RTL-SDR BLOG announced the first release of our new RTL-SDR Blog V4 dongle which is based on the R828D tuner chip.
The V4 comes with several improvements and changes that are listed below.
- Improved HF Reception. Now uses a built in upconverter instead of using a direct sampling circuit. This means no more Nyquist folding of signals around 14.4 MHz, improved sensitivity, and adjustable gain on HF.
- Improved filtering. The V4 makes use of the R828D tuner chip, which has three inputs. We triplex the SMA input into three bands, HF, VHF and UHF. This provides some isolation between the three bands, meaning out of band interference from strong broadcast stations is less likely to cause desensitization or imaging.
- Improved Filtering x2. In addition to the triplexing, we are also making use of the open drain pin on the R828D, which allows us to add simple notch filters for common interference bands such as broadcast AM, broadcast FM and the DAB bands. These only attenuate by a few dB, but may still help.
- Improved phase noise on strong signals. Due to an improved power supply design, phase noise from power supply noise has been significantly reduced.
- Less heat. Due to the improved power supply design the V4 uses slightly less current and generates slightly less heat compared to the V3.
- Cheaper price! The price of the R860 chip which is used in the V3 and most other RTL-SDR brands increased significantly at the beginning of 2023 which is part of the reason as to why RTL-SDR dongles have been increasing in price recently. For the V4 we are making use of an existing stockpile of R828D chips which are now priced cheaper than new productions of the R860. In a time when high inflation keeps pushing prices up this is incredibly welcome.
There are some other minor changes including a new bias tee LED and a small cutout hole in the enclosure so it’s easy to tell when the bias tee is on.
Of course the same innovations that we brought in with the V3 are still implemented such as the sleek conductive black metal enclosure which works as a shield and doubles as a heatsink, a thermal pad to sink heat away from the PCB, 1PPM TCXO, SMA connector, USB noise choking and improved ESD protection.
The V4 however does come with some disadvantages compared to the V3 that need to be noted:
- Due to the increased filtering there can be an average of 2-3 dB less sensitivity on some bands. Please see the MDS measurement graph below for the full picture.
- The V4 requires the use of our RTL-SDR Blog drivers. Our RTL-SDR blog drivers are on GitHub. Please be sure to follow the installation instructions on the quickstart guide carefully as the V4 will not work with default Osmocom drivers. In most cases using our drivers simply means running our install-rtlsdr-blog.bat file, or replacing a dll file. (We are working to upstream the main changes to the Osmocom repo too)
- The V4 is a Limited Edition Design. The R828D tuner chip is completely out of production now and the number of units we can produce is limited by the number of chips held by our contract manufacturer in China. They have indicated that there should be enough stockpile for about a years worth of production.
Because of these tradeoffs we will continue selling the V3 alongside the V4.
More About the V4 Design
The R828D
The core change on the RTL-SDR Blog V4 design is the change from the R860 tuner chip to the R828D tuner chip. The R828D was previously a more expensive chip, however with the huge price increases on the R860 which came in effect at the beginning of the year we have decided to make use of existing R828D stock which is now cheaper that the R860.
The R828D is very similar to the R820/R860 and shares much of the same circuitry. However, instead of just one input, it comes with three switchable inputs. We have used these three inputs together with a triplexer to create a dongle with some extra input filtering. In the past there have been some R828D based dongles on the market, but all designs are based on TV receiver circuits. Because our design is different, you will need to use our RTL-SDR Blog driver branch which has added compatibility for our R828D design.
Also please note that because the R828D chip stock is limited, and R828D is no longer in production, the V4 design is also a limited design which we expect to be able to sell for about a year.
HF Design
The HF design consists of a SA612 double-balanced mixer circuit with front end filtering, which is connected to the 28.8 MHz oscillator that is also used for the tuner and RTL2832U chip. This means that HF frequencies are upconverted by 28.8 MHz. Our drivers handle this upconversion seamlessly, so you just need to tune to 0 – 28.8 MHz in order to receive HF. There is no need to set any offset.
An upconverter design also means that unlike direct sampling full gain control is available, and also there is no folding of signals across 14.4 MHz due to Nyquist.
Adding Basic Input Filtering
One of the main problems with RTL-SDR dongles is overload from strong broadcast stations such as broadcast FM, broadcast AM and DAB. By using a triplexer circuit we can make use of the three inputs on the R828D tuner chip to provide some filtering. The triplexer splits the input signal into HF (0 – 28 MHz), VHF (28 MHz – 250 MHz), and UHF+ (250 MHz – 1.766 GHz). This means that interference from something like strong broadcast FM at 88-108 MHz is more isolated when we are tuned to the HF and UHF bands.
We’ve also made use of the open drain pin on the R828D (which is not on the R860) to implement a simple switchable notch filter for the main problem broadcast bands. These notch filters cover broadcast AM, broadcast FM and DAB, and reduce them about an additional 5-10 dB. By default the notch turns ON when tuned out of these bands, and is turned OFF when tuned within them.
In terms of sensitivity, the disadvantage of adding more filtering is that it can reduce sensitivity in some bands. However, sensitivity of the RTL-SDR is usually not a problem in most situations, as we’re usually limited by desensitization from strong out of band signals as mentioned above. If sensitivity is a priority an LNA such as our wideband LNA should be used anyway, for any RTL-SDR brand or model. Any front end LNA will totally dominate the sensitivity figures, making any sensitivity measurements of the RTL-SDR itself irrelevant.
Revised Power Design
The revised power design makes use of a more modern LDO with significantly better power supply noise rejection which results in much lower phase noise seen on strong narrow signals. There are also some PCB tweaks to reduce internally produced noise. The LDO improvement also has the effect of reducing power usage and lowering heat.
Other Changes
We’ve also added an LED to the bias tee, so it’s easier to tell if the bias tee has been activated in software or not.
MDS Measurements
The minimum discernable signal (MDS) is a test we can do to determine what is the minimum power level that a receiver can detect.
The results show that the MDS has significantly improved on the HF bands thanks to the upconverter design. However, there is some minor degradation in the VHF and UHF band.
Two Tone Isolation & Desensitization Test
Strong out of band signals can cause an SDR to desensitize on other bands. For example, very strong broadcast FM (which is common), can cause signals being received on other frequencies to be received with a lower signal to noise ratio.
In this test we injected an “interference” tone (Tone A) at 95 MHz, and injected a second tone (Tone B) at another frequency. We then slowly increased the power on Tone A. When we noticed a 3 dB drop in the signal strength of Tone B we recorded the power level of Tone A that this occurred at.
This gives us a way to see the effect of the triplexer filters and notch filters when compared against the Blog V3 which has no filtering. A higher recorded value means that a stronger signal is required to desensitize the receiver, meaning that the strong signal handling capability is improved.
From the difference graph we can see that isolation results within the same triplexer band are improved by about 8 dB thanks to the notch, and then out of band isolation is improved by 28 – 43 dB thanks to a combination of the triplexer filters and notch.
We note that between 305 – 1405 our measurements were limited by the max power out from our signal generator, and we believe the true results are roughly 5dB better than what was recorded at these frequencies.
