Hello from sunny southern California.

[KO6CDG]

Hi there!

I am a newly licensed ham; I have as of this moment only transmitted twice from my radios. Although I have existed on the periphery of radio communications for a while: as a software engineer at a telecommunications company and later at aerospace company operating synthetic aperture radar systems in orbit.

I was pleasantly surprised that many of the random bits of information that I picked up from my day jobs were things needed for the Technician Class license. And I am now in the process of reading up for my General Class.

My actual path to the hobby and here is meandering and wandering. But might make for an interesting read and be a good way to share some more details about myself. As I mentioned above I'm a software engineer, I'm also an open source advocate with a long interest in data encoding, compression and encryption.

So on that note (which will eventually get back to radio): A form of unbreakable "perfect" encryption is called one-time pad: it involves sharing single-use pre-shared keys equal in size to the messages (think sheets of paper with random numbers and letters). Because the keys are entirely random: cryptanalysis would be impossible. So during the cold war countries would give these pads to diplomats and spies. Then use radio "number stations" to send messages that would be decoded using the pads.

To outside observers receiving number station transmissions: the transmissions appear liminal, in an eerie, forlorn, often surreal sense. Which has resulted in number station recordings and samples (from the Conet Project) working their way into music and movies. An early example being Wilco's best-selling 2001 album "Yankee Hotel Foxtrot" (which is where I first learned of them) and has been followed by countless other acts.

Since the end of the cold war and the rise of the internet: number stations were largely historical curiosities that had mostly disappeared. That is until the Russian invasion of Ukraine, where we are now witnessing a strong reappearance. In an effort to hear some of these modern number stations I came across the Ringway Manchester YouTube channel and subscribed.

One of Ringway Manchester's videos was on the Thunderpole TX a CB Radio for the UK market that featured digital noise reduction. Having used a family CB radio as a child I was impressed with the review and started to explore options for something similar that would operate in the North American bands. With the goal of just being able to give a simple radio to a friend or family member (I did end up getting a GMRS license). During the COVID lock-down I setup a Mumble voice server for my team, it was hugely successful. And I was really hoping to find a radio that used Xiph's RNNoise noise suppression library; which I find superior to other digital noise reduction techniques.

I have yet to find my ideal radio with noise reduction/suppression (I'm hoping it might be something OpenGD77 might be able to do one day). But after looking at a number of options I settled on the Quansheng UV-K5, because it was just discovered that it's firmware could be replaced and the early open source firmware enabled the radio to receive CB bands. Since then the firmware has only gotten better and new features added.

I was having fun just scanning the local area for signals and wasn't really planning on doing anything else. But after listening to nets and hearing the various conversations, I started to think it would be fun to join in. And was curious to learn more about various things I heard. So I looked into what would be needed to getting a license and took my test.

While reading the GitHub tickets on one of the UV-K5 repos: I saw a user mention how his UV-K5 and his OpenGD77 [something] radio had replaced everything else in his kit. And how much he loves both projects. Which in turn led me to check out the OpenGD77 project. After doing some research over here I bought two TYT MD-UV390 Plus radios and flashed them to OpenGD77. (My wife is now studying for her Technician license. And wanted a water proof radio.)

Anyways, thanks for reading. Looking forward to exploring DMR with the OpenGD77 community. I can safely say the CPS software is an upgrade over the stock TYT software. And I've already created a GUI theme, I plan on sharing a little later. Sadly, now that I've been able to listen for a few days I think I'm in a DMR dead zone and will need to get a hotspot. There are some strong digital stations in the area, but I've discovered upon second look that they are in the commercial bands.

73

[F1RMB]

Welcome

[VK3KYY]

I have yet to find my ideal radio with noise reduction/suppression (I'm hoping it might be something OpenGD77 might be able to do one day).

The radio hardware is not SDR. We have no way to add any functionality like noise reduction.
The RF transceiver chip, used in all supported handhelds is a very low performance cheap device, only intended for FM use e.g. like in the cheap Baofeng radios etc. It literally costs $1 to the manufacturer, and has performance to match its price.

Its a miracle it can operate on DMR at all, and the transcriver chip is not actually capable of transmitting DMR. The DMR Tx is created by the DMR chip directly modulating the master RF reference oscillator, and the RF chip only produces the carrier frequency

[KO6CDG]

We have no way to add any functionality like noise reduction.

Thanks for the feedback. I know next to nothing when it comes to how the hardware works.

I assumed that the ARM CPU would be doing some sort of DSP on the signal from the RF transceiver before sending it to a DAC. Guessing that the ARM CPU was where you got things like the S-Meter, automatic gain control, etc. So my thought was: it might be fast enough to also run some sort of noise reduction/suppression logic. (The RF transceiver chip itself wouldn't be doing anything new.)

And don't be too hard on $1 parts. ESP32 modules are near that price point and are on par with ~20 year old desktops.

[VK3KYY]

I assumed that the ARM CPU would be doing some sort of DSP on the signal from the RF transceiver before sending it to a DAC. Guessing that the ARM CPU was where you got things like the S-Meter, automatic gain control, etc. So my thought was: it might be fast enough to also run some sort of noise reduction/suppression logic. (The RF transceiver chip itself wouldn't be doing anything new.)

No.

The CPU does not do any audio processing. The AF output of the TRx chip is directly connected to the audio PA chip (TDA something or other)

The S meter (RSSI) is read from a register in the TRx chip, as is the S/N ratio (aka "noise" value).

There is no RF gain control in of the TRx chip.
There is some limited gain control of the audio output from the TRx chip, in 2 different internal parts of the chip, one they call digital audio gain and the other they call analog audio gain, but as there is no clear internal block diagram its unknown what difference the difference is between digital and analogue audio gain. The 2 gains are summed / multiplied, so to get the full range of audio gain control it would be necessary to vary both gains,but on the GD77 style radio

The TRx chip does not have any documented AGC controls. It does have a load of parameters to control its internal IF, but there is no documentation about what each of the 16 paramaters do, simply a list of values that is required for 12.5kHz and 25kHz FM Rx

The contol of the TRx chip is quite slow via I2C and not designed for anything like attempting to turn the gain up and down to somehow cut out noise.

i.e the RSSI data from the chip only changes relatively slowly, taking around 150mS to settle after a signal is received.
S/N data is also similarly slow. These factors govern the scan speed to around 50ms settling time, and even slower for DMR (nearer 100ms settings time)

And with severely limited number of gain settings, (I think 16 in total), I doubt you could do any effective noise reduction.
i.e all you'd end up is cutouts from the audio which would not be in sync with the noise on the audio.

Anyway. FM is not especially prone to the sort of noise that you could reduce simply by turning the audio gain up and down in very course steps.



The DMR functionality is handled my a separate chip, which does all the 4FSK decodeing and encoding (aka MODEM) functionality, it also encodes and decodes and encodes the DMR frames. The DMR chip has direct AF connection to the TRx chip and to the master reference oscillator to modulate the 4FSK Tx.
There is very limited documentation on the DMR chip and a lot of the control and initialisation data needed to be sent to the DMR chip, were gathered by attaching a logic analyser to the internal data bus, i.e we recorded what the official firmware sent to the DMR chip to initialise it, and we send the same data

Bizarrely we can't send the same data to the same DMR chip in the UV380 as we do on the GD77, because the chip won't work correctly.
We have no idea why this is, as it doesn't make any sense.

Possibly its something to do with the audio coupling between the FM TRx chip and the DMR chip, being different between the GD77 and UV380 style radios, but as there is no official documentation for the DMR chip its impossible to know the function of the different intialisation sequences.

[KO6CDG]

Thanks for the detailed explanation! You also filled me in a number of smaller details I was curious about; such as to do with scanning and the spectrum display.

That's a bummer we don't have greater programmatic control over the signal.

Also let me commend you and the team for your investigative work on all of the undocumented components. The results are very impressive.

I read somewhere that Retevis was supporting the project in a semi-official way, and encouraging advanced users to upgrade their RT3S's with OpenGD77. (I read that on the internet, so it must be true! ) If that is true: I assume you guys haven't had any luck asking them for hardware documentation.

[G4EML]

Unfortunately the companies making and selling all of these radios don't design them or even write the firmware. As far as we know they buy it in from a mystery third party supplier who seems to be very secretive.

[VK3KYY]

Retevis are also just a brand.
They simply buy TYT and other manufacturers radios , and stick their label on them.

I have tried contacting TYT on several occasions but they are not interested in supplying any information about the hardware.

I have never hear from Baofeng, they seem to be even more secretive


It's an open secret that the firmware for radios from TYT and Boateng and other makes is written by an unknown mystery organisation, which points to it being from a state run company / organisation, i.e the Chinese goverment

[SA0BUX]

Retevis are also just a brand.
They simply buy TYT and other manufacturers radios , and stick their label on them.

I have tried contacting TYT on several occasions but they are not interested in supplying any information about the hardware.

I have never hear from Baofeng, they seem to be even more secretive


It's an open secret that the firmware for radios from TYT and Boateng and other makes is written by an unknown mystery organisation, which points to it being from a state run company / organisation, i.e the Chinese goverment

I guess that Baofeng is just another brandname, Baofeng DM-1701 and Retevis RT84 has the same internals.

Like Alluince which seems to have some relationship with Retevis.

[VK3KYY]

I guess that Baofeng is just another brandname, Baofeng DM-1701 and Retevis RT84 has the same internals.

Like Alluince which seems to have some relationship with Retevis.

No.

Baofeng are a manufacturer, like TYT.

Retevis buy the internal parts of the DM-1701 from Baofeng and put it in a different case. probably they just commissioned Baofeng to put it in a different case for them to have some brand differentiation.

The vertical resellers like Retevis and Radioddity get some small changes made to the variants of the radios they sell, this can be as little as a different logo, or a slight different plastic case, or slighly modified firmware or modified CPS.

Radioddity actually commissioned a 3rd party developer to write a completely new CPS for the GD-77, rather than use the TYT MD-760 CPS.
i.e. The GD-77 is a rebranded TYT MD-760 with a different CPS.

The Radioddity GD-88 is just a rebranded Kydera 880 with slighly modified firmware.


Retevis and Radioddity will source radios from many different companies and rebrand them.