K6JRF Auto Page

K6JRF's Page
formerly W6FZC

DSL Operation with DirecTV Sat Receiver
Photos by K6JRF

All pages on this site are best viewed with 1024 x 768 or greater resolution
(Update: Aug 5, 2015)

Click on any image to enlarge

DSL Operation with DirecTV Sat Receiver
DirecTV (DTV) has provided video and audio quality as well as reliable service for the many years. Compared to other HDTV providers, their video quality is the best IMO. Over the years, all satellite rcvrs improve as they do, I upgrade to the newer versions. Recently, an H24 replaced my H23-600 model.

Many today use an Ethernet connection directly into a DVR type rcvr and don't require a phone line connection. Unfortunately, I don't like to record events . . . if I can't watch it live, then I'll miss the program.

DSL Interconnection Diagram All DTV rcvrs can be tied into the phone line so that PPV (movies, special events) can be automatically ordered via the remote control. The phone line serves to provide DTV with the rcvr's location as well record the purchase on the "smart-card". Later, during the month, the rcvr "phones home" upon command to dump the PPV purchases and reset the card.

ADSL and POTS (Plain Old Telephone Service) share the same two copper wires coming into the home. As the sketch shows, the inside telephone wiring carries both POTS and ADSL service. Splitters and filters separate the signals so that there's no interference to the DSL signal, at least in principle.

In my home, the phone line carries both signals. The DSL service consists of 1.5Mbps ADSL which is top speed for my distant location from the Central Office (CO). The measured D/L speed is 1.35Mbps and U/L speed is 0.35Mbps. Not the speediest of DSL service but it works nicely for my needs. Note the "white-box" labeled Z-Blkr Filter. More about it later.

ADSL Full Frequency Spectrum POTS and ADSL service occupy different frequency spectrums. POTS exists in the 0-4 kHz region and ADSL service is present from 26kHz to approximately 1.1MHz.

The "upstream" spectrum lies between 26Khz and 130Khz. The "downstream" spectrum lies between 134Khz and 1.1Mhz. There's a guard band between 4Khz and 26Khz to minimize interference with POTs and upstream data.

The frequency layout can be summarized as:
0-4 Khz, voice,
4-25 Khz, unused guard band,
25-138 Khz, upstream,
130-1104 Khz, downstream.

DSL Interference
DirecTV H24 HD 3D Receiver The newly upgraded H24 rcvr replaced the older H23-600 rcvr. With the phone line connected to the H24, my DSL performance suffered; low speed, lost packets, line stalls so I investigated to find what differences there might be.

The H23 rcvr's modem output showed as 9.5Mohm where the H24 shows 2.5Mohm, a 7Mohm difference! With the H23, there were no DSL operating problems. Unplugging the phone line to the H24 also cleared up the DSL problems. So it appears that the impedance difference is the "problem".

So I made an "impedance expander" (simply add more resistance!) to test. Using an old Radio Shack Ring-Kill product, I modified it by soldering two (2) high value resistors, 3.5Mohm to the PCB. They were in series with the "green" and "red" lines, making the total measured impedance (plugged into the H24) as 10Mohm. My DSL performance returned to normal.

Please note that even though the problem SEEMED to solved, the 2.5Mohm impedance should be large enough to not adversely effect DSL performance, so there must be some other frequency anomalies output from the H24 rcvr that disrupts the DSL signal.

The 3.5M resistors in series with each line, severly attenuate the H24's modem output "tone" level so that the "phone-home" data is lost in the noise! The POTs line is apx 200 ohms impedance so the math shows a large signal drop. Obviously unacceptable!

POTS Filter Design
Since I have a few "Excelsus Z-Blocker DSL Filter" on hand, this would make a perfect item. The filter is designed to attenuate and separate an ADSL signal into the "POTS" voice band, 0 - 4Khz while passing the ADSL signal with no attenuation.

I tried it out and put it inline with the telco line going to the H24 rcvr. The white-box shown in the interconnection diagram is a stock filter labeled "Z-Blkr Filter". I then started an Spice analysis to see what might be improved.

Z-Block Filter Stock Analysis

Using 5Spice, the Z-Blocker filter was analyzed to see how it shapes the POTS frequency spectrum; a LPF with 4Khz as the corner frequency. What the analysis showed is precisely why that filter didn't work.

The "stock" filter analysis shows it doesn't just cover the POTS filter band but extends to 25Khz before attenuating signals.

At 18Khz, there's a peak of 0dB. Not what you expect. Note that at 100Khz, the filter stop band attenuation is only -40db. Compared to the "ideal" filter response (in BLUE), the spurious response between 5Khz and 30Khz shows that the circuit values are no where close to optimum!

Each filter element, L1 - L4 and C1 - C3 were de-soldered from the PCB and measured as shown in the schematic.

The value of each element including inductor's series resistance:
L1 = 4.95mH - Rs = 4.45 ohms
L2 = 4.92mH - Rs = 4.40 ohms
L3 = 4.47mH - Rs = 3.80 ohms
L4 = 4.46mH - Rs = 4.78 ohms
* The series resistance wasn't included in the ckt modeling b/c the values didn't effect the filter's response.

Now, having all circuit elements quantified, the 5Spice schematic was re-drawn into the program's schematic capture and then run to analyze it for improved circuit response.

Z-Block Filter Stock -Reverse- Analysis

Perhaps the filter performs better when the PHONE input is employed. The schematic shows the circuit layout where the DTV H24 rcvr is connected via the PHONE input.

Using 5Spice, this Z-Blocker configuration was analyzed to see how it shapes the POTS frequency spectrum. This analysis showed basically the same response as the previous analysis.

The "REVERSE" filter analysis shows it doesn't just cover the POTS filter band but extends to 30Khz before attenuating signals. At 18Khz, there's a peak of apx 0dB. Note that at 100Khz, the filter stop band attenuation is only -40db. Compared to the "ideal" filter response (in BLUE), the spurious response between 5Khz and 30Khz shows that the circuit values are no where close to optimum!

Z-Block Filter Modified Analysis
After a preliminary analysis, it was apparent that the original C1, C2 and C3 values, 0.01uf were not optimum. A little more showed that 0.2uf ceramic caps were the correct values. Parts (0.02uf, 100V) were purchased from Mouser Electronics here

It also become obvious, that swapping the inductors would materially improve the filter and the filter should PLUG INTO THE RCVR and not into the phone line! The inductors, L1 - L4 were swapped position wise as shown in the picture below.

The H24 modem is "on-hook" until it "phones-home" when the modem's output impedance is apx 200 ohms. The analysis is shows the response when the rcvr is active. The resultant performance curve DOES reflect a POTS response of 0 to 4Khz and approximates the ideal response chart (in BLUE). Also note -34dB attenuation @ 25Khz, the start of the "upstream" band.

Summary

The modified Z-Block Filter uses the "pig-tail" to plug into the sat rcvr as the picture shows. The PHONE jack takes an extension to plug into the normal wall phone jack. This is not the normal way it is used!

The stock Z-Block Filter design uses the guard band (4 - 25Khz) but IMO, it is best to NOT to design the POTS filter to use that band. Using it does NOT allow the ultimate attenuation in the "upstream" band to be achieved.

Please note that at the TERMINATOR, there is a ADSL + POTS filter installed to isolate the DSL signals from this special phone line to the DTV rcvr.

Over the last week, DSL performance has been superb, so the modified Z-Blocker filter is the best solution solving both the impedance loss and DSL line isolation at the DTV rcvr.

Send me with your comments