DMR Decoding Errors (DSD+ FastLane)

[thewraith2008]

No, they use a rubidium or cesium standard. You wouldn't use FMP24 to calibrate because it's the wrong tool for the job. Use something that has better resolution like SDR# and increase the FFT resolution.

While SDR# only uses integers for the PPM value, you can use an educated guess to set the factional part and even then it's only 0.2 increments for dongles. FMP24 auto center tuning will take care of the rest if needed.

Hams have been zero beating against WWV for decades to tune radios and other things with excellent results.

Here is a video showing that exact process to tune a frequency counter. With 0.5hz accuracy.
https://www.youtube.com/watch?v=QCJ4cQGOQLI

we2aew does some great videos

Anyway using WWV was only a suggestion because it's a know standard. (3x10ˉ16)

https://www.nist.gov/pml/time-and-frequency-division/primary-standard-nist-f1

Using kalibrate against GSM. It tells me the PPM to use: (5 different channels)
CH
43 = -1.4 ppm
47 = -1.6 ppm
80 = -1.0 ppm
90 = -0.6 ppm
92 = -1.6 ppm

Test repeated 5 times. So which one above is right?
While the difference is not much, I hardly find GSM to be a calibration standard.

 

[slicerwizard]

No, they use a rubidium or cesium standard. You wouldn't use FMP24 to calibrate because it's the wrong tool for the job. Use something that has better resolution like SDR# and increase the FFT resolution.

Never had any problem using FMP24 to accurately set PPM. Just have to use a high frequency reference. SDR# isn't going to do a better job.

While SDR# only uses integers for the PPM value, you can use an educated guess to set the factional part and even then it's only 0.2 increments for dongles. FMP24 auto center tuning will take care of the rest if needed.

So now you're saying that SDR# isn't the right tool. And where did you get 0.2 from?

Hams have been zero beating against WWV for decades to tune radios and other things with excellent results.

I doubt that the OP's dongle goes down to 10 MHz and I doubt he has an antenna tuned for that band.

Using kalibrate against GSM. It tells me the PPM to use: (5 different channels)
CH
43 = -1.4 ppm
47 = -1.6 ppm
80 = -1.0 ppm
90 = -0.6 ppm
92 = -1.6 ppm

Test repeated 5 times. So which one above is right?
While the difference is not much, I hardly find GSM to be a calibration standard.

You chose to use kalibrate. Not my problem if you don't like the results. And why are you choosing the complicated route anyway? Tune FMP24 to a 700 to 900 MHz signal and center the zoomed spectrum. Done.

 

[GTR8000]

Tune FMP24 to a 700 to 900 MHz signal and center the zoomed spectrum. Done.

^ This

If you're lucky enough to have a Motorola ASTRO 25 700/800 MHz system near you, calibrate the dongle to that control channel. The G Series repeaters are dead nuts on frequency. I'm not playing favorites with manufacturers, I'm simply speaking from firsthand experience with those particular systems.

Before the TCXO dongles were priced reasonably and readily available, my tool of choice to calibrate the drifty old dongles was to fire up HDSDR, tune to one of those ASTRO 25 control channels, let the dongle warm up for 10-15 minutes, then calibrate it. I found HDSDR's calibration tool quick and easy to use after using AFT to center the signal.

That's just one method, but the key is to calibrate against a signal that is rock steady and of high accuracy.

 

[tbailey1712]

How much is DSD+ performance affected running in Parallels under MacOS?

Could it be just a raw lack of CPU power causing decoding errors?

 

[thewraith2008]

Jesus, why are you fighting me on this. Using 10MHz is just an alternative. Use it, don't use it.

And I've never had any problem using SDR# to accurately set PPM.

So now you're saying that SDR# isn't the right tool

hmm, No I didn't

The 0.2PPM was from Unitrunker. adjusting by 0.1 made no different to window value, 0.2 did.

I don't used kalibrate to set my PPM.
I used kalibrate here to demonstrate the differences against 5 GSM frequencies and how they vary. Are they not supposed to be all the same?

Tune SDR# to WWV 10MHz signal and center the zoomed spectrum. Done.

And your probably right. His dongle probably doesn't go down to 10MHz. I assumed the "NooElec RTL-SDR" was like the RTL-SDR.com dongle.

Anyway. Back to the OPs problem.

 

[GTR8000]

And your probably right. His dongle probably doesn't go down to 10MHz. I assumed the "NooElec RTL-SDR" was like the RTL-SDR.com dongle.

Neither does the RTL-SDR.com dongle (begins at 24 MHz). In fact, no dongle based on the R820T/R820T2 tuner goes below 24 MHz without using a converter or direct sampling. That includes the NooElec, the RTL-SDR.com, the AirSpy, etc.

Recommending calibration to a frequency these dongles can't natively tune to is a bit silly, isn't it? :roll:

 

[thewraith2008]

@GTR8000

The RTL-SDR.com does goto 10MHz

Experimental: HF Direct Sampling Mode – Listen to 500kHz to 24 MHz with direct sampling. Simply connect an HF antenna to the SMA port, and choose the Q-branch to use. (Note that HF filters are required for optimum performance).

from here:
Buy RTL-SDR Dongles (RTL2832U) - rtl-sdr.com

What's silly is your statement with no facts. isn't it.

 

[GTR8000]

I guess you didn't fully read or comprehend my post, where I very clearly stated:

...no dongle based on the R820T/R820T2 tuner goes below 24 MHz without using a converter or direct sampling.

And if you actually bothered to read the entirety of their description of the dongle, you would have seen the stated frequency range:

The R820T2 has a frequency range of 24 – 1766 MHz

Do you understand the difference between the native range of the tuner, vs using direct sampling to get it below 24 MHz? Do you realize that the majority of users are not going to understand how to do that?

Don't lecture me on facts, when all you've done is cherry picked them in an attempt to reinforce your position, which doesn't have much merit to begin with.

 

[thewraith2008]

On a RTL-SDR.com dongle in SDR# setting "Sampling Mode" to "Direct sampling (Q branch)", done.

Do you understand the difference between the native range of the tuner, vs using direct sampling to get it below 24 MHz?

What's it matter. With the above settings, I receive under 24MHz just fine.

Don't lecture me on facts, when all you've done is cherry picked them in an attempt to reinforce your position, which doesn't have much merit to begin with

So you agree what I've mentioned are facts. Thanks. Nothing about calibrating using WWV is false. If so, please explain how so.

 

[slicerwizard]

That's just one method, but the key is to calibrate against a signal that is rock steady and of high accuracy.

And high frequency!

How much is DSD+ performance affected running in Parallels under MacOS?

Could it be just a raw lack of CPU power causing decoding errors?

Your screen shot shows minimal decoding errors and a small tuning error. I'd deal with the tuning error first. No idea if you have a lack of CPU power. If you do, it would be more evident during voice decoding, which uses considerably more CPU cycles. In Windows, one would check that using Task Manager. MacOS must have a similar utility.

Jesus, why are you fighting me on this. Using 10MHz is just an alternative. Use it, don't use it.

I doubt the OP has a longwire antenna and a matching transformer just lying around waiting to do a fine tuning job.

And I've never had any problem using SDR# to accurately set PPM.
hmm, No I didn't

Yes, you said SDR#'s PPM granularity is only 1 PPM and that you have to guesstimate the rest.

The 0.2PPM was from Unitrunker. adjusting by 0.1 made no different to window value, 0.2 did.

Now you're just dragging an unrelated program in. We don't know why 0.1 makes no difference in that particular program. Probably related to R820T tuning granularity, which might make the value frequency-dependent, not a fixed 0.2. Or it has something to do with how Rick coded things. I don't see how it's relevant to the OP and his FMP or FMP24 usage.

I don't used kalibrate to set my PPM.
I used kalibrate here to demonstrate the differences against 5 GSM frequencies and how they vary. Are they not supposed to be all the same?

Do we know exactly how kalibrate works? How does it deal with the R820T's tuning granularity? The differences you see could be strictly due to that. The GSM signals are probably all right on frequency.

When someone is using FMP or FMP24, the simplest, most logical thing to do is tune to a high frequency (770 MHz+) repeater and use the PPM adjustment keys to center the signal in the zoomed spectrum. Done and done.

 

[SCPD]

While SDR# only uses integers for the PPM value, you can use an educated guess to set the factional part and even then it's only 0.2 increments for dongles.

Any tool that tries to calibrate a Realtek USB SDR with the R820T(2) will run into two problems.

1. the tuning resolution offered by the SDM (sigma delta modulation) registers (prescaler and divisor) in the R820T is about 200 hertz at 800 Mhz.

2. the register on the RTL2832U controller that sets the warp correction value resolves to increments of 0.125 ppm (one eighth of a ppm).

Every time you change the center frequency - the tuning error of the R820T may change - between zero and 200 hertz. That means a 1 ppm warp correction error - which should be 800 hertz at 800 Mhz - may appear to be 800 to 1000 hertz in error (or 700 to 900 hertz - I don't remember if the register math rounds or truncates).

The point is - when you calibrate against different known good reference frequencies - expect to see slightly different results.

It is possible - in software - to completely compensate for both #1 and #2. So far, none of us lazy coders have actually gotten around to doing so.

 

[Mikek]

How much is DSD+ performance affected running in Parallels under MacOS?

Could it be just a raw lack of CPU power causing decoding errors?

I run DSD+ this way... and while it works, it's never 'clean'. I get constant errors.

Recently I switched to CubicSDR and DSD+ in a Wineskin wrapper and ALL of the errors are gone, there's no glitchy decodes and life is good.

 

[ChrisHansen]

Im Getting a Crash on a freq DMR DCC:1 464.6625
I can send the CC.log if that helps.
---------------------------
Error
---------------------------

CC DSD+ 2.39

Unrecoverable error encountered:

Exception number = 0xC0000005 (Access Violation)
Instruction address = 0x43AE67
Thread = #1
Action: Write to address 0x3F3F3F (4144959)

Register dump:
DS: 0000002B CS: 00000023 IP: 0043AE67 EAX: 00000000
ES: 0000002B SS: 0000002B SP: 0019CD40 EBX: 003F3F3F
FS: 00000053 Fl: 00010202 EBP: 0019CD48 ECX: 00000000
GS: 0000002B ESI: 0019CE44 EDI: 0019CDE8 EDX: 00000073
[Press Ctrl-C to copy this text to the clipboard]
---------------------------
OK
---------------------------