"The next step is to determine the theoretical limits of 8-VSB reception followed by implementation of circuitry to meet the criteria as closely as possible. Then I think another test might be in order. It is no longer a comparison between NTSC and DTV, it is now between two conceptually different DTV systems."

I visited the Sinclair COFDM/8-VSB tests in Baltimore yesterday (July 14, 1999).

I was among 11 others that visited that morning. Another demonstration was scheduled for the afternoon. Sinclair is accommodating up to 25 visitors a day and hopes to keep the facility on the air for the rest of July. With any luck they expect to have some HD material for the COFDM signal as they do for 8-VSB.

The transmission facility was reasonably well setup in a spacious and well equipped transmitter building. There is not much documentation but the high quality test equipment indicated both signals were in good form. About 28 dB SNR and 3.1% EVM for the 8-VSB signal - well within normal operating specifications. The COFDM system operated in 2K, 64QAM, 3/4 FEC, 1/8 spacing, 24.88 Mbps gross data rate mode. Useful data rate for 8-VSB data rate was 19.39 and COFDM was 18.66 Mbps. Close enough for useful comparison. Bandwidth occupancy was 5.7 MHz for COFDM and 5.38 for 8-VSB. The wider bandwidth for COFDM forced a compromise on the bandpass filter so it did not meet FCC specs - not needed for this test.

The DTV transmission was on channel 40 (the DTV assignment for Sinclair's channel 54 station) with an ERP of about 50 kW average for both signals using a 30 gain antenna at about 1000 ft above ground. Peak-to-average power was 8.5 dB for COFDM and 6.5 dB for 8-VSB. One of four 60 kW IOT power amplifiers normally used for Sinclair's channel 45 station was used for the DTV tests at an average output power of about 5.5 kW.

The COFDM signal consisted of a 7 second clip of standard definition PAL material with an encoded rate of about 3.5 Mbps followed by bit stuffing in the Rohde & Schwarz modulator for the 18.66 Mbps total. For 8-VSB some HD material was provided from a server. There was no real complaint about using SD on one and HD on the other but it did generate some discussion about what effect the difference could make in reception.

Ed Williams (R) With Nat Ostroff

The receiving setup seemed a bit complicated but there were good reasons for each element of the system all well explained by Mark Aitken, Ray Kiesel and Harvey Arnold, the Sinclair engineers. Two UHF antennas, a single stand alone bow-tie (comes with most TV sets) and double bow-tie with reflector (Radio Shack), were used for the demonstrations. A diagnostic display on one of the COFDM receivers provided indications of the transmit mode of the COFDM system.
As for the two sites visited my experience was along the lines of that reported by Mark Shubin in his earlier report to this group. COFDM is easy to receive and 8-VSB is more difficult. Both can be made to fail, of course. However, I found (Sinclair allowed us to conduct our own tests at each site) that at locations where both 8-VSB and COFDM could be received the margin to threshold was nearly identical although there were significant variations among the two COFDM receivers and the two 8-VSB receivers.

By the way, there was an NTSC receiver available (connected to an MATV system) at the first site (11th floor apartment) but no one in the group expressed an interest in using it for comparison on an antenna although we were told that NTSC pictures were not very good at this location because of the substantial amount of multipath.

Several of the observers (including a receiver mfg) commented that the signal level variations displayed on the spectrum analyzer under both receivable (+/- 5 dB) and non-receivable (+/- 10 dB) conditions for 8-VSB did not appear to be unreasonably distorted. That was my view as well having seen many DTV signals on the DTV Express and having decoded most of them but using different make receivers. So it may be that the difference in reception capability is in the design of the receivers.
My immediate reaction to the demonstration is there clearly is a difference in receivability. We would not have seen that difference without the efforts of Sinclair to mount this impressive demonstration.

The next step is to determine the theoretical limits of 8-VSB reception followed by implementation of circuitry to meet the criteria as closely as possible. Then I think another test might be in order. It is no longer a comparison between NTSC and DTV, it is now between two conceptually different DTV systems.

Ed Williams, PBS
email: ewilliams@pbs.org