Anyone using DVSI AMBE3000 chip for digital decoding?

[rumcajs_tr]

Everybody is using SW decoders like DSD+ for receiving DMR etc. I wonder, if anyone made some progress to use the HW DVSI AMBE3000 chip for decoding DMR. The decoding quality should be much improved (and with much weaker signals). The chips are available for around 30 USD and some of them can be connectet to USB.

Examples:
ThumbDV™ - ThumbDV™
Portable AMBE Server - Portable AMBE Server
DVstick 30 - DVstick 30

However, the relevant SW is for HAM only, did anybody find any SW used for monitoring/decoding the digital audio on the PC or Raspberry Pi?

 

[boatbod]

Everybody is using SW decoders like DSD+ for receiving DMR etc. I wonder, if anyone made some progress to use the HW DVSI AMBE3000 chip for decoding DMR. The decoding quality should be much improved (and with much weaker signals). The chips are available for around 30 USD and some of them can be connectet to USB.

Examples:
ThumbDV™ - ThumbDV™
Portable AMBE Server - Portable AMBE Server
DVstick 30 - DVstick 30

However, the relevant SW is for HAM only, did anybody find any SW used for monitoring/decoding the digital audio on the PC or Raspberry Pi?

I can see how using a hw chip may reduce cpu utilization, but how is it going to improve weaker signal performance? You still have to run all the demod, protocol decode, and a good portion of the FEC in software before feeding xMBE codewords to the codec hardware.

 

[slicerwizard]

how is it going to improve weaker signal performance?

It's not.

 

[rumcajs_tr]

The idea behind this is that the HW chip could reproduce the usefull data stream from less quality signal than the SW decoder. But I might be wrong. My expectation is just based on real experience, when decoding the same signal with professional DMR radio and with Airspy R2 and DDS+ gives different results. The decoded audio from professional radio is much better. So I assume that the SW decoders do not do the exact algorithms and error correction of data stream in the same way as the HW implementation.

 

[boatbod]

The idea behind this is that the HW chip could reproduce the usefull data stream from less quality signal than the SW decoder. But I might be wrong. My expectation is just based on real experience, when decoding the same signal with professional DMR radio and with Airspy R2 and DDS+ gives different results. The decoded audio from professional radio is much better. So I assume that the SW decoders do not do the exact algorithms and error correction of data stream in the same way as the HW implementation.

The FEC algorithms are pretty well documented and at least as far as op25 is concerned, have been implemented fairly comprehensively. Biggest difference between the pro radios and inexpensive sdr software is the demodulator. The better the demod, the lower the BER and better the recovered signal.

 

[slicerwizard]

The idea behind this is that the HW chip could reproduce the usefull data stream from less quality signal than the SW decoder. But I might be wrong.

The chips do not produce a data stream from a signal. The chips are fed a data stream produced by the radio they're in.

My expectation is just based on real experience, when decoding the same signal with professional DMR radio and with Airspy R2 and DDS+ gives different results. The decoded audio from professional radio is much better. So I assume that the SW decoders do not do the exact algorithms and error correction of data stream in the same way as the HW implementation.

You're ignoring all of the differences between a real radio and an R820T2 tuner.

 

[rumcajs_tr]

The FEC algorithms are pretty well documented and at least as far as op25 is concerned, have been implemented fairly comprehensively. Biggest difference between the pro radios and inexpensive sdr software is the demodulator. The better the demod, the lower the BER and better the recovered signal.

The chips do not produce a data stream from a signal. The chips are fed a data stream produced by the radio they're in.

So if I understand it correctly, a high level ADC would help more with better decode than the AMBE HW chip itself, right?

You're ignoring all of the differences between a real radio and an R820T2 tuner.

You are right, but the AirSpy is prety decent receiver, not just cheap RTL. But point taken.

However, my overall impression is that at least DSD+ compared to HW decode does not give such excelent results. Can't speak for OP25 (didn't test it). As DSD+ is not opensource, maybe the codec implementation is not exactly as specified in the standards.

Back to my original question - did anybody experiment wit using the HW AMBE 3000 chip for decode and if so, what software did you use? Thanks.

 

[slicerwizard]

So if I understand it correctly, a high level ADC would help more with better decode than the AMBE HW chip itself, right?

Are you dealing with bad decodes? Are you working weak signals? Does DSD+ indicate that there are decoding errors?

 

[hamradionl]

@rumcajs_tr
AOR build external radio modems with AMBE chip inside that decodes very well. The project started in 1998 or 1999 and the hardware based on the development from Charles Brain G4GUO and Andy Talbot G4JNT using a AMBE chip. That time i was one of the early BETA testers from that project using any standard HF, VHF or UHF radio. In 1999 knowing the AMBE chip could be used for DMR, NXDN, P25 and Tetra who not existing that time. Short after these test, the digital modes introduced in hamradio and later the first d-star introduced as fallowup. (time before SDR#, D-STAR, DMR, TETRA, NXDN and DSD existing or see the daylight).
Using a standard transceiver, the hardware decodes better on low level SNR as RTLSDR on the market, this moment.
For those who like reading about this older project this is a copy from the original page:

Digital Voice Project – Temple University Amateur Radio Club K3TU

AOR voice modems

OTHER PRODUCTS | AOR,LTD. Authority On Radio Communications

 

[Unitrunker2]

All three devices you linked only support DSTAR and DMR. DVSI chips are crap for the following reasons:

Limited protocol options.

To get more protocol variants (which differ in trivial ways) you must buy more chips.

Does not scale.

Want to record two, five, twenty calls at once in real-time? You need one chip per channel. I can see a rats nest of USB cables already.

 

[hamradionl]

Dont shoot the messenger, AOR 300 support digital modes like P25-1, NXDN, DMR, D-STAR, dPMR. But again, i only mention the project in 1999 about the hardware with there limitations.
We waiting for a receiver with build in chinese decoder hardware chip to be connected to a PC and record 2, 10 or 100 calls at once in real-time. The pricing level must be $10,00 or 10,00 Euro or less. The PC software is for free, no donations to author is possible.

 

[Unitrunker2]

Dont shoot the messenger,

i was referring to the three devices in the OP's opening post above.

 

[KA1RBI]

the codec implementation is not exactly as specified in the standards.

No, the DVSI implementation is not exactly as specified in the standards.

Max