Electronics > Projects, Designs, and Technical Stuff
Connecting a PCB to a slotted enclosure
<< < (3/4) > >>
T3sl4co1l:
If it's isolated then I fail to see what you're looking for -- nowhere for ESD/surge currents to flow, nothing wrong!

If it's not truly isolated (in the wideband sense, merely galvanically), then do what Ethernet does, add a ~1nF capacitor to ground (which would be front panel ground, the nearest accessible point -- and also the location the offending ESD is coming in through!).

Tim
cdev:
What are the parts on the top and where is the FPC on the back?

You probably have the best likelihood of successful RFI suppression if you use gold plated RFI contacts and have the amount they need to deform be exactly what is specified by their manufacturer for maximal longevity. Take advantage of whatever support they have. Don't have something that somebody who is closing up the box needs to muck around with in any unusual way because people never re-read instructions a few months or years down the line. They will invariably not do anything unusual like press the fingers down to position them properly for contact. That would be almost guaranteed to not be done by users.

The most effective RF shielding for aluminum boxes Ive encountered has been having lots of flat surfaces and lots of screws through them spaced closely together. (bringing us back to the beginning)

A good example of a problematic situation is a computer. Shielding most computer cases adequately is very difficult.
On YouTube Leif Asbrink, SM5BSZ has four videos entitled "Probes" or something like that. He shows how to make your own e-field and H-field probes and using them to locate where RFI is being created. He is very knowledgeable and I always learn useful things from his videos. Tips that work. Thats a really good practical set of tips on figuring out RFI problems from nearby devices unsing basic probes and thats just worth knowing.

You can make really small probes and use them with a variable attenuator(s) and broadband SDR to get a very fine grained level of RFI origin sensing.  Maybe you could first get or make the probes (if you dont already have them) You can make probes that can sense EMI and determine the 'sense" of the magnetic H-field. They are sold commercially too, but fairly expensive for something you can make yourself fairly well.

(but worth it if you do this a lot) around a year ago one of the fine people here, Frazier posted a deal on a set of probes, for very little. See if you can find his post, and even if the deal is gone you can see from the post the construction details and make one.

One you have suitable probes and some tools to sense them (you probably already do) Try the brushes out and see. If your device will see a lot of card insertion and 'desertion' (?) the brushes may fatigue and no longer work well, but if not and you use enough of them they are probably okay.  See if you can pick up the signal from them with various shielding methods including the one with the copper tape so you have a point of comarison. You may have RFI no matter how well its shielded there because the RFI is actually escaping some other way.

-----

Edit: If your signals can handle it they also sell flat ferrite cables in many different sizes as well as clamp on flat ferrite blocks to suppress EMI. I have a bunch of them also for ribbon cable. If you still have a problem from the back you can slap an appropriately sized block around the cable on the back. If your signals are differential, as I understand it (others here please correct me if this is wrong) the reactance (might only be a few hundred ohms and you need to pick the material by the frequency of EMI you want to suppress) should cancel out some well definable amount of the unbalanced EMI and leave nothing but your significantly cleaned up (balanced so unaffected) signal(s).

My best luck with probes has been when I made them very small and connected them up using multiple ferrite beads on the feed line to prevent the feedline picking up the signals. Look at the wildly expensive EMC sniffing gear and try to see if you can replicate its functionality. They often try to make it look esoteric in order to discourage reverse engineering it. But its not rocket science.

A simple clip with a ferrite split bead between its jaws and a few turns of wire connected to a coax can sniff RF riding on cables. Again decouple the feedline with ferrites.
max_torque:
Ok, looking at those connectors tells me your product is isolated thermocouple amplifier!   If you are doing the isolation for HV reasons, rather than noise, and the device is in a metal box, then you WANT to have a large, bolted ground stud or connector on the unit, and the instructions need to make it very clear indeed that this MUST be earthed at all times.  That earth point can then be simply connected to the pcb with a suitable earth wire or ground strap (which needs to carry any potentially large fault currents, ie be big!)

If you are isolated just for common mode noise rejection, then it would be better to just try to arrange your design to have the lowest possible capacitive coupling path out to the real world and between channels.  Here, every pF counts ime........
Dave:

--- Quote from: max_torque on December 05, 2018, 06:23:23 pm ---Ok, looking at those connectors tells me your product is isolated thermocouple amplifier!

--- End quote ---
Correctamundo! ;D

Excuse the awfulness of the drawing, this is basically what I'm doing here.

The thermocouple contacts are going to be exposed and prone to getting zapped here and there when connecting thermocouples to them. Each of them is connected to an isolated channel, which is what the long fingers on the PCB are for. The fault currents are going to be directed to the isolated ground through a pair of TVS diodes.
As the channel capacitance towards the system ground is going to be quite small, it wouldn't take much charge to move the potential of the isolated ground sky high and punch through the isolation barriers of my power and signal isolators, which are used between the isolated channel and the main system circuit (rear part of the PCB). Therefore, a GDT is placed between both grounds, to ensure that it sparks over before the isolators could get damaged. The fault current now needs to go from the system ground to the earth grounded enclosure, to complete the circuit. A Y-rated capacitor is going to be placed between the two for this purpose.

TL;DR I need to get the fault current from the rear of the PCB to the enclosure somehow.

The solder-on horizontal contacts (linked above) seem like the most viable option at the moment, but I'm always open to new ideas.


--- Quote from: cdev on December 05, 2018, 02:28:22 pm ---What are the parts on the top and where is the FPC on the back?

--- End quote ---
As max_torque already figured out, they are thermocouple connectors. The FPC connector is not visible, because I haven't put it in yet.
It's not really an RF issue, but I do appreciate you taking the time to share your RFI shielding techniques.

I have attached another image of the current state of the PCB design. A fair bit of the circuit is still missing (including FPC connectors), but I'm slowly getting there.
Ian.M:
Its hard to beat a machine screw with a zinc plated serrated washer under its head for getting a reliable ground to an aluminum case or chassis.

Take a look at https://www.pemnet.com/fastening-products/fasteners-for-mounting-into-printed-circuit-boards-2/
Right Angle Fastener - Type SMTRA
if you can tolerate an extra hole and screw in the case side, or
Right Angle Fastener - Type SMTRA
if you'd rather have it more central on the PCB with a scrw through the top or bottom of the case.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod