BER and Duty Cycle

Post Reply
User avatar
kd2lh
Posts: 171
Joined: Mon Dec 02, 2019 2:44 pm

BER and Duty Cycle

Post by kd2lh » Tue Dec 03, 2019 12:57 pm

I just got OpenGD77 up and running on my GD-77 radio using a Raspberry Pi 4b and PiStar R4.1.0 rc6.

It's operating well and appears fully functional.

I've noticed that I can still adjust RxOffset and TxOffset, and that the resulting frequency will show up on the GD-77 display confirming the frequency it's tuned to. The BER shows over 1% to 2% on receive, yet the reports I'm getting in conversation indicate no appreciable degradation of voice quality.

I have not tested this to the point of detuning it far enough to affect voice quality.

Does the SDR chip in this device have AFC features available to self tune - or, in fact, is this operating and the BER reporting how far it's correcting?

Also, does anyone know if specifications for duty cycle for this radio's transmitter have ever been published. What kind of heat dissipation is built into the radio? I'm operating at 1 watt (default) and not noticing any problems, but I'd like some idea of how hard this will be on the radio and thoughts on useful life in hotspot mode.

Thanks for some remarkable work. I'll be working to try and help the project, and testing the unit on a transmitter tester to see what the RF envelope looks like.

Marc KD2LH '73

G4EML
Posts: 171
Joined: Sat Nov 16, 2019 10:01 am

Re: BER and Duty Cycle

Post by G4EML » Tue Dec 03, 2019 1:14 pm

A properly implemented DMR decoder in a custom designed chip as used in the GD77 and other DMR radios is tolerant of quite large frequency errors, indeed this is a requirement to fully comply with the DMR specifications. Hotspots have always been excessively sensitive to fine tuning of the frequency. This is mainly due to the simple hardware used for decoding.

Therefore you should find that using the GD77 as a hotspot will produce much better results.

User avatar
kd2lh
Posts: 171
Joined: Mon Dec 02, 2019 2:44 pm

Re: BER and Duty Cycle

Post by kd2lh » Tue Dec 03, 2019 2:12 pm

Agreed... it is delivering superior results at the measured BER rates.

Would there ever be a need to use RxOffset or TxOffset with the OpenGD77 configuration?

I did try testing PiStar-MMDVMCal and the API it uses to control the radio is not present in OpenGD77. Is there a scenario where it would ever be needed?

G4EML
Posts: 171
Joined: Sat Nov 16, 2019 10:01 am

Re: BER and Duty Cycle

Post by G4EML » Tue Dec 03, 2019 4:48 pm

I don't think RxOffset or TxOffset would be needed for the OpenGD77. DMR Radios can talk to each other without the need for constant fine tuning. That was something that became necessary as a work-around for problems with the Hotspot hardware which are not present in the GD77.

For the same reason MMDVMCal is not needed as all calibration is implemented in the GD77 firmware.

VK3KYY
Posts: 1612
Joined: Sat Nov 16, 2019 3:25 am
Location: Melbourne, Australia
Contact:

Re: BER and Duty Cycle

Post by VK3KYY » Tue Dec 03, 2019 8:02 pm

Initially I did not implement the special case of the Tx and Rx offsets in the hotspot, because like Colin said, it should not be necessary, however some people complained that their setup needed offsets, so I ended up implementing it.

Unfortunately because of the way MMDVMHost works, the frequency and the offsets are not sent separately to the Hotspots, instead MMDVMHost adds or subtracts the offsets from the Tx and Rx frequency and sends the modified frequency to the Hotspot.
Hence the Hotspot has no way of knowing what the core frequency is.

Part if the problem people are having is because the cheap hardware in the JumboSpot type devices needs to be fine tuned to work, and people leave these crazy offsets in their PiStar config when they connect the GD77 Hotspot.

I should possibly display a warning when the Hotspot is sent different Tx and Rx frequencies or perhaps disable the hotspot under those conditions

User avatar
kd2lh
Posts: 171
Joined: Mon Dec 02, 2019 2:44 pm

Re: BER and Duty Cycle

Post by kd2lh » Thu Dec 05, 2019 12:13 am

When an offset has been left in the configuration (I cloned my PiStar SD card as the starting point for my OpenGD77 install) the GD77 frequency display shows the actual receive frequency as modified by the RXoffset. This typically shows a fractional frequency like 438.59985.

A BER is being reported to Pi-Star, but receive and transmit audio is much cleaner and more intelligible than the way the same radio sounds on a JumboSpot MMDVM Hat hotspot. I know that AFC is not implemented in the JumboSpot hats, which is why the RXoffset must be adjusted.

I'm not sure a warning about the offsets is needed since it's clear that an offset has been applied if you look at the radio display. Perhaps an explanation of what's happening?

- - - - -

Separately, on the duty cycle issue, I've asked Radioddity to contact the Chinese developers to see if they have done any duty cycle / power transistor survival testing on the radio. Until then, I'm running it at 1 watt, and only seeing moderate warming of the back plate. We'll see if they get back from the lab there.

With the higher time spent in transmit, especially with a number of static talkgroups configured, deterioration could be an issue.

I'll also do some testing of the transmitter soon to ensure it remains within spec for spurious energy radiation.

VK3KYY
Posts: 1612
Joined: Sat Nov 16, 2019 3:25 am
Location: Melbourne, Australia
Contact:

Re: BER and Duty Cycle

Post by VK3KYY » Thu Dec 05, 2019 12:29 am

Thanks

I agree if people look at the display, its clear what frequency/frequencies they are using, so I guess I don't need to go through the hassle of adding a warning.

Its just a shame that MMDVMHost does not send the offset separately.

Re: Duty cycle.

In DMR its only 50% duty cycle of FM as its only transmitting on one TS, unlike the Jumbospots which actually transmit on both slots but only have data in one slot (normally).

I agree, the radio was not designed to operate with high duty cycles and the here is no direct heat sink on the PA fets, they just reply on heat dissipation via the PCB and then into the central aluminium chassis , and then partially into the battery as a thermal mass.

User avatar
kd2lh
Posts: 171
Joined: Mon Dec 02, 2019 2:44 pm

Re: BER and Duty Cycle

Post by kd2lh » Thu Dec 05, 2019 1:02 am

If the Chinese lab responds with any test data, I'll post it.

VK3KYY
Posts: 1612
Joined: Sat Nov 16, 2019 3:25 am
Location: Melbourne, Australia
Contact:

Re: BER and Duty Cycle

Post by VK3KYY » Thu Dec 05, 2019 1:11 am

kd2lh wrote:
Thu Dec 05, 2019 1:02 am
If the Chinese lab responds with any test data, I'll post it.
OK

Thanks

Post Reply