clock rate, you need to provide more power to turn on and off transistors faster. This means higher current and higher current means stronger EMI (Electromagnetic interference) emissions. So signals from one part of the circuit can enter other parts without sufficient isolation or attenuation and can cause additional noise or totally unexpected behaviour.
In may ways it is like driving a car designed to go at 60 miles a hour at 120 miles a hour.
1 Like
I measured the board temperature (40M samplerate + 2x oversampling) and the FPGA was running over 90C. We built a heat-sink and got it under control, now i am rolling with 4x oversampling. I guess mini with aluminium-closing would be fine aswell.
Not unless it has direct contact at the right places and also acts as a heat sink.
If it doesn’t it will just act like an oven.
1 Like
does the case have contact in the right places?
Hi
Can someone respond to the questions please ?
Did the limesdr mini really support 60Mhz “officially” ?
If not, why ? Power dissipation ? How is the power dissipation of the aluminium enclosed limesdr mini ? Is the aluminium shield in contact with the hot components ? Did you have an inside photography of limesdr with aluminium enclosure ?
Thanks
On the crowdsupply product page they explicitly say that the sample rate is 30.72MSPS, If you want to run it outside of specification without adding some major cooling, expect the lifetime of the product to drop exponentially. The relationship between MTBF and temperature is an exponential curve ( The 10 degree rule: Every 10°C increase in temperature reduces the life of electronics by half [or the 18 Fahrenheit rule]). Also from a RF perspective the higher the temperature of a device (and the larger the bandwidth) the higher the noise floor will be due to Johnson–Nyquist noise. In summary - More bandwidth equals more noise and higher temperature equals more noise.
Like I said already, because a LimeSDR-mini has 4 layers of copper less than the LimeSDR-USB/LimeSDR-PCI there will be less grounding/shielding/attenuation between signal lines so random glitches due to cross talk at a higher clock rate are entirely possible.
I don’t think Lime Microsystems are going to explicitly say that you can run a mini at a higher sample rate than 30.72 MSPS, that would legally open them up to extra costs.
In my mind the LimeSDR-mini’s were made as well as they could be made, but with the price point locked in stone. So you could squeeze slightly more performance out of them by spending more money on enhancing them, but there is an upper limit.
True,
knowing how LimSDR USB performing regarding cooling,
my expectations was exactly the same like described.
I am trying to run my both units bellow 45 C whole time, in lab or in the filed.
BRG
Djani
1 Like
can anyone tell me what is the lowest sampling rate of limeSDR mini?