Yes, it's possible. You can use any of the example flow graphs, but I recommend using a smaller FFT size. At the lower bandwidth, the carriers are very close together and phase noise starts to become a problem.

A good starting point would be the example flow graph vv010-2kfft.grc

The first thing you need to do is change the samples rate to 131000000.0 / 71.0.

But that will make the frame time too long, so the number of symbols has to be reduced. To get back to 250 milliseconds, reduce the number of symbols from 983 to 190. Since there's less symbols, the number of FEC blocks also has to be reduced from 93 to 18.

That will give you a bitrate of about 2.8 Mbps. You can change the FEC rate to get more bitrate. The bitrate calculator is at:

https://github.com/drmpeg/dtv-utils/blob/master/dvbt2rate.c

Use a bandwidth of 0 to select 1.7 MHz. Here's what it looks like for the example above.

./dvbt2rate 0 2 2 190 18 2 2 0 0 2 1
FFT size = 2048
guard interval = 1/8
number of data symbols = 190
number of FEC blocks = 18
code rate = 3/5
constellation = 16QAM
frame size = normal
carrier mode = normal
pilot pattern = PP2
L1 constellation = QPSK

clock rate = 1845070.422535, TF = 248.359084 ms
Normal mode bitrate = 2798142.064679
High Efficiency mode bitrate = 2813105.391228

max symbols = 200, max blocks = 18
symbols = 198, max blocks = 18
cells = 299912, stream = 291600, L1 = 2592, dummy = 5609, unmodulated = 111

PAPR max symbols = 200, max blocks = 18
symbols = 198, max blocks = 18
cells = 296492, stream = 291600, L1 = 2592, dummy = 2189, unmodulated = 111

The rules are that the frame time TF has to less than 250 ms and the number of dummy cells has to be positive. The guard interval and pilot pattern have to be valid from table 58 in the DVB-T2 specification.

You could get more bitrate by switching to 1/32 guard interval, but you have to use pilot pattern 4 or 7. Here's what 1/32 and PP7 looks like:

/dvbt2rate 0 2 0 190 18 2 2 0 0 7 1
FFT size = 2048
guard interval = 1/32
number of data symbols = 190
number of FEC blocks = 18
code rate = 3/5
constellation = 16QAM
frame size = normal
carrier mode = normal
pilot pattern = PP7
L1 constellation = QPSK

clock rate = 1845070.422535, TF = 227.754992 ms
Normal mode bitrate = 3051278.888684
High Efficiency mode bitrate = 3067595.888089

max symbols = 218, max blocks = 21
symbols = 198, max blocks = 19
cells = 321670, stream = 291600, L1 = 2592, dummy = 27242, unmodulated = 236

PAPR max symbols = 218, max blocks = 21
symbols = 198, max blocks = 19
cells = 318250, stream = 291600, L1 = 2592, dummy = 23822, unmodulated = 236

Since the guard interval is shorter, the frame time is less. So we can optimize this new configuration by going to 209 symbols and 21 FEC blocks.

./dvbt2rate 0 2 0 209 21 2 2 0 0 7 1
FFT size = 2048
guard interval = 1/32
number of data symbols = 209
number of FEC blocks = 21
code rate = 3/5
constellation = 16QAM
frame size = normal
carrier mode = normal
pilot pattern = PP7
L1 constellation = QPSK

clock rate = 1845070.422535, TF = 249.503756 ms
Normal mode bitrate = 3249522.227204
High Efficiency mode bitrate = 3266899.351414

max symbols = 218, max blocks = 21
symbols = 217, max blocks = 21
cells = 352944, stream = 340200, L1 = 2592, dummy = 9916, unmodulated = 236

PAPR max symbols = 218, max blocks = 21
symbols = 217, max blocks = 21
cells = 349182, stream = 340200, L1 = 2592, dummy = 6154, unmodulated = 236

You can also use slightly shorter frame times to optimize the number of dummy cells. For example, we can get a little more bitrate at 194 symbols and 20 FEC blocks.

./dvbt2rate 0 2 0 194 20 2 2 0 0 7 1
FFT size = 2048
guard interval = 1/32
number of data symbols = 194
number of FEC blocks = 20
code rate = 3/5
constellation = 16QAM
frame size = normal
carrier mode = normal
pilot pattern = PP7
L1 constellation = QPSK

clock rate = 1845070.422535, TF = 232.333679 ms
Normal mode bitrate = 3323495.767078
High Efficiency mode bitrate = 3341268.471715

max symbols = 218, max blocks = 21
symbols = 202, max blocks = 20
cells = 328254, stream = 324000, L1 = 2592, dummy = 1426, unmodulated = 236

PAPR max symbols = 218, max blocks = 21
symbols = 202, max blocks = 19
cells = 324762, stream = 324000, L1 = 2592, dummy = -2066, unmodulated = 236

If you change the constellation, then you have to change the number of FEC blocks. For example, going to 64QAM would require 30 blocks.

./dvbt2rate 0 2 0 194 30 2 3 0 0 7 1
FFT size = 2048
guard interval = 1/32
number of data symbols = 194
number of FEC blocks = 30
code rate = 3/5
constellation = 64QAM
frame size = normal
carrier mode = normal
pilot pattern = PP7
L1 constellation = QPSK

clock rate = 1845070.422535, TF = 232.333679 ms
Normal mode bitrate = 4985243.650617
High Efficiency mode bitrate = 5011902.707572

max symbols = 218, max blocks = 32
symbols = 202, max blocks = 30
cells = 328254, stream = 324000, L1 = 2592, dummy = 1426, unmodulated = 236

PAPR max symbols = 218, max blocks = 32
symbols = 202, max blocks = 29
cells = 324762, stream = 324000, L1 = 2592, dummy = -2066, unmodulated = 236

I hope this makes sense for you.

Ron W6RZ

I've created a couple of test bitstreams for you. The first one is at the correct bitrate for 1.7 MHz mode.

https://www.w6rz.net/smiling-2798142.ts

The second one is a sanity check for that flow graph. Since you told me you had 8 MHz mode working before, it's at the correct bitrate for 8 MHz mode. So just set the sample rate back to (8000000.0 * 8) / 7 and see if those parameters work.

https://www.w6rz.net/smiling-13865603.ts

-muxrate should work for ffmpeg. For example:

ffmpeg -i test.mp4 -c:v copy -c:a copy -muxrate 2798142 test.ts

Ron W6RZ

При таких параметрах

фрагмент записи trp
https://cloud.mail.ru/public/uFzq/D2zMkxk14