A few days ago I received an MSI-SDR stick and accessories, which I had bought from the “Chinese” for a horrific 42US $ . Only a few technical data were known about the stick:
10kHz to 2GHz, 12Bit ADC, -0.5ppm TCXO and 50Ω input.
That is why I wanted to find out more about the “Blue Miracle” that was unknown to me. I had the SDR console from Simon Brown and SDRuno available as operating software. A comparison at the same test frequencies showed strong secondary reception points when using the console , which did not occur with SDRuno.
Likewise, the console shows completely wrong level relationships at different bandwidths, which is obviously a general oneProblem of this software is. SDRuno is physically correct in this regard. All subsequent measurements and tests were therefore carried out with SDRuno.
The stick shows many whistle points when the entrance is complete. Theinputlevels above -90dBm equivalentare subsequently listed in the form “f in MHz / level in dBm”. Thelevel information refers to the displayed level, which, however, can differconsiderably from theactual equivalent leveldepending on the frequency(see frequency response). 8 / -89 24 / -59 32 / -85 48 / -69 60 / -74 96 / -85 120 / -81 192 / -74 216 / -78 240 / -73 264 / -76 288 / -79360 / -70 408 / -75
To measure the frequency response, a constant -80dBm unmodulated without cable was fed directly from a Marconi 2041 into the stick. The measured level was read on the S-Meter of the Uno software. The resulting level error is shown below.
In order to measure the reflection loss at the input of the stick, only a low measuring level may be used so as not to overdrive it. All measurements therefore took place at -10dBm source level and 40dB attenuation for Port1 of the HP8753.
Together with the bridge attenuation of the S-parameter test set, this results in a level at the input of the stick of less than -60dBm. The measuring arrangement can then still determine reflection attenuations down to -30dB with sufficient accuracy .
The picture visible here is obtained for frequencies between 60 and 119.999999MHz.
This image applies to frequencies from 120 to 249.999999MHz. At all higher frequencies that isReflection loss similarly bad. For this reason, an illustration has been omitted.
The user should always be aware of the poor input adjustment and the resulting consequences!
Due to the poor input matching , all measurements are carried out with a 10dB attenuator connected upstream of the stick , whose attenuation is calculated out. 500Hz are selected as the evaluation bandwidth. From this, the minimum detectable signal per 1 Hz bandwidth and the noise figure are calculated.
At 60MHz and 1000MHz there are violent jumps in the limit sensitivity.
2nd order intermodulation up to 60MHz
With a noise figure of 27dB in the entire AFU shortwave range, the stick is of course interesting for applications on amateur antennas . This works without additional band passes, however, only if it has good properties with regard to second-order product formation. A critical situation arises when products from several FM radio stations fall into the KW band.
As a representative, 93MHz and 96.7MHz are fed in each with -20dBm. We receive on the difference frequency of 3.7MHz. The pleasing result is a product level of only -101dBm. Since there is no converter overload, a low-pass filter must be effective. A weakening of the FMSignals around 10dB delivers -111dBm on the receiving frequency.
This signals a behavior that stands for a classic analog front end with an IP2 of 61dBm. You can live with that well.
But what happens if the interference signals are below 60MHz? For this purpose 21MHz and 24.7MHz with -40dBm each are fed in. Again the difference is 3.7MHz. The measured product level is -100dBm.
This results in an IP2 of only 20dBm. Other frequency combinations and levels also result in an IP2 of just 20dBm.
The conclusion from this is: For operation in the KW range, sub-octave bandpasses must be switched between antenna and stick.
2nd order intermodulation above 60MHz
The reception scenario was selected as 435MHz. Once 201MHz and 234MHz were applied with -40dBm each, so that their sum product was received. Furthermore, tests were carried out at the same level at 565MHz and 1000MHz.
In the first case, a product level of -111dBm occurred, which corresponds to an IP2 of 31dBm. This is a typical value for a front end with single-ended signal processing . Here, too, suboctave bandpasses should be used to improve the data.
The second case was even worse with a product level of -92dBm, which corresponds to an IP2 of only 12dBm .
3rd order intermodulation 3rd order
Product formation was measured in characteristic amateur radio bands. Thetest signals were each 10kHz apart and had a level of -40dBm. A check regardingthe behavior of the productlevelsaccording to the radiation set wascarried outat -35dBm or even at -30dBmdepending on the controloption. All results show that the analog front end in front ofthe AD converter determines the properties.
f in MHz IP3 in dBm
The sideband noise of the stick was determined by feeding a test signal from a Marconi 2041 transmitter in low-noise mode. At frequencies> 1GHz, the measuring transmitter still manages more than 135dBc / Hz within a few kHz. The noise levels were read at a distance of 10 kHz from the carrier with a 500 Hz CW bandwidth and calculated back to dBc / Hz.
f in MHz dBc / Hz
Overdrive limit of the AD converter
The overdrive limit was assessed as the input level at which there were just no characteristic secondary lines in the spectrogram. However, SDRuno shows “overload” at levels that are a few dB lower.
f in MHz level in dBm
Evaluation of the measurement
Results Unfortunately, the assumption that the low frequency limit suggested that the signal is passed directly to the AD converter in the baseband has not been confirmed. This can be seen from the rather poor values for product formation and sideband noise in the KW range.
The stick should only be operated on an antenna via sub-octave bandpasses in order to reliably avoid unwanted ghost signals or an increase in the background noise. For the purpose of HF measurements, the stick should only be used by amateurs with sufficient experience in measurement technology.
You always have to switch additional damping and by observing the displayed signals, convince yourself that what you see is really there. Even signals that are not classified as a product in this way can still appear in the stick as alias signals or mirrors. These can usually be recognized by the behavior when the frequency changes. I did not address this problem in my measurements because of the extremely complex behavior of these signals.
Despite all the restrictions, the MSI stick is an inexpensive enrichment for every radio amateur who likes to experiment with HF signals. I hope to have contributed to exploring the limits of this tool.