Standoffs far apart that pass tech ground (chassis)

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

boji

Well-known member
Joined
Jan 6, 2010
Messages
2,375
Location
Maryland, USA
Hello. There's many posts about proper grounding around here, but I could not find one on chassis loops between two PCB's.  My concern is this: I have taken chassis ground from the backplane, passed it to a 10ohm resistor, and flooded the outside plane of PCB 1. To pass shield to PCB 2, I'm using M3 aluminum risers, (standoffs) that rest on pads in two spots on the inside of the cards. Each standoff is pretty far apart from each other on opposite ends, as they serve double duty as structural support. My question is, should I only use one standoff to pass the chassis signal or is two ok? Or forgo the whole riser-as-signal-connector thing, and simply add another couple of pin connectors and make the risers null.

Thanks for your suggestions!
 
Am no expert but your scheme sounds like it could have issues. I try to keep ground connection's to a minimum, one ideally, by using a star ground and if I have to use a bus I try and make sure the higher current circuit's are nearest the ground connection, with the lower current circuit's further up the  bus.

Andy.
 
As a rule you should keep the chassis and analogue 0V completely separate because they serve different functions. They should then be connected together at one point only.

The purpose of the chassis is to catch external interference and route it safely away. You don't want to share this route with analogue 0V because you are inviting interference currents to inject themselves into the analogue circuits. It looks to me like you are following this rule but using a PCB flood as a screen. This is fine. However, for a screen to work properly it needs to be an equipotential which means it should have as low a resistance as possible. You do not need to worry about loops in the chassis because there is no loop. The chassis itself floats and it is connected to earth at one point only. So I would ditch the 10 ohm resistors and use as many standoffs as you can to connect the two PCB chassis planes.

Cheers

Ian
 
Basically what Ian said. But I don't really understand why you are using a PCB plane as a screen connected to chassis.
Assuming it's in a metal box ?
You could use the plane as 'signal' 0V reference to improve the PCB performance.
 
Newmarket said:
Basically what Ian said. But I don't really understand why you are using a PCB plane as a screen connected to chassis.
Assuming it's in a metal box ?
You could use the plane as 'signal' 0V reference to improve the PCB performance.
+1

That's what I do. The bottom side of all my PCBs is basically analogue 0V (with the odd track routed on that side when I absolutely have to).

Cheers

ian
 
There is a lot written about grounds/grounding. The first thing to remember is that grounds are not 0V but some arbitrary conductive node.

I like to think of chassis ground as a sewer pipe (dirty ground) while internal grounds are more like sink drains (less dirty) and still not 0V. As long as there is poop in the pipe (or water) there will be voltage drops (think Ohms law).

Successful audio design is keeping audio referenced to local grounds and passed differentially.

Of course this is over simplified...

JR

PS: Back in the day when we put heavy power transformers inside steel amplifier chassis there were indeed ground loops in the chassis but so what, we don't wash our hands in the sewer water.  :eek:
 
But I don't really understand why you are using a PCB plane as a screen connected to chassis.

Well, mostly because there's no metal panels between channels in the desk they are going into.  I figured it would help with crosstalk. I've run a 3.8mm audio 0v as one large-ish trace that follows and sinks all signal.

 
boji said:
Well, mostly because there's no metal panels between channels in the desk they are going into.  I figured it would help with crosstalk. I've run a 3.8mm audio 0v as one large-ish trace that follows and sinks all signal.
Brute force rarely beats finesse.

JR
 
The omission of the 10 ohm resistor is interesting, as to be completely frank, I only added it because I've seen people do it with some other professional products in the same vein. I assumed the rationale was you wanted some resistive channel isolation to avoid backfeed, but Ian suggests the resistor effectively reduces/negates the shield's utility!

Brute force < Finesse
So basically my attention to signal 0V's singular fat trace is not as good as replacing it with a flood.  That seems brutish.  >.<

Edit: So much to learn.
 
0V ground pours act like traces above or below signal traces in RF.  Wow ok. 
Do we treat audio as DC for simplicity's sake?


Assume now a solid ground plane with the ground connection at each IC near the trace connection point. The return currents have to go from the ground connection of one IC to the ground connection of the other. Since we have a solid ground plane, the path of least resistance, and thus the path of DC current, will be a straight line (the blue arrow in Figure '8'). At high frequency the mutual inductance between the trace and the ground plane beneath it make the ground path of least impedance directly under the trace (the red trace in Figure '8').
rZOzuwt.png


(Successful PCB Grounding with Mixed-Signal Chips - Follow the Path of Least Impedance)
https://www.maximintegrated.com/en/app-notes/index.mvp/id/5450
 
boji said:
0V ground pours act like traces above or below signal traces in RF.  Wow ok. 
Do we treat audio as DC for simplicity's sake?

Basically yes. AF and RF are totally different ball games.

Cheers

Ian
 
Forgive the constant pepering of questions and hand holding guys, I really appreciate all your comments thus far.

I'll try to close this thread with one last q:  Might someone point me towards a layout (or link) to study that in your estimation best embodies the principal 'Signal follows ground'?

Thank you!

Edit: No pictures of a mic cable, please. =P

So I would ditch the 10 ohm resistors and use as many standoffs as you can to connect the two PCB chassis planes.

Great thanks, I'm on it.
 
boji said:
The omission of the 10 ohm resistor is interesting, as to be completely frank, I only added it because I've seen people do it with some other professional products in the same vein. I assumed the rationale was you wanted some resistive channel isolation to avoid backfeed, but Ian suggests the resistor effectively reduces/negates the shield's utility!
So basically my attention to signal 0V's singular fat trace is not as good as replacing it with a flood.  That seems brutish.  >.<

Edit: So much to learn.
A flood is also brute force.... Finesse is using things like differential amps to cancel (subtract ) ground voltage errors.

There is room for both, but finesse beats brute force most of the time.

JR
 
boji said:
Well, mostly because there's no metal panels between channels in the desk they are going into.  I figured it would help with crosstalk. I've run a 3.8mm audio 0v as one large-ish trace that follows and sinks all signal.

OK. I can see what you mean and if you did have metal panels they would likely attach to chassis.
But 0V signal ground planes will also similarly attenuate crosstalk.
Then you get 0V signal impedance benefits and, arguably, less chance of capacative coupling from the chassis ground currents.
You can still employ 'finesse' techniques as mentioned by JR.

The '10 Ohm' thing is sometimes 'deployed' to 'break' ground paths and 'encourage' ground currents to flow the right way whilst the grounds either side stay close to each others potential. Not sure how well I've articulated that ! Often used where 'Pin 1' (Screen) comes onto the PCB but hopefully that's not happening here  :D
 
"OK. I can see what you mean and if you did have metal panels they would likely attach to chassis.
But 0V signal ground planes will also similarly attenuate crosstalk.
Then you get 0V signal impedance benefits and, arguably, less chance of capacative coupling from the chassis ground currents.
You can still employ 'finesse' techniques as mentioned by JR."

Thank you Newmarket.
Now I just have to decide if I want to redo the darn thing and swap chassis for 0v.  I can say doing so would get rid of about 30 short, 1.25mm traces that tie to the main 0v trace that snakes its way around the board.

Following common sense, wherever dg or dv lives, I kept 0v well away from it. If I understand everyone correctly, if I switch from chassis to 0v pour, nothing changes in this respect; that pour ought not lie below or adjacent to dv (12v & 5v) and dg.
 
boji said:
Thank you Newmarket.
Now I just have to decide if I want to redo the darn thing and swap chassis for 0v.  I can say doing so would get rid of about 30 short, 1.25mm traces that tie to the main 0v trace that snakes its way around the board.

Following common sense, wherever dg or dv lives, I kept 0v well away from it. If I understand everyone correctly, if I switch from chassis to 0v pour, nothing changes in this respect; that pour ought not lie below or adjacent to dv (12v & 5v) and dg.

Should I take it that dg / dv are 'digital' voltage rails ?
If so, what are they being used for - Logic / DAC / ADC /micro  etc ?
 
"If so, what are they being used for - Logic / DAC / ADC /micro  etc ?"

Logic-ish. Just some old-fashioned relays and leds. The microcontrollers will be off console.

Edit: I decided to forego (temporarily?) the ccr's for all the button led's that many people kindly and wisely recommended.
I'm trying to keep it simple.
 
So, per advice of a few folks, I've vastly simplified my layout by adding shielded freewire points where signals run across the board. It's gotten so clean on the non DOA PCB,  there's now a ton of room for an Audio 0v pour on the inside (top), or opposite of the Chassis pour, which is on the outside (bottom) of the card.  So my question is this- does anyone see a problem with the chassis pour  being opposite a AGND pour? I assume not, as I've see a few products do it this way, but i wanted to get your opinions on it before I go ahead and make the polygon.
 
I don't really understand what  "shielded freewire points" are ?

And I still don't get why you want to put a 'Chassis Pour' on a pcb ?
An 'AGND pour' will attenuate crosstalk between channels. As will any Copper area held at a low impedance fixed potential eg a voltage rail - but 0V is usually easily available and offers other electrical benefits.

Is it actually a problem in itselfd to have AGND opposite CHASSIS GND ?
Well. as always, it depends...
Assuming you have a double sided 'standard' FR4 1.6mm PCB then you are forming a capacitor with 1.6mm of FR4 as the dielectric. The value of the capacitor will depend on the exact geometries of the copper pour areas.

By my calculation 1 cm sq areas at 1.6mm spacing will give approx 2.5pF ( I am assuming FR4 has permittivity similar to the range given for Epoxy Glass @ 4.2 - 4.7 -ref EMC for Product Designers (Tim Williams) ).
Welcome to check my maths (note it's maths btw  :) - If I were doing this for money I'd be triple checking it !
So you can take it from there.

The capacitance is relatively small but it does mean that you are getting progressively less impedance between the two planes as frequency increases. It may not seem relevant at anywhere near audioland. But the chassis is subject to the whole frequency range and you want to be shunting that away - not giving it a chance to 'jump' over to your AGND via capacitance (let alone produce EM fields inside the equipment).
There is a case for rf filtering the signal  to Chassis  but this is best done as near the physical input connector as possible.

Whether or not it matters in reality depends on too many things to think about. But as a general rule - where multiple Ground / Power Planes are used on the same PCB layer and separation is desired a common distance to start at is 3mm. (It's a difficult situation to quantify as the dielectric in that case can be considered to be made up of air / solder resist / FR4 - Good Luck with that  :)

I say - keep CHASSIS on the chassis  :)
 
Back
Top