Asher Guitars WD Music Products Amplified Parts Mod Kits DIY Nordstarnd Pickups darrenriley.com
Asher Guitars WD Music Products Amplified Parts Mod Kits DIY Nordstarnd Pickups Warmoth.com
Asher Guitars WD Music Products Amplified Parts Mod Kits DIY Nordstarnd Pickups Warmoth.com

Ground. (What it isn't)

Discussion in 'Amp Tech Center' started by elpico, Jun 5, 2018.

  1. Snfoilhat

    Snfoilhat Tele-Holic

    Age:
    37
    621
    Apr 8, 2016
    Oakland, CA
    Everything you have asked about is in this PDF. I look forward to seeing elpico's approach to the topic, too, but yall could meet him halfway and do some reading in advance.

    I'll try to summarize in one sentence for the impatient: some "grounds" serve as a reference voltage with little or no current, but other "grounds" are completing circuits and may carry a great deal of current, and these especially may have some unwanted residual signal or noise in them which can intrude into the signal path.


    Bonus sentence: traditional amp building used the chassis itself as a big awkward wire, as the photo in post #34 displays so well, and I think we only got away with it because amps were relatively low gain and low fidelity. Cheers
     
    moosie and SSL9000J like this.

  2. LudwigvonBirk

    LudwigvonBirk Tele-Holic

    Age:
    116
    561
    Aug 26, 2017
    Madison
    Sorry for the straw-manish pita questions. I can make all kinds of stuff work and debug and de-noise most things ok, but I'm weak at precise accurate descriptions of stuff like this.

    1) If white and green are "ultimately connected" (at the house box or anywhere), in my amp chassis, will my meter beep and declare continuity, if one probe is connected to white and the other is connected to green?

    1a) Does the amp being powered on or off change anything, wrt 1)?

    Anticipating the informed responses to 1) and 1a), then

    2) What is the difference between "ultimately connected" and simply "connected"?
     
    Last edited: Jun 8, 2018

  3. LudwigvonBirk

    LudwigvonBirk Tele-Holic

    Age:
    116
    561
    Aug 26, 2017
    Madison
    Yep. Have already read (multiple times) and will re-read again. Valvewizard's articles are very well done and thought-through, every one.
     
    the fatch likes this.

  4. elpico

    elpico Tele-Holic

    674
    Sep 14, 2011
    Vancouver BC
    Sure we can dig into the topic of noise, but fair warning: even the experts writing the textbooks admit noise fighting is part black magic... and I'm no expert :lol:

    I had a few foamy beverages while watching the Stanley Cup final this evening (I'm Canadian so that part of my life wasn't optional) so probably we should leave that topic for tomorrow night.
     
    SSL9000J likes this.

  5. Speedy454

    Speedy454 Tele-Meister

    487
    Oct 1, 2013
    Highland, IL
    I would expect you to have continuity from the white (neutral) wire to the green wire (safety ground).

    To be plain and simple, in any device running on household AC , think of the black wire as the source.
    The white is the return. It should never tie to the chassis or anything other than the line cord.

    The green wire is there strictly for safety. If there is a component failure that shorts voltage/current to the chassis, that voltage/current is safely directed back to the main panel instead of through you. It should also cause the fuse on the device to blow, and possibly the fuse or circuit breaker at the main panel.

    Also, if there is a problem in the path of the white neutral, whether it be a connection in the device, open spot in the cord or plug or outlet, the green safety will still keep you from getting shocked through the chassis.

    Keep in mind that the white wire is carrying current back to the source.
    From the main panel, through the black wire, through the load (transformer in the case of an amplifier) through the white wire back to the source is a series circuit.

    Current is the same in a series circuit.

    Take a battery and a light bulb.

    Hook it up so the light shines.
    Measure the current from the + to the bulb.
    Measure the current from the bulb to the -.

    The current will be the same.

    So if the white wire ties to the chassis, or accidentally shorts to the chassis, there is a potential to be shocked if there is no safety ground.

    If this happens, and you are plugged into a GFI outlet, it will trip, preventing a potential shock to the user.
     

  6. elpico

    elpico Tele-Holic

    674
    Sep 14, 2011
    Vancouver BC
    Although someone linked to that on page one I'd never read it before tonight. This merlin fellow wasn't part of the online scene back when I learned this stuff in the late 90s... but wow, that hits on so many of the points I was intending to talk about that if I'd known about it my first post should've just been a link to that pdf with the word "questions?" after it.:D

    Being serious though, he left us enough room to add on to that. It seems to be concerned mainly with conducted noise (if radiated noise was mentioned I skimmed past it). And my biggest gripe about old school guitar amp layouts, loop area, doesn't seem to have much treatment there. We can get into those.
     
    SSL9000J likes this.

  7. Bendyha

    Bendyha Tele-Afflicted Ad Free Member

    Mar 26, 2014
    Northern Germany
    I would also like to reccommend reading this Hum.pdf article written by Tim Robbins at Dalmura. Although it's theme is HUM, there is much to do with grounding issues, lead dress is part of the grounding...or vice-versa, as is the magnetic-flux managment.
    Filament-hum...also part of the grounding issue.
     
    Last edited: Jun 8, 2018
    SSL9000J and the fatch like this.

  8. robrob

    robrob Poster Extraordinaire Ad Free Member

    Dec 29, 2012
    United States
    Calling Fender amps with chassis signal grounds "junk" is hyperbole. It's not optimal but it works just fine. I don't recommend it for new builds but I definitely don't recommend changing 50's, 60's and 70's amps' grounding schemes.

    This is a really good thread though.

    The 3rd power cord prong is a parallel path to ground for safety. It is connected to the neutral wire at the service entrance.

    This is the "typical" 5E3 layout that uses a split ground bus. The power amp's B+1 and B+2 and transformer center tap are grounded at the upper right power transformer bolt. The problem is the preamp is grounded at the upper right input jack. All the current flowing through the preamp tube cathodes must return to the power transformer center tap so it flows through the chassis from input jack to center tap bolt.

    [​IMG]
    Thousands of amps have been built like this and it works just fine.

    Here's an EE approved unified ground bus 5E3 layout. The only chassis ground point is the input jack and no current should flow through the chassis. Arrows show the current return flow (opposite direction of actual electron flow):
    [​IMG]
    Electrons are pulled off the rectifier tube plates and pushed out the center tap to the ground bus. Electrons flow from the ground bus through the 6 cathodes to the plates where they are returned to the power transformer through the B+ wire to the rectifier cathode. It takes about a week of constant running for an electron to make a lap around the circuit.

    It would be easy to build using the single ground bus then clip the jumper between the 2nd and 3rd filter caps and tie the center tap and ground bus to the transformer bolt and compare the two for noise and hum.

    [​IMG]
     
    Last edited: Jun 8, 2018
    SSL9000J likes this.

  9. peteb

    peteb Friend of Leo's

    Apr 25, 2003
    Cascadia
    Another important thing about ground.




    Zero volts represents zero shock hazard.



    Not only is the earth at zero electrostatic potential, but so are you.
     

  10. robrob

    robrob Poster Extraordinaire Ad Free Member

    Dec 29, 2012
    United States
    Google "line man helicopter" and see a human at 50,000 volts of potential not get shocked.
     
    Mr BC, Bill Moore, Bendyha and 2 others like this.

  11. Bendyha

    Bendyha Tele-Afflicted Ad Free Member

    Mar 26, 2014
    Northern Germany
    Depends where you are coming from, ground potential can be at a lethal level to another ground level.
    Kind of like falling from a cliff, it's still the ground that kills you.
     
    dsutton24, Bill Moore and robrob like this.

  12. elpico

    elpico Tele-Holic

    674
    Sep 14, 2011
    Vancouver BC
    To me the difference between "ultimately connected" and simply "connected" is that there are many potential hazards for the white wire between it's ground connection at the panel and it's arrival at your amp. Like I said it's been connected to the black wire in various ways by other devices in your house before it's arrival at your amp. The green wire has not. The white wire is carrying the return current of those other devices. The green wire is not. If you made them "simply connected" by connecting both white and green to your chassis or something you'd violate electrical code, form a ground loop, and potentially conduct the return currents of other appliances through your amp's chassis.
     
    LudwigvonBirk likes this.

  13. moosie

    moosie Doctor of Teleocity Silver Supporter

    Age:
    61
    Jul 18, 2010
    Western Connecticut
    I read Merlin's grounding treatise countless times over about four years, AND THEN finally had to implement it in an amp, before I understood the concepts: ground vs common; elevating DC ground above the AC; and noise reduction by routing high current away from low - of which ground loops are one example, and using the chassis as a common ground plane is another.

    And the word "understood" might be a bit grand in this context...

    It's all there, just keep thinking about it.


    Regarding noise, I find it useful to think about like this:

    Imagine two waterways inside your amp. The power supply is a huge river, with fast currents and a large volume of water. The preamp / guitar signal is a small brook, barely deep enough to cover the stones, chuckling it's way down the mountain. Imagine what happens when the brook meets the river. The delicate preamp is going to be affected by the power supply's higher current and voltage, which will result in noise. Which then gets amplified, because that's what tubes do.

    As @Snfoilhat mentioned, historically the metal chassis was seen as a convenient common ground plane, with both power supply and preamp currents running through it.

    Now imagine a conduit is built that routes the brook away from the river, until the very end, when they both dump into the bay. They won't have any effect on each other. That's what's happening in the typical two bus design, where the preamp common is physically separate from the power supply common (call them ground if you like, but since it's not a safety ground, I prefer to think of them as the common, the return path for the circuit). To really isolate the two, one uses insulated bushings where the jacks meet the chassis. Often good enough is the half-measure of routing the DC all to one side of the chassis (bus connects to chassis at input jack), and the power to the other (near the PT). The chassis is still used as a common ground plane, but it's less likely that the currents will stomp on each other.
     
    peteb, SSL9000J and LudwigvonBirk like this.

  14. elpico

    elpico Tele-Holic

    674
    Sep 14, 2011
    Vancouver BC
    Some of you wanted to talk about noise so let's start on that. (Sorry for the delay, weekend stuff) "Noise" is quite a wide topic and I'm certainly not qualified to write the book on that one, but I'm happy to share what I know and hopefully I can learn from what you guys know as well.

    We call any undesired sound coming from our amps a noise, but we'll need to be a little more specific than that here. There are many sources of "hiss" noises in an amp, notably the resistors and the tubes themselves. This is typically white or pink noise with no specific frequency or character, although these devices can cause a crackle noise or bursts like popcorn popping as well which is harder to identify as coming from inside your circuit. Hiss has no relationship to circuit layout though, it's intrinsic to the devices themselves, so let's set that aside.

    The noises we'd like to talk about here can be more specifically called interference. They occur when a signal generated in one electronic circuit affects another. There are many ways this can happen but they fall into two broad groups:

    - Conducted noise
    If we've brought the noise into our circuit with a wire, it's conducted noise. Power supply ripple is one example of a conducted noise. If two or more devices are supplied by a shared length of wire then we have a conductive path between them that can couple noise from one into the other as well.

    - Radiated noise
    Every conductor is also an antenna. Wanted or not, that's the way the world works. The wires in your amp can both receive and emit unwanted signals. A signal from one part of your circuit can couple through the air to an unconnected part, particularly if the two circuits are close to each other. Or a stronger signal, intentionally broadcast into the world, can couple to your circuits from far away.

    Radiated noise also goes by the initials EMI (electro-magnetic interference) and the art of building circuits that can operate in a world that contains other circuits is called EMC (electro-magnetic compatibility). This is one area of electronics that's developed radically since the time these old amps were designed. I'm certain the designers of the time were aware of EMI, Leo was a radio repairman after all, but the idea that the guitar player would have a phone in his pocket, and so would every member of the audience, and that every device in the room would be pinging each other over bluetooth, mapping out the local wi-fi networks, and checking in with the local cell towers constantly while he played? That was understandably a situation nobody saw coming. But it did come, and once it did we all started hearing things like this coming out of our audio devices:



    Hopefully you're hearing that particular galloping sound less often than you used to because both the audio device designers and the cellular network designers have had a few decades to improve their EMC. The audio guys improved their layouts to make them more immune to EMI. You started to see audio products hit the market with stickers that proclaimed "no cell phone buzz!" etc. At the same time the RF engineers worked on their communication schemes to reduce the amount of work everyone else was having to do so the newest generation of devices play nicer with our circuits.

    Our tools for avoiding interference are filtering (works for both types), shielding (not effective against conducted noise), and carefully controlling the layout of our components and conductors. That last one is the most important here and everything about understanding that tool comes down to one thing: loops. This is why I started the thread where I did. We've covered how the amplifier is not a block of components powered by a red wire and drawing electrons from an endless sea in the ground, but maybe we haven't said enough about what it IS yet: a loop.

    Let's extend our basic amplifier drawing to include the power supply. I'm going to draw a half-wave rectifier for clarity. I know our amps typically use a full wave rectifier, but that's just a duplication of this to reduce ripple. The concepts don't change:

    [​IMG]

    The power transformer's traditional lead colours should be familiar, but if they're hard to see here you've got the wall power coming in from the left via the black and white wires and the high voltage secondary is represented on the right by the red wires. (with a yellow stripe to further mark the center tap lead)

    Past the end of the transformer leads the wire colour is chosen by the builder. Let's extend the traditional scheme to make it clear what's going on here:

    [​IMG]
    The voltage at point A is changing constantly, but when it's voltage is high relative to point B the diode is open and electrons travel in a loop like this. It's easy to think of the transformer as a black box with the red lead as something separate from the red-yellow, but they are one continuous piece of wire that enters the transformer, wraps around the steel core several times and comes back out. When we connect our devices across those leads we are closing the loop. The loop is why we call them electronic "circuits" after all, like a racing circuit - it's a closed loop that things circle continuously. Note that the ground is not part of the loop. No electrons are flowing through that path, everything is going through the loop of red wire that goes through the transformer.

    When the voltage at point A swings low relative to point B the diode closes and things change:

    [​IMG]
    Now the power supply filter capacitor forms the left side of the loop and the electrons circle this smaller loop. This is actually the situation for the majority of the time the amp is operating. The diode only opens for a very short period during each AC cycle. The rest of the time the capacitor is the power source for the amplifier.

    So we really have two loops here, and they behave very differently. Everything to the right of the capacitor is steady and unchanging. Because the tube's grid is grounded only a constant DC bias current is flowing around this loop. To the left of the capacitor we have the least constant current in the entire amplifier. Current is zero most of the time, spiking very violently to a high value once per cycle. (twice per cycle in a full-wave rectifier, but you know what I mean). Here's what that might look like in a small amp:

    [​IMG]

    How sharp the spikes are depends on the resistance in the transformer and rectifier used and the size of the first filter cap. A later amp with solid state diodes and large filter caps produces the harshest, noisiest spikes. The root note of this noise will be at 120Hz in north america, but the harmonics extend through the entire audio band. Even past the audio band - short intense current spikes are great at creating EMI. (Maybe you remember reading about Marconi's spark gap transmitter? Same principle).

    So we've arrived at our first noise source, the rectifier current loop, and it's capable of producing both conducted and radiated noise. Next time loop area.
     
    Last edited: Jun 11, 2018
    SSL9000J, D'tar and LudwigvonBirk like this.

  15. LudwigvonBirk

    LudwigvonBirk Tele-Holic

    Age:
    116
    561
    Aug 26, 2017
    Madison
    Quick question on this in case something important might blow by me...

    When referring to size above do you mean the physical dimension of the filter cap, or the filter cap rating (e.g., 22uf/50v)? Or some combination of both?
     
    Last edited: Jun 12, 2018

  16. elpico

    elpico Tele-Holic

    674
    Sep 14, 2011
    Vancouver BC
    It's a good thing you asked because that's wrong anyways. I was referring to capacitance value only, but I was forgetting that a higher capacitance first cap only causes higher current spikes after you first turn the amp on and they're still charging up.

    Once things settle down to business as usual the capacitance value shouldn't affect the size of the spikes anymore so I was thinking of the wrong variable there. The correct one is: How sharp the spikes are (during normal operation) depends on the resistance of the transformer and rectifier used and the total current draw of the tubes.
     
    SSL9000J likes this.

  17. SSL9000J

    SSL9000J Tele-Meister

    Age:
    46
    102
    Dec 18, 2016
    North of Atlanta, GA
    Very concise and well put!
     

  18. elpico

    elpico Tele-Holic

    674
    Sep 14, 2011
    Vancouver BC
    Our drawing needs to have a third loop added to it for the next part. Each tube's grid circuit is also a loop. I think I better flip the drawing around so things are in the traditional order with the input signal at the left:

    [​IMG]
    It turns out that the ability of a current loop to emit unwanted interference, or to receive interference itself, is related to the physical size of the loop. In particular the area (in square inches etc) that's enclosed by it, which we call simply the loop area. The distance between the emitting loop and it's potential victim is also very significant, so for best rejection of this type of interference we always want to keep loop areas to a minimum and keep the noisy loops far from the susceptible receiving loops.

    How do we know who's a victim and who's a noise maker? Potential victims are the loops that have either low signal level, high impedance, or both. A loop's potential to emit noise is related to the rate of change of the current flowing through it. Our rectifier loop with it's sudden spikes from 0mA to a high current and back is an example of a noisy loop.

    When you look at an old guitar amplifier layout it's easy to see that these ideals were only haphazardly applied. For example one of the easiest and most effective ways to reduce loop area to it's minimum is to twist the wires that form the loop together. Instead of a tall, wide box enclosing a lot of area like our rectifier loop above, a twisted pair encloses a minimally narrow area and the twists further help reduce unwanted emission and coupling. In your average guitar amp this principle has been followed for some loops, notably the heater loop is almost always twisted together to reduce it's ability to emit hum, but it's been completely ignored for other loops.

    The rectifier loop and the input signal loop are the most egregious examples I see in old amp layouts. As we've seen the rectifier current has a particularly noisy waveform but instead of keeping this loop small it's sometimes spread halfway across the amp, often to reach a front panel stand-by switch which is a completely unnecessary control to put on an amplifier in the first place. Making a bad situation worse sometimes this noisy loop isn't even fully formed by wires but connects to the chassis at two points instead. When you run audio frequency current through a conductive plate like the chassis it doesn't travel in a line from point A to point B, it spreads out across the entire surface of the plate like a wide river, so in effect you've added the entire surface area of the chassis between these points to your loop area.

    The input signal loop tends to also be unnecessarily spread out with those 68k mixing resistors mounted on the board far from the tube and no shielding of these low level leads. Running shielded wire from the jack directly to the tube and locating the mix or grid stop resistors there, directly connected to the grid pin, reduces the loop area to a minimum and the shielding further reduces potential interference.
     
    SSL9000J and peteb like this.

  19. peteb

    peteb Friend of Leo's

    Apr 25, 2003
    Cascadia
    That's interesting about the area of the loop.



    Would it be flux crossing thru the loop that can cause the noise problems?
     

  20. Speedy454

    Speedy454 Tele-Meister

    487
    Oct 1, 2013
    Highland, IL
    \

    There is a difference between inrush current, the several AC cycles required to initially charge the capacitors and the diode conduction period / peak diode current.

    As filter capacitance increases in a DC power supply, the diode conduction time decreases.
    The diode must conduct a higher current to provide the energy needed to recharge the power supply.

    This is from my old college textbook "Electronic Devices and Circuit Theory" by Boylestad & Nashelsky. C1972,1978.

    The diagram shows a half wave circuit, but the same applies for a full wave. The higher the capacitance, the lower the ripple, but the higher the current needed to recharge the capacitor.

    This is for normal operating condition.

    Fortunately, most diode makers know this, and buried deep in the diode ratings are the impulse current ratings.

    Take a common 1N400X, 1 amp diode. It also has this rating...

    Peak forward surge current for 8.3ms single half sine-wave superimposed on rated load (JEDEC Method)
    IFSM 30.0 Amps



    Diode current.jpg
     
    elpico likes this.

IMPORTANT: Treat everyone here with respect, no matter how difficult!
No sex, drug, political, religion or hate discussion permitted here.