Arrived last Friday. The Sunday before I had brought the signal generator with me from the KONG HQ in anticipation of its arrival, since it was unlikely that I could go there this weekend for testing. A 10k run on Saturday got all the attention (and rightly so since I did a PB on the run).
Its blue box with yellow elements such as the five SMA connectors creates a nice tribute to its Ukrainian heritage.
Anyway: I have tested the device. Installation was almost as straight-forward as I expected. I have tested the Fobos SDR with HDSDR 2.80 and 2.81 beta 6, and SDR# (version 1716). It weighs in at 143 grams. Less than one-tenth of the 2-MHz Perseus22.
A few pictures follow. I wonder how they're going to open the box if they need to service it. There is nothing for a screwdriver or torxdriver to connect to. "No serviceable parts inside"?
I should have checked the block diagram more carefully before I bought it. I was quite surprised that sensitivity on the HF1 and HF2 antenna inputs is pedestrian, to say the least. I measured around -85 dBm on 1000 kHz, actually up to 20 dB worse than other SDRs I have! Now, the block diagram indicates that the LNA only activates in "RF mode", i.e. from 25 MHz and up. So, there is no amplification in the <25 MHz (HF1/HF2) range except the LTC6401 chip.
 |
| So, that's where the LNA is hiding |
What a strange omission. And why on earth didn't I notice.
It does sample 50 MHz. But the lowest sampling rate is 8 MSPS, not 4 as indicated on their web page. They need to offer lower bandwidths as well.
It is very light on CPU. Sampling 50 MHz was handled with ease on my four years old HP AMD PC. Recording 50 MHz - well I'm not sure that it was stutter-free, but it didn't crash the application or the PC.
It is quite warm to touch when sampling 50 MHz. 8 MHz was a lot cooler.
Here at home I have no antennas, so it's really quite limited what I can do. And I really need to talk to the people at RigExpert about this. Hopefully, one of the coming weekends I will be able to test how it performs on FM at the KONG HQ, compared with the RSPdx.
More as it happens.
I got one. It works reasonably well on FM once I added a Kitz Technologies preamp to the antenna; was fairly deaf before that. The HF direct conversion part of the radio from 0-25 MHz is badly broken, and RigExpert seems to have no interest in fixing it. All of the software I use with it mirrors the spectrum, which is to say, if I select 8 MHz as my bandwidth and I tune to 700 kHz, I will also hear 8000 - 700 kHz, or 7300 kHz; I will also hear whatever is broadcasting at 700 kHz at 7300 kHz. RigExpert claims that this is sampling theorem issues, but it works this way through the entire spectrum, not just on the edges the way aliasing actually happens. So if I tune to 3900 in this scenario, whatever I hear there is also heard at 4100. As a result, the radio is pretty much useless for HF and MW.
ReplyDeleteOne other issue; you can’t set the center frequency in HF1 or HF2 mode. I dug through the code; it’s hard coded to be the sampling frequency divided by 2. They’ve made some really curious choices in implementing the direct conversion part of the receiver. It really feels like an afterthought.
ReplyDeleteThanks Ralph, and your observations match mine exactly after testing mine last weekend with antennas. I have described my problems in detail for RigExpert, but I have received no response.
ReplyDeleteWhat they are really selling for HF is just an A/D with an adjustable sample rate clock, a 25 MHz. anti-aliasing filter and a USB 3 interface capability. The Fobos SDR is missing several components that limits its performance.. These are present in other high performance direct sampling SDR products.
ReplyDelete- There is no RF gain adjustment capability (this requires variable gain LNA or fixed gain LNA & attenuators). The external LNA and attenuator are required to increase sensitivity or reduce it to avoid A/D overload. And these will have to be manually controlled which is far from ideal.
- The only filter is a 25 MHz anti-aliasing filter. Preselector filters are missing which will significantly affect dynamic range.
- There is no FPGA chip to do complex digital down convert (DDC) and decimation. This means the full sample rate of 50 to 80 megasamples per second will be sent over the USB 3 connection. More on this below.
The FPGA in other direct sampling SDR's is what takes the high rate digital sample stream and converts it into an IQ sample stream at a lower sample rate. First a complex digital multiplication is made that shifts the entire spectrum down based on the desired band of interest (20 M for example). Then a digital filter is applied to the signal (to avoid aliasing) and finally decimation is used to reduce the IQ sample rate for transmission to the SDR program for demod and display. This takes a lot of processing since each sample is being operated on and that is why a hardware FPGA is used. Doing this type of processing on the host computer is very processor intensive.
And they have removed the ADC LPF bank on board Rev 3
ReplyDelete