Allright, I have been reading a lot of posts about star grounding in a pedal and this is what I found:
Best practice is to ground the input jack ( so Maxwells law sends the rfi riding along the cable to the chassis). Then we connect the sleeve of the input jack to a point on the pcb where the different traces carrying grounds meet ( the star). This seperates the ground currents so they all will be connected directly to a clean ground and they will all be at the same potential, eliminating hum and other ground related artifacts.
The part that I don't understand is why it is also neccesary to use an isolated output jack and run a wire from the sleeve to the star ground.
Why can't this be an unisolated jack without the wire? Isn't the chassis also providing a single connection to ground using an unisolated jack?
Nope; it could create a ground loop (in theory). Whether it WILL do so (in any way you can hear) is open to interpretation, though.
The main reason for not grounding the jack to the enclosure and using the wire is, however...because over time it will work loose, and you'll get a god-awful BUZZZ when that happens!
....usually, what we end up with is some sort of 'modified star', in practice...at least in MY practice, LOL...The current levels we're dealing with aren't high enough to worry as much as one would with an amp, for example.
Quote from: Max999 on June 04, 2018, 06:29:43 PM
Why can't this be an unisolated jack without the wire? Isn't the chassis also providing a single connection to ground using an unisolated jack?
Because then it isn't a star. The output jack ground would be connected to the input jack ground which then goes to the common star ground.
It's because the star point is not in the chassis, and you need to directly connect every ground to the star point, including the output jack ground. If you use an unisolated output jack, the output jack ground will be indirectly connected to the star ground by the enclosure.
Btw, usually in pedals a star grounding doesn't really make a huge difference. A proper ground planning is required when building high gain preamps and/or circuits that use noise sensitive parts (J201, LM386, CD4049, to name a few), but in most of the cases, just avoid ground loops and you'll be fine. Ground loops are bad for any pedal.
Ideally, you use the chassis as a shield, but not a ground current conductor. This means only a single point on the chassis gets connected to ground, and you aren't using the chassis to provide a ground for anything else.
(I prefer to ignore the fact that we easily get away with ignoring these types of "best practices" and the fact that following them does not make any audible difference. )
Thanks everybody.
I see now that in my example the output jack ground is daisy chained with the input ground to the star point.
But what if I make the input jack ground the star point ... still wrong?
In audio gear as small as a pedal, "star grounding" is pointless.
You do want to be sure high-current paths don't share the wire with low-current paths. Otherwise none of the wires are long enough that "star" or "bus" or "mish-mosh" really matters.
I like high gain PRR. What I noticed is that humfree circuit followed by a clean amp/pedal will suddenly have worse perceived noise specs because of the compression of the high gain. Offcourse high gain will always have a high noise floor, but if I can change 5% of it I will try to. (Emphasis on "try") :)
Honestly, in my experience, the only place I've found star grounding to be useful is in a tube amp.
I use the DC jack as a star ground in my pedals, despite the fact that they would sound exactly the same if I didn't. For me, it is part of the artistry/craftsmanship* and makes me feel more proud of my work.
*A personal aesthetic choice -- I don't feel that my stuff is better that anyone else's just because I choose to do it my particular way.
Quote from: EBK on June 04, 2018, 07:13:45 PM
Ideally, you use the chassis as a shield, but not a ground current conductor. This means only a single point on the chassis gets connected to ground, and you aren't using the chassis to provide a ground for anything else.
True, but that was not common practice for a long time. Almost all of the older effects used the chassis as a common ground. But back then tape hiss and pops on records were just accepted, as well as some hum from amp/pedal/guitar.
I completely ignore star grounding convention in pedals and just ground everything everywhere. :icon_twisted:
It tends to work.
I imagine if you purposely tried to create an infinite number of ground loops, you'd end up with effectively zero ground loops.... :icon_confused:
Quote from: EBK on June 06, 2018, 06:34:01 PM
I imagine if you purposely tried to create an infinite number of ground loops, you'd end up with effectively zero ground loops.... :icon_confused:
That's exactly my strategy when building pedals. In amps I absolutely go for star grounding, but in pedals I find the Infinite Ground Loop Cancellation convention to be extremely useful.
forgetting all concerns about ground loops, IMO star grounding is way more reliable for enclosure-shielding generally. use a coupla tooth washers along with your crimp, nut and bolt and that thing will never come loose.
RG's system of using a PNP to power up charge pump circuits via. a stereo input jack often makes me wonder if there are other scenarios in which it's bad to pass circuit grounds so close to the high-impedance input. That's another reason i always avoid grounding everything to the input jack.
IMO, star grounding is a well-tested tube amp practice that has made its way into pedals with no particularly good engineering reason.
i contend that as the input jack is a mechanical component which may be used fairly frequently and has some risk of loosening, its role should be purely that of carrying the input signal and its specific ground, and shouldn't be relied on for the entire ground network to reach the enclosure. So while the intention may often be wrong, i think star grounding is superior for that reason.
Quote from: GibsonGM on June 04, 2018, 06:46:59 PM
Nope; it could create a ground loop (in theory). Whether it WILL do so (in any way you can hear) is open to interpretation, though.
The main reason for not grounding the jack to the enclosure and using the wire is, however...because over time it will work loose, and you'll get a god-awful BUZZZ when that happens!
....usually, what we end up with is some sort of 'modified star', in practice...at least in MY practice, LOL...The current levels we're dealing with aren't high enough to worry as much as one would with an amp, for example.
i should have re-read other posts before opening my beak! anyone got a cracker?
Star grounding can really be fruitful in building a studio. In fact, you'd be crazy not to. There usually is a couple "sub-stars" going on, but you can't stray too far.
I believe the size of the loops is one of the factors- in the sense of accidentally fabricating an antenna, and ground runs that have a little resistance, thus potential difference, between ends, are the kinds of factors that lead to problems in the high- fidelity atmosphere of a studio. It's hard to believe these problems can happen in a little aluminum box.
Quote from: MrStab on June 07, 2018, 10:40:41 AM
i should have re-read other posts before opening my beak! anyone got a cracker?
No problem, I just wanted to stick my nose back in!! :)
Yes, for the 'trouble' of doing it, star grounding is a good idea. Sometimes we end up doing the 'modified' approach, but anything you do to make a low-resistance groundpath that 'flows' correctly - will always yield less noise! It becomes more and more important as your power levels increase.
Thanks everybody. I made a star ground with the star at the big filter capacitor. The casing however is only connected at the input jack for rfi protection, so they go together to the star. I hope this is not an issue.
Quote from: GibsonGM on June 04, 2018, 06:46:59 PM
, however...because over time it will work loose, and you'll get a god-awful BUZZZ when that happens!
Just to drive this point home, if you have a ground loop issue, you'll likely know when you build the pedal, when the output comes loose you probably won't know until the most embarrassing and inconvenient time. IMHO - ask me how I learned this valuable lesson. :D
Star grounding is a means to an end, not an end in itself. It is not the only way to minimize cross-contamination of signal from different sources, nor the only way to cure hum issues, but it is the only way known that definitely will cure such issues before actually building and testing (or tediously modelling) a circuit.
True star grounding is probably never done in practice, because there would simply be too many wires. Modified star grounding, which clumps together "related" circuit elements in a local "star" and then star grounds the clumps is much more practical.
I've typed in my polemic on the number and kinds of ground conductors and currents before several times, so I won't do it again. But the discussion in this thread has highlighted some areas that are confused, so I'll hit those a bit. DC and audio grounding are close to identical, as the frequencies where electromagnetic field effects happen are mostly above audio. This matters because hum is down in the audio range where you just need to have a low-ohms connection in for low-current circuits. For pedal circuits, unless you have some odd cases like high current paths, you can ignore it.
One counter example where you can't ignore this is high pulse current sections in a pedal, and the best simple example of this is a power converter running at audio frequencies, like a 555 or a charge pump. The 555 in particular has the nasty habit of sucking an amp or two for a microsecond or so every time it changes state. So its ground wire - and power wire, for that matter - had better not be part of your signal ground path, or you'll hear the "ticks". Charge pumps do a similar thing as they charge and discharge their caps, but most charge pumps run above audio where it's much easier to filter the stream of ticks out and you can't hear that high in any case.
At yet higher frequencies, the electromagnetic fields force electrical currents to flow in ways that you don't expect until you've studied RF setups. When you get into radio bands, you need to think of the shielding type of ground. Shield grounds keep radiated RF out of circuits. That's one reason we put effects in metal boxes. We also live in a sea of power-line-frequency EM fields, so metal boxes prevent both the capacitive transfer of power-line hum out of circuits, and the RF-buzz of power line related spikes of RF as generated by fluorescent lights and such out of circuits.
Inputs are special. They are the circuit's most sensitive point and they're exposed to the world outside the box. They need special consideration. Input jack grounds must connect to the shielding ground to keep RF out, and much connect to the signal ground to get the signal to the circuit. For pedals, the shield ground of the box CAN act as a signal ground, as long as the jack is connected tightly to the shield/box ground. But as noted, it doesn't hurt to have a backup wire. Backup wires can break, but this happens less often that fixing nuts come loose.
Input ground wires are special too. They are the conductor that carries the tiny but real currents back to wherever they were generated. So mucking up an input ground wire with carrying power supply currents is a not-too-great idea. And many pedals do this by using the trick of using a stereo input jack to ground a battery or DC negative wire to carry DC power to the circuit board. It is true that the currents are small, and that the wire resistance is low, so nearly always you get away with it.
Nearly always.
I vastly prefer using a PNP transistor or a P-channel MOSFET inserted in the positive power line to switch power to a circuit, and connecting the base/gate to the stereo contact to turn the device on/off. It makes any current carried on the input ground be not only dramatically smaller, but also constant. Of course, you could do the same thing by just using the output jack for your power switching.
Thank you R.G. for your very complete answer. It seems that you are very much in line with PRR's answer, by stating that the currents are too low to really matter for hum, except as mentioned by you in oddball cases and directly at the input sleeve ( by using the power switch trick).
Regarding the input jack connecting to the shielding ground for RF protection, would it be a good safety measure to also connect a wire from the input sleeve to the chassis with a bolt?
Or would this destroy the RF protection and thus need the low value capacitor from sleeve to shield you mentioned in another grounding thread?
In little circuits and boxes, the differences are small.
For the very, very noise conducted in, you'd insulate the input jack and carry the signal ground lead all the way to the input of the first stage. This is likely to be unnoticeable in most pedals. I could design one to show it up, but that's not the point of pedals. If you did an insulated input jack, you want a small ceramic cap to the chassis from the input jack's 'ground" contact. This bleeds RF off to the shielding shell instead of conducting it into the circuit. If you connect the input jack to the chassis by the bushing, a cap is useless. If you like belt-and-suspenders, you can do a ground wire from the circuit board to the bushing of the input jack, and be assured that if one fails open, the other will keep the big buzz from happening.
I would consider it overkill to make a wire connection from the jack to a bolt on the chassis. Sure, there are some situations where this would help if something else is failing, but those situations are way out on the edges of rarely happening. Not worth the worry. Shoot, if you're on stage and something like this happens, replug your effects to get rid of that pedal, and get back in and play. As Frank Zappa famously said, shut up and play your guitar. :icon_wink:
Thank you again R.G. but I am a bit confused about this sentence: "If you like belt-and-suspenders, you can do a ground wire from the circuit board to the bushing of the input jack, and be assured that if one fails open, the other will keep the big buzz from happening".
We always need a wire from the input sleeve to the circuit board when the circuit board is not in contact with the shield, right?
I've had my Peterson strobe tuner embarass me on stage due to an input jack that worked its way loose, resulting in loud popping every time I retuned. During a break, I tightened it back up with a socket wrench from my gig toolkit and stopped the popping.
It does not matter how a pedal is built until it does. :icon_wink:
I had that happen exactly ONE time. Total loss of contact..."BZZZZZZZZZZZZZZZZZZZZZZZZZZ!" for like 20 seconds, at stage volumes and thru the PA.
That will NEVER allow that to happen again ;)
This could be useful for pedal building:
(https://images-na.ssl-images-amazon.com/images/I/51tslS1SbWL._SY450_.jpg)
I've used it to secure strap locks to straps. Never thought to use it on jacks, but I might give it a go sometime.
Quote from: Max999 on June 10, 2018, 01:52:16 AM
Thank you again R.G. but I am a bit confused about this sentence: "If you like belt-and-suspenders, you can do a ground wire from the circuit board to the bushing of the input jack, and be assured that if one fails open, the other will keep the big buzz from happening".
We always need a wire from the input sleeve to the circuit board when the circuit board is not in contact with the shield, right?
Sorry - I was not very clear there. I was distinguishing in a muddy, indistinct fashion, between a wire to the jack's bushing contact only, and two wires, one from the circuit board to the jack, and another from the jack's bushing to a dedicated bolt/nut on the chassis.
There are lots of ways to ground the metal enclsure, including wires from input and/or output jacks, dedicated wire from a ground pad on the PCB, wire from the power entry, etc.