Measurements of Pre-Emphasis on Altera® Stratix® GX with the BERTScope 12500A

Author: Guy Foster, Director of Marketing, SyntheSys Research, Inc.

June 2005, ver. 1.0 1

CP-STGX05-01

Abstract

This paper gives a brief overview of signal integrity measurements made with the Stratix GX device from Altera. It gives quantifiable measurement of the effectiveness of the pre-emphasis function that is available on the Stratix GX family. It employs Eye Diagram, Jitter and BER Contour functions of the BSA 12500A to reveal the part’s performance.

Features of the Arrow/Altera Stratix GX Evaluation Board

The Altera part may be evaluated on several different boards. The one used here, from Arrow Electronics, provided the following features:

4 electrical transmitters, 4 electrical receivers Ability to run at 2.50 or 3.125 Gb/s Internal clocking, or facility to use an external clock at ÷16 or ÷20 of the data rate. Transmitter variable output amplitude from 400 – 1600mV in increments Pre-emphasis and equalization can be added in increments (0 to 5) Several test lengths of FR4 to act as channels to demonstrate the effectiveness of pre-emphasis. PRBS patterns PN7 to PN23 and simple predetermined RAM patterns. Easily installed USB control software with straight forward GUI.

In this paper, the measurements were made using a PN7 pattern at 2.5 Gb/s. The evaluation board is pictured in Figure 1.

Measurements of Pre-Emphasis on Altera® Stratix® GX with the BERTScope 12500A Altera Corporation

2

Figure 1: The Arrow/Altera Evaluation Board Used. The Stratix GX device is visible in the center, the external clock input has coax connected on the left. Two data output cables are visible top right. The SMA connectors give access to other inputs and outputs, some through different lengths of FR4.

Test Set Up

Schematically the set up is shown in Figure 2, and pictured in Figure 3. The BERTScope pattern generator supplied a RAM pattern mimicking a clock divided by 16. The device under test (DUT) was configured to give a PN7 pattern (See Figure 4 for this and other settings of the GUI).

Figure 2: Schematic of the test set up, showing the configuration of the BERTScope giving out a data pattern that acts as a ÷16 clock signal. (PG = Pattern Generator, ED = Error Detector)

Altera Corporation Measurements of Pre-Emphasis on Altera® Stratix® GX with the BERTScope 12500A

3

Figure 3: The physical test setup showing the evaluation board and BERTScope 12500A with looped full rate clock.

Figure 4: PC-based GUI showing configuration settings.

Measurements of Pre-Emphasis on Altera® Stratix® GX with the BERTScope 12500A Altera Corporation

4

The evaluation board was measured in two configurations – with the Stratix transmitter having no channel (short path to the output SMA connectors), and with 50” of FR4 on-board path. This allowed an easy ability to see the effect of the pre-emphasis with and without the channel dispersive effects it is designed to counter-act.

Results

The effects of pre-emphasis can clearly be seen in Figure 5. The top row of BERTScope eye diagrams shows the 6 steps of pre-emphasis (from 0, none applied, to 5, full amount applied). On the screenshots with no channel applied, the eye deviates more and more from the conventional NRZ eye diagram as more and more signal is added to the first in a sequence of one or more identical bits before returning to the more normal level for any remaining identical bits. This all makes more sense once the channel is inserted (lower row of eye diagrams). With little or no signal conditioning, the channel causes frequency limiting (low pass filtering) which results in a dispersive closure of the eye as can be seen on the left of the sequence. As will be seen later, the link was not capable of error-free operation without signal conditioning. Increasing pre-emphasis had a radical effect on the eye, as can easily be seen lower right of figure 5.

Figure 6 shows jitter results measured with the BERTScope. The BERTScope measures jitter down to specified BER levels according to the Dual Dirac methodology described in MJSQ. As can be seen from the graph, with no channel in place, we are looking at the jitter of the top line of Figure 6. The results are given as percentage of the unit interval (UI), and although the eye starts to look different with increasing pre-emphasis, the jitter shows little effect. Once again, the picture is very different after the FR4, i.e. the lower row of eyes in Figure 5. The jitter is un-measurable with no signal conditioning as the link will not operate down to 1x10-12 BER. It is not until 2 of the 5 steps of pre-emphasis have been applied that we get TJ below 1 UI. As with the eye diagrams, the improvement in jitter is dramatic, until with the full amount switched in the residual jitter is approaching the value apparent with no channel.

The most revealing view of the performance available is given by BER Contour. This is given in Figure 7, with the same layout as Figure 5. Here the outer most contour equates with 1x10-6, and the inner (white) contour with an estimated 1x10-16 BER operating point. Inside this white contour, the link can be assumed to run error free parametrically. Looking at the lower row where the 50” of FR4 is part of the link, it is easily seen that the eye opening increases from closed through to a relatively healthy picture where the white contour bounds an area which would be straight forward to place a receiver decision point and achieve error-free operation.

Altera Corporation Measurements of Pre-Emphasis on Altera® Stratix® GX with the BERTScope 12500A

5

Figure 5: The effect of pre-emphasis, before and after 50” of FR4.

Figure 6: Using Jitter Peak on the BERTScope, the graph shows measurement of the total jitter (TJ) at 1x10-12 BER level, with and without the 50” FR4 channel, and with varying degrees of pre-emphasis applied.

Measurements of Pre-Emphasis on Altera® Stratix® GX with the BERTScope 12500A Altera Corporation

6

Figure 7: BERTScope BER Contour view of the eye opening.

Conclusions

The Stratix GX pre-emphasis was very effective opening the eye up to be error free, and significantly reducing the jitter, even after 50” FR4. The measurements were made at 2.5 Gb/s using a PN7 pattern, 800 mVp-p.

The BERTScope BER Contour easily showed the area within the eye where error-free operation was achievable and provided unique insight into the performance of the applied pre-emphasis.

Altera Corporation Measurements of Pre-Emphasis on Altera® Stratix® GX with the BERTScope 12500A

7

Acknowledgements

SyntheSys Research would like to thank engineers at Altera Corporation for their assistance with these measurements. Further information on the Stratix GX device can be found at http://www.altera.com/products/devices/stratixgx/sgx-index.jsp. Stratix GX Evaluation Boards may be purchased from Arrow Electronics: http://www.arrownac.com.

References

MJSQ - Methodologies for Jitter and Signal Quality Specification is a document written as part of the INCITS project T11.2. http://www.t11.org/index.htm.

101 Innovation Drive San Jose, CA 95134 (408) 544-7000 www.altera.com

Copyright © 2005 Altera Corporation. All rights reserved. Altera, The Programmable Solutions Company, the stylized Altera logo, specific device designations, and all other words and logos that are identified as trademarks and/or service marks are, unless noted otherwise, the trademarks and service marks of Altera Corporation in the U.S. and other countries.* All other product or service names are the property of their respective holders. Altera products are protected under numerous U.S. and foreign patents and pending applications, maskwork rights, and copyrights. Altera warrants performance of its semiconductor products to current specifications in accordance with Altera’s standard warranty, but reserves the right to make changes to any products and services at any time without notice. Altera assumes no responsibility or liability arising out of the application or use of any information, product, or service described herein except as expressly agreed to in writing by Altera Corporation. Altera customers are advised to obtain the latest version of device specifications before relying on any published information and before placing orders for products or services. Copyright © 2005 SyntheSys Research, Inc. SyntheSys is a registered trademark of SyntheSys Research, Inc. All rights reserved."