LMH0387: Extended 3G Reach Mode Using 75 ohm RG179 cable and LMH0387 on SD387EVK Dev Board Kit

I would like to add a few notes on our recent testing:

We generated our own CLI graph for our RG179 cable. The graph is shown below. It should be noted that this graph was not created with single cables, we had to piece together segments based on cables we had built up in our lab. The table for the number of breaks is shown below too. In all, we would see a change in the CLI bit value about every 8' cable increase.

 

Actual Cable Length (ft)	Breaks	CLI (HEX)	CLI (DEC)
1	0	8	8
4	1	8	8
11	0	10	16
14	1	10	16
15.5	1	10	16
17	1	18	24
21.33	2	18	24
28.08333333	2	20	32
28.5	2	20	32
33.16666667	2	28	40
36.5	2	30	48
43.5	2	30	48
47.66666667	2	38	56
51.5	2	38	56
57	2	40	64
65	2	48	72
68	2	50	80
74.5	2	58	88
79.5	3	58	88
90.66666667	3	60	96
94.5	3	68	104
98.66666667	3	68	104
105.0833333	3	70	112
113.0833333	3	78	120
125.4166667	3	80	128
134.8333333	4	88	136
140.3333333	4	88	136
147	2	90	144
151.3333333	4	90	144
155	2	90	144
162.75	2	98	152
173.75	2	A0	160
180.5	3	A8	168
190	3	A8	168
205	3	B0	176
209.1666667	3	B8	184
218.5	3	B8	184
223.25	3	B8	184
245.75	4	C8	200
264.9166667	4	C8	200

We used the Xilinx IP iBERT demo to measure Bit Error Rate and determined our max length would be when we started to see errors. The iBERT demo continuously sends bits with a Pseudo-Random (PR) pattern, tests to make sure the link speed is at least 1.25 Gbps, and measures the number of bit errors and BER. We set up iBERT on a demo FPGA board and ran a loopback on that board and attached various cable lengths running to/from the SD387 line driver dev boards. With the iBERT testing, we were able to send 1e12 bits with a PR 31-bit pattern over 161' of the RG179 cable and receive 0 errors. Our 161' test cable was setup as follows: 3.08' | 12.5' | 41.66' | 37.5' | 37.5' |15.4' | 14' (each "|" represents a connector break).

We then tested 164' setup as follows: 3.08' | 18.75' | 41.66' | 41.66' | 22.5' | 22.5' | 14' and sent 1e12 bits again with the PR 31-bit pattern. We received 1.5e1 errors and said our limit was around 161' without reconfiguring the line drivers for extended 3G reach. We tried running this setup again, after enabling extended 3G reach mode, and we still received errors over this 164' setup.

To test sending Ethernet over the RG179 coax cable, we ran a similar loopback test on an FPGA board and ran the RG179 cables out from the FPGA board to the SD387 Line Driver dev boards, then SMA cables connected our TX/RX lines from the SD387 boards to the DP83867E-SGMII-EVM dev board, and the Ethernet from the DP83867E board ran to a PC. From the PC, we used a simple network Ping test with the ip address of the FPGA board and noticed successful pings with a 154' RG179 setup as follows: 3.08' | 12.5' | 41.66' | 37.5' | 22.5' | 2' | 20.83' | 14'. Once we added a couple more feet to this run, the Ping tests failed and kept timing out. The setup for the 156' run was 3.08' | 18.75' | 41.66' | 37.5' | 22.5' | 19.16' | 14'. We tried reconfiguring the line drivers to enable 3G reach mode again, and we were still unsuccessful with our Ping test at the 156' run.

The example setup of the dev boards and cabling is shown below.

Are we at our limit for extended 3G reach mode being used, or should we be able to use extended 3G reach mode at this length with our RG179 cable?

Hi Nasser,

Answers to your questions are listed below.

1) We have already incorporated the LMH0387 driver on a couple of designs. We are conducting this additional testing for a follow-on device that would connect to a board that already has the LMH0387. We are looking at the possibility of testing the LMH0397 chip as well, but if we were to use the 0397 device, the previously developed boards would have to be changed. Are there any other improved devices we should look into testing?

2) When we were conducting the iBERT tests, the speeds were ~1.28 Gbps. We aren't tied to any specific rate for our SGMII signals coming out the Line driver, but we would like to maintain the Gb Ethernet speeds out of the PHY.

3) We are using PN E132276-A Cat 6 Patch cable. This cable is rated for 550 MHz/1 Gbps communications.

Thanks,
Charlie

Nasser,

LMH0342 is not showing up on the TI website. Is this a discontinued product?

Out of curiosity, what would happen if we had an LMH0387 line driver on one end and the LMH0397 on the other end? Would they be able to support communication to/from eachother?

Thanks,
Charlie

Nasser,

1) Would I be able to run further lengths if only one end were the better chip (either LMH0397 or LMH0324), or would I need both ends to be either LMH0397 devices or LMH0324 devices?

2) For the LMH0397, are we tied to using 1.485G, or are you just saying that's where the best performance is?

Thanks,
Charlie