I always appreciate how clearly you explain your PCB layout choices, Zach. Since we know you're an analog guy, would you consider doing a video on best practices for layout of an AF power amplifier? As I understand it, output currents become a major consideration at low frequencies.
I love your videos. Many thanks! One thing to be aware of is that quickly dragging schematics and board layouts back and forth can trigger motion sickness.
Really enjoy watching your videos and really thankful for the efforts you have put in so far in producing so many useful and technically enriching videos. I am an RF guy essentially, but your videos have certainly went a long way in helping me learn and obtain practical insights on few other arenas like pcb design, digital section, power components, etc. As an engineer and a learner, I always look up to you as one of my teachers (apologies if its sounding too childish😄)
Hello @Zachariah Peterson, do you recommend, in this case, to remove the solder mask on the RF tracks? If yes, should I remove the solder mask from the GND as well? Thank you!
Yes to remove some additional losses along the interconnect you can also remove the solder mask along edge of the GND region closest to the signal line. There is no need to remove all of the solder mask everywhere, but it is still interesting that I see some designs where they remove ALL of the solder mask from the entire GND polygon around the signal line, that is not necessary. If you wanted to do it in this example design you could just span the solder mask opening to the far edges of the via array.
Great video Zach! When is it appropriate to use a coplanar waveguide in a pcb compared to just using an impedance matched microstrip? For example between a GNSS receiver and an SMA antenna connector, what would you recommend and why?
The main reason you would want to use coplanar waveguide is because you get high shielding effectiveness even when using a thicker substrate or dielectric layer. For example, I could have put a ground cutout through the internal layers and have ground pour on the bottom layer, and this would require using a wider line. This would give me lower loss, but when I put the pour on the top layer with the stitching vias to form the coplanar waveguide I could still get a reasonable trace width (about 30 Ohms). The shielding effectiveness I would expect would be comparable to having a narrow microstrip over ground on a thin dielectric. In this project I went with the thinner microstrip because the operating frequency is lower, and because we need a thinner line in order to route into the output of the power amplifier. I could have had this in sections too, and then we can look at the S-parameters across the entire interconnect.
Amazing video as always. Btw @16:08 what hotkey did you use to quickly switch directly from the Top layer into the Bottom layer? I don't think you use Ctrl+Shift scroll hotkey for that.
You can use the * key to cycle through signal layers from top to bottom. You can also use Shift + * to cycle backwards through layers. One thing to note is that * and Shift + * will skip plane layers.
You should look at reference designs from power amplifier IC manufacturers. Analog devices has released a lot of their evaluation board files as reference designs that you can use to get started. Try looking their for help with the circuitry and an example PCB layout.
Last week I wrote a long review with mistakes done in this video/design. But it never got published. This is really sad. Either it was flagged as spam or altium or @Zachariah Peterson doesn't want to deal with professional critical reviews. Cheers Mel
I don't flag stuff as spam. This is not a hobby for me, I run a design firm and I get to deal with plenty of professional critical reviews. Portions of this design were based on a system that was fabbed, tested, and is deployed in the field. There were some variations of the design that would still need to be checked/simulated, like the via fence on the CPW going to the output. And of course we would want to get an actual stackup from a fab house but we wanted to show something just as an example. If your review did not appear then I apologize but I am not the one who flagged or deleted anything.
Hi, If the edge connector pads are moved to the edge as shown in video, does it not result in exposed copper pads during manufacturing of boards? Could you please let me know, if there is any workaround by avoiding exposed copper along edges? Thanks for the video, It was really helpful.
I'm not sure I understand, those pads will always be exposed because they are surface mount pads, it does not matter if they are pulled away from the edge. If you're worried about plating along the edge, then yes the through-hole plating process can deposit copper across that gap if the pads run right up to the edge of the board and the plating throwing strength is correct. If you do not specify edge plating options in your PCB fab notes, but you place the pads along the edge anyways (maybe on accident), your fabrication house should warn you that this might happen, or they may apply a pullback around the board edge. Finally, even if there was some plating across the top and bottom pads on that SMA edge connector, those pads are connected to the same net anyways so that plating does not hurt anything.
In regard to the Stack-up material. Most RF PCBs use CORE on the Outer Layers? NOT Pre-Preg.......... Due to Pre-Preg "press-out" thickness Fab issues. (I think)
Yes the stack can also be inverted. Not required but if you cannot tolerate any variations along the outer layer then consider using core on the outer layer. I've also had to do that when the outer layer was intentionally much thicker so that the RF line could be made wider and coplanar routing was used, then it would reduce the conduction loss. But yes press-out can be an issue with prepreg on an outer layer, gotta verify that stackup from a fab house first.
Hello, i enjoyed watching your video. the rf amplifier that i need has to be in a frequency rang 900mhz to 5Ghz . can i get it done using thisdesign? if yes do you have any suggestions on component change?
I do not know if you can do a direct swap for a power amplifier with different frequency range. There are many different HMC components, so I'm sure you can find something in the microwave range. The HMC components use a variety of different biasing and startup procedures, so because of that you most likely will not be able to do direct swap, even if the packages and pinouts match.
Which variation of FR4 are you referring to? FR4 is a category of materials specifically referring to flammability rating, it is not one specific material. There are many different versions of it, including low-loss FR4 that has lower dielectric loss but not much more cost. That being said, 6 GHz is about the upper frequency limit where most FR4 materials might want to be swapped for a low-loss type FR4 or a different material system altogether. There are several reasons for this which I have explained across many different videos.
the rf amplifier that i need has to be in a frequency rang 100khz to 5mhz and rf amplitude upto 250Vpp. can i get it done using thisdesign? if yes do you have any suggestions on component change?
I am not aware of any part numbers that can handle those voltages. Usually when you are getting to those levels, you should buy an off-the-shelf module, or if you want to build it yourself then you might need to use discrete components.
![Diggy Graves - Red Vineyard [Official Lyric Video]](http://i.ytimg.com/vi/8Ro085T0mnQ/mqdefault.jpg)
I always appreciate how clearly you explain your PCB layout choices, Zach.
Since we know you're an analog guy, would you consider doing a video on best practices for layout of an AF power amplifier? As I understand it, output currents become a major consideration at low frequencies.
I'll put it on the list! Can't promise it will be quick though as there have been a lot of requests lately.
Good stuff! Power amps aren’t always as simple and straightforward to design and layout as a lot of people assume
What an absolute dude, thanks for these videos!
I love your videos. Many thanks! One thing to be aware of is that quickly dragging schematics and board layouts back and forth can trigger motion sickness.
Really enjoy watching your videos and really thankful for the efforts you have put in so far in producing so many useful and technically enriching videos. I am an RF guy essentially, but your videos have certainly went a long way in helping me learn and obtain practical insights on few other arenas like pcb design, digital section, power components, etc. As an engineer and a learner, I always look up to you as one of my teachers (apologies if its sounding too childish😄)
Thank you very much! I started doing stuff with RF but later had to get good at digital design because everything uses digital.
Thanks Zach!
Hello @Zachariah Peterson, do you recommend, in this case, to remove the solder mask on the RF tracks? If yes, should I remove the solder mask from the GND as well? Thank you!
Yes to remove some additional losses along the interconnect you can also remove the solder mask along edge of the GND region closest to the signal line. There is no need to remove all of the solder mask everywhere, but it is still interesting that I see some designs where they remove ALL of the solder mask from the entire GND polygon around the signal line, that is not necessary. If you wanted to do it in this example design you could just span the solder mask opening to the far edges of the via array.
Great video Zach! When is it appropriate to use a coplanar waveguide in a pcb compared to just using an impedance matched microstrip? For example between a GNSS receiver and an SMA antenna connector, what would you recommend and why?
The main reason you would want to use coplanar waveguide is because you get high shielding effectiveness even when using a thicker substrate or dielectric layer. For example, I could have put a ground cutout through the internal layers and have ground pour on the bottom layer, and this would require using a wider line. This would give me lower loss, but when I put the pour on the top layer with the stitching vias to form the coplanar waveguide I could still get a reasonable trace width (about 30 Ohms). The shielding effectiveness I would expect would be comparable to having a narrow microstrip over ground on a thin dielectric. In this project I went with the thinner microstrip because the operating frequency is lower, and because we need a thinner line in order to route into the output of the power amplifier. I could have had this in sections too, and then we can look at the S-parameters across the entire interconnect.
Amazing video as always. Btw @16:08 what hotkey did you use to quickly switch directly from the Top layer into the Bottom layer? I don't think you use Ctrl+Shift scroll hotkey for that.
You can use the * key to cycle through signal layers from top to bottom. You can also use Shift + * to cycle backwards through layers. One thing to note is that * and Shift + * will skip plane layers.
@@Zachariah-Peterson Sweet! Thanks
Are there ready-made designs for high-frequency power amplifiers? Can you help me with this? I want examples of it
You should look at reference designs from power amplifier IC manufacturers. Analog devices has released a lot of their evaluation board files as reference designs that you can use to get started. Try looking their for help with the circuitry and an example PCB layout.
@@Zachariah-Peterson
Thank you, engineer. I want a good book on that. I searched and found a lot, but I want to leave me the basics on that?
Last week I wrote a long review with mistakes done in this video/design.
But it never got published. This is really sad.
Either it was flagged as spam or altium or @Zachariah Peterson doesn't want to deal with professional critical reviews.
Cheers
Mel
I don't flag stuff as spam. This is not a hobby for me, I run a design firm and I get to deal with plenty of professional critical reviews. Portions of this design were based on a system that was fabbed, tested, and is deployed in the field. There were some variations of the design that would still need to be checked/simulated, like the via fence on the CPW going to the output. And of course we would want to get an actual stackup from a fab house but we wanted to show something just as an example. If your review did not appear then I apologize but I am not the one who flagged or deleted anything.
Hi,
If the edge connector pads are moved to the edge as shown in video, does it not result in exposed copper pads during manufacturing of boards?
Could you please let me know, if there is any workaround by avoiding exposed copper along edges?
Thanks for the video, It was really helpful.
I'm not sure I understand, those pads will always be exposed because they are surface mount pads, it does not matter if they are pulled away from the edge. If you're worried about plating along the edge, then yes the through-hole plating process can deposit copper across that gap if the pads run right up to the edge of the board and the plating throwing strength is correct. If you do not specify edge plating options in your PCB fab notes, but you place the pads along the edge anyways (maybe on accident), your fabrication house should warn you that this might happen, or they may apply a pullback around the board edge. Finally, even if there was some plating across the top and bottom pads on that SMA edge connector, those pads are connected to the same net anyways so that plating does not hurt anything.
In regard to the Stack-up material. Most RF PCBs use CORE on the Outer Layers? NOT Pre-Preg.......... Due to Pre-Preg "press-out" thickness Fab issues. (I think)
Yes the stack can also be inverted. Not required but if you cannot tolerate any variations along the outer layer then consider using core on the outer layer. I've also had to do that when the outer layer was intentionally much thicker so that the RF line could be made wider and coplanar routing was used, then it would reduce the conduction loss. But yes press-out can be an issue with prepreg on an outer layer, gotta verify that stackup from a fab house first.
Hello, i enjoyed watching your video. the rf amplifier that i need has to be in a frequency rang 900mhz to 5Ghz . can i get it done using thisdesign? if yes do you have any suggestions on component change?
I do not know if you can do a direct swap for a power amplifier with different frequency range. There are many different HMC components, so I'm sure you can find something in the microwave range. The HMC components use a variety of different biasing and startup procedures, so because of that you most likely will not be able to do direct swap, even if the packages and pinouts match.
Is FR4 good choice for 6Ghz design?
Which variation of FR4 are you referring to? FR4 is a category of materials specifically referring to flammability rating, it is not one specific material. There are many different versions of it, including low-loss FR4 that has lower dielectric loss but not much more cost. That being said, 6 GHz is about the upper frequency limit where most FR4 materials might want to be swapped for a low-loss type FR4 or a different material system altogether. There are several reasons for this which I have explained across many different videos.
the rf amplifier that i need has to be in a frequency rang 100khz to 5mhz and rf amplitude upto 250Vpp. can i get it done using thisdesign? if yes do you have any suggestions on component change?
I am not aware of any part numbers that can handle those voltages. Usually when you are getting to those levels, you should buy an off-the-shelf module, or if you want to build it yourself then you might need to use discrete components.