New SDR radio does not seem to work

[WaveyDipole]

I recently purchased an SDR radio, one of these cheap RTL8232U/R820T2 based gizmos from the far east (see picture below). This one is in a nice aluminium case and I paid a little more for it than for a standard USB dongle. it arrived the other day but unfortunately it does not seem to be working as expected.

Initially I plugged in and the light came on so all seemed good to go.

I then tried several things without much success and have reached the conclusion that is must be a hardware fault. These are the things I have tried:

- initially tried windows software (SDR#, HDSDR, SDR-Radio) but although the device was recognized, none of these programs could communicate with it.

- switched to Linux for diagnostic purposes. Connected device to the computer, installed rlt_sdr and ran rtl_test -t. The kernel driver has been blacklisted as instructed. The device is recognized as 'Using device 0: Generic RTL2832U OE'. The device is detected as idVendor=0bda and idProduct=2838. The product id is rather confusing as it ought perhaps to be 2832. The radio IC is an R820T2, not an E4000. rtl_test seems to identify the R820T2 tuner correctly but then seems to be looking for an E4000 tuner?

- checked the device current usage which is around 100mA. This should be well within the capability of a typical USB port. Tried with no other devices connected.

- tested the dongle on another computer. Same result.

I have now also noticed that the external blue light no longer comes on, or sometimes just flashes briefly. I opened the aluminium case, primarily to confirm the chipset, and found that it does indeed have a T820T2 tuner. The case contains a main PCB and a daughterboard which appears to be an actual SDR board from the original dongle package. There is another LED on this board which does light up and stay on. The case gets only slightly warm during operation but not warm enough - I would not describe it as hot - to warrant concern.

I have contacted the seller to request a refund/return, although sending it back to China will cost nearly 9gpb according to Royal Mail so I'm not sure this would be worthwhile, although I am still waiting for the seller to get back to me and tell me what they are willing to do.

So as already stated, although it is possible I missed sommething in configuration, since this has failed in both Windows and Linux and two different computers, I think it reasonable to assume that this is a hardware fault. However, I also wondered whether it could be something as simple as a dodgy joint, particularly between main board and daughter board, and whether it might be worth re-flowing this using a hot air station? If I don't have any luck with the seller, might that be worth a try?

Below is the output from rtl_test (when it can actually 'see' the dongle):

$ rtl_test -t
Found 1 device(s):
  0:  Realtek, RTL2838UHIDIR, SN: 00000001

Using device 0: Generic RTL2832U OEM
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_read_reg failed with -1
rtlsdr_write_reg failed with -1
rtlsdr_read_reg failed with -1
rtlsdr_write_reg failed with -1
rtlsdr_read_reg failed with -1
rtlsdr_write_reg failed with -1
rtlsdr_read_reg failed with -1
rtlsdr_write_reg failed with -1
No supported tuner found
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
Enabled direct sampling mode, input 1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
Supported gain values (1): 0.0
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -9
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
rtlsdr_demod_write_reg failed with -1
rtlsdr_demod_read_reg failed with -1
WARNING: Failed to set sample rate.
No E4000 tuner found, aborting.
rtlsdr_write_reg failed with -1

[denverpilot]

You broke the rule! “Don’t turn it on, take it apart!” LOL.

Kidding of course. Sorry about the infant mortality... never any fun when new kit doesn’t work right out of the box, and you’re looking forward to having a play!

[RoGeorge]

Did you installed "ExtIO_RTL2832.dll'?

My installation notes for R820T2 based USB dongle (from some years ago):
1. zadig
2. hdsdr.de
3. ExtIO_RTL2832.dll

[WaveyDipole]

You broke the rule! “Don’t turn it on, take it apart!” LOL.

Well yes, I did do it the other way around so I guess I got what I deserved 
I did have to open it up eventually if only to confirm which chips were present.

Kidding of course. Sorry about the infant mortality... never any fun when new kit doesn’t work right out of the box, and you’re looking forward to having a play!

Yes, rather disappointing but even more so when factoring in the the cost of sending it back halfway around the world. Still waiting on the seller to respond regarding that however.

Did you installed "ExtIO_RTL2832.dll'?

My installation notes for R820T2 based USB dongle (from some years ago):
1. zadig
2. hdsdr.de
3. ExtIO_RTL2832.dll

Yes, that is exactly what I discovered also. HDSDR requires that one runs zadig and installs the driver. It also requires that ExtIO_RTL2832.dll is present in the HDSDR folder and loaded. I did follow both of these steps. When I load the DLL manually, the results are rather unpredictable although the program asks whether I want to load another ExIO_RTL2832.dll. Sometimes the program hangs. Sometimes it fails to load. Sometimes it appears to load successfully but the device is not shown as connected. Either way, it doesn't work.

[cdev]

It is a direct sampling unit, so, did you switch it to "direct sampling, q input"  in your software, for use on HF ? 

And use a decent antenna and ground?

Should be in the tuner settings area of your software,

or if you use gnuradio- in the connect string.
You need to use this string.

rtl=0, direct_samp=2 to receive HF (below 24 MHz) Or else your receiver will seem dead.

Look up "direct sampling rtl2832" on youtube to see how that works.

Your device uses has a unique hack that turns a TV dongle into perhaps one of the most widely tunable range receivers around with very few additional parts.

[WaveyDipole]

It seems I may have been blindly following the numerous examples I found in the Internet and using the wrong flag for rtl_test command. Apparently the '-t' flag is specifically to enable tests for the E4000 tuner, which this dongle does not have and the correct command is to run just 'rtl_test' without any flags. However, the poutput of this is still not very encouraging:

# rtl_test
Found 1 device(s):
  0:  Realtek, RTL2838UHIDIR, SN: 00000001

Using device 0: Generic RTL2832U OEM
usb_claim_interface error -5
Failed to open rtlsdr device #0.

I also ran lsmod to check whether the post is being held by the kernel driver, but the dvb_usb_rtl28xxu is not listed in the output so it would appear that it is successfully blacklisted and not being loaded.

I have also had a reply for the seller asking if I can solder the back of the board. Not quite sure what they mean, but they have also offered to send a replacement board if this does not work.

[cdev]

I own the same board you have.. I can help you get it working. Its dongle is no different than any other dongle except that it uses the normally unused pins 4+5 on the RTL chip for HF input.


If I were you, unless you already have, I would not open it's case because the wire connection is fragile and can easily be damaged. And re-soldering it is a PITA. That is what the seller is trying to explain to you. He or she is trying to ask you are you skillful enough to solder those two little wires yourself.

But it sounds to me as if you never were able to switch to the "Q" input. So its likely not broken. Google "direct sampling" RTLSDR.

To use it on HF, you need to select the Q input

Then, with a good antenna, it works surprisingly well, considering that its using the differential input to the RTL chip. The design is actually quite good, it maintains ESD protection and matches the RF impedance. It could probably be optimized using a network analyzer but my own experience with it is that it works fairly well.

I bought it out of curiosity and wasn't expecting it to work as well as it does. I bought it as a kit.

Note that you need to use the HF antenna input and tune it to an HF frequency.

Try to see if you can receive the 20 or 40 meter ham bands.. 14 MHz or 7 Mhz.

If your rtl_* apps are broken I would remove and then reinstall the rtlsdr library.  Thats not the hardware's fault.

Since its on windows, you're on your own there.

[WaveyDipole]

cdev, thanks for your comments.

The plot thickens with this unit. The seller sent me some instructions indicating the 5 solder points he wanted me to re-solder. There are the big 5 points under the antenna jack on the internal board. I refloweed these but it made no difference. The seller has offered to send a replacement board if the soldering didn't work. Since it didn't, I replied to that effect and am awaiting a response.

Later on I had another look at the board under the microscope to see whether there were any bad connections etc. I couldn't see anything obvious, only a tiny blob of solder next to a resistor which was easy to remove. I then cleaned the board with IPA removing any traces of flux. After that I probed with a DMM to check that power was reaching the daughterboard - it was and at this point to my surprise, the blue LED lit up as well.

I now tried the board on the computer - but still the sme problem. I substituted the USB cable once again and I was surprised when rtl_test now retuned a sane result:

$ rtl_test
Found 1 device(s):
  0:  Realtek, RTL2838UHIDIR, SN: 00000001

Using device 0: Generic RTL2832U OEM
Found Rafael Micro R820T tuner
Supported gain values (29): 0.0 0.9 1.4 2.7 3.7 7.7 8.7 12.5 14.4 15.7 16.6 19.7 20.7 22.9 25.4 28.0 29.7 32.8 33.8 36.4 37.2 38.6 40.2 42.1 43.4 43.9 44.5 48.0 49.6
[R82XX] PLL not locked!
Sampling at 2048000 S/s.

Info: This tool will continuously read from the device, and report if
samples get lost. If you observe no further output, everything is fine.

Reading samples in async mode...

Now I had tried that subsitute cable before without success. I also found and tested a third one - still working. Went back to the originally supplied one and the errors came back. Sure enough, the substiitute restored normal operation again. So it seems that I definitely have a duff cable, but why did the substitute one not work before?

After a while the blue LED stopped working again, but the radio continued to work and I was now running gqrx. However, the receiption on FM was absolutely awfull. The signal was weak and getting lost in the hiss and crackle. The signal strength was low regardless of whether I used the supplied antenna, an external one or a length of wire. I couldn't tune into HF as gqrx doesn't seem to allow me to tune below 24.897Mhz. I tried adding 'direct_samp=2' to the device string, but that causes the program to crash for some reason. I also tried HDSDR on Windows. It did work this time, but the reception was even worse than in gqrx on Linux. I couldn't copy any FM stations at all. It was all lost in the noise.

Maybe cleaning it up removed some debris that was shorting something although I am still baffled about the LED. If you have the same box, does the LED stay on all the time?

I still think the hardware is dodgy as the front end of the receiver seems suspect and the LED does not stay on.

UPDATE:
I had a look on the gqrx website and found these instructions to install the latest version.

http://gqrx.dk/download/install-ubuntu

Following these I got the latest version up and running and this now does accept 'direct_samp=2' in the device string. However, I cannot figure out where its tuning to. I don't get any AM stations on LW, MW although I do get various beeps on these bands where AM stations should be. BTW, I am selecting AM mode and using the normal setting for filter width and shape.

[cdev]

I think GQRX has an "ignore limits" checkbox or something like that you need to check..

Try CubicSDR. Or since it sounds like you are on Windows, HDSDR.

[cdev]

You'll have better luck if you use a USB extension and a couple of ferrite common mode chokes to quench the RFI that is an unavoidable rider on the outside of USB cables..


You are using a device that doesnt have a Elonics tuner so you should NOT be using rtl_test -t
 here..  here is the help for the program.

rtl_test -h
rtl_test, a benchmark tool for RTL2832 based DVB-T receivers

Usage:
   [-s samplerate (default: 2048000 Hz)]
   [-d device_index (default: 0)]
   [-t enable Elonics E4000 tuner benchmark]
   [-p[seconds] enable PPM error measurement (default: 10 seconds)]
   [-b output_block_size (default: 16 * 16384)]
   [-S force sync output (default: async)]


Try "rtl_test -p10" or
"rtl_test -p60" or similar...

Make sure your long wire is up and in the clear and that you have a decent ground. I would also suggest wiring up a 9:1 unun if you can..

[WaveyDipole]

I think GQRX has an "ignore limits" checkbox or something like that you need to check..

Try CubicSDR. Or since it sounds like you are on Windows, HDSDR.

I am actually playing with software on both platforms. I am currently booted into Windows and have just had a look at CubicSDR which is quite nice. Same issue with the HF bands though so it doesn't seem to be the software but the radio. I will have a look for that "ignore limits" option when I go back into Linux.

You'll have better luck if you use a USB extension and a couple of ferrite common mode chokes to quench the RFI that is an unavoidable rider on the outside of USB cables..

Something to look at when I get the SDR box working properly - or get a properly working one - whichever comes first!

You are using a device that doesnt have a Elonics tuner so you should NOT be using rtl_test -t
 here..  here is the help for the program.

rtl_test -h
rtl_test, a benchmark tool for RTL2832 based DVB-T receivers

Usage:
   [-s samplerate (default: 2048000 Hz)]
   [-d device_index (default: 0)]
   [-t enable Elonics E4000 tuner benchmark]
   [-p[seconds] enable PPM error measurement (default: 10 seconds)]
   [-b output_block_size (default: 16 * 16384)]
   [-S force sync output (default: async)]

Try "rtl_test -p10" or
"rtl_test -p60" or similar...

Yes, thanks for pointing it out. Someone on the osmacom.org rtl_sdr mailing list pointed it out as well so now I just use 'rtl_test' without any parameters which is sufficient.
 

Make sure your long wire is up and in the clear and that you have a decent ground. I would also suggest wiring up a 9:1 unun if you can..

My external antenna has a proper ground and I did add a self-made 9:1 unun shortly after I put it up. It seems that the dongle is quite fussy about having a proper ground. I also have a proper FM dipole antenna in the loft, but I don't have the correct connector to use it with the SDR dongle. I appreciate that the supplied antenna is not worth bothering with.

I was also thinking of building a discone at some point as I would like to pick up aircraft and possibly other stuff, but only once I have the SDR setup working properly. I might also set it up on a Raspberry PI and set it up to be accessible via TCP.

[cdev]

Try this- $ rtl_test -p100
Found 1 device(s):
  0:  Realtek, RTL2838UHIDIR, SN: 00000001

Using device 0: Generic RTL2832U OEM
Found Rafael Micro R820T tuner
Supported gain values (29): 0.0 0.9 1.4 2.7 3.7 7.7 8.7 12.5 14.4 15.7 16.6 19.7 20.7 22.9 25.4 28.0 29.7 32.8 33.8 36.4 37.2 38.6 40.2 42.1 43.4 43.9 44.5 48.0 49.6
[R82XX] PLL not locked!
Sampling at 2048000 S/s.
Reporting PPM error measurement every 100 seconds...
Press ^C after a few minutes.
Reading samples in async mode...
real sample rate: 2048006 current PPM: 3 cumulative PPM: 3
real sample rate: 2048007 current PPM: 4 cumulative PPM: 3
real sample rate: 2047993 current PPM: -3 cumulative PPM: 1
real sample rate: 2047998 current PPM: -1 cumulative PPM: 1
real sample rate: 2047993 current PPM: -3 cumulative PPM: 0
real sample rate: 2047987 current PPM: -6 cumulative PPM: -1
real sample rate: 2047998 current PPM: -1 cumulative PPM: -1
real sample rate: 2047993 current PPM: -3 cumulative PPM: -1
real sample rate: 2047991 current PPM: -4 cumulative PPM: -2
real sample rate: 2047992 current PPM: -3 cumulative PPM: -2
real sample rate: 2047995 current PPM: -2 cumulative PPM: -2
real sample rate: 2047999 current PPM: 0 cumulative PPM: -2
real sample rate: 2047982 current PPM: -9 cumulative PPM: -2
cb transfer status: 5, canceling...
cb transfer status: 5, canceling...
cb transfer status: 5, canceling...
cb transfer status: 5, canceling...


If your computer's clock is being set via a network time server it should be quite accurate and then the PPM figure you get is likely to be fairly close, after it runs for a few minutes the wandering will stop and it will just stay on the same number.

You'll get a ballpark idea of how far off its frequency settings might be.

[cdev]

Since you have it open could you photograph how they anchor the two little wires to the RTL chip and keep them there? Also be careful very very careful not to knock them off.

I think you'll be pleasantly surprised at the reception. For what you paid its hard to beat.

Although its not quite as sensitive as some of the upconverters, particularly ones that contain a low noise amplifier, the direct sampling is free-er of birdies than the typical upconverted HF output.

All RTLSDRs running on HF need a good ground to work well, and seem quite vulnerable to noise of all kinds. or perhaps a better term is less forgiving of it.

Considering all that they work pretty well.

[Beamin]

Did you installed "ExtIO_RTL2832.dll'?

My installation notes for R820T2 based USB dongle (from some years ago):
1. zadig
2. hdsdr.de
3. ExtIO_RTL2832.dll

I almost threw mine away but this fixed the problem.

I also had issued on windows 10 with the comport being not setup right. There is a program with an installer on the sdr# website.

Be careful not to get the comport wrong.

[WaveyDipole]

I own the same board you have.. I can help you get it working. Its dongle is no different than any other dongle except that it uses the normally unused pins 4+5 on the RTL chip for HF input.

If I were you, unless you already have, I would not open it's case because the wire connection is fragile and can easily be damaged. And re-soldering it is a PITA. That is what the seller is trying to explain to you. He or she is trying to ask you are you skillful enough to solder those two little wires yourself.

Since you have it open could you photograph how they anchor the two little wires to the RTL chip and keep them there? Also be careful very very careful not to knock them off.

I think you'll be pleasantly surprised at the reception. For what you paid its hard to beat.

I am curious about this myself, because I can't see any wires attached! Each pin is simply connected via a capacitor to GND and there are not plated through junctions to the underside. Apparently it is possible to use pin1 was well, but visually this appears to be n/c. From the other end, the HF antenna connector connects to a couple of inductors printed on the PCB and some additional circuitry, but I would need to trace out how the connections to the radio are made. The only obvious connections to the main PCB from the daughterboard seem to be the antenna socket and 3 out of the 4 USB pins. If it was necessary to add these wires, then I suppose they could be connected to the capacitor pads. I have attached a pic of the connections.

Did you installed "ExtIO_RTL2832.dll'?

My installation notes for R820T2 based USB dongle (from some years ago):
1. zadig
2. hdsdr.de
3. ExtIO_RTL2832.dll

I almost threw mine away but this fixed the problem.

I also had issued on windows 10 with the comport being not setup right. There is a program with an installer on the sdr# website.

Be careful not to get the comport wrong.

Beamin, it turns out that you have to run Zadig against each com port. I managed to get HDSDR to work but SDR# stubbornly refused to, until I tried the 32bit version that is. I initially downloaded the 64bit version as I have a 64bit Windows.

[cdev]

Thats the Rafael Micro tuner chip, the HF direct sampling is done by the RTL2832 chip.

We want the other chip, the one with the wires connected to it!

Also, how is it working?

Adjust the RF gain for the maximum spread between the peaks and valleys.  You should be able to get 50 db approximately, with the AGC turned off. Any less and you need to improve the common mode filter on the USB and the boards bypass caps.. There is a bunch of additional improvement which can be done.  A bigger cap across the USB power rails helps, but not so big that it is seen as a short when the power goes on. Then the dongle wont start.

I also added some SMD caps to further decouple the power going to the VHF/UHF antenna. the "bias-tee".

[WaveyDipole]

Thats the Rafael Micro tuner chip, the HF direct sampling is done by the RTL2832 chip.

We want the other chip, the one with the wires connected to it!

Also, how is it working?

Sorry, let's try that again with the correct chip 

I have uploaded a pic of the RTL2832 chip. It looks like pins 4 and 5 are n/c, but there are plated through holes and a pad provided to facilitate a connection which makes sense. Still, regrettably, this does not help you out with a picture of the wires. Looking at the wider pictures of the main board, the blue lines show where the points on the main board connect with two further points underneath the daughterboard which correspond to the two holes topside on the daughterboard. So it would seem that this is where connections are supposed to be made. I verified with a DMM that these points are definitely not connected. I also checked that the larger hole is connected with the smaller one nearer the IC as indicated by the white line. The other connection point has a clearly visible track to the IC. Perhaps its an assembly omission or maybe its up to the buyer to research and make those connections? Since the antenna circuit is disconnected this would explain the abscence of any AM band signals, but then what were the various high pitch noises I was getting? Might they have been PLL noise or birdies perhaps?

My board does not have the ferrite ring that your board has. On the other hand the antenna connector, IR sensor and power LED are absent from your SDR board. I'm not sure really why these were put in since they are inaccessible inside the case and serve no useful purpose.

Adjust the RF gain for the maximum spread between the peaks and valleys.  You should be able to get 50 db approximately, with the AGC turned off. Any less and you need to improve the common mode filter on the USB and the boards bypass caps.. There is a bunch of additional improvement which can be done.  A bigger cap across the USB power rails helps, but not so big that it is seen as a short when the power goes on. Then the dongle wont start.

I also added some SMD caps to further decouple the power going to the VHF/UHF antenna. the "bias-tee".

Thanks. These are all these are good points for me to think about. I'm waiting on confirmation but hopefully I will be getting a replacement board. If they don't want the old one back then I will have a play with it and connect those two points to see what happens. I also found that the power LED that does light up has a 2.2k resistor in series with it from the 5v power rail which looked rather high and sure enough there is only 0.8v across the LED. I reckon it was probably meant to be a 220 ohm.

[cdev]

Wow, pretty cool!

Looks like the various vendors selling them are all now selling enough of them to get their own custom dongle mods from the folks at Newskysz.

I'm surprised that it doesn't work. Are you sure?

[WaveyDipole]

Its getting rather late here now, but hopefully I will be able to do a bit more work on it tomorrow then I should know for sure.

BTW, just noticed your country of origin. Antartica?

[cdev]

Thats in honour of my Aunt.

[WaveyDipole]

Today I connected up those two links with a short pieces of wire and I am now getting AM broadcast signals. I could barely pick up anything on LW. I can usually pick up R4 and a couple of French stations quite strongly, but I just barely got a very weak R4 and no French stations. On MW frequencies the story was a bit better, but it seems I can only pick up strong local stations there as well. I would have to say that sensitivity is quite poor on MW and abysmal at LW frequencies. I'm not expecting much for the cost, but I guess I expected it to perform rather than just being able to pick up a few strong local stations. However, in normal mode FM sensitivity is a bit poor as well. I couldn't pick up much at all on HF frequencies. I should at least be able to pick up the Chinese station. I will wait until the evening when the bands come to life a bit more and then try again.

Incidentally, the problem with the LED turned out to be a faulty intermittent LED as well as the wrong  series resistor being used. Another LED from my bits box and a 500ohm resistor work perfectly OK. I also checked and the current drawn is still around 100mA.

One thing that I don't like about the board I have is that the RTL dongle board is soldered directly on top of the main board. The only insulation is the screen. Not only would this create potential for shorts, but must also add capacitance between RTL dongle board and the main board, the substantial part of this area essentially creating a ground plane. Could this adversely affect performance? Your board on the other hand has a neat cut-out for the TRL dongle board to fit into so this would not be an issue. I have been thinking about de-soldering the RTL dongle board and mounting it on wire 'stilts' to give it a bit of clearance from the main board.

The seller confirmed that he will be sending me a replacement board, so it will be interesting to see how that performs by comparison and whether those HF antenna connections are connected up.

In the meantime, do you have details of the windings required on that ferrite please? I have a couple of rings and can probably find some copper enamelled wire and would like to try adding that to see if it makes a difference.

[Whales]

> gqrx .... bad fm reception

My first experience was the same.  Things barely stood out from the noise and I was quite disappointed.  No prior SDR experience to compare it to.

It turns out I needed to pump up the gain in one of the UI tabs.  Not sure where the gain is implemented on-chip, but boosting it to somewhere in the 20-30dB range increased my overall perception of the performance dramatically.  FM stations were now clear and the world escaped the floor.

There is an AGC mode, but it seemed overly aggressive to me.  Probably suited for the dongle's original purpose (digital TV), not hunting through the ether for quiet curiosities.

[WaveyDipole]

With AM using direct sampling mode, having hardware AGC enabled seems to make a substantial improvement. With it disabled and and slider set to max (49db) weak stations disappear into the noise and stronger ones get very crackly. Even with AGC enabled, the stronger peaks are only at around -60db and there is still substantial noise on these stations. These are stations I get loud and clear on any AM radio. The weaker ones -80 or below. The audio gain, as expected, just makes everything louder including the noise. On the HF connector, connecting the antenna earth actually makes things a lot worse so I leave it disconnected.

With FM, the graphs sits at the -60 make and the peaks reach into maybe -30db, but the signals generally disappear into the noise. With hardware AGC unchecked and the slider set to 49DB, the graph sinks to -90DB and the signals disappear altogether. There are a number of signals I can pick up loud and clear on an ordinary FM radio that are just not present or very noisy on the RTLSDR. On FM via the UV connector, connecting the antenna earth does improve reception, but there is still a lot of noise.

Switching the presents between Fast/Medium/Slow/Off makes no difference.

[cdev]

The antenna you use is all important..

But reading through the stuff online it appears that the design of this device likely could be improved substantially - that they could have done better.

I have the device they were selling before, but I assembled it myself. They didn't assemble it.  And it was a major PITA, so much so that I have avoided doing anything with it since then.

Generally, it holds its own against any of the similarly priced (consumer level) portable shortwave radios - including the quite decent better Tecsun models, with a decent antenna. Which they also need, to be honest.

But from what I am reading it likely could be improve quite a bit.
Here is their doc file..

Maybe this link will work:
http://115.28.16.44:81/file/3156.doc

That is their instructions for it.

There also seem to be quite a few videos online of the earlier device, its construction and operation. Look at how it works in the Youtube videos of it.  It is what it is.
It seems that the bypassing and their choice of parts could be improved a lot.


Start out learning a bit more about rtlsdrs, notably that they are vulnerable to noise and USB connections are inherently noisy already with VHF. On HF they are even more noisy.

I put lots of ferrite cores on my USBs when I use SDR.

Also to get decent reception I needed a long wire, a good ground, and a 9:1 unun.

To be honest with you, despite some moments of decent reception everything RTLSDR have always made for marginal shortwave receptionfor me where I live because its so noisy.

Ive had much better luck with a different SDR probably because it has much more dynamic range and better rejection of noise and out of band signals.

Lacking an antenna tuner, a unun makes all the difference.

That applies pretty much to all my receivers, all of which are SDRs. I live in a very noisy suburban location.

All of that said, likely anybody with any decent amount of RF experience could likely find areas for improvement on the Ba5SBA design.

Russian web pages go into length about this but not having actually trying any of these changes.

I would try using a different RF transformer. I would use a binocular core, or perhaps two or even four side by side instead of a toroid. Thats a better broadband transformer. And I would try more turns on the balanced input side

You should make a post asking for other people's opinions on what the best transformer to use would be. 

Maybe you might also want to experiment with different materials, or maybe even commercial RF transformers..

I bet you could improve the performance significantly. Iam reluctant to on mine because there si a limit to how many times the little wires can be resoldered I am sure. It will become impossible to fix.

[WaveyDipole]

I have now traced out the HF antenna circuit. There is a semicondictor device in the circuit and I'm not sure whether this is a transistor or a pair of diodes so I have attached a images showing both possibilities. The circuit does have some similarity to your hand drawn one in that it has two inductors on the input and a number of capacitors to form a network. However there is no 1:2 impedance matching transformer. Instead we have a 3 pin semiconductor marked J3.  There seems to be no ESR protection on the input either. In addition to the two connection points for pins 4 & 5 of the RTL2832, there appears to be a possibility for a 3.3v power feed taken from the regulator on the SDR dongle board. The points marked 'X' were all left disconnected in this example.

I note that your hand schematic has some additional caps on the USB feed and you also suggested adding clip on ferrites. There is a 22uF cap drawn next to the torroid inductor. Is it in parrallel with it? The primary of the torroid in your schematic seems to have a power feed - was this to drive an active antenna? There is also a capacitor at the top of your diagram, one end of which seems to connect to earth. Was the other connected to the 5v line for filtering?

I would apprciate it is someone could give me some idea what the SMD part marked 'J3' might be? The references I found show this to be a two pin device, either a varicap or other type of diode. The device here has 3 connected pins.

[cdev]

The device I have is a couple of years old and indeed, it does look a lot different. You should poke around on the Chinese ham web sites. I bet you could find some discussions there as whatever changes were made occurred.

I gather that you have not owned an RTLSDR before? Common mode chokes on the USB are necessary.

Yes, the one I have does let you send power to the antenna, as I explained its a functional but quite rudimentary "bias tee".

It can supply DC to the (VHF/UHF) antenna to power a LNA. Its enabled by changing a jumper. All in all, I thought it was a pretty good deal, considering I paid - maybe $28 for it.
I don't remember.

But you need a decent antenna and ground. And the device I have looks completely different from yours.

[WaveyDipole]

Yes, it was difficult to make out under the magnifier lamp, but under the microscope it was quite clear - just 2 characters: J3. I am trying to understand the difference between the two circuits, and the purpose of the semiconductor part. If its a transistor, that might explain the need for a 3.3v line input and there is a DC path back to GND via R3, however 10meg would seem rather high to provide sufficient base current to turn the transistor on? If it is a package with two TVS diodes, that might also make sense. I believe TVS diodes are sometimes used in hard drive PCBs and those I have seen are usually two pin devices. Do TVS diodes read like a regular diode on the DMM? This package reads like a regular transistor, i.e. like a diode from base to C/E and infinity between C and E. Red probe on B to get a reading of the usual approx 600ohms.

You are quite correct - I have not owned an RTL-SDR before and my first experience does not turn out to be a very good one! In your earlier post you mentioned that performance of the RTL-SDR holds its own against a Tecsun radio? The performance of this particular unit is a long way short of that unfortunately.

I was reading about bias tees following your mention the other evening and was just trying to confirm that the configuration here is, in fact, an example of a bias tee.

[chris_leyson]

Common mode noise is always going to be a problem with SDRs especially when using E-field antennas, you need a quite ground connection at the antenna input. The ground plane on the PCB is your reference point and any noise on the ground plane just gets added to the received signal. A shielded magnetic antenna might give better results.
I have a QS1R SDR and to get the best from it I have to use a linear bench supply to power it, no switching wall wart rubbish and a small notebook PC running off batteries or a linear bench supply. I could probably get much better performance with a tuned and screened mag loop but haven't got around to building it yet.

[cdev]

You may want to consider completely redoing the input stage with something different.

Since you have access to the two differential inputs without the risk of knocking it off its your oyster so to speak. There is a lot of opportunity to learn about RF visually with that setup.

Its really almost an ideal learning setup.

The input on the RTLSDR at pins 4 and 5 is around 3000 ohm differential input.

Its suceptible to ESD, so if you build your own whatever, keep that risk in mind.

Try to filter out everything above 24 MHz (since your dongle by itself can receive above 24 Mhz on its own.)

I would concentrate on the lower part of that range since we're near the minimum of the sunspot cycle.  An LC network can match the antenna impedance to the input impedance.

Especially do whatever it takes to electrically decouple your receiving setup somewhat from the noise thats often found riding on power lines.

In a pinch, even just a bunch of wire (i.e. an air core inductor) will work. 

But its best if you have an appropriate magnetic material you can use for a core.

Try to keep the length of wire between your unun and your ground connection as short as possible.

Another thing that may potentially help a lot is to decouple the coax from your receiver by wrapping it as many times as you can around a ferrite rod or toroid or even just make a tightly would coil using the coax.. creating an RF choke. before it gets to the receiver.

I have done this with RG-174 and Ive been able to get a lot of turns in there because its so thin. (The loss in RG-174 is still pretty modest for HF, so this helps a lot.)

If its convenient, try to put your ground rod somewhere where the soil stays wet year round. (like under a downspout)

An interesting antenna thats a fairly good choice for people without the room to erect a really long wire is the so called "mini-whip" e-field probe antenna.

Its an amplified antenna and it usually gets its power via its feedline.

Its feedline is used in part of the antenna but it needs to have a big RF choke to prevent the nise indoors from making its way out to it.

The active unit is tiny, its basically a capacitive probe that sticks up into the air as high as possible. The antenna itself has a very high impedance preamp.

Its performance depends on it being put in a quiet location.. (outside of a house, and as far away from e-stuff as possible.)

Then ground the coaxial cable outside before it enters the house, then put a common mode choke on the coax before it gets to the receiver.

https://www.google.com/search?hl=en&as_q=pa0rdt+mini+whip

[Beamin]

I now cost is an issue but the difference in a dongle and an SDRplay is night and day difference below 50MHz. For example at night every single channel on the MW band has two to three stations every 10 KHz apart from 540 to 1600 on it from all the skip. Every multiple of allocated AM broad cast has a huge sharp peak on it. Some times its fun just to start one end of the band and count how many stations total you can pick up to two to three hundred different AM stations with a big long wire antenna. Lets see your car radio do that.

This thread makes me want to take my old dongle apart but I have a feeling it going into to little rasp pi box and do something useful like a self contained SDR with a 3" screen if I can find all off the shelf parts. If you only care about one frequency you could mount the pi/ dongle and battery/ solar cell in some high remote location with a small amp and have it broad cast the audio back. The pi gets choked with CPU usage when you add bandwidth to it.

[WaveyDipole]

I decided to mod the antenna circuit to make it similar to the BA5SRA circuit by adding a torroid 1:4 transformer which I wound as per the recommendations here:

http://mikikg.files.wordpress.com/2012/08/rtl2832u-dc-mods.pdf

Furrther information regarding transformers can be found here under the heading 'Balun/Matching Transformer Improvement':

https://www.rtl-sdr.com/rtl-sdr-direct-sampling-mode/

It states that the transformer should be wound on a FT37-43 core using 5 turns wound in trifilar fashion. Since I only had a couple of FT37-61 ferrites left over from a previous project, I had to adjust the number of windings. It so happens that this other project was originally designed for FT37-43 cores but came supplied with FT37-61 cores and required the same adjustement and the number of turns required for both types of core were provided. Calculating for 5 turns proportionally worked out to 13-14 turns.

I first removed the two links I had put in from the main board to RTL2832 pins 4 & 5. I then also removed the series capacitor (C4 on previous diagram) and also the 47ohm resistor (R1). I then used a couple of spare pads to connect the primary of the transformer. The secondary was connected to the two pads on the SDR dongle board. Finally I added two protection diodes (1N4148) to the inside of the antenna connector.

The difference was huge! The MW band is now alive with stations, Radio 4 came in clear on 198kHz LW and I can pick up various foreign stations on SW. I could almost call this 'holding its own' against the Tecsun. The dongle still has more noise than the radio and further experimentation with filtering may improve things, but reception is now quite acceptable and connecting the antenna earth no longer has a detrimental effect. I also found in gqrx, that using the narrow filter width on AM helped considerably with filtering out background noise.

The second reference I linked above mentions a T16-6T-KK81 transformer. So far I haven't found one available online to the UK, but I if I could find one then I wouldn't mind giving it a try to see whether it improves things further.

In the meantime I still await the replacement board and will report back when it arrives, however it would seem that this newer blue box appears to have an inferior HF circuit layout to the BA5SRA design.

[cdev]

This is great!


Glad to hear this!

I don't know what kind of bypass caps he uses in his board but I had good luck adding a few additional caps to my board. I added a number of additional large value SMT cps across the power rails in different points and a tantalum cap across the 3.3 volt output of the voltage regulator.

There are a bunch of good ideas here if you scroll down to "Reception of VLF frequencies using RTL820T2 Dongle":
http://www.g8jnj.net/softwaredefinedradio.htm

(also his "active Antennas" page collects a lot of good links to limited space antennas which would work well with a VLF enabled dongle.. Also look at what Martin says about the transformer, right at the bottom.

The optimization of the dongle for VLF will bring improvements to HF reception too. Its kind of a worst case scenario for noise and interference.

Once you have the transformer working at its best, then look at reducing the DC to DC converter noise..

Its even possible to eliminate it and replace it with several diodes. See the link to "Toshi"s web site in japan for a schematic.  http://ggtoshi.at.webry.info/201406/article_6.html

Its amazing how much can be done and learned, with a cheap dongle.. Really, they must break some kind of record for most bang per buck educational tool.

People from all around the world have been adding to the fun.

http://userdisk.webry.biglobe.ne.jp/019/963/01/N000/000/000/140317102568976083225_RtL2832U.PNG

http://userdisk.webry.biglobe.ne.jp/019/963/01/N000/000/000/140317075718660037225_R820T_20140619183917.PNG

http://blog.livedoor.jp/bh5ea20tb

Small improvements in bypassing bring big improvement in signals on HF.

Do you have an oscilloscope? If you do that will make it easier figuring out if some change represents a a likely improvement and why.

Its impossible to do this without a good connection to the board.
i would solder on some grabber-friendly loops of wire to use as test points to the points where the various voltages are used on the board to make it easy to attach your probes.

I think I am going to put a dip socket in there so i can swap out transformers easily.

I have always thought I could improve the low frequency response with a better transformer but its scary because the connection is so iffy..

The dongle kit came with a lot of parts but like everything it was made to sell at a certain price point and make a profit. So a little bit of investment in a better transformer will probably pay off.

See the images below, which are all kind of old but they use the Newsky dongle- likely some changes have been made since then and also for this specific dongle but the general schematic is still likely very similar.


There is also a bunch of helpful info here - but you will need to use Google Translate: http://ggtoshi.at.webry.info/201501/article_1.html

Also, here...

http://blog.livedoor.jp/bh5ea20tb/archives/cat_120700.html

also look at Chris Trask's varactor tuned active antenna design and especially its double transformer..

http://home.earthlink.net/~christrask/

[WaveyDipole]

I managed to obtain a mini 1:16 transformer which I tried today. This thing is tiny and the piece of ferrite inside must be miniscule - significantly smaller than my ferrite ring. It performed reasonably well and gain seemed to be increaed a little but the noise level was also considerably increased. I tried re-placing the series capacitor, reconnecting the 3.3v supply and placing the transformer across the original link points to the RTL2832 IC but curiously this did not work at all - I got no signal whatsoever. I then re-instated my orignal mod and hand wound transformer as on the whole that gave me the best performance. I have also ordered some clip on ferrite beads to put on the USB cable.

[cdev]

If you have not had the clip on ferrite beads up until now you'll likely see a big improvement once they arrive and you put them on there!

Also, more filtering/bypass capacitors help. At some point however it refuses to start up (when the cap isnt charged up so you need to discharge it to see this- because it will draw current at the beginning) eventually you will have added too much and it then wont work, the USB circuitry doesnt like it, then back off a bit. (this is from memory and its been a while but thatsbasically what I remember)

It seemed when I was looking at your boards design that it perhaps had fewer add on caps than mine.. anyway, more bypassing if its short path to ground and there are not loops will bring improvement. Tantalum capacitors work the best in my experience but thre may be better. SMD caps are better than through hole because of lead length. As the 'manual' says a tantalum is a good addition.. if you can an additional .22 uf (or .33 uf) tantalum cap and put it across the 3.3 volt rail thats good.. Make sure its the right polarity!

I am in the process of installing a little dip socket in my modded rtlsdr so that I can plug in different transformers without tearing out the wires.

I then re-instated my original mod and hand wound transformer as on the whole that gave me the best performance. I have also ordered some clip on ferrite beads to put on the USB cable.

This web page contains some good ways to build HF transformers:

http://www.qsl.net/in3otd/ham_radio/160m_transformers/160m_trafos.html