Cheap "MSI SDR" (SDRPlay RSP1) clones, anybody have one?

[radiolistener]

Input anti-aliasing and preselector filters do indeed affect IMD-related and aliasing spurious responses.  But of course so do the A/D and NCO and datapath.

I'm talking about self spurs, when there is no signal on the input (open input). They are depends on LO frequency.

Regarding to IMD distortion, I tried to measure it, but MSi chipset works very strange. At a glance IMD of MSi chipset is much worse than  RTLSDR, but it very depends on the gain, so it's hard to measure it's exact value, because it changes significantly on different gain, signal level and frequency. RTLSDR works more predictable. May be this is because I was used two-tone mix obtained from two square waves, so there is possible mirror channel and aliasing influence.

[cdev]

Also I hear that msi001 tuner has just 85dB dynamic range, so this is the reason why MSi.SDR cannot have better dynamic range than RTLSDR. Just because dynamic range of analog frontend of MSi.SDR is limited to 85 dB. So, 12 bit cannot help here, and spurious LO makes things even worse...

For example:

RTLSDR:
8 bit ADC = 6.02 * 8 + 1.76 = 49.92  dB
Processing gain: 28.8 MHz => 3 kHz = 10*log(28800/(3*2)) = 36.8 dB
SNR = 49.92 + 36.8 = 86.72 dB

MSi.SDR
12 bit ADC = 6.02 * 12 + 1.76 = 74 dB
Processing gain: 24 MHz => 3 kHz = 10*log(24000/(3*2)) = 36 dB
SNR = 74+36 = 110 dB
But it is limited with msi001 tuner to 85 dB, so the total SNR = 85 dB
Almost the same as RTLSDR...

Neither the RTLSDR or hackrf has dynamic range much above 60db. Which isnt very good at all.  85 db even would be a huge improvement. 110 db is decent, what I would expect from a decent SDR. A good sound card based SDR can often get that. Its a world of difference.

[A.Z.]

the better dynamic range of MSi.SDR allowed me to handle conditions which a bare RTLSDR

Here is comparison for dynamic range left is RTLSDRv3, right is MSi.SDR.
RTLSDRv3 is connected through 30 dB attenuator (because RTLSDRv3 has better sensitivity).
MSi.SDR is connected directly (with no attenuator).

Thank you for the test, as I wrote I didn't run real tests, all I can say is what I experienced; months ago I still had the RTL dongle (the MSi just got delivered), I was listening on the 20m band when a local ham started transmitting there (on a different frequency on the same band), the RTL dongle became totally deaf when that happened; being curious, I switched to the MSi unit and that one was able to deal with the local transmission and allowed me to keep listening the other signals, now I believe this does make a difference, doesn't it  ?

As for comparing the units, I thought we were discussing about cheap commercial SDR units, and I believe that comparing those with highly priced ones would be like comparing apples and oranges, otherwise we may just compare an RTL or MSi unit with an ETTUS USRP and conclude that the RTL/MSi are crap

[radiolistener]

Neither the RTLSDR or hackrf has dynamic range much above 60db. Which isnt very good at all.  85 db even would be a huge improvement.

I don't have hackrf, so I cannot compare it. But RTLSDR has dynamic range about 80 dB. You can see it on the screenshot.

Left is RTLSDR, right is MSi.SDR. There is 40 MHz signal which is shown with -5 dBFS level on both. And both shows noise floor about -85...95 dB.

RTLSDR shows SNR = 80 dB
MSi.SDR shows SNR = 80-85 dB

Some small difference in dynamic rage is really present, it's about 5 dB. But this is almost nothing.
Because you can lose these 5 dB on bandpass filter for example

But again LO beat and noise which is present on MSi.SDR ruins all advantage.
I think it's better to use silent RTLSDR with fair 80 dB, than noisy MSi.SDR with 85 dB and these annoying pulses and buzzing with about 20 dB above noise floor and which you cannot remove.

[radiolistener]

As for comparing the units, I thought we were discussing about cheap commercial SDR units

yes, we're talking about cheap SDR with price < 50 USD.

I believe that comparing those with highly priced ones would be like comparing apples and oranges, otherwise we may just compare an RTL or MSi unit with an ETTUS USRP and conclude that the RTL/MSi are crap

MSi.SDR cost about 40-45 USD. This is a lot. By spending just a little more money (about 70 USD) you can build fair 12 bit SDR setup with much better performance.

For example: https://www.aliexpress.com/item/4000796488236.html

[A.Z.]

MSi.SDR cost about 40-45 USD. This is a lot. By spending just a little more money (about 70 USD) you can build fair 12 bit SDR setup with much better performance.

Well, I paid mine around 40 euros, so I believe it's still in the "cheap units" range; as for building ... yes, one can go for that, but then it won't be a "commercial" unit anymore and then one may decide to look elsewhere and, for example follow this design

https://circuitsalad.com/2020/01/06/compact-si5351-based-sdr/

which also uses a Spin Semiconductors FV-1 chip

http://www.spinsemi.com/products.html

 

[A.Z.]

Just in case, here's mine with the FM bandstop filter and the chokes

[cdev]

With the caveat that it was designed for TV with radio as an afterthought, I am petty sure that the MSI sdr is a 12 bit SDR.

The RTLSDR is definitely nothing like a high bit depth SDR. Weak signals cant be heard. This is obvious on the ham bands - any frequency really.

That said, the idea of using it for the US digital TV system as well as an SDR is interesting. Can it demodulate TV, say off of the air in an area with broadcast HDTV, either European, Asian or multi TV systems with an MSI dongle, I wonder?

 That was what the original literature on it that I have holds out the promise of. (as well as using it as a radio receiver)

Also, MSI has a web site. This is all likely explained there.

I wonder if one can clock the MSI SDR externally?  It would be great if you could clock all your gear using the same GPSDO-derived oscillator signal(s)

[fourfathom]

The RTLSDR is definitely nothing like a high bit depth SDR. Weak signals cant be heard. This is obvious on the ham bands - any frequency really.

I agree that the RTLSDR isn't as good as a higher bit depth one, but in my testing of the RTLSDR-V3 I saw a noise floor of about 0.005 uV.  My input was a 10.001 MHz CW signal and I was using the "csdr" program as a SSB demodulator.  I set it to 10 MHz, giving a 1 KHz SSB beat-note, and a 5 KHz audio low-pass filter.  This was all running on a Raspberry Pi.  I used a 10 MHz bandpass filter and a 30 dB preamp at the filter (total gain about 24 dB).  With a 1 uV input signal I measured about 55dB SNR.  This was using the direct-sampling mode of the RTLSDR.

This is pretty decent sensitivity at HF.  In fact I may have too much preamp gain, since there will be a tradeoff when it comes to overload.  I was more worried about the aliasing though, given the 28.8 MHz sampling clock.

If you're curious, here's a ham club presentation I gave on that project (http://wb6cxc.com/wp-content/uploads/2019/04/Presentation-1-0.pdf).  Since then I've made a small combo filter/preamp board that I can stuff for any frequency (well into the VHF bands).

[radiolistener]

With the caveat that it was designed for TV with radio as an afterthought, I am petty sure that the MSI sdr is a 12 bit SDR.

The RTLSDR is definitely nothing like a high bit depth SDR.

No, they both have the same performance, both use the same technology. And both have SNR about 80 dB. But MSi.SDR has more noise issue, more unstable and has more bugs.

As I said before, "bits" is just marketing here. MSi.SDR has performance of usual 8 bit ADC. It may use 12 bit inside, but dynamic range of analog frontend (tuner chip MSi001) is about 70-90 dB (the same as RTLSDR), so you cannot get advantage of 12 bit. At some conditions with MSi.SDR you will get even worse dynamic range than 8 bit ADC.

But resolution bits is not the full story. There are a lot of other important things. ADC speed, DC offset, spurious performance, phase noise, LO purity, SNR of RF tuner and many other things. The resolution bits don't play important role in cheap SDR dongle.

Weak signals cant be heard. This is obvious on the ham bands - any frequency really.

This is false information. RTLSDR has better sensitivity than MSi.SDR.

For example, with RTLSDRv3 I can hear very weak signals on 145 MHz. But with MSi.SDR the same weak signals are not detectable at all, I even don't see any indication of carrier presence on the waterfall.

In addition, MSi.SDR have much worse dynamic range when you select high gain.

Also, MSI has a web site. This is all likely explained there.

No. MSi.SDR is Chinese clone. SdrPlay said that MSi.SDR is not their product and they don't support it. They even deleted all messages about problems with MSi.SDR from their forum.

I wonder if one can clock the MSI SDR externally?  It would be great if you could clock all your gear using the same GPSDO-derived oscillator signal(s)

This is completely useless, both RTLSDRv3 and MSi.SDR use TCXO and frequency error is about 1-2 Hz on 10 MHz. This is almost nothing. There is no frequency drift, just a little frequency error at factory calibration.

[radiolistener]

Here is my test setup - both RTLSDR and MSi.SDR connected through T connector to antenna. I performed S-level calibration to keep the same signal level on both devices, so you can compare sensitivity level. And the third screenshot with disconnected antenna.

It shows receivers in realtime simultaneously.
Left part of screen is RTLSDRv3.
Right part of screen is MSi.SDR.

The first screenshot with connected antenna.
The second screenshot with disconnected antenna.
Both screenshots taken with the same settings. So, you can compare sensitivity and noise floor.

As you can see RTLSDR has a little better noise floor and a little better sensitivity than MSi.SDR.

[A.Z.]

Since my MSi unit is serving me better (at least in my experience, till now) than the RTL one I had, and since I'm just marginally interested (if it works, I'm just going to try squeezing out as much performance as I can, instead of debating), I'll try to stay away from this, that said, it would be interesting to see what happens to the noise level (at bottom of spectrum display) when you connect the MSi SMA antenna connector ground to a good ground, I'm writing this since in some cases it made a LOT of difference

[radiolistener]

Since my MSi unit is serving me better (at least in my experience, till now) than the RTL

I think it's better to understand the root of cause why it serving you better. And I think the root of cause is because when you put very strong signal on the input RTLSDR it is overloaded much earlier than MSi.SDR. Just because RTLSDR has better sensitivity.

You can fix it just by adding some attenuator on the input of RTLSDR and you will get the same result.

[radiolistener]

it would be interesting to see what happens to the noise level (at bottom of spectrum display) when you connect the MSi SMA antenna connector ground to a good ground, I'm writing this since in some cases it made a LOT of difference

It won't affect result. Both receiver will show you exactly the same result.
Because such connections don't affect receiver performance. It affects antenna performance.

And I don't recommend you to connect "good ground" it will leads to bad antenna noise performance. The proper way is to use RF chokes on your coax cable and put your antenna and it's counter poise as far as possible from your home and any electric equipment.

[A.Z.]

Since my MSi unit is serving me better (at least in my experience, till now) than the RTL

I think it's better to understand the root of cause why it serving you better. And I think the root of cause is because when you put very strong signal on the input RTLSDR it is overloaded much earlier than MSi.SDR. Just because RTLSDR has better sensitivity.

You can fix it just by adding some attenuator on the input of RTLSDR and you will get the same result.

Did that (the attenuator) when I still had the RTL unit, but that didn't solve the issue and in any case didn't help with DX signals (you usually don't want to attenuate them  ), again, I decided to buy the MSi.SDR dongle time ago, compared it with the RTL I had (same setup), and decided to give away the RTL and keep the MSi, and till now I'm a happy camper, then if it doesn't work for you and you prefer the RTL units, it's fine with me, but I don't think I'll go back to an RTL dongle and then maybe I'll go forward to some better SDR unit (I'm considering the AirSpy HF+ and the RSPduo), but then, it's still just a maybe

[A.Z.]

it would be interesting to see what happens to the noise level (at bottom of spectrum display) when you connect the MSi SMA antenna connector ground to a good ground, I'm writing this since in some cases it made a LOT of difference

It won't affect result. Both receiver will show you exactly the same result.

And trying that would be easy, so... why not ?

[radiolistener]

Did that (the attenuator) when I still had the RTL unit, but that didn't solve the issue and in any case didn't help with DX signals (you usually don't want to attenuate them  )

I believe you, this is not so easy to properly setup and configure all things, it requires to understand a lot of things. This is a common problem when equipment with better performance gives worse results with incorrect settings. I compared both devices very deep and can say that this is almost impossible to get better performance with MSi.SDR instead of RTLSDRv3. With proper setting of course.

Of course RTLSDR has it's own cons and pros. For example it suffers from aliasing issue (between 0-14.4 MHz and 14.4-28.8 MHz), you can find these issues on the screenshot that I posted above. But you can fix it with proper LPF/HPF. But overall performance of RTLSDR is much better.

[radiolistener]

And trying that would be easy, so... why not ?

I prefer to understand what is going on under the hood instead of random trying. I understand what happens and why, when you connect "good ground". Antenna don't needs "good ground", it needs good counterpoise.

This is like Ohms law, when you know it, there is no magic why current depends on resistance and voltage. You can predict it and your calculation will be exactly the same when you try to measure it. There is no magic. The same thing is here.

When you trying to connect ground, your're change your antenna. Your receiver performance don't depends on the ground, it remains exactly the same. But the ground affects the signal source on the input of your receiver. By connecting and disconnecting ground, you're trying to compare different antennas. Not different receivers.

[cdev]

You need to use some means of isolating the two receivers from one another. The way you have connected them is not going to be representative of either of them because you will get spurs from one SDR being presented at the input of the other. I would simply use an A/B antenna switch. Thats much better than a splitter because a splitter will give you the same problem, plus attenuation of the signals. Even a cheap CATV antenna switch will work. I would also turn off the SDR you are not using and anything like a upconverter you may be using. Turn off anything that produces LO signals.


You do need to use a good ground. Ideally a good RF ground. That is a whole subject in itself. But it is very important. Otherwise your receiver will use its power source and all wires connected to itself. That could bring in a lot of crap. I use a 12 volt battery to power my HF receiver. Doing that right off the bat cleans up a lot of junk.  The USB connection brings in its own problems with dirty power. I have always added multiple additional capacitors to the power rails on my RTLSDRs.


Also the picture you show, is not showing any way of differentiating the two by dynamic range. You need images that contain more strong signals, next to weak ones.  You'll notice that the dynamic range is less than 50 db. Also, the performance of the RTLSDR (the only one of the two I am familiar with) varies a great deal as you move around the bands. Especially on direct sampling. As you mention, the performance at certain bands is very bad, I also find my RTLSDR is almost useless for 10 and 6 meters and VHF low band. Also the bottom of 2 meters which I am very interested in receiving, is horrible.

Here is my test setup - both RTLSDR and MSi.SDR connected through T connector to antenna. I performed S-level calibration to keep the same signal level on both devices, so you can compare sensitivity level. And the third screenshot with disconnected antenna.

It shows receivers in realtime simultaneously.
Left part of screen is RTLSDRv3.
Right part of screen is MSi.SDR.

The first screenshot with connected antenna.
The second screenshot with disconnected antenna.
Both screenshots taken with the same settings. So, you can compare sensitivity and noise floor.

As you can see RTLSDR has a little better noise floor and a little better sensitivity than MSi.SDR.

[radiolistener]

You need to use some means of isolating the two receivers from one another. The way you have connected them is not going to be representative of either of them because you will get spurs from one SDR being presented at the input of the other.

yes, but I didn't compared spur performance on that screen. I compared sensitivity and noise floor.
Such setup is good enough for that goal, because it provides exactly the same signal level for both receivers

Signal amplitude is exactly the same on both receiver's input and this is exactly what I wanted for the test.

Also the picture you show, is not showing any way of differentiating the two by dynamic range.

Yes, this is not dynamic range test. This is sensitivity and noise floor test.

Dynamic range of both receivers almost the same. It's about 80 dB.
For MSi.SDR if may vary from 45 to 92 dB depends on the gain and frequency.

[cdev]

I have been unable to receive a bunch of weak shortwave stations using anything on my RTLSDR setup that come in clearly on other more-bit SDRs.

That said, for the price, which is very low, they cant be beat.

With a good upconverter they compete well with tabletop consumer radios with shortwave capabilities. On direct sampling they can work really well with a mag loop. Which adds preselection.

I would be surprised if the SDRplay had only 42 bits of dynamic range. If that was the case I think I would have heard people bitching about it!

But it seems to me like you are either accessing some capabilities which I have never seen before or maybe mistaken as far as RTLSDRs having that much dynamic range.

Unless they have new drivers that unlock more bits. Which I think is very unlikely.  And as I said, HackRF is around the same. What people like about HackRF is the very wide frequency range they support.  Not as a high quality solution, as a quick and dirty way of scanning a very wide range of frequencies.

Like the RTLSDR, they get you involved by forcing you to spend a lot of time thinking of workarounds so you can get the most performace out of them despite their many shortcomings..

After all they were made to sell, as a mass market item, to workers so they could catch a lottle TV on their laptops.

Sell profitably for $10.

And so, being so cheap, for learning they are ideal.

[radiolistener]

I have been unable to receive a bunch of weak shortwave stations using anything on my RTLSDR setup that come in clearly on other more-bit SDRs.

Don't confuse ADC resolution bits, dynamic range and sensitivity. These are different things.

If your RTLSDR cannot hear some weak signal while other receiver can do it, it doesn't related with ADC resolution bits. This is just because your other receiver has better sensitivity. Your other receiver may use 5 bit ADC or even 1 bit ADC. It doesn't matter. This is sensitivity issue, not ADC resolution bits.

You can use Low Noise Amplifier (LNA) with proper filters in order to listen weak signals on RTLSDR.

But regarding to MSi.SDR, it has even worse sensitivity than RTLSDR.


Regarding to the bits, you can consider it in complex with ADC speed and analog frontend dynamic range. Because total dynamic range depends on all these things.

For example you can get 100 dB dynamic range on 6 bit ADC and 40 dB dynamic range on 12 bit ADC. Just because 6 bit ADC has processing gain and 12 bit ADC has analog frontend and clock phase noise limitations.

This is why when you're talking "12 bit ADC" this don't means that it is better than "8 bit ADC". You're needs to consider many other factors.

With a good upconverter they compete well with tabletop consumer radios with shortwave capabilities.

I don't agree. Direct sampling on RTLSDRv3 works good enough. And in my opinion it works better than upconverters.
The only thing that you can take care is just proper bandpass filter and LNA on the input.

But in order to get best performance you're needs to understand how to configure it properly and how to tune it to avoid problems. This is not so easy, especially if you don't know all details how it works under the hood.

For example, I never use AGC (automatic gain control) on RTLSDR.
For me it's better to setup gain manually, it gives much better result and more control.

[cdev]

I never use AGC either, and I also have a couple of e4000 RTLSDRs which are much better for UHF than the R820T/T2 SDRs.

By all accounts, what the RTLSDRs do, all the many tricks they use to make the most of the bits they have is truly phenomenal.


Additionally, on top of that I have done a great many things, to try to stretch their performance even makde a bunch of different LNAs with their front ends optimized (as far as choice of components) for different frequency bands.

Still, there are a lot of signals which my RTLSDRs cannot receive, even when put on a good antenna. I would say that as far as the things i want to listen to, most of them are not accessible to an RTLSDR.   In the past I didnt have any really good antennas and I had no way of evaluating the ones I made myself, but now I do.

Some of it is living in a very noisy RF environment, some of it is also (now) the summer heat. Which raises the noise level substantially in the daytime.
But some of it is the radio. I do have a substantially better SDR which I use for HF. And its dramatic. I really do see that on HF.

[radiolistener]

I would be surprised if the SDRplay had only 42 bits of dynamic range. If that was the case I think I would have heard people bitching about it!

Don't confuse ADC resolution bits and dynamic range. These are different things.
And resolution bits is not equals to dynamic range of receiver.

SdrPlay use MSi2500 chip it uses 12 bit ADC but it doesn't means that it provide you the better dynamic range, because you're needs to take into account process gain, phase noise, and other things.

And the things going even worse, because SdrPlay also use MSi001 tuner chip (the same as MSi.SDR) which dynamic range is limited to about 45-92 dB, depends on the gain settings. It means that you will get just about 80-90 dB dynamic range for receiver. Almost the same as RTLSDR (80 dB).

[radiolistener]

Still, there are a lot of signals which my RTLSDRs cannot receive, even when put on a good antenna.

A long time ago I also had the same thoughts that sensitivity and reception quality depends on ADC bits. But when I tried better ADC with higher resolution and higher clock, I found that it don't provide you better sensitivity. It just provide you more dynamic range (it depends on clock speed and resolution bits).

Dynamic range just allows you to keep low noise floor with very strong signals on the input. It can be an issue when you have broadcast transmitters near your home. But also you can fix it with bandbass filter.

If you want to get better sensitivity, you're needs to work on LNA, bandpass filters. With proper LNA and bandpass filter you can get almost the same result as very expensive top receiver.

Better dynamic range just allows you to not use bandpass filters on the input.

With RTLSDR and proper filter and LNA you can receive any weak station, if it even possible at all

But if you're want to add LNA, first make sure it has low noise figure. Because noisy amplifier cannot amplify weak signals, it will flood the singal in self noise of amplifier. So, noise figure of amplifier is a critical parameter.