Another. What am I doing wrong DSD post

Status
Not open for further replies.

dubs0980

Member
Premium Subscriber
Joined
Apr 4, 2012
Messages
116
Hey guys. I got a RTL SDR# VBCable DSD+ setup with a confirmed DMR signal and I am only getting very partial decode. It cuts in a out and very garbled when it does decode any voice. Any thoughts on what to check?

I think I modified the SDRsharp.exe.config to the 4800 sample rate like the SDR-RTL tutorual showed. (Just opened in notepad changed the value and saved.)

I didnt touch any VB cable settings, and havent changed anything in DSD+. I'm brand new to all of this and SDR# but the data signals sounds fairly strong to my ears but maybe something isnt tuned right for the decoder.

thanks for any help guys.






 

br0adband

Member
Joined
Apr 8, 2005
Messages
1,569
Location
Springfield MO
(long post with a lot of info, so be warned...)

Try these settings for SDR# and see what happens - they don't have to be permanent, just try them and go from there (and of course use the two tips above about unfiltered audio and the Correct IQ checkboxes):

Under Configure (the button to configure sample rate/ppm/etc):
- 1024 MSPS
- set your ppm to as accurate as possible after a 5-10 minute period of warmup
- uncheck all 3 boxes, set the RF Gain to 28 dB

Main window:
- NFM
- Filter type is typically irrelevant since we're not filtering the audio signal being fed to DSD+
- Filter order set to 10
- Squelch as required (I set mine at 60 almost always and that's that)
- Correct IQ checked
- AF Gain set to roughly the next to last notch on the right side of the slider

As far as DSD+, if you intend to decode DMR/MOTOTRBO specifically then you should always use:

-fr

at the bare minimum so it ensures DSD+ is only going to focus on detecting and decoding DMR/MOTOTRBO traffic and ignoring anything else. Obviously you're using Auto (defaults) for detection as shown by the info line on the command prompt windows you pictured.

While I'm not going to tell you what you're looking at is all wrong or whatever, I will say this: if you alter the settings for the spectrum display and the waterfall so you're getting actually useful information instead of a "wall of noise" as you currently appear to be doing, I promise you it'll make things easier. Try these tips if you wish, or not, it's up to you:

- you want to adjust the FFT spectrum so that it's only going to display the signal(s) you actually care to look for or listen to, that means having all the noise under the noise floor is pretty much useless to just have there being displayed. I can't tell what your apparent noise floor is from the pictures you've shared because you've zoomed in so much - zoom all the way out and then look at where the spectrum shows your average noise floor is. On the image I've attached below the average noise floor is about -65dB which is typical for most RTL sticks, it's just a side-effect of them being "cheap USB TV tuners" and not having such great noise suppression. The point being that - just for the image I took - I adjusted the range of my FFT display so it "cuts off" at -70dB since anything below that is for most all intents and purposes going to be drowned out by that noise floor.

You've got your FFT display range set for -130dB which means you're going to be forcing the FFT display to show you "nothin' but noise" below your average noise floor. The images show a roughly -30dB "floor" across the visible spectrum which means one of two things to me: 1) you've got the RTL AGC on + you've got the RF Gain set to at least 30 dB or higher, or 2) you've got the Tuner AGC on which will basically max it out across the spectrum and it's a very bad thing.

So try this: set the Zoom to the lowest level (maximum Zoom out, or slide it to the bottom), and look at the spectrum. If you're using a typical RTL stick, it'll have an average noise floor around -50 to -65dB or so, somewhere in there is what these sticks provide and there's not much that can be done to lower it (they're just noisy, period). Once you can see what your current hardware's average noise floor is, adjust your FFT display Range slider so it shows the level just below the average - as you can see in the example image below I've got mine set for -70dB at the moment, that's the bottom of the range. I normally adjust it to -60, sometimes -50 actually and by doing so what I am getting are the signals that matter. If I want to "see" more activity I don't have to adjust the Range again, I just increase the Contrast of the waterfall and the signals will "pop out" visually if they're there.

Uncheck all three of the boxes for Offset Tuning, RTL AGC, and Tuner AGC (meaning those options are off - I don't even know what Offset Tuning is used for but I figure it's there for some reason, uncheck the box anyway for the time being). Set the slider to 30dB for testing purposes and that's your only signal gain for now and see what you get. DMR/MOTOTRBO signals are typically easy to spot on the waterfall but... that's the next tip. If you're overloading the front end or saturating the receiver by maxing out the gain with those gain settings (checkboxes checked) then that's another reason the DMR/MOTOTRBO will be affected and not decode properly.

- your waterfall is set to basically set to "something" that I can't see because the command prompt window is blocking it. As you can see from my example below, I keep mine at about the halfway point - I am NOT saying my way is the best way, so please understand this, I'm simply saying that if you're using this method of communications monitoring, it really helps to be able to "see" the signals apart from all the noise. You can adjust the sliders on the right side to kick out the Zoom and see what's going on as well as reducing the Contrast so you can discern the signals. In my example image the DMR/MOTOTRBO signal is right there, absolutely impossible to miss. If I crank the Contrast up I can still see it of course but it gets drowned out by a "wall of visual noise" for the most part.

You can also spot little "blips" of transmissions all across the waterfall - those are data bursts from water heaters and the like (the longer bluish ones), the thin reddish orange one just to the left of the DMR/MOTOTRBO signal I'm monitoring/decoding is an NXDN control channel, and the very tiny bluish and orange ones are squelch breaks from LTR trunking systems here in my area.

The visual spectrum and the waterfall are really what make this so much different from "the good old days" and a big reason for the appeal and quick adoption of SDR using these "cheap USB TV tuners" - it seems logical that if you adjust things so you can actually see what's going on in the spectrum, as well as being able to discern signal/transmission types visually by how they look on the waterfall is a big big plus to things.

Seems like most people don't grasp they can adjust things so they get actually useful visual info they can interpret nearly instantly instead of having to try and "decode" what they're seeing. Makes sense to me, hope that makes sense to you, dubs0980, or anyone else. ;)



EDIT:
One last tip: for the VB-Cable Audio settings, you do NOT want to pass the VB-Cable Audio to your speakers, that means you'd be hearing the actual digital data stream on top of hearing what DSD+ is providing you as decoded audio. Uncheck that box in the VB-Audio Cable options for "Listen to this device" - it's unnecessary and just means you get that annoying data stream mixed in with the decoded audio (when it works, of course).

One last last tip: Check the frequency or frequencies to verify what you're getting is a DMR/MOTOTRBO signal. If you had a strong signal, even with that apparently over-saturation in the image below (a wall of -30dB signal strength across the spectrum even as zoomed) it would show a nice peak above the rest, but I don't see a peak, at least not like the one(s) in my example image - there are two DMR/ MOTOTRBO transmissions in mine, one to the far left and one near the middle that I'm tuned to, both a very very easily spotted visually.

Based on those images above, and the spectrum (even with the -30dB signal blasting across the spectrum) and coupled with the over-contrasted waterfall, I don't actually see a signal there at the frequency you're tuned to - there's nothing there but noise. The closest signal is a remnant over at 463.725 that's already passed down the waterfall.
 
Last edited:

BM82557

Member
Joined
Aug 28, 2006
Messages
2,700
Location
Berkeley Co WV
Another thing I meant to ask yesterday was what kind of antenna are you using? The one included with the dongle isn't the greatest and is basically useless.
 

moonbounce

Member
Joined
Nov 5, 2002
Messages
1,257
Get a program called VB Cable all you do is download and run none of the hassles of a VAC Cable.
 

dubs0980

Member
Premium Subscriber
Joined
Apr 4, 2012
Messages
116
Awesome post! thanks for taking the time, it's much apprecaited (and on my earlier posts as well.) I'm not sure why I had it zoomed in so far. Lowering the FFT range is a great idea and didn't see that adjustment yesterday. I set things up with your setting and the decode is much closer. I can hear the occasional clear word but it's spotty. I'm playing in the office today so it might be the signal but last night I was getting good signals from home and it was still having the same type of decode. I set DSD to -fr, and the tuning seems correct after warming up. It's pretty high in the ppm but I'm not sure whats normal. I did have the tuner AGC checked too (now unchecked.) Antenna is a wide band (25-1900) Watson w-881 bnc portable anteanna I use for my 396xt.

I think it may have something to do with my VB cable setup, or my levels I hear the motorboating and the voice that actually comes through when DSD+ decodes. I'll be frank that the VBcable has confused me with all the inputs and listen to boxes. I don't know if its possible to feed two signals into DSD and confuse it or have a hot signal that clips it and hurts decode.

Thanks again guys for the help and any ideas are more than appreciated. Let me know if I can get some more screen shots or options that might help.



 

W4KRR

Member
Premium Subscriber
Joined
Apr 1, 2001
Messages
2,930
Location
Coconut Creek, FL
Rather than start a new thread, I'll ask this here as the topic fits....

I use DSD+ on a Windows 7 machine with a scanner using a discriminator tap as the signal source. I get very good voice decodes in NXDN, DMR, and P-25.

Now, I have a Windows 7 laptop with which I use a USB dongle and SDR# for the signal source, and the decodes are terrible at best, if at all, with lots of error indications (I've only attempted DMR on the laptop so far). Is this an expected difference between using a discriminator tapped scanner and using a USB dongle and SDR#? Or is it a matter of fine tuning SDR# based on the instructions detailed in post #6 above?

(The desktop machine's processor is Intel i3, the laptop is Intel i5, if this matters)
 

dubs0980

Member
Premium Subscriber
Joined
Apr 4, 2012
Messages
116
Well not sure what happened but I came back from lunch, plugged things in and boom its decoding pretty well for being inside my building (which is an RF black hole.) I found more volume options when I clicked on the volume button on the bottom right toolbar. playing with those might have helped something. That allowed me to turn down the cable output and the separate DSD up so I don't hear the vbcable output motor-boating. I have no idea how that works. It's kind of weird but I hear quiet voices along with a more pronounced conversations. I'm sure there is still room for improvement but this is so cool. Already found another DMR system for city transportation, right next to it.

 

BM82557

Member
Joined
Aug 28, 2006
Messages
2,700
Location
Berkeley Co WV
Rather than start a new thread, I'll ask this here as the topic fits....

I use DSD+ on a Windows 7 machine with a scanner using a discriminator tap as the signal source. I get very good voice decodes in NXDN, DMR, and P-25.

Now, I have a Windows 7 laptop with which I use a USB dongle and SDR# for the signal source, and the decodes are terrible at best, if at all, with lots of error indications (I've only attempted DMR on the laptop so far). Is this an expected difference between using a discriminator tapped scanner and using a USB dongle and SDR#? Or is it a matter of fine tuning SDR# based on the instructions detailed in post #6 above?

(The desktop machine's processor is Intel i3, the laptop is Intel i5, if this matters)

Using SDR# > VB Cable > DSD+ on my laptop (Win 7 64-bit) I get very good decodes on P25 systems.
 

br0adband

Member
Joined
Apr 8, 2005
Messages
1,569
Location
Springfield MO
Here's a tip: if you're using VB-Cable Audio (as I am, and I just provided the link) there is really nothing to configure at all since it's already set correctly when you install it - and the product I'm talking about is the one I just provided a link to. The problem for many is the confusing names/terminology which even the actual website does as well - in one instance they call it "VB-Cable Audio" and then on the same exact page they refer to it as "VB-Audio Cable" but the basic gist of it is that it's free and it works by default with SDR# as noted.

Install the driver (and as noted in another thread about this subject there are three available ones identified as VB-Audio Virtual Cable (the one linked above), VB-Audio Cable A (provided you make a donation), and VB-Audio Cable B (again, provided you make a donation). I use all three myself but that's not relevant: all that's required is the single one VB-Audio Virtual Cable, from the link above, and itself being donationware it doesn't hurt to drop a buck or whatever to the developer since it's an excellent product.

Ok, having said that, again there's no configuration necessary. I see that control panel applet above and I swear I can't even find that myself with my installation of the same software; I even verified the version number of the driver several times and I still can't find that control panel applet (which I don't need anyway).

It seems dubs0980 did get it working better, and as visually noted the signal there is somewhat strong (I still don't know what the gain settings are but at least you can easily spot the signal peak - lowering the Contrast allows you to spot such signals like DMR/MOTOTRBO almost instantly in a single glance. When the Contrast is set higher it gets a bit muddled but still readily identifiable. After a while of doing this you'll be able to tell what transmissions are what just at a glance by what they look like on the waterfall. Also, the reason I use 8000 for the Filter bandwidth is because hardly any signals require more than that. If you're not actively using the audio filtering, that won't have much effect at all on the end result. Besides, the "digital" signals that are broadcast by DMR/MOTOTRBO/P25/NXDN/etc are narrow anyway, just analog bitstreams one could say.

The ppm setting varies wildly on these RTL sticks - my two sticks are 56 and 50 ppm respectively, so 62 ppm ain't too far off. As long as you get a known good strong signal and you nail the signal down the dead center or as close to it as you can, it'll help a lot. Use the NOAA weather signals, or 470.309 MHZ which should still be available in most parts (holdout from the analog TV signals) - it's fairly consistent in most areas and provides a good signal to tune the ppm with. DMR/MOTOTRBO aren't seriously sensitive to the ppm setting, it can still be off by 1-2 ppm but DSD+ will still decode - NXDN on the other hand is incredibly sensitive to the ppm setting and if it's off by 1 ppm the decodes will simply not happen, at least in my experience, so yes it's important when working with these "digital" signals to get that ppm adjusted as accurately as possible after a warm-up.

As for the "quiet voices" what you're hearing are side-effects of the digital encoding and decoding process, basically inherent echoes in the processing; it could also be actual people in the background with respect to whoever is transmitting (like a security person in a crowd, etc). It's just something you end up ignoring sooner or later - an actual DMR/MOTOTRBO device may or may not sound exactly the same if you were using one, probably sounds better considering it's a native device compared to using DSD or DSD+ which aren't quite as "perfect" with respect to the AMBE/IMBE encoders/decoders (the actual algorithms used to do the audio encoding/decoding).

W4KRR:

It's a little bit of everything I suppose, tweaking the settings to get the best results. It's all about experimentation for the most part. A discriminator tap provides the nearly perfect signal to be decoded, it's raw and purely unfiltered in any respect right off the tuner whereas using the signal pulled from the RTL stick using SDR# is processed in some respects by comparison and it's going through a digital to analog conversion at least once (it's a digital signal across the USB that to get passed through the DAC to get to the audio hardware) so yes, it's not going to be nearly as "perfect" as a true discriminator tap will ever be, but it can get pretty damned good when you tweak those settings as required - no audio filtering at all, volume levels set properly for a good decoding level, using the particular switches to lock in the decoding specifically, and even going so far as to use dsdtune and "tuning" the additional parameters that DSD+ offers. The tuning isn't specifically necessary for all signals, but in some instances it CAN provide a massive difference in decoding quality.

I guess for me personally I'm lucky: in the past few weeks I've picked up about two dozen DMR/MOTOTRBO systems here in my area, but whenever I do a tuning run with dsdtune I end up getting zilch for results meaning I'm apparently getting the best signals I can already. I do a raw recording (roughly 60 seconds worth) of busy DMR/MOTOTRBO traffic, then do a full dsdtune pass and I end up getting barely any improvement(s) at all so, there goes that idea. But there are examples where using dsdtune has dramatically improved the decoding process and resulting in vastly improved audio quality.

Twist, twist, click, check, uncheck, tune, optimize, adjust, tweak some more, tune, etc... experiment. ;)

Last last tip: some DMR/MOTOTRBO and even NXDN transmissions are control channels, you won't hear any voice traffic on them at all, or if you do it'll be very infrequent. If you're focusing on DMR/MOTOTRBO traffic specifically, and the signal is constant and never stops/starts/stops/starts like traditional radio traffic, it's most likely a control channel and then you'd want to use DMRDecode at the same time to get info from it. In fact, using DMRDecode + DSD+ is recommended I suppose so you can see what's going on with the data streams. It'll tell you whenever there's actual voice traffic (Group calls) as well as Data only (Data calls unit to unit usually imply the "Privacy" mode of operation but it can be actual data sent to a device).
 
Last edited:

W4KRR

Member
Premium Subscriber
Joined
Apr 1, 2001
Messages
2,930
Location
Coconut Creek, FL
br0adband: I tried the settings you suggested above, and it didn't make much difference, until I used the switches that forces DSD+ to decode only one mode at a time; at that point there was tremendous signal decode improvement.

What I would like to do is create a batch file that would launch DSD+ from the desktop with a double click. I could have three separate batch files, one for DMR, one for NXDN, and one for P-25.

It's been a long time since I dealt with batch files; maybe someone could write a sample batch file that would do what I want?

Thanks!
 

br0adband

Member
Joined
Apr 8, 2005
Messages
1,569
Location
Springfield MO
It would be as simple as this where dsdplus.exe is the actual filename (that's what I use, some people alter it, YMMV):

Code:
dsdplus.exe -fr
pause
and save that as <whatever>.bat - obviously name them so you know which is which (one for DMR/MOTOTRBO, one for NXDN, etc). When you execute that batch file, it'll run DSD+ with the given command line switch(es) and when you're done with it, one tap on Esc will end the program gracefully so it closes out the audio files it may be creating and saves them, then the command prompt will stay available with the pause command until you press any key which would then terminate the batch file execution.

Replace -fr with what protocol you're wanting to decode, add other switches as required, and so on.
 
Last edited:

W4KRR

Member
Premium Subscriber
Joined
Apr 1, 2001
Messages
2,930
Location
Coconut Creek, FL
Well, that is what I would think, but all I get is:

'dsd.exe' is not recognized as an internal or external command,
operable program or batch file.

On my system, DSD+ is located at:
C:\DSD Plus

And the program executable is:
DSD.exe

I've tried to alter the commands within the batch file with no luck.
 

br0adband

Member
Joined
Apr 8, 2005
Messages
1,569
Location
Springfield MO
Might be causing you trouble because you have a space in the directory name: C:\DSD Plus

Try renaming that directory to just DSDPlus (no space) and see what happens. If you can't, try using the whole command as:

Code:
C:\DSD Plus\dsd.exe -fr
pause
or whatever. Easier to just rename the directory as DSDPlus I'd say. ;)
 

W4KRR

Member
Premium Subscriber
Joined
Apr 1, 2001
Messages
2,930
Location
Coconut Creek, FL
Yep, the space in the folder name was the issue. Thanks! It's been 15 years since I last dabbled with batch files. :D
 
Status
Not open for further replies.
Top