Saturday, March 12, 2005

Hyperlink HGLN-DSL surge protection

My DSL service is with SBC (formerly SNET). I seem to live at an electrically "long" distance from the Branford, CT central office. At any rate, I had quite a bit of trouble getting my modem to sync when the service was originally installed. Even now, while the service has been very stable, I am "capped" at 384 kbs download and upload.

[Note, 7/2008: I got upgraded not long after this. I now have AT&T "Elite" DSL, which is 6 Mb/s down and 768 kb/s up.]

As part of my station grounding project, I wanted to include my computer network in the "safe" area protected by a single point ground (SPG) and appropriate lightning/surge protectors. Devices for DSL circuits shared with POTS telephone ringing currents are not very easy to find. I decided on a Model HGLN-DSL-1S from HyperLink Technologies, available at

The unit appears well constructed, but when inserted in my DSL line, my modem immediately dropped sync and would not come back. There is a problem, at least with my rather weak DSL connection.

My modem, an Efficient Networks Model 5260, reports the following line parameters when the filter is not in the circuit:

Current SNR Margin 8.5 dB
Current Attenuation 66.0 dB
Current Output Power 12.0 dB

The modem also claims a maximum supportable transfer rate of 640 kbs with this SNR. This apparently is bad enough that SBC gives me a 384 kbs rate cap.

I tried a lot of things to get DSL back. The most significant was to replace my phone jack connection temporarily with 250 ft of Cat 5e cable -- using only one twisted pair of the four available. I didn't want to cut my 250 ft spool of cable, but the real run length from my telco entrance to the shack is more like 60 ft. Without the DSL filter, I now see the following modem report:

Current SNR Margin 10.5 dB
Current Attenuation 66.5 dB
Current Output Power 12.0 dB

Two dB better SNR is very good. That corresponds to a 711 kbs line limit, according to the modem. However, with the filter, the modem would still not lock up.

Time to take a look at the schematic: (Click for larger view.)

I was able to identify the components via Google. The Spark Gap SG is a type 3R-230 from Beijing Tehy. The spec says it has a trigger voltage of 230 V +/- 20% and an inter-electrode capacitance of <5 pF. The resistors are 3.3 ohms, 1 watt (?). The transient absorber is not actually as drawn. It is a composite of three devices, thus:

The transient suppressor seems to be similar to a Bourns CD214B "ER" type. The Bourns part has a breakdown voltage of 189 - 218 V and a junction capacitance of 2000 pF at 5 V standoff running down to 70 pF at 200 V. At typical phone line bias, 50 V, we should have 200 pF. The two cross-connected diodes seem to be similar to Diodes Inc. type FR1J, 600 PIV.

The short story is that the diode suppressors load my marginal DSL line too much. Unsoldering one end of the ER device allows the modem to sync up. I haven't run full tests to see if it is as reliable as without the filter, but there seems to be almost nothing left in the circuit that could plausibly degrade the signal.

Why does the surge suppressor cause so much trouble? The capacitance and resistance seem immaterial. Perhaps it is the non-linear characteristic with voltage that may add too much distortion.

So, will we have enough protection without the second stage of suppression? The modem faces a greater risk of being fried in an impulse event, but I expect that most of the "downstream" computer gear is protected well enough. One never really knows without full-scale and expensive tests.

No comments: