MSi.SDR sound crackle pulses and spurs

[radiolistener]

Just got MSi.SDR dongle from aliexpress. It was described here.

But unfortunately I got defective one. It has annoying sound crackle pulses across entire bandwidth.
Pulse period depends on LO frequency. For example this defect is most loud at frequencies like 4000000, 5000000, 6000000 Hz (but not limited to these frequencies, it's just an example).

Here is screenshot which shows these crackle pulses on the waterfall (input connector is open): download/file.php?id=934

These sound pulses are so annoying, it's like beating into your head.
I recorded IQ record at 9 MHz bandwidth with SDRuno 1.31, so you can check it here (400 MB zip file and 600 MB unzipped WAV):
https://drive.google.com/file/d/1E3YQ6e ... sp=sharing

Here is audio record: https://drive.google.com/file/d/1dxD3Gh ... sp=sharing

Is it just me so unlucky to get defective item, or these MSi.SDR from aliexpress are all defective?

* * *

Also, there are a lot of spurs, and I found that these sound pulses are most loud when these spurs merging into single peak.
Here is screenshot which shows these spurs: download/file.php?id=935

I can understand spur on 24 MHz, this is XTAL leakage. But what is the reason for spurs at 30 MHz, 60 Mhz, ...?

As you can see, there is also too high DC offset spike. It is more than 80 dB. And the worse thing, is that this DC spike cannot be removed with IIR filter even with 200 Hz bandwidth, because it is unstable.

Does RSP1 and other MSi.SDR dongles has such issues with spurs?

Thanks

[rtlsdrblog]

I've actually seen that crackle before on the RSP1, so I don't think it's a defective unit. The problem was something to do with the USB port dropping samples. I changed USB ports and used a better USB cable, and then it was gone. Could also be that your PC is too slow?

For spurs, yeah the MSI.SDR is not a very good design. I did see a lot of spurs myself, and the lack of filtering really hurts the performance. Without filtering these MSi chips are often worse than the RTL-SDR.

The RSP line is successful because they use a switched bank of filters on the front end to overcome those issues.

[radiolistener]

no, in my instance it is not related to USB at all. These pulses are present with any gain and their pattern depends on LO frequency value. It doesn't depends on sample rate. I also tested with clean external low-noise lab power supply and ferrite beads to make sure that there is no issue with power supply. Tested it with RFI filters on USB cable. Nothing of my attempts doesn't affect that "pulse sound" issue at all. Also I tested it on different PC and notebook. There is no even a little correlation with power supply, USB noise or PC.

RTLSDRv3 at the same USB port doesn't have such sound pulses and crackles at all. I even tried to connect both MSI.SDR and RTLSDRv3 to the same antenna and used them simultaneously. Both devices shows the same sensitivity, but there is no pulses on RTLSDRv3 and these pulses are present on MSI.SDR.

I also found that when you change LO, spurs are moving across bandwidth and at some frequencies they are merged together. Near these frequencies, these random pulses appears across entire waterfall and going to very loud. Also I tested EMI/RFI with RTLSDRv3 around this MSI.SDR pcb and there is no such pulses in reality. It happens inside chip due to some defect. So, this is definitely hardware defect.

Also, my MSI.SDR cannot work stable at 10 MHz sample rate. When I switch to 10 MHz, noise floor starts to jumping like crazy and all what I hear is just terrible noise, random oscillations, pulses and crackles

Also, I found that sometimes with 600 kHz filter and 2 MHz sample rate, sound periodically disappears and appears again. For example it plays sound for 2-3 seconds and then sound stops for a second and then playing again. It doesn't related with PC hardware, because there is powerful i5 processor working at 2800 MHz and CPU load is smaller than 5%.

Also I found that crystal oscillator in MSI.SDR has very high phase noise. It leads to high "Batman ears" effect around clean sine source on the spectrum. These "batman ears" is just 45 dBc. While RTLSDRv3 can show it with about 80 dBс noise floor around the signal.

And also I found that my MSI.SDR has very high IMD distortions, it is much worse than RTLSDRv3. I connected both devices together, and MSI.SDR has distortions and curvature noise floor around power signals (45 dB above noise floor), where RTLSDR shows it with no distortions and with flat noise floor. I tried to connect MSI.SDR through bandpass filter which cuts just a very narrow piece of spectrum. But these IMD distortions are still present.

I'm very disappointed, MSI.SDR that I got works much-much worse than RTLSDRv3 (which costs just 21 USD).

Probably these MSI.SDR are attempt to sell factory rejected chips which is rejected from RSP manufacturing due to some defects. I'm just interesting, if other MSI.SDR also has such defects? Or it's just a random defect in my instance.

Another assumption is that these MSI.SDR using fake Mirics chipsets (cloned). And this may be the reason why my instance have such defects. I think in such case all these MSI.SDR will have such defects. Unfortunately a lot of people is not familiar with SDR technology and will just don't understand why it works so bad. So, if you have such defects, you should know that this is critical defect. Normal SDR receiver should not have random pulses all around full bandwidth.

Here two pictures taken in a row, they shows what sound pulses I'm talking about.
They randomly flashing all around full bandwidth (input RF connector is open):

[rtlsdrblog]

Yeah a while back we looked into designing something like the MSi.SDR with Mirics chips, but decided against it because the Mirics chips have really bad performance without filtering and an LNA (which is what the SDRplay units have to make them work well).

A while ago I asked SDRplay/Mirics (SDRplay and Mirics are run by the same people) where these bootleg Mirics chips are coming from, since SDRplay/Mirics is supposed to be fully in control of the chip distribution, and they don't sell to third parties. But they don't know where they came from.

I don't think they're cloned chips. This market is too small to start a clone. They could indeed be rejected chips that were stolen from bins.

Best to avoid the MSi.SDR IMO, and by a proper SDRplay unit if that's what you're after. The difference in performance is huge.

[artm]

I have this msi.sdr now and tested it. I don't see these crackle pulses and no big dc spike. I have quite lot spurs, they are higher than even plastic case rtl-sdr. Msi seems to handle quite strong signals S9+60dB on hdsdr (gain setting 59) but noise floor is also high S4. Increasing gain to full noise floor jumps to S9+20 !! and dynamic range drops then dramatically about to 50dB. I dont have original RSP but guess it not behaving that way.

Sensitivity is enought to 3.7Mhz HF, but on VHF it can't copy signals at all when rtlsdr v3 read loud and clear (gain 30).
On rtl v3 gain does not affect much to noise floor, usually S1 to S2.

My msi001 chip is marked 191043.1 1038.

On alibaba/aliexpress, there seems to be alot these MSi chips, they claims to can deliver hundred thousands of them..
Here on example: https://www.alibaba.com/product-detail/ ... 75989.html

[artm]

I have some more observations. 24Mhz oscillator harmonics are very strong even to 2Ghz. Most are over S9 and worst S9+30dB. Is it something about pcb layout or something? The front end have some filtering, there is 3pcs 3Ghz rf switches, but haven't look how these filters are made. There's no external LNA at all.
DC spike, on hdsdr v2.76a zeroIF it's about 10dB over noise, lowIF no spike. Rtlsdr v3 small spike but DC removal helps.
On SDR# (1500) 10dB spike and bouncing? same on both IF mode. Rtl v3 no spikes at all.
Ubuntu and gqrx 2.11 there is steady 60dB spike!!

Btw, here one testing for original RSP1.
http://www.dc4ku.darc.de/SDRplay_RSP1.pdf

[radiolistener]

My msi001 chip is marked 191043.1 1038.

This is my one:

MSiSDR-front.jpg (297.87 KiB) Viewed 741 times

I have this msi.sdr now and tested it. I don't see these crackle pulses and no big dc spike.

Does your MSi.SDR works ok at 10 MHz sample rate?
When I switch to 10 MHz on my one, the noise floor starts to jumps like crazy, and I hear some pulses and highly distorted sounds.
Other sample rate frequencies (2,5,6,7,8,9 MHz) works ok (noise floor doesn't jumps).
But crackle pulses (for example at LO = 6'000'000 kHz) are present at any sample rate.

May be someone knows - Is it possible to fix it with some element replacement?

[artm]

I got SDRuno so i did some experiments with it. At 10Mhz after adjusting IF gain noise floor jumps s1 to s7 slowly for some time and then stabilize. Does somehow high noise affect chip AGC? although AGC was off. The noise on display was the flat, no drops at edges until gain is down, when hopping nott occur. Lower sample rates edges was normal. I didn't see crackles but sometimes noise was rought and noise floor bouncing a few dB's. When using low-IF everythin was better. This roughness was on hdsdr too.

Also with sdruno sometimes tuning causes some spikes all over display, these was not seen with hdsdr.

Interesting, in your dongle component markings are scratced off, mine not except one. No electrolyte neither but seems not necessary.

DSCN2110.JPG (874.34 KiB) Viewed 706 times

[artm]

More testing, when (hdsdr) 10Mhz sample rate and tune to 24Mhz (same occurs other frequencies, but worst on xtal and harmonics) then slow and huge oscillation occurs. If tuning one herz up or down and oscillation change, interference is audible on AM too. Also when put finger over input filters or terminate input oscillation change. The spur "bump" also goes around.
I think if whole system oscillates, due to pcb design or is something to bad 24Mhz leak and maybe these affect vco? Voltage at regulator seems to be steady. At 8Mhz sample rate same occurs but much less, spur bump also goes around.

[radiolistener]

Hm... your PCB has different layout than my one... It seems that they are different versions...

Can you show the photo of back side? Do you have marking on the flash memory chip?

Here is my: