Feedback Request: New RTL-SDR Product, Ideas and Interest Check

We are considering building a new multi-purpose RTL-SDR product. The idea is to make several difficult to achieve applications and projects much more accessible. We are looking to implement the following ideas:

  • 3x on-board coherent RTL-SDRs built into the PCB
    • 4x SMA inputs: 3x individual inputs, 1x common input (switched between the two). 
    • All RTL-SDRs connected to the same clock source – enables coherent experiments
    • All RTL-SDR feature sets and performance equivalent to RTL-SDR V3 or better
  • On-board noise source and directional coupler
    • Useful for correlation with rtl_coherent
    • Measure filter characteristics, and get rough SWR antenna readings.
  • Noise source able to be switched in and out via silicon switches
    • Useful with rtl_coherent and other coherent experiments for cross correlation timing correction. This allows for accurate direction finding.
  • Ability to mount onto a Raspberry Pi 3, and provide an ESD protected, buffered and filtered output for RpiTX transmissions. (a PCB plugin filter specific to the transmission frequency would need to be installed onto PCB to use this feature)
    • With a filter installed the board can be connected to an antenna and used with RpiTX for simple transmissions.
    • Go portable with an Raspberry Pi 3 compatible HDMI LCD screen and a battery pack. Possible HackRF portapack alternative.

Possible applications:

  • Multi-band RTL-SDR applications
    • One RTL-SDR receiving NOAA, one receiving ADS-B, one scanning the air band.
    • Easy trunk tracking with 2x RTL-SDR. Third RTL-SDR used for something else.
    • One streaming NOAA weather, one scheduled to receive NOAA/Meteor sats and weather balloons, one receiving Outernet weather updates.
  • Coherent applications
    • RF direction finding
    • Passive radar
    • Possible radio astronomy applications?
  • Noise source applications
    • Characterize filters
    • VSWR meter with directional coupler
  • Raspberry Pi mount applications
    • Replay attacks and security analysis of ISM band devices with RpiTX and an ISM band filter.
    • Transmitting WSPR with WSPRpi.
    • Portable if used with a small HDMI screen and battery pack.
    • Possible control of board via an Android app.
    • Similar applications to the HackRF Portapack idea.
    • Multi-band noise locator if a GPS is added to the Pi. e.g. See Tim Havens’ ‘Driveby’ concept.

The idea is still in the concept stages so we’re looking for any feedback from the community to see if this is even something that people would want.

Would a receiver board like this interest anyone? We would also work on providing basic ready to go software on a downloadable image file for the Raspberry Pi 3 so starting an app would be as easy as using a launcher. We would also provide various tutorials as well.

The target price would be $99 USD. If you think this is too much, please let us know what you would expect to pay in the comments.

Are there any additional features that anyone requests? Please let us know in the comments.

Would you pay $99 USD for a 3-input RTL-SDR coherent receiver with built in noise source, antenna switcher and filtered RpiTX output?

View Results

Loading ... Loading ...
Notify of

Inline Feedbacks
View all comments
Peter Sheerin

I would definitely buy this for $99. My main interests are:
* direction finding
* Multi-band reception

I have a few questions, though:
1) In your bullet point “4x SMA inputs: 3x individual inputs, 1x common input (switched between the two).”, can you clarify the “switched between the two” means? Since there are 3 receivers, which two is it switched among, and why? Or is it switched among all 3?
2) Can you provide more details on the RDF capabilities? That’s the nature of my question about switching between 2 or 3 receivers, since using only 2 receivers leaves you ambiguity in most cases.

Alan Corey

If you have 2 or 3 receivers it’s going to need 2 or 3 times as much processing power and getting all that through USB seems like a bottleneck. It almost seems like what you ‘re looking at would be better as a motherboard that could host N plugin receivers, each with its own digital connnection. I like the original just fine, I’ve got 3 of them now.

Rob T.

Currently I’m not interested in multiple coherent receivers. If I was it would be for direction finding.

I’m interested in the broad band noise source and directional couplers that would allow for characterizing antennas and filters to take the place of more expensive test gear.

Also I’d be interested in a two antenna / two path — HF and everything else, solution similar to the HF+ but closer to the rest of th RTL characteristics in range, price, etc. with full ESD protection.

Also psychologically at a $100 USD price point, I’d expect something that is more of a polished product that isn’t a work in progress. I have a lot more tolerance in the <= $50 range for experimental stuff. But that might be just me.


Very interesting, especially the transmitting feature !


The coherent receiver sounds interesting. I’d be interested in something capable of low data rate S-Band reception so the modular downconverters as part of the package would be good. I believe the current units can handle 38k4 BPSK reception but handling 2200-2300 transmissions with the DC modules is nice. If the data rates could handle higher data rates then extending cubesat and similar reception becomes interesting e.g. 40mbps QPSK on ~720MHz X-Band but maybe that’s too specialised or covered adequately by the mid range SDRs.


Is it not possible to request RTL manufacturer to improve their rtl2832u chipset (according to SDR) with same I/O interface. If I am right then their are most users using these chips for SDR purposes only rather then intended purposes by manufacturer.


It costs around $1 million to fab a design into silicon. And that is excluding any patent license royalties that would need to be paid. And that money is required up front. Yea you could request it, if the market size was big enough that they could swallow the cost and then make their losses back in the first three months and then pure profit after that. But is the SDR market big enough for that ? (maybe you couuld find a fab that will do it for less using older 350nm technology and delay expecting profits after 6 months). The bottleneck with the ultimate perfect custom silicon would move to the the R820T2 which has a maximum filter width of 15MHz.

So if you got the rtl2832u chip modified to be the ultra cool rtl2832u-sdr which ran a 7-bit ADC at 28.8msps and decimated that down to 15msps and transferred that that across USB 2.0 HS at 30MB/sec that would instantly remove the RPi line of hardware, because they would drop USB packets every second of ever minute of ever hour on every day with the letter “A” in it. They just can not handle that much data, so that would drop the potential market size by a factor of 100.

Knowing this if you decided to press on the other option might be to do a really small run and hop into bed with the U.S. Department of Defence using , but I’ve no idea how big the license fees payable to .RealTek would be for that, or whether they would even allow it to happen. If I was betting person I would lay good odds on “No” being their answer.

And on top of all that redesigning silicon to work first time, every time, is a highly specialised job, mucho mucho dinero. At a million a run a failed batch is bad, you got to spend another million on your bug fix run and hope you did not make a different mistake.

My personal option is that at this exact moment in time custom, ASIC is not a viable option for SDR.


I messed up that URL it shoudl be


After hackrf and limesdr project campaign we have already see how many people support this project. I think after some improvement in rtl2832u chip, chip price go double or thrice time more but customer may be happily pay for this improvment.


Design: ~20 weeks (which could easily could go up to 2 years with a complex analog and digital design, has to work first time, one tiny fault = a lot of money wasted ).
Simulation (at around 0.5-50 kHz depending on the complexity of the design): ~20 weeks (more is learned in the very first second that the final device is powered up then all of the time spent in simulation)
Manufacture (from final design): dope, etch, masking, and other processing to the wafer (physically shipped to a remote site – vibrations from processing are bad for the previous stage), testing wafer, wafer cutting, mounting, wire bonding, encapsulation, testing, weed out early failures (power up and run basic tests for several hours inside an oven to weed out early MTBF devices): ~8 weeks

You could be talking 1 to 3 years from concept to having a physical chip in your hand. Yes it is possible, but what other chips are coming to market during that time that will make yours obsolete before you even have it. It is much safer to piggy back on existing chips than trying to make a new one.


Could you provide more information about technical parameteter and design you intend to use like frequency range and design, type of noise source and coupler you intend to use

Here a few thoughts, additional ideas
– Modular sounds good,
— what about need/option for shielding e.g. noise source
— alternative connectors to SMA, SMB,SMC?

– which blocks/components will be available, which should be external, shielded
— filter (-bank)
— LNA and divider
— (pin diode) attenuator
— noise source or VCO
— directional coupler

– reference clock 1 ppm,
— output of reference clock,
— option for external clock

– how about IQ-output
— to allow use of higher resolution AD-converter
— combination of IQ for time domain analysis

– down converter to extend reception to 6GHz e.g. LTC5577 – High Signal Level Active Downconverting Mixer $8.20/100
— also for reception of frequencies above 1 GHz

-upconverter starting from 0 MHz, e.g. by using the if-port of a double balance diode ring


Were you thinking about a simple DC like Minicircuits offers or one similar to the design HP described in AN which later DJ7VY described in UKW Berichte or another design?
From what I could find about the BG7TBL Noise Source it’s a Zehner Diode with amplifier stages added. As a cuation I measured a roll of hundreds of Zehner-Diodes long ago until I found some 3 that generated lots of broadband noise. Unless the Zehner-Diodes are preselected, which may be the case for the BG7TBL Noise Source, I would assume this approach to provide a not to consistend noise source. Other designs using part of transistors may be a better choise.
If you have a Noise source you may think about an

– Antenna Noise Bridge for antenna measuring

vy 73
-.. ..-. …… .–. -.–


I would DEFINITELY BE INTERESTED in this board(s)!


1) 100% Linux compatible, on PI3’s or AMD64 desktops. If it supports non Linux too, don’t care… but so long as its Raspbian, Debian, *buntu compatible, GREAT.

2) I am not sure that the Pi’s even a Pi3, is up to running 3 of these at a time, say with one for trunk controll channel, and the other 2 for decoding voice.. the 2.4Mhz BW, would require the need for the other 2 to decode voice due to the spread of channels within the 851-860Mhz range for a f ew systems. Throw in streaming to my IceCast server via DarkIce or EZStream, and the Pi’s start to use 100% CPU.

Additionally on the Pi3’s it probably would need some sort of cooling with all that type of stuff going on. If this is a PiCrust that covers over the board… possible cooling issues???? Or this is a USB connected device to make it controller independent ie: Pi3’s and PC’s.

3) I have NO interest in the PiTX side of this, and honestly can see this being a problem an a source for issues with regulatory agencies from the FCC to others. Its a neat party trick that you can get a Pi to do this. But it requires a lot of filtering on the outside to do much. I’ve got $30 dual band radios to $3K P25 to $20k service monitors that can do all the transmitting I need. I peronsally am only interested in the SDR’s and getting them running rtl-fm to receive NOAA NWR and pipe that to my IceCast server. Using SDR Trunk or Trunk Recorder to pipe the local P25 systems to IceCast. In a few other spots using rtl-fm in scan mode, ie: multiple -f and pipe that to IceCast. If any thing to support the PiTX stuff could be removed and get the cost below $99 and be profitiable, GREAT!

4) In regards to this: “f someone could put together an easy-to-use dedicated scanning app that doesn’t require the use of virtual audio cables and half a dozen different software packages/plugins, I think that could really boost sales and usage.

Make scanning on the PC as easy as using Uniden’s HomePatrol scanner should be possible! It just isn’t. Yet”

I would LOVE to see a single LINUX application that is a “soft” scanner.. BUT what I definitely do not want is that software to be tied to that stupid database that Uniden is forced tied to now! The newest radio to get ProVoice support is totally useless! It can NOT BE PROGRAMMED MANUALLY! I don’t want or use that stupid database source, its outdated, and I have no intention of supporting that site to correct it for them to PROFITEER FROM ME! Programming a scanner for a P25 system is pretty dead simple. Put in at most 4 channels have them set for P25, and BAM it spits out audio. For my main system I need one channel programmed. There is a leanring curve to everything… For example the EDACS systems in my area need ALL CHANNELS programmed in a specific order, aka LCN for EDACS to work. Its not that big a deal.

Still even if the board came with other features I have no use for at $99/each, I would snap them up! Just for the 3 SDR’s on a board.

So Two thumbs! I’ll take 10!


I think there may be better platforms to base it on than a raspberry pi… Its peripheral architecture (everything on USB) would probably cause a bottleneck. I would want to be able to sample all 3 RX at the full 2.4 Msps and still have the ability to stream the samples over a network. I have doubts that a pi3 could keep up with that (happy to be wrong though)

I also really like the idea of having an integrated noise source and directional coupler. That would bump up the value significantly, especially when you look at the cost of similar test instruments. It would be nice if the directional coupler was dual directional and it used 2 receivers so it received the forward and reflected simultaneously. You can get away with a single direction though.


I was curious so I did a little more research on the pi3’s USB throughput. It looks like it should be able to support 3x 2.4 Msps while simultaneously streaming it over the ethernet interface. I see where one user is hitting a sustained 321 Mbps throughput over USB (to a gigabit USB adapter.) They’re only using one USB device though. I have encountered problems with stability of older pi boards while trying to use more than one USB device. Looks promising though.


I am not able to run 2 SDR’s at 2msps each at the same time on a pi3 without dropping packets as seen in rtl_test… Maybe its my USB HUB, haven’t tried putting each on its own on-board port instead yet.


I’m not sure about which feature generates how much production cost but from my point of view the RPiTX does not fit into a 3 channel coherent receiver application.

100$ might be ok if the product is a low risk hardware (=> if software samples are available for direction finding for example) and if the board is stackable (sync, clock lines). Without 70$ – 80$ might be a interessting price (from my point of view) because in this case i would rate it like 3 single rtlsdr v3. If cohrent software is available later fine, if not i have 3 independent sdr channels.
Maybe it helps to keep the price down if some parts most of the users might not need are not fitted but in this case the parts should be standardized, easily available and easy to solder (like a sma-connector for the external clock input).


RpiTX filters are not needed “built in” on your boards as they can already be fitted directly to a raspberry Pi.

I do not want to spend more than $50 on a board that I can accidentally blow up whilst experimenting.

Joining $50 and $30 modules together would be acceptable.

On the Outernet Forum one of the comments as to why someone did not buy a new $99.00 (now $89) Outernet Dreamcatcher board was that as a single board that did everything it would be a expensive boat anchor if you plugged it in wrong.. But with the Old Outernet modular system it was more likely if you made a mistake you would only brake a module ie a LNA or RTL Dongle etc.


I voted yes, but there are a few caveats. $100 is too much with the few features listed here. However, I’d be willing to pay around $100 or more if you include more hardware features (build into the unit, or modular). A solution like the LNA4All or another similar LNA would be a must. Including an up-converter and or band-pass filters, e.g. FM, would sweeten the deal even more.

I think software is best left to the community to develop, not necessarily the OEM, so I don’t really care if it comes with PI support or anything special. I’d rather see more hardware features in a device like this, because without them nothing makes it more special than just three synced RTL-SDRs (which I could hack together myself).


A LNA in a receiver is bad, and can cause overloads, ideal placement of a LNA is at the antenna.


May be provided separately.


Now, I am interested in lna4all make sure they are in water proof case for outdoor.


Yeah, for the LNA and whatnot I meant more to have it included as a separate unit in the package, not on the board.

Mike Frisco

I love that you’re considering making a more powerful revision!

I wouldn’t want to pay more than $50.
I could get by with only two tuners, just enough to scan trunked systems.

In my opinion as a scanner user, not as a HAM or whatever, is that the biggest deficiency in the RTL-SDR world today is software. It feels like you need to use a bunch of half-finished apps to use a dongle as a scanner.

If someone could put together an easy-to-use dedicated scanning app that doesn’t require the use of virtual audio cables and half a dozen different software packages/plugins, I think that could really boost sales and usage.

Make scanning on the PC as easy as using Uniden’s HomePatrol scanner should be possible! It just isn’t. Yet.


Qutoing Mike Frisco “If someone could put together an easy-to-use dedicated scanning app that doesn’t require the use of virtual audio cables and half a dozen different software packages/plugins, I think that could really boost sales and usage.”

I too would love to see this on, the Linux platform.

Quoting Mike Frisco “Make scanning on the PC as easy as using Uniden’s HomePatrol scanner should be possible! It just isn’t. Yet”

Definitely NOT interested in a dumbed down radio, like the the HP and the newer units from Uniden where you can not program them by hand/manually by entering a frequency(s) and mode and/or talkgroup info as needed.

The site that provides the data to the HP etc. is the bane of existence to many scanner enthusiats… and is making a PROFIT from YOU! Its also not the be all end all correct source of data for areas. It takes local knowledge on some things, which said knowledge at times is not easily translated into some turn it on and forget it, it works automagically device.


I agree on the scanner. That would be cool.


Where’s the software to make all these things listed? start talking about GNU radio and DIY developement and 95% of those who voted YES will disappear untill the software is released and well tested. Do you want to check? start a crowfunding for this product.


Your point is right, well said on ground, however feasibility is possible and it is now only at concept stage but not return true value of money.


I would rather be interested in a cheaper SDR with 1 Rx + 1 Tx, in the 30-40 USD price range


I vote NO: Keep it cheap, make it modular, so that anyone can connect 3, 5 or 10 RTLSDR’s together for their specific application.

For coherent applications with a budget of $100-$150, I would rather use a fully coherent receiver based on the AD9363 ($80) or the like.

For a new RTLSDR version, I would prefer to see the frequency range extended to 2.7GHz+ (2.4GHz ISM band + all cellular bands). This can be done for example by using the 5th LO harmonic of the R820T, or by using a integrated downconverter.


I agreed


Could you please elaborate on how to use the 5th harmonic of the R820T2 LO? Searching for a cheap WiFi way to receive the 2.4 band.


How about making the bias-Tee switchable on/off via software?


it’s alreays implemented.


Time to upgrade the V3 version hi hi! Thanks admin and Luigi!


From a commercial point of view i would give the priority to a band pass filter bank that covers all the rtl’s frequency range. And also individual BPF boards for specific bands of interest. This is something that sells to any radio owner, not just SDR, not just your SDR.
There are already some filter banks on the market but it think they are overpriced, i don’t think that smd L and C cost that much.


good idea, switchable and controlled via software. a filterbank for rtl-sdrs.


Every filter adds attenuation and every switch adds attenuation. This attenuation can best be compensated for traditionally by having a tiny amount of gain at the antenna.

But a coherent receiver with a phased array of antennas can have the antennas aligned (manually or with enough antennas by adjusting phase in software) to attenuate a strong signal (destructive interference) while at the same time amplifying a signal of interest (constructive interference). And the really neat part is that if a FFT is used ALL the IQ signals after alignment (lots of CPU power is required) that different phases adjustments can be applied to different parts of the RF spectrum. So that you can cancel out signals that are not of interest and enhance signals that are of interest.

But if you are looking at a more traditional use treating the receiver as 3+ separate unrelated RTL-SDR dongles, and ignoring the any advantages that can be gained by a coherent receiver, then far better signals would result from a tiny amount of gain at the antennas (and an attenuator for the band of the local source of strong signals) directly in front of the receiver.


a sdr-based direction finder ( every type, even doppler based) would be cool.


A doppler based DF requires nut just receiver, it requires a large (to the wavelength) antenna base and the necessary switching network to achieve detection of the direction. If you’re interested you can find more information in HAM radio literature that describe their Doppler DF that they use for fox hunting


I agree that filters are a must, otherwise you can’t use a RTL-SDR receiver in a congested RF-Envrionment.
However the cost of a filter bank covereing the complete RT-SDR-Rx range having high Q to provide sufficiently high attenuation outside their passband, will cost more then the planned bord itself.
Use of high Q, frequency selective antennas like tuned magnetic loops eliminate the need for good Filters. They work well in the SW- to UHF-range, but they require retuning after each modest frequency change. Their bandwidth is 1 to 2 Octaves so you will need a number of them.

Michael Robertson

This would be a real game changer if it could be achieved along with decent software (SDR# etc) integration.

Alex Mule

I would be extremely interested in this product and would certainly pay 99 USD. My school’s astronomical society has shown some interest in radio astronomy and forward scatter meteor bounce, and this would be an excellent (and competitively low cost) way to begin exploring these fields.

Alex Mule

The compatibility with RpiTX is also intriguing. Honestly that, along with the noise source and the other improvements would probably justify me buying another one for myself… if I had one request, it would probably be the inclusion/production of a practical downconverter. My soldering skills are just not good enough for the direcTV mod, and I think use of ubiquitous 2.4 GHz wifi signals for passive radar is intriguing.

Also, the software side development is much appreciated. Any chance you’d be collaborating with the people behind simple ra?


“I think use of ubiquitous 2.4 GHz wifi signals for passive radar is intriguing.”

Why would 2.4ghz be so intriguing? Please elaborate.

Alexander Mule

I didn’t mean to imply the frequency itself would be particularly intriguing. Indeed, lower frequency passive radars have led to some very interesting developments due to the antenna properties of the objects being observed (can’t find the article right now, but it seemed some were able to identify planes by their radar signatures)

Anyway, I find 2.4GHz intriguing for that purpose because of the ubiquity of TX sources literally everywhere. If I wanted to reveive echoes off of cars for doppler observation for example, there could be 10 sources (all on different channels, hopefully) available for use just along the street. With a lot more software side effort, multiple antennas, and coherent receivers, I might even be able to take advantage of the multitude of sources at once and perform more accurate distance/direction finding, maybe even in real time.


One application would be to test and troubleshoot drone (sUAS) transmit signals; am doing this currently using a HackRF One. It helps to make sure the drone’s transmitter is operating and on frequency, ensuring your are using the correct channel (many channels to choose from) assessing the bandwidth of the signal, etc. Very handy indeed.


Well, it may be a simple low-powered VHF-UHF transceiver, if being used with Tx-amplifier and good antennas (e.g. dual-band Yagi). I don’t see any application for other 2 receivers though, however it would be nice to mess with some passive radar stuff without any necessity to awaken an iron. $99 is not very high, but $85 should do definitely.

Waiting for your dipole antenna, guys! It is absolutely the thing.


Yes, I’d definitely pay $100 for such a system. To the scrooges wanting it for $70, I’d say lay off the beer for a few days and you’ll be set. My only disappointment with RTL’s in general is the 1.7GHz upper frequency limit. There’s such a lot going on between that and, say, 2.5 or 3GHz. that these devices miss out on.

Jim Charr


Jim Charr

” Similar applications to the HackRF Portapack idea. ”

PLEASE YES !!!!!!!!!!!!!!!!!!!!!!!!


I have a P.S. on the new radio , who is going to be the wizard or wizards what ever the case may be of the black arts who is going wright the software ??????????????????????


How about some degree of modularity similar to how XBee sockets work?

It’d be great to have a single baseboard that you can pop modules of various frequencies and send/receive capabilities into.


Well I am going get my two cents in on this . I like the Measuring filter characteristics, and get rough SWR antenna readings. That would be nice for experimenting with rx antennas to find whats good . Now some were I read that the HF part of this new radio is still direct sampling like the V3 dongle is . Well that means that you will still need something like a MFJ-1020c to control the receiver amplitude and frequency separation . Well that is it .

Mark - listener

How about a bus for stacking to make 3, 6, 9 etc receivers? I don’t know why but why not?
Also a switch-inable mixer front end down converter for higher reception, deep space mission stuff?
Excuse my naivety…

Mark - listener

oh and how about a Peltier cooler?


You would need a lot more power for a Peltier cooler than USB can provide, because they are really really inefficient. And you will still need to cool the hot side with an even bigger fan because you also need to get rid of all the additional heat generated by using so much extra power.


Unfortunately deep space is very very far away. The ITU (International Telecommunications Union) defines deep space as starting 2 million km from the Earth’s surface.

NASA have kit out much much further with a 32 GHz downlink frequency.

If you plug 32 GHz and 2 million km into the FSPL equation ( ) you end up with signals being attenuated by ~248.6 dB and more. So unless you have a gigantic parabolic dish in your back yard, NASA use 70 meter dishes (for TX and RX), I don’t think that you will be picking up the DSN directly. You would not need a 70 meter dish for RX only but you would need a decent sized parabolic dish, like a couple of meters.


I think you need to sit down and see what you can do for $50 – $60 dollars.

I have a lot of bits and pieces that I can kludge together with hardware and software to make something or experiment with a new idea.

But for $50.00 I would probably just buy it.

Ie I just bought the $39.00 (29 +10 postage) WSPR PiTX board .

I am very interested in Passive radar and would definitely buy something that could be kludged to work.


I actually have a project that uses 4 RTL-SDRs and a Raspberry Pi, so I’d welcome this.

Don’t forget an external MicroUSB connector specifically to power it!


As long as it improves SNR and has a nice noise floor.


I’m interested, the only part that turns me off a bit are the limitations for easier RPi support. I understand why, but I’d be in the same boat as David Taylor above, add one or more extra inputs. Basically upgrade the USB hub chip you plan to use from a 3 port (14.4MB/sec) to more, which would mean that if all 4+ devices were running at their full sample rates then the RPi hardware would drop samples right, left and centre (if the NIC was in use or a keyboard/mouse was attached, or the day had the letter “A” in it.).

And selling a board with 4+xRTL-SDR’s in it and explaining to people who wanted to use all 4+ at maximum sample rate in RPi hardware why they could not would get frustrating.

To sum up my opinion – 2 coherent devices is Ok, 3 is starting to be interesting, 4+ is really interesting.


Agree on the sample issue, I have problems dropping samples when using just 2 RTL’s running at 2MSPS each on a single RPi3 right now with not much else running on it…


Great idea ! I’m actually using to devices on a Rpi3 for ADSB and ACARS. Thanks !


Please add ability to synchronize LO clock to a 10 MHz source


Frequency calibration of other signal sources? frequency counting?

It’s always a good thing to be sure of the received signal frequency.

that’s the same idea of improvement than going from xtal to tcxo, one step farther.


Locking to 10 MHz is required for coherent sampling applications such as precision frequency and time measurement, where you want the sampler to be frequency stable. In a professional system, you would also have a 1 PPS input that would then allow time-tagging of data with absolute calibration using e.g. a GPS unit, but that might be too much for this application.

David Taylor

Four rather than three devices might be useful, e.g. Direction finding.


5 would be even better, but they are directly targeting the RPi, with it’s one and only USB port that is shared by everything and the kitchen sink, so even 3 might be a problem with most configurations.
comment image
They are aiming for the largest potential market and sticking three letters in there (rpi) will increase the potential market by at least 100 fold, which I don’t see as a bad thing.


What is freq range is planned for directional coupler? (may be better as optional unit to lower price?)

Will have HF directs sampling RX mode?

It would be very nice to have place to install few hi-q BPF’s like:

If no direct sampling planned, may be instead of 3 RTL-SDRs, place two RTL’s + HF converter, aimed from same LO 57.6 MHz (28.8 TCXO mult. by 2)

Price should be around 50-60$.


Will the upconverter be built in?


>Are you wanting the BPFs for TX with RPiTX or on RX?

First for RX! 8bit RTL’s ADC with wide input working hard on big antennas.
BPF’s will save dynamic range by filtering strong out of band signals.
But cleaning RPITX spurious emissions will be good too.

Also please consider attenuators resistor pads with simple DIP switches, 0 / -10 / -20 dB will be good choice.
Because on practice RTL-SDR V3 need that attenuator on HF full size antennas.
Without ATT it could saturate most time on evenings, when HF and MF broadcast signals is strong.

P.S. If you will produce next RTLSDR dongle like ver 4, I’m vote to include switchable attenuator too, for HF direct sampling mode.


Those parts are quite bulky at 38.1 x 12.7mm (1.5 x 0.5 inches) maybe they would be better suited to an external enclosure with two SMA connectors.

Coherent Receiver

It is nice that our ideas ( will lead to the development of the new RTL-SDR product. The cheap entry level receiver will probably lead to the growing market size of coherent systems.


Your ideas? The concept and hardware existed long before you started. Also, the last time I checked, your hardware lacks some of the required parts for synchronization.



You were the very first person to think of one clock and 2 or more ADC’s ? Was phased arrays, your idea as well ?


I don’t think they mean anything negative in their comment, their stuff has been covered here before. Looks like its a German group, so I’d expect their english to be choppy, and not PR-perfect.


You are right, probably lost in translation.

Coherent Receiver

Sorry, but I cannot understand your negative comments:
1. We have provided credits and links to the relevant patents, publications and public projects (incl. rtl_coherent) on our site from the project start.
2. We decided to make our project public and made a comprehensive description of all developed components. It was the first commercial project in this area or more pr-correct: we didn’t know about other rtl2832U unlimited-channel receivers that were commercial available at that time.
3. It is really nice that the new price-affordable multi-channel receiver will be available on the market. Hopefully, it will boost the development in this segment and will lead to the new interesting projects and ideas.
P.S. It was very interesting to read about the user requirements for the new receiver.

Alyx J McCown

I can see the use in it, everything but the TX could be of use to me one time or another … at least to a point where i should have one on the shelf
I hope for some proper shielding (more than just a case) with removable cans (makes it easy to cut holes for heat sinks)


Hey carl,
I think rtlsdr v3 is a product and now at paramount incomparison to other rtl2832U chip based SDR dongle. As far as your proposal does not attracted to me because this is not wise to do those thing which div community already doing . You should focus to develop some other higher end SDR if possible in mid price segments.

73, virender


Is not possible to built a 12 bit SDR? Just adding 4 bits change the game.


please stop the 8-bit direction. always use a 12bit sampling…


Not when using the Realtek RTL2832U in debug mode (which keeps the price low). But you can have 12-bits by setting the sample rate to 2.4 MSPS (8-bits with no drops) and decimating by 4^4 (256) to increase the bits to 12 at the cost of dropping the bandwidth to 9375 SPS.


What do you mean by debug mode?

Steve S

“The RTL2832 itself is a TV chip that turned out to have a lot more flexiblility than originally intended..

Its primarily designed for use with DVB-T the european/asian digital TV system. It would have just ended there, but a few years ago it was discovered it could be put into an undocumented debug test mode where yes, essentially that is what it does in that debug test mode, it forwards a bitstream of captured data in the form of numbers to other software of your choosing.”


The RTL2832U as far as is known has a 7-bit ADC that samples at 28.8 MSPS. But this data is only available for it’s primary function DVB-T COFDM demodulation which is implemented in silicon. There is special debug access, or a design feature, where 8-bit samples are available but at a much lower sample rate. This interface is the bottleneck in getting data out of the chip. It can run at 3.2MSPS if your application is tolerant to dropped samples, or 2.4 MSPS (or less) with no dropped sample.


Haan…, I know it but from improving SDR hardware perspective I don’t think decimation feature increase sensitivity of dongle. This project may be good for SDR application perspective but not for developing chip standard SDR hardware.


Yes and no is my answer. If you have enough noise and signal to tickle the least significant bit of the ADC (one of the very few cases where noise is good) then with decimation you can receive signals that were almost invisible. Oversampling and decimating can be good ( It is not exactly this, but this gives a basic concept – ). But if there is no signal (or noise+signal) inside the dynamic range of the ADC, then yes there is nothing that can be improved, but that is where additional gain may be used. The additional dynamic range from decimation is useful, because you can turn the gain up higher and it does not cause overloading (until the extended dynamic range provided by decimation is exceeded).


Thanks for dilated explanation but you don’t think that 8 bit ADC and noise floor should be improved because at HF exploring rtlsdr dongle far behind. Adjustments of gains for different frequency at different spectrum make rtlsdr more irritating thing.


The problem at HF is that you will have overloading local signals and extremely weak distant signals, one of which will always sit outside the dynamic range of 8-bits. And the lower you go in frequency the worse things get (24-bit audiophile sound cards offer the best price/performance at VLF frequencies).
HF and below in terms of price/performance more bits in the ADC is the cheapest option to get better performance.

You may be able to get improved performance with a complex array of many well spaced HF antennas and use the features of a coherent receiver to cancel out sources digitally by adjusting the phase of the signal in different parts of the frequency range through destructive interference, but it will be more expensive. And having a requirement for an array of well spaced HF antennas is just crazy for most people. But you could do things like pick up one of two transmissions on the exact same frequency if the phase was different for both sources. But if I was designing such a complex setup I would go with something much higher than 8-bits.


Brilliant. The hobby gets better every day.


Brilliant. The hobby gets beter every day.