Signal ground/chassis ground: theory vs pratice

Discussion in 'General Electronics Chat' started by CurlsOnKeys, Mar 1, 2016.

  1. CurlsOnKeys

    Thread Starter Member

    Jan 3, 2015
    47
    2
    Hello all,

    I've been reading and studying a lot about these topics, but I still find it hard to fundamentally understand some of the key concepts, especially when I look at practical design examples. I'll try to bundle some questions here:

    In this great article on Rane (http://www.rane.com/note151.html) one reads that in electrical units you have signal ground and chassis ground. Signal ground is used as a the reference point for internal electronics, chassis ground is, well basically, all connections to the chassis (which in a 3-prong line cords is connected to the mains ground plug). Signal ground should, at one point, be connected to chassis ground. The Rane-note gives two reasons for this: if any noise currents/voltages are induced directly on the chassis or indirectly on the shield of an interconnecting cable (which - in turn - connects to the chassis ground), signal ground should be allowed to fluctuate along with the now modulated chassis ground. If .2 mV of electrostatic charge is induced in the chassis ground, signal ground should also be allowed to "raise" .2 mV as to maintain the relation between them, correct? The second reason would be, quote: to keep the signal grounds of two interconnected units at very nearly the same voltage potential. Doing so prevents the loss of system dynamic range where the incoming peak voltage levels exceed the power supply rails of the receiving unit.
    I think I also get this: the chassis grounds of two interconnecting units are connected using the shield cable, which keeps both chassis grounds at nearly the same potential and tying signal ground to chassis ground in turn keeps both signal grounds at the same potential (relative to chassis ground), which avoids unwanted clipping and loss of dynamic range, correct?

    Now, if I more or less get that (correct me if I'm wrong), on to a practical example, I thought! And I opened up my old Roland SH101 synthesizer of which schematics can be found here: http://www.synthfool.com/docs/Rolan...Servicemanual/Roland-SH101-Service-Manual.pdf. As you can see on page 5 (attached here as a .gif), real life seems to be a bit different. Things that puzzle me:
    - for starters (you can't see that on the schematic, but take it from me): nowhere in the SH101 is - what's indicated in the schematic as "chassis ground" - really connected to the synth's chassis. Only the bottom part of the synth has a metallic chassis and only two screws attach this chassis to the synth, but these screws are not connected to any part of the synth's electronics. Thus, chassis ground in this case simply means "the 0V ground reference plane", right? To which various, lots of, nearly all of the internal circuitry is directly connected, as you can see on the schematics. In fact, would it be correct to think that this is actually the synth's signal ground? (Signal ground is used as a the reference point for internal electronics)?
    - according to the schematics, there IS a signal ground, a -5V ground. Some parts of the circuit are connected to this -5V (I assume they need another reference level to function properly, take IC2B on the bender board for example), some are not. Would it be correct to think of this ground as a second signal ground, a second reference point, to which soms IC's are connected?
    - as far as I can tell, and here real world seems to follow theory, there's only one place where the SH's signal ground (-5V) and chassis ground (0V) connect: starting from the left hand side of the schematic: connection 11: signal ground is tied to chassis ground, with a cap in between (C6). So if stray currents or charges modulate chassis ground, signal ground is modulated "along", right?

    Final question, once again referring to the Rane note: It all comes down to paying close attention to where currents flow. The key issue is that these noise currents do not flow through a path shared by any audio currents: if noise currents induced in the shield of an unbalanced audio cable flow in the my SH, they will happily flow through almost EVERY path of audio current, since the chassis ground is here also the reference point/ground for all internal electronics. Why would you design this as such? Even if this is a consumer synth with 2-prong power plug, without a real "chassis" and connection to chassis ground, why not separate the signal ground, the reference ground to which all electronics are connected, from another "chassis" ground, that's tied next to the power supply and that serves to divert noise from entering the synth? Is this "bad" designing or is Rane over-emphasising things?

    Thanks for the long read!


     
  2. MaxHeadRoom

    Expert

    Jul 18, 2013
    10,553
    2,375
    One of the problems is that the terminology and the related 'symbology' has got rather out of kilter.
    Chassis does not necessarily mean Earth Ground connection, also there is a complete abandoning of correct symbol use for each type of 'Common' connection, as is illustrated in works that such as The Art Of Electronics which use the earth symbol (wrongly IMO) throughout the publication.
    http://www.brucearch.com/videos.html#vid2
    http://sites.ieee.org/ctx-emcs/files/2010/09/Archambeault-Ground-Myth.pdf
    Max.
     
  3. MaxHeadRoom

    Expert

    Jul 18, 2013
    10,553
    2,375
    Looking at the schematic they just show a Chassis ground/common with the correct symbol as far as I could see.
    Max.
     
  4. WBahn

    Moderator

    Mar 31, 2012
    17,757
    4,800
    I haven't read your post in detail or looked at the Rane article you linked to, so consider this just a general observation for your consideration. The issue of proper grounding is very complicated because there are many competing goals that need to be addressed, including safety, noise, ground loops, EMI/EMC, frequency dependence, power vs signal, etc. Each of these argue for a particular approach to grounding and those approaches are often incompatible -- so you have to make compromises. A couple of examples -- should the shield of a cable be connected to the chassis at both ends (to establish a common reference voltage) or should it only be connected at one end (to prevent ground loops)? Answer: It depends -- which issue is the bigger problem. Another example, should your grounding strategy be single-point (to reduce interaction of sensitive circuits via ground noise) or multi-point (to create a more solid ground plane)? Answer: It depends -- often a hybrid approach is take were DC is a single-point but, through the use of stitching capacitors, high frequency is multi-point.
     
  5. WBahn

    Moderator

    Mar 31, 2012
    17,757
    4,800
    I'm definitely sympathetic to your cause and try to use the right terminology where practical -- but the fact is that it is a ship that has sailed and sometimes using the correct terminology actually increases the amount of confusion. Here I think making the distinction is quite worthwhile.
     
  6. MaxHeadRoom

    Expert

    Jul 18, 2013
    10,553
    2,375
    The terminology is slightly exacerbated in N.A. that uses the term 'Ground' for both chassis and earth, rather than many other jurisdictions that make a distinction between them.
    Where Earth is used for a grounded conductor, whereas in N.A. ground is used in two situations, as Dr Bruce points out in his video.
    Max..
     
  7. MaxHeadRoom

    Expert

    Jul 18, 2013
    10,553
    2,375
    It is not so much in the terminology as the symbols used, as in the total disregard of the International symbol standards.
    Max.
     
  8. WBahn

    Moderator

    Mar 31, 2012
    17,757
    4,800
    I think the two are related and would suggest that the abandonment of proper symbol usage arose primarily as a result of the abandonment of proper terminology usage. Could be wrong.
     
  9. MaxHeadRoom

    Expert

    Jul 18, 2013
    10,553
    2,375
    That ship may have sailed, but one (OP) just sailed into port seemingly confused!
    I don't see by using the correct terminology should be termed confusing, I know it is for me when I see posts and designs that contravene the standard, trying to make out what is actually intended!.
    I am with Dr Bruce all the way.
    Max.
     
  10. CurlsOnKeys

    Thread Starter Member

    Jan 3, 2015
    47
    2
    Thanks both of your for your insights and help! However, speaking of sailing ships, we're maybe drifting of target a bit here :) I'm well aware of the different grounding symbols and their meaning, that's not really the issue (I guess).
    I know my original topic was elaborate, but to summarise, I was mainly confused about the following things:

    - chassis ground (fork-like-symbol) doesn't necessarily have to be connected to the chassis? It's not the case with my SH101 and it just seems as a 0V reference to which electronic circuitry is connected.
    - there can be multiple common/signal grounds (triangle-like-symbol); in my synth the common/signal ground is at -5V for example
    - as the Rane note pointed out, for stray currents or charges not to intervene with the audio-currents, they should be "tied" to the chassis ground and the chassis ground should only be connected to signal ground at one place: here this is not the case: I have multiple connections to chassis ground, so "noise" can easily couple into the circuit.

    Considering that last point, WBahn's first reply was helpful, providing me with the insight that one ground connection vs multiple depends on design choices, so there is no absolute right or wrong way to do things. Although to me it seems that it would be fairly simple to design in such a way that on a PCB you have multiple ground connections with a solid ground plane (as is the case with the SH-synth) and yet SEPARATE the shield from an interconnecting by not connecting it to the "main ground" plane to which al of your electronics are also connected, but by connecting it to a separate ground plane of some kind that you couple to the chassis or another low impedance point. That way stray currents/voltages would take that separate low impedance path instead of mingling in with all the connections via the electronics ground plane, no?
     
    Last edited: Mar 1, 2016
Loading...