0
\$\begingroup\$

I have a problem - this is my created layout for the HMC414 amplifier.

enter image description here

I'm wondering, if looking at this layout you can see that this circuit is at 50 Hz or at 2.4 GHz? Apparently, you can tell just by the look of the layout if it is at a particular frequency. I don't know how to check it - the amplifier operates at 2.4 GHz, but I don't know if the whole circuit does either.

Hence my second question - is it possible to set this in PADS Layout software (from Siemens). If so, where? I was wondering about the thickness/width of the tracks - but this is probably not it. Could someone who has experience in this topic help, I would be very grateful. Thank you in advance for any reply.

I've now made corrections to the layout, as recommended.

Layout w/o ground plane:

enter image description here

With added ground plane:

enter image description here

\$\endgroup\$
2
  • \$\begingroup\$ You could clean this design up a lot by just rotating a few parts 180 degrees. \$\endgroup\$
    – The Photon
    Jan 27 at 2:12
  • \$\begingroup\$ Greetings! Thank you for comment! As recommended in the replies and comments, I have modified the PCB Layout. I will also edit my post and add in the EDIT section the current look of the Layout. W/o groud plane: i.imgur.com/3bG1tLx.png With added groud plane: i.imgur.com/pc6FlPg.png I also added Vias to the Layout, generated Via Shield and added 5-pin DC Header. Additionally, I changed the position of some components and their rotations. Thanks to this, I got rid of some vias which were too much on some paths (there is one Via left which I can't get rid of). \$\endgroup\$
    – MagicMan
    Jan 28 at 21:46

3 Answers 3

1
\$\begingroup\$

The layout likely works better at 50 Hz as it looks like it has no chance at working at 2.4 GHz. It does not resemble the example layout for the RF amplifier in any way, which would maybe be a good starting point.

The ground planes are also missing, so it is impossible to tell if you have a ground plane or not.

So you don't set a frequency to your layout program, if you need a certain impedance then you should use a tool to calculate the track width based on the PCB parameters.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you so much for the comment! The circuit is not grounded yet, because using a ground plane blocks further changes to the design - so I wanted to do that last. I wanted to deal with the width of the paths first - after moving the components to the board. \$\endgroup\$
    – MagicMan
    Jan 26 at 23:08
1
\$\begingroup\$

Umm, that is vanishingly unlikely to work at RF.

I would guess that you slapped parts down at random and hit auto route? That nearly never gives a viable layout, and NEVER works for a high frequency design.

Impedances are defined by track width and stackup height to the reference plane (And the software probably includes a way to make it handle this), and seriously pretty much all microwave amplifier data sheets include diagrams showing a suggested layout, which I can absolutely guarantee looks nothing like that!

\$\endgroup\$
5
  • \$\begingroup\$ Thank you for the comment. True - I've arranged the parts randomly as I wanted them to be as close to each other as possible (this was my recommendation), but I didn't use auto route - I don't trust this feature too much and I preferred to connect the components myself. \$\endgroup\$
    – MagicMan
    Jan 26 at 23:13
  • \$\begingroup\$ @MagicMan That is not even close to being a dense layout. Also, you do know that you can rotate components in the cad tool right? I mean your input and output DC blocks are insane... The parts that want to be close (where close means short, fat traces) are the supply decoupling for the gain block, and the input and output networks. \$\endgroup\$
    – Dan Mills
    Jan 26 at 23:19
  • 1
    \$\begingroup\$ @MagicMan I just looked at the datasheet, and not only is there a perfectly good example layout, but it also reveals that one of those stupid little tracks of yours is GROUND, rip all of that up work over a ground plane and just drop vias to it as required. It will be much better. Also, note the number of vias on the datasheet example, it means it... \$\endgroup\$
    – Dan Mills
    Jan 26 at 23:26
  • \$\begingroup\$ Greetings! Thank you for comment. As recommended in the replies and comments, I have modified the PCB Layout. I will also edit my post and add in the EDIT section the current look of the Layout. W/o groud plane: i.imgur.com/3bG1tLx.png With added groud plane: i.imgur.com/pc6FlPg.png I also added Vias to the Layout, generated Via Shield and added 5-pin DC Header. Additionally, I changed the position of some components and their rotations. Thanks to this, I got rid of some vias which were too much on some paths (there is one Via left which I can't get rid of). \$\endgroup\$
    – MagicMan
    Jan 28 at 21:46
  • \$\begingroup\$ Objectively - after corrections and different placement of components, Layout looks better (maybe even way better). The question is, whether I did not forget about anything else? \$\endgroup\$
    – MagicMan
    Jan 28 at 21:51
1
\$\begingroup\$

You can't tell just from looking at a layout whether or not a design will work properly at it's intended frequencies, particularly when you're in the GHz or higher frequency range. The only way to do that is to model/simulate the parts of the design (interconnects, RF components) with an appropriate tool such as HFSS, CST Microwave (your tool of choice). Then you need to stitch them together with another tool like ADS and see if it's performance meets your requirements. Even then you may not get it right.

We had a design that needed to work up to 16 GHz. It consisted of several RF chips that were placed in a cavity and wired together with wirebonds and traces. The first model looked good, so we built some prototypes. In testing we found some unexpected resonances across the needed frequency band - not good. Refining the model showed us that the nice rectangular cavity we had chosen supported the resonances we saw, because they were related to the dimensions of the cavity. In order to fix the problem, one of the things we did was to make the nice rectangular cavity pseudo-rectangular by moving the sides around and making the sides irregular (wiggly).

GHz designs is no place for seat-of-the-pants design techniques.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.