11

I am designing a PCB for LNA with an SPF5189z. I found a reference schematic design in the SPF5189z datasheet:

enter image description here

(Image excerpted from the previously linked datasheet.)

What do they mean by those angles of 8.4° and 13.3°, etc? I know how to design a 50 Ω trace on a PCB and what the track thickness should be, but what does the degree given in the image mean?

Andrew Morton
  • 2,389
  • 1
  • 17
  • 26
Asif Rahman
  • 223
  • 1
  • 6
  • 1
    It looks like the data sheet for that is no longer available on the manufacturer's website. Could you please link to a **live** datasheet, or give detail on the test board, most importantly what design frequency they used. – TimWescott Apr 02 '23 at 19:47
  • You are designing a PCB with a discontinued part? – StainlessSteelRat Apr 03 '23 at 14:41

2 Answers2

18

It means that the section of \$50\Omega\$ line pictured should cause the specified \$13.3^\circ\$ at the design frequency of the board. Because different board materials have different velocity factors, they give this to you in degrees rather than mm or inches.

The reason they're calling out specific lengths is because all of the mentioned lump impedances are significant enough to modify the circuit impedance; thus, they, and the lengths of transmission line, become part of the board's matching circuits.

TimWescott
  • 44,867
  • 1
  • 41
  • 104
  • Thanks Tim. So I am using fr4. Not certain how to convert 13.3 degree to length in mm to calculate length of transmission line. – Asif Rahman Apr 02 '23 at 21:32
  • 1
    @AsifRahman It depends on more than just board material, but also transmission line geometry. "FR-4" isn't enough information to know what the material in question is, though--check the actual parameters of the specific FR-4 you're using. Cheap PCB fabs might not specify because they use whatever FR-4 they can get; you may need to pay extra for a specific material. – Hearth Apr 02 '23 at 22:01
  • Lets say i am using rogers 4350b. 50 ohm Impedance can be matched based on copper depth of board and other info. So whats the 13.3 degree? How it is related to length of trasmission line? – Asif Rahman Apr 02 '23 at 22:08
  • 1
    Do you mean how it's related to transmission line length _in general_, or _on a PCB_. _In general_, 360 degrees of delay is one wavelength at your design frequency. If that still doesn't make sense, perhaps you need to hit a book on transmission line theory, or do a search, or ask a _separate_ question here. – TimWescott Apr 02 '23 at 22:22
  • I agree that in this schematic it means a specific length of 50 ohm transmission line. Interestingly the same notation is used for completely different purpose: to represent a complex impedance, e.g. [this](https://www.allaboutcircuits.com/uploads/articles/component-values-replaced-by-impedances.png). – jpa Apr 03 '23 at 10:06
  • 1
    @AsifRahman this should be a good starting point: 360 degrees is one wavelength – user253751 Apr 03 '23 at 11:31
  • @jpa Both are just arguments of certain complex number. In one case, the S21 parameter of the transmission line section, in the other the impedance. – krollspell Apr 03 '23 at 12:36
  • 2
    @TimWescott Would it be clearer to say "cause the specified 13.3 degree **phase shift** at the design frequency" ? – jonathanjo Apr 03 '23 at 13:03
  • I guess I don't see the usefulness of this without a frequency definition. And it's totally useless for a wide band system. – SteveSh Apr 03 '23 at 14:21
  • 1
    @SteveSh "electrical lengths refer to 2GHz" is *just below* where that screenshot was cropped. But they have another example for 900MHz, where the TL angles are proportionally shorter. Which does lend credence to the idea that they're not at all critical, and they're just included for the sake of "this was the setup our graphs refer to". – hobbs Apr 03 '23 at 14:34
  • 6
    I thinks the answer is missing the proper terminology. The parameter is called **electrical length** or **phase shift**. See https://en.wikipedia.org/wiki/Electrical_length – pabouk - Ukraine stay strong Apr 03 '23 at 15:26
3

In the design provided in the datasheet the electrical length of the transmission line sections are not critical: the lines are 50 ohm and the electrical lengths are rather short (everything below 22.5º or even 45º is almost always negligible). My advice: just keep them short, like the ones in the layout above the schematic, and do not worry too much.

krollspell
  • 131
  • 2
  • This sounds logical to me, but could you explain what would be detrimental when these three transmission lines were very long? And why do the schematics for 900 MHz and 1900 MHz recommend different lengths then ? – tobalt Apr 03 '23 at 13:03
  • A fixed length (think time delay) translates to different phase shifts for different frequencies. From eyeballing the phase shifts for the 900 MHz and 1900 MHz designs, it looks like the lengths are the same. – SteveSh Apr 03 '23 at 14:52
  • These are electrical lengths. Furthermore, they are describing the evaluation board at 900MHz and 2GHz per the notes (it is the same board in both cases, although for 900MHz, they cut a trace and add an inductor on the input side). If you are running narrowband near one of these frequencies, then you can perhaps use these values as a reference for your design. If you are running at a different frequency, you may want to use the eval board and a VNA to see how the match holds up. – Troutdog Apr 03 '23 at 14:58
  • @tobalt Electrical length is essential when working with transmission lines, since standing waves of voltage and current appear. For example, 90º is the electrical length between a maximum and a minimum of the standing wave. As a result, A 90º section of transmission line makes a short circuit behave as an open circuit (and vice versa). Chose poorly your length, and no power will flow from one side of your line to the other side. Chose it wisely, and you will be able to design circuits with complex behavior, like filters, just with traces and no additional components. – krollspell Apr 03 '23 at 17:00
  • @krollspell Thanks. I see now, that these lines are "unterminated". Now the standing wave pattern and length make sense. – tobalt Apr 03 '23 at 17:36