Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Impedance/Length Matching Tracks in PCB Layout

Status
Not open for further replies.

Nick Arsi

Newbie level 5
Joined
Nov 22, 2013
Messages
8
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
60
Dear All,
Kindly guide me which criteria should follow to do length and impedance matching in PCB layout?
Right now I have a design with different type of flip flops, Do I have to length match of data pins of Flip Flops or just rout them as normal signals?
In addition to my above query at which frequency range do we need to length match in layout?

Thanks
 

ArticCynda

Advanced Member level 1
Joined
May 8, 2013
Messages
424
Helped
82
Reputation
164
Reaction score
81
Trophy points
1,308
Activity points
4,047
Technically you need length matching all all frequencies, except for low frequencies the acceptable tolerance is much larger than the PCB size so it's not usually an issue to take into account. So whether you'll have to match the trace lengths only depends on the frequency of your signals.

Matching impedances is done to maximize signal transfer and to avoid reflections from the destination back to the source.
 

SunnySkyguy

Advanced Member level 5
Joined
Sep 26, 2007
Messages
6,744
Helped
1,675
Reputation
3,348
Reaction score
1,644
Trophy points
1,413
Location
Richmond Hill, ON, Canada
Activity points
50,737
If the delay time on the track length is more than 1/3 to 1/6 of the rise time, then you can expect overshoot from mismatched transmission line impedance and load. The track width to height (w/h) above ground plane dominates complex math for low impedance. so lower h or wider w reduces impedance. CMOS technology determines ESR or RdsOn of drivers, which are typically 50 Ohms to 25 Ohms for ALV type 2.

PECL always uses controlled impedances due to rise time.

Vp = c/√(μ*εr) depending on dielectric Vp= 1/2 to 2/3 *c typ.
where c= 3e8m/s or Vp = 1.5 to 2 e8 [m/s] or 150 to 200 mm/ns

Using a 3x rule of thumb , this reduces ... 50 to 66 mm/ns of rise time.
USing a 6x RUle of thumb, this reduces ... 25 to 33 mm per ns of rise time on your Flip Flop outputs.

The reason for concern on FF is for reasons of lowest latency the outputs are internally fed back, so any transient spikes or ripple may cause false operation and should be buffered, which can be controlled with load and/or transmission line impedances and if not, buffered for external signals that may get crosstalk.

So with slow Rise/fall time, the tolerance on length is wide.


This is from my quick calculations.... not old memories.
 

binlch.lv

Junior Member level 1
Joined
Jul 2, 2015
Messages
15
Helped
2
Reputation
4
Reaction score
2
Trophy points
3
Activity points
89
SunnySkyguy's answer is good.
I simplified the question.
1. length matching .make sure your setup and hold time margin greater than pcb delay,pcb delay 1ns/6inch.
2.impedance matching .make sure your pcb line impedance and load is the same value.pcb line setup you can use si9000.
 

FvM

Super Moderator
Staff member
Joined
Jan 22, 2008
Messages
48,430
Helped
14,252
Reputation
28,765
Reaction score
12,944
Trophy points
1,393
Location
Bochum, Germany
Activity points
280,238
You are asking about digital logic PCB. Impedance matching makes only sense if you are using an I/O standard with termination. Is it so?

Length matching is useless unless the basic signal travel time is somehow compensated in the link. Examples

1. DDR2 expects all data bits arriving at the same time because it utilizes a common delay compensation
2. A source synchronous high speed transmission with spearate clock and data. The relation between clock and data has to be maintained
 

SunnySkyguy

Advanced Member level 5
Joined
Sep 26, 2007
Messages
6,744
Helped
1,675
Reputation
3,348
Reaction score
1,644
Trophy points
1,413
Location
Richmond Hill, ON, Canada
Activity points
50,737
If there are forced timing conditions in logic where you expect data to be stable at the clock edge but somehow latency from unequal logic thresholds or track length delay the clock, it may cause a race condition.

But it's all in the details for worst case timing of the design.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top