Author Topic: AC Coupling Question  (Read 4161 times)

0 Members and 1 Guest are viewing this topic.

Offline jmsiglerTopic starter

  • Regular Contributor
  • *
  • Posts: 57
  • Country: us
AC Coupling Question
« on: August 06, 2016, 05:13:37 pm »
Hello, If i am ac coupling a signal output from a clock gen ic say, do I need a 50ohm resistor right after the capacitor? My thought is that I need to present a 50ohm source impedance? Same goes for when ac coupling into an input.
 

Offline voltz

  • Frequent Contributor
  • **
  • Posts: 267
  • Country: gb
Re: AC Coupling Question
« Reply #1 on: August 06, 2016, 05:39:19 pm »
Assuming the output impedance from the clock generator is low impedance (like less than a few ohms) then yes, you could place a 50 ohm resistor in series with any ac de-coupling capacitor to give you a 50 ohm output (source) impedance. The capacitor can basically be considered a short circuit at AC.
 

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21675
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: AC Coupling Question
« Reply #2 on: August 06, 2016, 09:11:16 pm »
An average CMOS logic-level clock generator will have an output impedance comparable to 50 ohms. Don't expect precise source termination, even if you add extra resistance.  Better to use an attenuator (if you can tolerate the loss), or load termination (if the source can put out that much power without overheating).

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline jmsiglerTopic starter

  • Regular Contributor
  • *
  • Posts: 57
  • Country: us
Re: AC Coupling Question
« Reply #3 on: August 07, 2016, 03:27:15 am »
Okay, that is what I was looking for. The output impedance is listed at 45ohms suggesting I need a 5ohm resistor.

Of course if the line is less than a fraction of the transition time long then it isn't really a transmission line and you don't have to worry about impedance matching for SI so much.
As a rule of thumb, how do you determine when a trace will begin to act as a transmission line for both sin/square waves? I've seen .5/rise time as an approximation for knee frequency for square waves and can find wavelength from that, but I'm not sure how to relate the wavelength to where impedance matching becomes important.
 

Offline voltz

  • Frequent Contributor
  • **
  • Posts: 267
  • Country: gb
Re: AC Coupling Question
« Reply #4 on: August 07, 2016, 08:22:04 am »
Okay, that is what I was looking for. The output impedance is listed at 45ohms suggesting I need a 5ohm resistor.

Correct. May not even be necessary at all depending on what tolerance you need.

What frequency is the clock generator producing? is the rise time very high? PCB track design only matters if you're into 20+ Mhz rather than Khz or less. (roughly). Does it matter in this application?
« Last Edit: August 07, 2016, 08:23:57 am by voltz »
 

Offline jmsiglerTopic starter

  • Regular Contributor
  • *
  • Posts: 57
  • Country: us
Re: AC Coupling Question
« Reply #5 on: August 07, 2016, 07:59:07 pm »
Okay, that is what I was looking for. The output impedance is listed at 45ohms suggesting I need a 5ohm resistor.

Correct. May not even be necessary at all depending on what tolerance you need.

What frequency is the clock generator producing? is the rise time very high? PCB track design only matters if you're into 20+ Mhz rather than Khz or less. (roughly). Does it matter in this application?

The clock gen its self is producing 12.8MHz and is feeding a PLL (LMX2531) that is producing 1.57GHz. I am currently ac terminating the pll with 100nF on the input and 100pF on the output, which was suggested by the app note.
 

Offline voltz

  • Frequent Contributor
  • **
  • Posts: 267
  • Country: gb
Re: AC Coupling Question
« Reply #6 on: August 07, 2016, 08:09:00 pm »
Quote
The clock gen its self is producing 12.8MHz and is feeding a PLL (LMX2531) that is producing 1.57GHz. I am currently ac terminating the pll with 100nF on the input and 100pF on the output, which was suggested by the app note.

ok, so the PCB tracks between the clock gen (12.8Mhz) and the PLL input are not that critical. But keeping tracks as short as possible is always good practice anyway. However your 1.57Ghz PLL output very much IS critical in the PCB design. And good knowledge of strip line design techniques will be needed if you want to keep your 50 ohm output impedance from the PLL stage correct to the following stage. presumably a mixer of some kind?..

Hope this helps.
 

Offline jmsiglerTopic starter

  • Regular Contributor
  • *
  • Posts: 57
  • Country: us
Re: AC Coupling Question
« Reply #7 on: August 08, 2016, 05:14:21 am »

And good knowledge of strip line design techniques will be needed if you want to keep your 50 ohm output impedance from the PLL stage correct to the following stage. presumably a mixer of some kind?..

Correct, its feeding a mixer on a different board though a short length of sma cable. I used saturn's pcb calculator and got about 14.5mil trace for a 4 layer osh park FR408 board with gnd plane on the second layer.

Besides trace impedance and length, are there any other things I should look out for at higher speeds? For example I've seen you want to use "curvy" traces rather than straight lines with 45deg angles.
« Last Edit: August 08, 2016, 09:42:22 pm by jmsigler »
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 19493
  • Country: gb
  • Numbers, not adjectives
    • Having fun doing more, with less
Re: AC Coupling Question
« Reply #8 on: August 08, 2016, 08:02:28 am »
Okay, that is what I was looking for. The output impedance is listed at 45ohms suggesting I need a 5ohm resistor.

Of course if the line is less than a fraction of the transition time long then it isn't really a transmission line and you don't have to worry about impedance matching for SI so much.
As a rule of thumb, how do you determine when a trace will begin to act as a transmission line for both sin/square waves? I've seen .5/rise time as an approximation for knee frequency for square waves and can find wavelength from that, but I'm not sure how to relate the wavelength to where impedance matching becomes important.

http://www.edn.com/collections/4435129/Bogatin-s-Rules-of-Thumb

Start at number 0, and then read whichever seem relevant; then read the others.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 
The following users thanked this post: voltz

Offline voltz

  • Frequent Contributor
  • **
  • Posts: 267
  • Country: gb
Re: AC Coupling Question
« Reply #9 on: August 08, 2016, 09:22:38 am »
@tggzzz , great series of rule of thumbs there. Very helpful.  :-+
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf