ADALM-PLUTO SDR

[borjam]

Don't take this the wrong way but what are you talking about? And maybe also why is it important?

Suitability as a spectrum analyzer, for instance

[CopperCone]

For finding infrequent burst transmissions and transient events.


Lets say you key a transmitter for 4 microseconds. If you use a sweeping generator, that takes 10mS to perform the spectrum sweep, you have a rather low chance of hitting it. If you have something that takes 10 microseconds to do the same sweep, you have a fairly good chance of hitting it.

It's described mathematically as the chance you will see the full signal on your instrument. Similar to waveforms per second on an oscilloscope, so you can see glitches and shit.

[dcarr]

This is a device that continuously streams samples at fixed (maximum) rate.  If we can assume that it's not dropping samples which it shouldn't be, then the POI should be largely determined by math and the software.  That part will depend upon the width of FFT bins you choose, window function, overlap, signal duration, and how well centered your signal is in the bin(s).  If you're interested in the contribution from the noise, you should be able to figure that out from the combination of the device noise figure (see the front end datasheet) and the quantization noise.  If you do some googling, there are some good app notes out there on this topic.

Hope that helps,
David

[CopperCone]

How come on narrow bandwidths it looks like the pluto sdr noise floor has a 'bow' to it?

[G0MJW]

Amateurs in the UK have just discovered the LO harmonics mix with baseband and are providing usable test signals on 10 GHz - so that's another band to tick off. Sometimes spurious outputs can be useful, as long as filtered.

http://www.batc.org.uk/forum/viewtopic.php?t=5129&p=13565#p13565

Mike

[CD4007UB]

The Pluto RF output has quite strong harmonics. With Tx output attenuation of 10dB (in IIO Oscilloscope) at 500MHz, the low-order harmonics measured on a spectrum analyser are:

1st 500Mhz -1.8dBm
2nd 1GHz -51dBm
3rd 1.5GHz -11.2dBm
4th 2GHz -54dBm
5th 2.5GHz -18.6dBm

The strong odd and weak even harmonics suggest the waveform is approximately a square wave at this frequency. With a 1GHz fundamental, I find -16.3dB at 3GHz (so, a bit more sinusoidal).

[CopperCone]

I notice when I have a bicone hooked up to it, this happens

I am at center frequency 2GHz with a span of 1Ghz. I see a strong signal at 1600MHz.

I center it at 1500MHz with a bandwith of 500MHz, the 1600MHz signal goes away.

What is going on?>

[borjam]

I have successfully compiled iio-oscilloscope on Mac OS X. It was quite a nightmare but it seems to work

I'll reproduce it on a virgin machine and try to contribute a set of diffs to AD. It involved installing some dependencies using Macports (gtk, glib, etc), installing gcc (Clang won't accept the compiler flags used by AD) and editing several files mostly to fix paths to include files.

I have one question, however: does the spectrum analyzer tab actually work? No matter what I try I get this error:

Failed to set the rx sampling rate to 20.000000in configure_data_capture

(Ignore the "20" value, it was originally 61.44 MSPS but it doesn't work either.

I can confirm that the bandwidth hack worked on mine. This is a screenshot of the FM broadcast band.

[CopperCone]

I noticed something, its probably not too important due to the high frequencies used, but unlike most lab equipment, the plutosdr has a 1kohm resistance between ground and USB. (unlike most equipment that has bulkhead connectors, and is directly grounded to the chassis earth.

Also, with the hack, since sampling is increased, should any other parameters be varied to satisfy FFT equations and theorum? Like the sample size for a FFT? (since it goes from 20 to 60 megsamples).. I assume with the wrong settings its not exploiting the sample rate to the max.

[borjam]

I have noticed a lot of interference on my SDR_Pluto when I tune it to 430 MHz.

Is this normal? I have cross checked with another SDR (SDRPlay) and an AR8600 receiver and I don't see such interference here.

[CD4007UB]

I just checked my Pluto SDR with MATLAB at 433MHz with 60MHz sample rate and 70dB gain. The spectrum is clean, except for a small spur at 440MHz about 7dB above the noise. (That may be a harmonic from the 40MHz onboard crystal.) IIO Oscilloscope shows basically the same spectrum. The Rx input was terminated with 50 ohms.

[borjam]

Thanks. Silly me, must be interference from a nearby cell phone mast (there's one on my roof, another one 50 metres away).

I'll try at a different location.

[CopperCone]

I have a similar harmonic on mine, I read complaints about that frequency on the RTLSDR reddit forum, some people think it might be from USB (might be a common problem among cheap SDR). I see it at 460 MHz. with nothing plugged in.

My HP equipment does not show it, even with a large antenna, with a noise floor of -144dBm, i'm pretty sure its not cell phone or anything like that, since I have the units are physically very close together.

[SNGLinks]

A video of a talk at BATC's CAT17 Convention.

https://youtu.be/idW3ysqvMno

Edit
Sorry I can't seem to link to the start of the video. Please rewind to the start.

[davorin]

I know it's an old thread...but recently a new batch arrived at Digikey so I had to order one....Mouser postponed their shipping date from May 24th to now somewhere in July....

It still uses the AD9363 but it can't be tweaked for a wider range...I was able to receive FM broadcast with SDRAngel on Linux...
Also the Dual-Core hack works...as it uses the real Zynq 7010 dual-core device...

If you need one...hurry as Digikey has only 60 pieces left at the moment (o;

[tsman]

It still uses the AD9363 but it can't be tweaked for a wider range...I was able to receive FM broadcast with SDRAngel on Linux...

Huh? It can't be tweaked but you're receiving FM broadcast? AFAIK there is no change on the AD9363.

Also the Dual-Core hack works...as it uses the real Zynq 7010 dual-core device...

Yes. It is a real Zynq 7010 with two CPU cores but they're paying a lower price to Xilinx to only use a single core. The FPGA design doesn't fit in the cheapest single core 7007S and the next step would be the 7012S which is more expensive than the 7010.

The end user can enable both cores with no problems.

[davorin]

Yes...I was receiving FM broadcast last week after setting those u-boot bootloader parameters...dunno if it goes higher than 3.8GHz as my R&S generator only goes up to 3GHz (o;

I could test with the R&S with FM if it can go down to 70MHz....

Attached screenshot with FM broadcast receiving...

[Qw3rtzuiop]

AD published a nice book that is about SDRs featuring the Pluto. The pdf can be downloaded for free.

http://www.analog.com/en/education/education-library/software-defined-radio-for-engineers.html

[Geoff-AU]

3rd 1.5GHz -11.2dBm

ouch.  needs lots of filtering.

I notice when I have a bicone hooked up to it, this happens

I am at center frequency 2GHz with a span of 1Ghz. I see a strong signal at 1600MHz.

I center it at 1500MHz with a bandwith of 500MHz, the 1600MHz signal goes away.

What is going on?>

Aliasing.  It's a strong out of band signal at another frequency that's not being filtered out.  Often if you move the centre frequency slowly you can see an aliased image move in the opposite direction to all the other signals in the span.

[Taucher]

Just a small post regarding my experience regarding crashes and lockups with the ADALM-PLUTO board (Rev. B):

My Pluto board sometimes seemed to stall when connected to the antenna - SDR# also crashed when such a stall happened. Using an USB-ETH-Interface the situation did not change at all - the board would still stall from time to time.

Some measurements later the problem was narrowed down to electric pulses due to switching of lamps and (capacitor-based) motors.
Those devices caused a spike for in the coax-shield (approx 60V, 5-20µs duration, 1-2 MHz freq.). 
To check if the device was frozen or just the USB connection I decided to connect an CP2104-MINIEK serial adaptor to the internal RX/TX/GND and see if the device continues to run when the connection is lost.... but the problem was gone - no chance to reproduce it while the serial cable was fully attached.
Well, that was a start - and according to the Analog wiki there were reports for EMC problems with the USB-OTG-detection circuit... so I decided to check out the grounds. 

Diagnosed problem:
Power-GND and GND have a common mode choke as only connection, but that choke does more harm than good when there is a sudden rise in the GND "level".

Solution:
My fix for the situation was simple: I added a bypass/bridge that directly connects GND and Power-GND. In addition I also bridged the 1k resistors to the USB-shields, but I suppose that measure was unnecessary.

Attached you can find the schematic.

[blackbird]

I know it's an old thread...but recently a new batch arrived at Digikey so I had to order one....Mouser postponed their shipping date from May 24th to now somewhere in July....

This is odd, I ordered a Pluto last week at Mouser. Order placed Tuesday (12-6) and received the package two days later (14-6).

[RoGeorge]

Indeed, now there are plenty on stock at Mouser EU. Quite in sync with the new Analog Devices free book about SDR.

That made me ordered an ADALM-PLUTO last night.
Can hardly wait for it to arrive 

Thanks for both the link to the book and the news about the replenished stock.

[CopperCone]

Just a small post regarding my experience regarding crashes and lockups with the ADALM-PLUTO board (Rev. B):

My Pluto board sometimes seemed to stall when connected to the antenna - SDR# also crashed when such a stall happened. Using an USB-ETH-Interface the situation did not change at all - the board would still stall from time to time.

Some measurements later the problem was narrowed down to electric pulses due to switching of lamps and (capacitor-based) motors.
Those devices caused a spike for in the coax-shield (approx 60V, 5-20µs duration, 1-2 MHz freq.). 
To check if the device was frozen or just the USB connection I decided to connect an CP2104-MINIEK serial adaptor to the internal RX/TX/GND and see if the device continues to run when the connection is lost.... but the problem was gone - no chance to reproduce it while the serial cable was fully attached.
Well, that was a start - and according to the Analog wiki there were reports for EMC problems with the USB-OTG-detection circuit... so I decided to check out the grounds. 

Diagnosed problem:
Power-GND and GND have a common mode choke as only connection, but that choke does more harm than good when there is a sudden rise in the GND "level".

Solution:
My fix for the situation was simple: I added a bypass/bridge that directly connects GND and Power-GND. In addition I also bridged the 1k resistors to the USB-shields, but I suppose that measure was unnecessary.

Attached you can find the schematic.

I measured it before and I thought there was also a series resistor between the grounds ? IIRC my measurement was in the kiloohms.


BTW as a mod for my plutosdr, purely based on hunch, I wrapped the thing in two layers of copper tape and sandwiched the tape down to the star washer on the SMA connector. I made a cut away around the signal USB port to maintain their designed ground isolation but I was tempted to short it out to the shield. It seemed really dumb to have a plastic enclosure on a USB device that works to 6GHz.

I suspect that choke will improve things so long its not being overloaded or whatever is going on, I am sure AD put it there as a result of a measurement they are RF experts, but it is a bad design, and it does not sound like chassis shielding would help much?

Can you explain what the mechanism is behind the failure? Is the impedance of the ground plane not low enough relative to the power and signal planes, so a impulse on the shield/circuit ground causes a rise in the signal layer? Might it be better to completely float the USB shield and use isolation?

[Taucher]

I measured it before and I thought there was also a series resistor between the grounds ? IIRC my measurement was in the kiloohms.

BTW as a mod for my plutosdr, purely based on hunch, I wrapped the thing in two layers of copper tape and sandwiched the tape down to the star washer on the SMA connector. I made a cut away around the signal USB port to maintain their designed ground isolation but I was tempted to short it out to the shield. It seemed really dumb to have a plastic enclosure on a USB device that works to 6GHz.

I suspect that choke will improve things so long its not being overloaded or whatever is going on, I am sure AD put it there as a result of a measurement they are RF experts, but it is a bad design, and it does not sound like chassis shielding would help much?

Can you explain what the mechanism is behind the failure? Is the impedance of the ground plane not low enough relative to the power and signal planes, so a impulse on the shield/circuit ground causes a rise in the signal layer? Might it be better to completely float the USB shield and use isolation?

The problem is not that the Pluto is crashing - it's just that a sudden change in any GND causes an edge on the USB-OTG-detection (and that leads to an unwanted switchover between host-mode and client-mode for the USB port). Such a rise happens when connecting different GND potentials together (antenna to Pluto is one example) or when there is travelling disturbance in the incoming GND due to some household devices or nearby lightning.

Now any such disturbance would usually be a job for the choke ... and it would do a good job - but it is also an inductor ... and inductors cause a phase-delay ... on it's own that's not a problem, but the OTG-measurement-path is not choked and hence there is no phase-delay there.
Now feel free to guess what happens when you offset a 60V AC-spike by 90°

Quite possibly there are better solutions than mine (post them please  ). My guess would be some really strong low-pass filtering to the detection pin (but adding a C_filter next to an L_choke could cause some nasty byproducts as resonances and ringing... I didn't bother to try). The correct fix would probably be an additional choke which allows for (co-)choking of the OTG-sense, but adding such a part to an existing Pluto is nontrivial

The posted solution has been tested to work (I've also received feedback from one other person who had Pluto crashes as well - and not a single "crash" since the mod was applied). After all, risking some PowerGND-noise (haven't measured) is much less an issue than the whole system losing it's USB connectivity.  My Pluto shows superior performance compared to any other SDR in my posession, so I'm quite happy with it

Regarding R between grounds: please consult the schematic... my proposed fix creates a low impedance bypass so any existing parts between both grounds will continue to work in respect of the ratio created by the respective voltage-divider.. as a wire is nearly 0 ohms it becomes the strongly preferred path for the current (basically it gets all of it). In my device I also bridged some ground-separating R like R55 ... but that's not necessary and I was too lazy to undo it as it has it didn't affect my Plutos performance.

Regarding any added shielding: I experienced some RF-reflection problems - in my case I tried adding copper planes of single sided PCBs (cut to fit the enclosure) ... such shielding didn't help with the OTG-loss but caused the noise-level to worsen significantly - especially when the shield was close to the RF-input, chip and the chokes in between both. It's not necessarily bad to radiate some unwanted RF away - I didn't have RF-absorbant material around, but even then I'd expect bad HF voodoo to happen when adding anything with a dielectric constant near any RF part

PS: What also helped (a bit) was just creating a really good GND-connection (I used alligator-clips and lab cables) from the SMA port to my local grid's earth conductor - this reduced the GND-interference conducted by the coax-shield arriving at my pluto by providing an alternative path to earth. For obvious reasons this just helped with mild interference as it just weakened the incoming stray signals (at risk of ground-loops).
Choking with a clip-on ferrite didn't help as the frequency is quite low and the intensity too strong to simply choke away.

[CopperCone]

how did you ground the shield you made? I don't see any other method then binding it to the RF i/o if you want to not mod the chassis?