Continue to Site

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.

what is the maximum amount of solder needed between SMA connector and microstrip?

Status
Not open for further replies.

skatefast08

Full Member level 3
Joined
Jan 24, 2018
Messages
189
Helped
2
Reputation
4
Reaction score
2
Trophy points
18
Activity points
1,779
I am building a 2.4 GHz LNA (2 layers), I want to be able to use the maximum amount of solder to connect a larger end-launch SMA to my PCB that includes the FR-4 and top and bottom layer of copper ( https://imgur.com/a/k7u2WPu ). The smallest SMA fitting I can find is 0.032''(0.813mm) and I want to attach this to my 0.4mm PCB board.

Due to the equation for calculating Zo for microstrip (needing 50 ohms). The more I decrease the thickness of my PCB (decreasing FR-4 size; H; dielectric), this will allow me to decrease the width of my t-line in order to get a closer width of my SMD (0402) components.

The main question, how much solder (in height) to attach SMA with PCB would keep me from decreasing the performance of my LNA? (I know this sounds a bit silly, but im not interested in the noise, only the gain and stability)

What h(height; dielectric), w(width; t-line), t(thickness of copper) and SMA connector size, using FR-4 would you use to design 50 ohm microstrip with 0402 components? (if I have this answer, this would be the best)
 
Last edited:

this will allow me to decrease the width of my t-line in order to get a closer width of my SMD (0402) components.

There is no need to decrease substrate thickness, because there is no need for 50 Ohm lines between the SMD in your design.

You misunderstood the comments on using smaller routing dimensions in your previous thread. As already mentioned there, you just need to make the interconnects short & small, to built what you have simulated: a lumped circuit. There is no need to make the interconnecet line impedance 50 Ohm, if they are short!! And if they are not short, your design will fail even if the lines are 50 Ohm, because you have not included the lines in your design.

Actually, by using a thinner substrate you will increase shunt capacitance of the routing, which is not desired.
 
There is no need to decrease substrate thickness, because there is no need for 50 Ohm lines between the SMD in your design.

You misunderstood the comments on using smaller routing dimensions in your previous thread. As already mentioned there, you just need to make the interconnects short & small, to built what you have simulated: a lumped circuit. There is no need to make the interconnecet line impedance 50 Ohm, if they are short!! And if they are not short, your design will fail even if the lines are 50 Ohm, because you have not included the lines in your design.

Actually, by using a thinner substrate you will increase shunt capacitance of the routing, which is not desired.

I've heard from other edaboard users, if my t-lines are less than 1/10 of a wavelength, then I should be fine, which would be negligible in performance using lumped components.

Would you think I'd be able to just use a 1 layer board and have ground on top if I bunched the components as close as possible and this would produce the same results?or be able to use autoroute on eagle?

What size fr4 would you use?
 
Last edited:

I've heard from other edaboard users, if my t-lines are less than 1/10 of a wavelength, then I should be fine, which would be negligible in performance using lumped components.

Yes, we discussed that, and that's why you should keep these interconnect small & short, and not worry about the width too much. Only the external long lines need to be 50 Ohm.

Would you think I'd be able to just use a 1 layer board and have ground on top if I bunched the components as close as possible and this would produce the same results?or be able to use autoroute on eagle?

No autorouter please, use manual routing for short path length.

What size fr4 would you use?

Without access to simulation, I would use 0.8mm or so, but that matters only for via inductance. If your circuit design doesn't require low inductance to ground, 1.5mm substrate should be ok. If you have a simualtion tool like ADS, you can easily check what the substrate thickness does to your inductance -> circuit performance.
 
Yes, we discussed that, and that's why you should keep these interconnect small & short, and not worry about the width too much. Only the external long lines need to be 50 Ohm

So even if the t-line at Zo=100 ohms and are very close together (1mm apart or something), then I should still be fine?

If I was going to include 50 ohm lines. How would I design the microstrip with 0402 components? Would the width of the line not matter when connecting to the pads of the components?

- - - Updated - - -

So even if the t-line at Zo=100 ohms and are very close together (1mm apart or something), then I should still be fine?

If I was going to include 50 ohm lines. How would I design the microstrip with 0402 components? Would the width of the line not matter when connecting to the pads of the components?

I mean 1mm in length
 

So even if the t-line at Zo=100 ohms and are very close together (1mm apart or something), then I should still be fine?

There is no general rule that applies always to everything, but applied to your LNA layout that you showed: yes, short narrow lines (100 Ohm) will work better than large wide lines (50 Ohm), because the resulting circuit layout will be smaller. The concept of using matched transmission lines does to apply to the inter-component routing in your lumped circuit. The rule for lumped circuit is: make it as small as possible.

Inside your LNA, there is no reason to prefer 50 Ohm interconnects over other dimensions, because none of the nodes inside your lumped circuit has 50 Ohm impedance, and the interconnects are short anyway. Only feedlines should be 50 Ohm.

You asked about 1mm length. 1mm is almost nothing at 2.4Ghz. When the line is short compared to the wavelength, the line impedance doesn't make a difference. You can calculate the resulting impedance transformation yourself, here's an extract from my transmission line appnote:

Zin.PNG

So your 1mm line at 2.4 GHz will simply behave as a little series inductance with a little shunt capacitance.
 
Last edited:
If I was going use 50 ohm transmission lines to match my circuit, what width of a transmission line and height (dielectric) would be best to use then? Even though I did not put that into my design.
 

When the line is short compared to the wavelength, the line impedance doesn't make a difference.
So your 1mm line at 2.4 GHz will simply behave as a little series inductance with a little shunt capacitance.
And the difference is: Equivalent inductance is scaled with line impedance.

You can either ignore the small parasitic elements, or consider it in the matching network calculation.

- - - Updated - - -

If I was going use 50 ohm transmission lines to match my circuit, what width of a transmission line and height (dielectric) would be best to use then?
I would use a substrate height that is mechanically stable and fits available end launch connectors, e.g. 0.6 to 0.8 mm. Line width for the 50 ohm feed line respectively. "Lumped" circuit part as small as possible, with parasitics considered in the circuit calculation.
 
Inside your LNA, there is no reason to prefer 50 Ohm interconnects over other dimensions, because none of the nodes inside your lumped circuit has 50 Ohm impedance, and the interconnects are short anyway. Only feedlines should be 50 Ohm.

If I was going use 50 ohm transmission lines to match my circuit with lumped components (0402), what width of a transmission line and height (dielectric) would be best to use then? Even though I did not put that into my design.
 

You can't MATCH (in terms of impedance matching) because NOTHING is matched INSIDE your circuit. But anyway, I give up, draw whatever you like best. Good luck!
 

You can't MATCH (in terms of impedance matching) because NOTHING is matched INSIDE your circuit. But anyway, I give up, draw whatever you like best. Good luck!


Im sorry, I already matched my circuit with L matched networks at the input and output. I should have asked about using 50 ohm transmission lines to stabilize the circuit instead of matching; I already matched my circuit for 2.4 GHz. :cry:
 
Last edited:

"Stabilize"? I don't understand your question. What lines? Between the components? => goto post #2 above
 

"Stabilize"? I don't understand your question. What lines? Between the components? => goto post #2 above

What if I match my circuit using only transmission lines, would the width and height matter? that's all im trying to get at about the question

- - - Updated - - -

"Stabilize"? I don't understand your question. What lines? Between the components? => goto post #2 above

I want to have k-factor greater than 1 on all frequencies, so in order for me to achieve good stability, then I could either add a series or shunt resistor (I added a 500 ohm resistor in shunt at the input and output of circuit). I have heard from other people that I could just use transmission line to in place of the resistors. Also, using a t-line at the collector instead of using capacitor and resistor you could achieve stability that way.
 

Sorry, your question is not clear, because it is not clear what the term "matching" means to you. I've tried my best to explain that you should route interconnect metal as small as possible inside your lumped circuit, and forget about matched lines inside your circuit. No idea how to explain it any better.

Ok, I see that you added a comment on "stabilizing". I am not aware that there are "best" line impedances for that purpose, but that's not my area of expertise.
 
Last edited:

Sorry, your question is not clear, because it is not clear what the term "matching" means to you. I've tried my best to explain that you should route interconnect metal as small as possible inside your lumped circuit, and forget about matched lines inside your circuit. No idea how to explain it any better.

Ok, I see that you added a comment on "stabilizing". I am not aware that there are "best" line impedances for that purpose, but that's not my area of expertise.

matching means putting the correct components (lumped) or using stub matching (using t-lines) in a particular fashion (L, pi networks etc.) to reduce reflection and increase gain at a particular frequency. I understand that I should keep the interconnected lines as small as possible inside my lumped circuit, which will keep my matched circuits from increasing parasitics (which will effect the response I intended for). If I use a transmission line at the collector to ground, rather than using bypass cap and resistor to ground, this will help my stabilization and also help reduce noise. Also, instead of using resistors to stabilize the input and output of the amplifier, I could use a t-line (50 ohm) at a certain length and be able to obtain the same stability and also suppress noise. So, if I was going to use those 50-ohm lines (for either stability or for stub matching), would you think the width of the line and the height of the dielectric be an important factor? Maybe there are (or used to be) standard width and height parameters for microstrip lines that work at 2.4 GHz?
 
Last edited:

The transmission line at the collector will act as an inductor. Maybe start from this 1.9GHZ LNA application note where they applied this technique and provide a lot of details: **broken link removed**
 

I already matched my circuit with L matched networks at the input and output. I should have asked about using 50 ohm transmission lines to stabilize the circuit instead of matching; I already matched my circuit for 2.4 GHz.

But did you put the parasitic PCB inductances and capacitances into the matching circuit calculation, and also that of the component packages?
 

But did you put the parasitic PCB inductances and capacitances into the matching circuit calculation, and also that of the component packages?

How can I put parasitic into my design? How do I know the parasitic of my components and PCB?
 

for reliability reasons, you need a SMALL fillet up both sides of the center pin. MIL Standards for soldering will explain how much is desired.
Solder on TOP of the pin does nothing good, so you do not need any on top.
 
Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top