Edge-Mount Coaxial Plug Implementations for Quick and Dirty Prototypes?

Thread Starter

Twigg

Joined Jul 25, 2015
23
On a recent-ish project, I was almost able to implement a 50-200MHz VCO and amplifier circuit without a PCB or etched board. It wasn't pretty (see image), and at the end of the day it didn't work, but I was surprised by how close it was to working. The only issue was a crosstalk noise which I pinpointed to a panel-mount BNC plug which I had epoxied onto the board as an edge-mount connector. The crosstalk was between a solder cup connected to the core conductor of the BNC and a grounding lug connected to the shielding. This was just the way the BNC receptacle came out of the box, I didn't add any of these contacts.

I get that I was not using the right parts for the job, but how are edge-mounted coax/BNC/SMA receptacles supposed to be attached to avoid crosstalk issues? How can I approximate those standards on a quick-and-dirty prototype like the one in the picture? Thanks!
 

Attachments

AnalogKid

Joined Aug 1, 2013
10,986
Edge-mount RF connectors do not have insulated bodies. Where a vertical or right angle SMA might have 1 center conductor pin surrounded by 4 ground (shell) pins that go to a ground plane on the pc board, and edge mount has the center pin flanked by two machined tabs designed to sit on pads to the sides of the center contact pad. Sometimes the tabs have slots machined in so an 0.062" thick pc board slips in between.

http://storefront.brackemfg.com/sto...rQty=1&cust_item=BM60941&whs_id=1&orderUom=EA

ak
 

RichardO

Joined May 4, 2013
2,270
I can't tell from your picture but is there a ground plane on the bottom of your perf-board? This is the ground plane that the BNC ground pin must make a solid connection to.

Also, how is the ground pin of the connector connected to its mounting plate? In the picture it looks like aluminum.
 

SLK001

Joined Nov 29, 2011
1,549
To begin with, your BNC should have been all metal, like this one shown below (without the extra wire shown). The one you used was an isolated connector.

DSCI1737.JPG

A second problem, is that you have wires running all over the place. I can't tell if you're trying to create a wire farm or prototype a circuit. If you don't want loads of crosstalk and interference, then you are going to have to use a ground plane under your circuit. Try a different type of prototyping, such as the Manhattan, or "dead bug" style (look them up). Also, keep your wires as short as possible.

That's all I can say without better pictures and a schematic.
 

Attachments

Thread Starter

Twigg

Joined Jul 25, 2015
23
Hey all,

I apologize for the very brief and rushed post. I've put some better pictures here and am working on finding the schematic.

I should have clarified. Everything on the perf board is low frequency, an AC input in the audio range for frequency modulating an RF carrier signal via a VCO, which is the IC which has been deadbugged and epoxied on the sheet aluminum. In short, everything in the DC or audio range is located on the perf board, and everything on the RF range is on the aluminum sheet. That aluminum plate was an attempt at a ground plane. I tried to epoxy everything on the aluminum, but I learned that epoxy expands as it dries, pushing contacts apart, so that didn't work as well as I hoped. During testing, I had to manually clamp the grounding lug of the BNC to the aluminum. A piece of copper sheet would've been much more convenient in hindsight, as then I could solder to it.

In the attached picture of the aluminum sheet, BNC connector, and VCO chip, you can see a little better what's going on with that BNC plug. There are two contacts sticking out of the insulation. One is a solder cup in the center, which connects to the core conductor. The other is a grounding tab-like contact which I've been calling a grounding lug because it has a hole in it. I bent this downwards towards the aluminum sheet and forced it to make mechanical contact. Definitely a bad plan on my part, I agree. I learnt from that mistake.

But even when I had a solid pressed contact between the grounding contact of the BNC plug and the aluminum sheet, I still saw crosstalk. That's when I gave up. To clarify, I detected the crosstalk by measuring the voltage difference between the shielding of the plug and the aluminum sheet on an oscilloscope and seeing a lot of high frequency noise. Thinking about it now, I guess that means that the parasitic coupling between those two contacts sticking out of the insulation was a better electrical contact than I had made between the grounding tab and the aluminum plate. Oops.

Edge-mount RF connectors do not have insulated bodies. Where a vertical or right angle SMA might have 1 center conductor pin surrounded by 4 ground (shell) pins that go to a ground plane on the pc board, and edge mount has the center pin flanked by two machined tabs designed to sit on pads to the sides of the center contact pad. Sometimes the tabs have slots machined in so an 0.062" thick pc board slips in between.

http://storefront.brackemfg.com/sto...rQty=1&cust_item=BM60941&whs_id=1&orderUom=EA

ak
This is very helpful, thanks! How are the tabs normally fastened to the ground plane when they aren't machined to the precise thickness of a PCB? Is it all done with solder? Also, what does the footprint look like? Thanks!

I can't tell from your picture but is there a ground plane on the bottom of your perf-board? This is the ground plane that the BNC ground pin must make a solid connection to.

Also, how is the ground pin of the connector connected to its mounting plate? In the picture it looks like aluminum.
The only area with a makeshift ground plane is the stuff sitting on the aluminum sheet. In a nutshell, the stuff on perfboard is a voltage regulator to keep a fixed DC supply going to the VCO and a maze of amplifiers for signals under 100kHz. The only RF connection is the bare copper wire from the upside-down VCO chip to the BNC plug, which is positioned over the aluminum sheet. The aluminum is just what I had lying around, and at the time I figured I would just have to make it work somehow. I lost that battle, even with the aid of epoxy and pliers. I will absolutely go out of my way to use copper next time.

To begin with, your BNC should have been all metal, like this one shown below (without the extra wire shown). The one you used was an isolated connector.
DSCI1737.JPG
I wish I had known these existed. Am I right to say that's a panel-mount plug? I may have been able to make a good contact mechanically with one of those. Thank you!!

A second problem, is that you have wires running all over the place. I can't tell if you're trying to create a wire farm or prototype a circuit. If you don't want loads of crosstalk and interference, then you are going to have to use a ground plane under your circuit. Try a different type of prototyping, such as the Manhattan, or "dead bug" style (look them up). Also, keep your wires as short as possible.
I was afraid that I would get crosstalk with a ~77kHz signal going through that bird's nest of wires by the IC sockets. I was very surprised when I had no problems there. My hypothesis is that at these frequencies wires aren't as much of a problem as densely packed copper traces. That yellow perfboard with all the wires sticking out of it is traceless. I think that may have saved the circuit from crosstalk issues at these low frequencies. Either way, I completely agree with you, and I would say I got very lucky. By the way, I was trying to deadbug the VCO IC that's epoxied to the aluminum sheet. It was my first attempt. It didn't work very well because the epoxy lifted the metallic case of the IC off the board as it hardened, breaking the continuity of the metal case to the ground plane. That aside, did I do it right? Any critique?
 

Attachments

SLK001

Joined Nov 29, 2011
1,549
By the way, I was trying to deadbug the VCO IC that's epoxied to the aluminum sheet. It was my first attempt. It didn't work very well because the epoxy lifted the metallic case of the IC off the board as it hardened, breaking the continuity of the metal case to the ground plane. That aside, did I do it right? Any critique?
You should have soldered it to a ground plane. The epoxy just isolated the case (which should be grounded) from any semblance of a ground plane that you had. I don't think that your amplifier ever worked.
 

AnalogKid

Joined Aug 1, 2013
10,986
This is very helpful, thanks! How are the tabs normally fastened to the ground plane when they aren't machined to the precise thickness of a PCB?
The connector is available pre-machined for one of three different pcb thicknesses. The normal and only fastening method is to solder the tabs to matching pcb pads. For the ground tabs, the pads are exposed areas of ground plane.

ak
 

Thread Starter

Twigg

Joined Jul 25, 2015
23
You should have soldered it to a ground plane. The epoxy just isolated the case (which should be grounded) from any semblance of a ground plane that you had. I don't think that your amplifier ever worked.
For next time, I figure I should use a piece of single clad copper board and use the copper layer as a ground plane. Then I could solder the IC case directly to the ground plane, and I could maybe do the same with a metallic panel-mount BNC connector (unless it's made of aluminum). Does that sound like a reasonable plan?

The connector is available pre-machined for one of three different pcb thicknesses. The normal and only fastening method is to solder the tabs to matching pcb pads. For the ground tabs, the pads are exposed areas of ground plane.
Could that footprint be etched on single-clad board? Would it be challenging? I've never etched. On the drawing of the SMA edge-mount connectors, the signal pin seems to sit lower than the upper grounding tabs. How does the signal pin sit flush on the board?
 

AnalogKid

Joined Aug 1, 2013
10,986
On the drawing of the SMA edge-mount connectors, the signal pin seems to sit lower than the upper grounding tabs. How does the signal pin sit flush on the board?
One surface of the signal pin is co-planar with one inside edge of each slot. All three of these sit on the same side of the board. For a single-sided board, these are the only two ground pins that would be soldered. It can be either the top or bottom side of the board; the connector doesn't care.

ak
 

RichardO

Joined May 4, 2013
2,270
I doubt that you ever had any connection to the aluminum "ground" plane. Aluminum oxide is an insulator (except when you want it such as on a heat sink).
Your VCO and the ground lug of the BNC were just floating above the aluminum plate.

There appear to be 4 pins on the VCO that I would expect to connect to a ground plane. In fact, I don't see any ground wire to the VCO at all. Is the bent pin on the lower right supposed to be the modulation input? I assume it was disconnected for debugging??

In addition, I don't see any power supply bypassing on the VCO. What is the part number, by the way? A schematic would go a long ways toward getting you better anse
 

SLK001

Joined Nov 29, 2011
1,549
Your VCO and the ground lug of the BNC were just floating above the aluminum plate.
That "VCO" is his amplifier. But I agree that it probably wasn't working.

For next time, I figure I should use a piece of single clad copper board and use the copper layer as a ground plane. Then I could solder the IC case directly to the ground plane, and I could maybe do the same with a metallic panel-mount BNC connector (unless it's made of aluminum). Does that sound like a reasonable plan?
It's better to use double-sided clad for your base. Use an X-ACTO knife to trim away from your circuit (your circuit is already there on the PCB, you just have to trim away the excess copper!). Here's how I usually connect SMA connectors to a PCB. The connector is soldered on both sides. A flange mount BNC could attach the same.

BD-SMA.jpg
 
Top