Connecting grounds in guitar FX Line mixer

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Phil smith

Active member
Joined
Jan 15, 2014
Messages
34
Location
Earth
Hey guys!

Recently I've started designing/building a guitar fx mini line mixer clone for my rack with some extra features - 4 stereo in's (each has separate level control and mute switch) and one stereo out driving by OPA2134 amps with output level control.
Power supply - linear with step-down transformer, 240VAC in, and two 18VAC secondaries, connected in series (18-0-0-18) for providing +18/-18VDC for opamps (LM317/337 regulated).
I've got a question related to grounds - how should I connect them to minimize noises and loops? (see the attached schematic for reference).
1) AC Mains earth tab to chassis, the transformer center tap is opamps and output jack's ground (GND) and each input's ground (GND1 thru GND4) are connected separability to ac mains earth tab. Input grounds can be disconnected from the common point via switch (not shown on the diagram).
2) Opamp / output ground (GND) and input grounds (GND1 thru GND4) are tied together but isolated from the chassis and ac mains earth. Input grounds can be disconnected from the common point via switch ("Lift GRND") (not shown on the diagram).
Or any other better option :)

Very appreciate any help!
Thanks!
 

Attachments

  • Mini line Mixer CD.pdf
    36.7 KB · Views: 62
If you want to minimize ground loops you would have to make the inputs balanced. Meaning instead of the ring connecting to ground it would connect to the other input of an "op amp difference amplifier" with 10K input Z maybe. Of course then the jacks would have to be isolated. Balanced inputs could be important. Especially if you want to connect to gear that has it's own wall supply. If there's even a few mV difference, you can imagine you might get a significant current running in/out of one of the inputs. That is probably the primary source of hum and noise in a device like this.

And for guitar, you probably want high impedance. Unfortunately getting both balanced and high impedance can be tricky. If you're ok with 10K inputs, it's a lot easier.

The second most common source of noise would be if high currents are returning on a ground that is also the ground reference for a high gain circuit. In this case, it can pay to have use a separate wire back to the ground at the filter caps in the PS. But you don't really have high current or high gain so I don't think it will matter to much how you do it.

Otherwise your circuit looks fine. You could use a fuse maybe. It's not crystal clear that you need the inverting stages.

 
Is this intended for 'pedal' type guitar fx ? Many of these have their output from an unbuffered pot wiper (like your output stage).
This means the output impedance is changing with output level - meaning you need to be careful with the input pot value - or buffer the input.
For similar reasons I'd recommend making the output stage an active volume control around the opamp so the output comes from the opamp output via just the 100 Ohm 'build out' resistor (and the coupling cap).

Other points:
If space allows I'd say increase the value of the output cap for less LF distortion - if that's a consideration with this mixer.
Might be an idea  to put a pull down resistor on the 'outside' of the output cap to minimise pops on plugging in...
And I'd say go for +/-17 max on the voltage rails to give a bit of margin.
 
Phil smith said:
1) AC Mains earth tab to chassis, the transformer center tap is opamps and output jack's ground (GND) and each input's ground (GND1 thru GND4) are connected separability to ac mains earth tab. Input grounds can be disconnected from the common point via switch (not shown on the diagram).
The most polluted point is the junction of the center-tap and the smoothing caps. You don't want that point to be allowed to pollute the audio. The point that must be connected to chassis is the junction of C119/125, as you have drawn it. Then the ground distribution must obey the ancient adage "ground follows signal", or hierarchical ground. Connecting te input jacks' ground to the main ground is not teh best idea.
Whatever ground switches you have, you'll be facing the problem that grounds will follow multiple paths when all your paraphernalia are connected (unless they have floating connections, which I doubt very much).

Ideally, only one piece of your gear should be earthed, the rest should be powered via non-grounded PSU's (real double-insulation, not a cheater adapter); since I suspect your amp is earthed, and your mixer as well, you must make sure they are powered from the same outlet and not too far from each other.
 
I may have shared this before but over the decades I have killed many brain cells thinking about human safety around live music equipment.

While a power supply will need a true DC coupled ground return path, audio, especially guitar audio can work with an AC coupled ground. In fact you can size the cap to get effective audio shielding while looking higher impedance at hum frequencies so avoiding loops or conduction.

I experimented with using a GFCI power drop but with the hard ground connection lifted and replaced by a capacitor (IIRC something like a 0.15uF***).

I had a friend who designs and sells guitar pedals (amptweaker) test it out and he said in some cases it reduced hum from powered guitar effects, and never made it worse.

CAVEAT... lifting the safety ground is a no-no and not approved by UL... In combination with a GFCI outlet it should be safe.

JR

PS: My main concern was about musicians getting hurt by energized grounds which is pretty rare but does still happen occasionally.

*** for highest safety the cap should be Y rated to not fail as a short.
 
squarewave said:
If you want to minimize ground loops you would have to make the inputs balanced. Meaning instead of the ring connecting to ground it would connect to the other input of an "op amp difference amplifier" with 10K input Z maybe. Of course then the jacks would have to be isolated. Balanced inputs could be important. Especially if you want to connect to gear that has it's own wall supply. If there's even a few mV difference, you can imagine you might get a significant current running in/out of one of the inputs. That is probably the primary source of hum and noise in a device like this.

And for guitar, you probably want high impedance. Unfortunately getting both balanced and high impedance can be tricky. If you're ok with 10K inputs, it's a lot easier.

The second most common source of noise would be if high currents are returning on a ground that is also the ground reference for a high gain circuit. In this case, it can pay to have use a separate wire back to the ground at the filter caps in the PS. But you don't really have high current or high gain so I don't think it will matter to much how you do it.

Otherwise your circuit looks fine. You could use a fuse maybe. It's not crystal clear that you need the inverting stages.

The input source is line level signal from FX processor (up to 3 stereo units and one mono/stereo dry signal), so I think 10k is ok.
And, as the matter of fact, they are unbalanced.
All units in the rack are powered from Furman power conditioner (Furman PL-PRO DMC E)
Yep, it's got the 0.25A fuse on live wire, just after the plug. (Sorry, I should update the schematic, cause now it shows just PCB circuit).

Newmarket said:
Is this intended for 'pedal' type guitar fx ? Many of these have their output from an unbuffered pot wiper (like your output stage).
This means the output impedance is changing with output level - meaning you need to be careful with the input pot value - or buffer the input.
For similar reasons I'd recommend making the output stage an active volume control around the opamp so the output comes from the opamp output via just the 100 Ohm 'build out' resistor (and the coupling cap).

Other points:
If space allows I'd say increase the value of the output cap for less LF distortion - if that's a consideration with this mixer.
Might be an idea  to put a pull down resistor on the 'outside' of the output cap to minimise pops on plugging in...
And I'd say go for +/-17 max on the voltage rails to give a bit of margin.

Active volume control, you mean something like this? (after C5):



Also they got pull-down resistor after cap as you suggested  ;)
I got some free room to install bigger caps, so I'll experiment with values.

abbey road d enfer said:
The most polluted point is the junction of the center-tap and the smoothing caps. You don't want that point to be allowed to pollute the audio. The point that must be connected to chassis is the junction of C119/125, as you have drawn it. Then the ground distribution must obey the ancient adage "ground follows signal", or hierarchical ground. Connecting te input jacks' ground to the main ground is not teh best idea.
Whatever ground switches you have, you'll be facing the problem that grounds will follow multiple paths when all your paraphernalia are connected (unless they have floating connections, which I doubt very much).

Ideally, only one piece of your gear should be earthed, the rest should be powered via non-grounded PSU's (real double-insulation, not a cheater adapter); since I suspect your amp is earthed, and your mixer as well, you must make sure they are powered from the same outlet and not too far from each other.

Yep, they powered from the one power conditioner witch provided AC mains to all devices in the rack.
So do I need to keep each input grounds separately from each other and from PSU ground and C119/C124 junction to the chassis along with ac mains earth tab?

JohnRoberts said:
I may have shared this before but over the decades I have killed many brain cells thinking about human safety around live music equipment.

While a power supply will need a true DC coupled ground return path, audio, especially guitar audio can work with an AC coupled ground. In fact you can size the cap to get effective audio shielding while looking higher impedance at hum frequencies so avoiding loops or conduction.

I experimented with using a GFCI power drop but with the hard ground connection lifted and replaced by a capacitor (IIRC something like a 0.15uF***).

I had a friend who designs and sells guitar pedals (amptweaker) test it out and he said in some cases it reduced hum from powered guitar effects, and never made it worse.

CAVEAT... lifting the safety ground is a no-no and not approved by UL... In combination with a GFCI outlet it should be safe.

JR

PS: My main concern was about musicians getting hurt by energized grounds which is pretty rare but does still happen occasionally.

*** for highest safety the cap should be Y rated to not fail as a short.

This is interesting. Can you please show some example circuits?

PS. Thank you guys for your input!
 
Phil smith said:
So do I need to keep each input grounds separately from each other and from PSU ground and C119/C124 junction to the chassis along with ac mains earth tab?
That's the big dilemma with unbalanced connections.  One is tempted to use the chassis as a strong connection between the different "grounds", which reduces the production of parasitic voltages resulting from external ground loops; OTOH it increases the risks of noise due to magnetic fields.
Since the circuit is a quite simple one, it is possible to concile both by making the loop as short as physically possible.
As mentioned before, it is possible, at the cost of some added complexity, to use differential inputs. A parallel RC circuit between sleeve and chassis (10 ohms//100nF) is necessary because interference are not distributed equally between shield and conductor in an unbalanced cable, so must be dumped into chassis before entering the rest of the circuit.
 
Phil smith said:
Active volume control, you mean something like this? (after C5):



Also they got pull-down resistor after cap as you suggested  ;)
I got some free room to install bigger caps, so I'll experiment with values.
PS. Thank you guys for your input!

Yes that sort of thing re Active Volume Control. Various ways and component values to implement it. Check you get a reasonable law so that control isn't all 'cramped up' at one end of the pot.
It would also be simple and cheap to to make the outputs impedance balanced - just a resistor (and cap if being extra thorough) from the unused 'R' connection on the output jacks to 0V.
Gives advantage using TRS jack going into balanced input stage. No downside using TS jack / unbalanced input.
 
Newmarket said:
If space allows I'd say increase the value of the output cap for less LF distortion - if that's a consideration with this mixer.
That would be a tad silly... Assuming a 5kohms load (10k input Z of the FX box and 10k for the "pull-down" res)  and an output voltage of 1V (which is a lot in the realm of guitar signals), the voltage across the capacitor would be 7mV, i.e. 43dB below. If the distortion across the cap was 1% (a very pessimistic figure), the resulting distortion on the output voltage would be 0.07mV,  -83dB, or 0.007%. Considering 20Hz is outside the guitar frequency range, I don't see how this could be a problem. At 62Hz (low-B on 7-string), THD would be 10dB lower.
Using unduly oversized components has some drawbacks: increasing parasitic capacitance, resulting in non-controlled coupling between different parts of the circuit, and increase of the loop area, which increases sensitivity to magnetic fields.
 
abbey road d enfer said:
That would be a tad silly... Assuming a 5kohms load (10k input Z of the FX box and 10k for the "pull-down" res)  and an output voltage of 1V (which is a lot in the realm of guitar signals), the voltage across the capacitor would be 7mV, i.e. 43dB below. If the distortion across the cap was 1% (a very pessimistic figure), the resulting distortion on the output voltage would be 0.07mV,  -83dB, or 0.007%. Considering 20Hz is outside the guitar frequency range, I don't see how this could be a problem. At 62Hz (low-B on 7-string), THD would be 10dB lower.
Using unduly oversized components has some drawbacks: increasing parasitic capacitance, resulting in non-controlled coupling between different parts of the circuit, and increase of the loop area, which increases sensitivity to magnetic fields.

I take your points (without checking the figures). You could say that concern with tiny levels of distortion in a circuit for electric guitar use is of no practical consequence.
But I'm not assuming too much about what the output level or load on the output will be.
I do tend to think along the lines of - we've got a circuit here so let's make it generally applicable and so useful for any signal source and output destination - mixer / amp / DI etc.

Large component geometries can indeed introduce reactive impedance and loop area issues. However, in increasing the output cap from 10uF to, say, 100uF we are looking at a minimal, if any, increase in these.
On that point I just saw that the schematic has the caps notated as 160V rating.
Not needed (larger size) and it might just be unintended. Might want to go 63V min if it might inadvertently come into contact with phantom.
 
Newmarket said:
Large component geometries can indeed introduce reactive impedance and loop area issues. However, in increasing the output cap from 10uF to, say, 100uF we are looking at a minimal, if any, increase in these.
I see 220uF on the schemo... My estimates are based on this.
 
Phil smith said:
This is interesting. Can you please show some example circuits?
No schematic, cut the fat green wire (safety ground) and replace with capacitor (0.16uF  "Y" cap).  In the picture I added a switch so I could alternate between cap coupled ground vs. hard ground. Note: UL will not approve cutting the safety ground.

The GFCI is what keeps this safe around humans. The cap is sized large enough to pass adequate current to still trip the GFCI (5mA) in case of a mains fault, while simultaneously keeping that fault current low enough to not harm humans, and preventing low impedance ground loops.

JR
 

Attachments

  • P1010096.JPG
    P1010096.JPG
    58.3 KB · Views: 19
JohnRoberts said:
The GFCI is what keeps this safe around humans. The cap is sized large enough to pass adequate current to still trip the GFCI (5mA) in case of a mains fault
In most of Europe, with 230Vac, the current would be about 10-12mA, which is not enough to trip our GFCI's, rated at 30mA!
 
abbey road d enfer said:
I see 220uF on the schemo... My estimates are based on this.

I think that's a schematic that the OP pulled up (I assume from the interweb - OP free to correct me if not) wrt my suggestion for active volume control rather than taking the outputs from fader wipers.
 
abbey road d enfer said:
In most of Europe, with 230Vac, the current would be about 10-12mA, which is not enough to trip our GFCI's, rated at 30mA!
Yes... the capacitor calculus is different for 230V and different for 50Hz....

RCDs are generally protecting an entire branch circuit not just a single outlet like in US, so RCD use the higher current threshold to reduce nuisance trips.

I DO NOT advocate increasing the capacitance enough to trip a RCD as the human could get stuck to that much current.

That said a 0.15uF cap could be used in series with a guitar ground to reduce the shock hazard from grounded strings and a hot microphone or other electrical fault encountered on stage.  Some people have put capacitors inside guitars to isolate the exposed metal parts but few want to modify their guitars this way.

JR
 
> our GFCI's, rated at 30mA!

I have wondered about the large values for UK RCBs. My understanding is different agencies take different ends of the (time)*(current) curve. In the US we often say "5mA" but it takes a "long" time for a US GFI to trip at 5mA. It trips much faster at 50mA. It may be the UK and EUR breakers are rated at the fast end of the curve.

Look at attached.

5mA in 2 seconds. 30mA in 0.2 seconds. Either would be a reasonable "nominal" number for the name-plate.

Anybody know?
 

Attachments

  • GFI-time-curve.jpg
    GFI-time-curve.jpg
    127.5 KB · Views: 12
PRR said:
> our GFCI's, rated at 30mA!

I have wondered about the large values for UK RCBs. My understanding is different agencies take different ends of the (time)*(current) curve. In the US we often say "5mA" but it takes a "long" time for a US GFI to trip at 5mA. It trips much faster at 50mA. It may be the UK and EUR breakers are rated at the fast end of the curve.

Look at attached.

5mA in 2 seconds. 30mA in 0.2 seconds. Either would be a reasonable "nominal" number for the name-plate.

Anybody know?
I wonder what causes this time dependancy. As far as I can tell, "our" RCB's are "instantaneous". Looking at the guts, they are purely magnetic-driven, which supports this. Unfortunately, there are no documents easily available, and DIY experimenting is quite tedious.
 
yes response time will depend on the fault current vs the set point,

you can not have it instantaneous because you will have nuisance tripping,

so you walk the line of "how long can this human take it" vs "how many nuisance trips do we want per month?"

where does this nuisance tripping come from? not all the gremlins that come down the line are common mode, therefore they do not cancel when they pass thru the sensing coil. so if GM is stamping fenders at the plant down the street and the line voltage is getting pounded on the hot lead only, then the GFI will trip. So you build a little delay in there, which can be in the form of an RC circuit  as is the case with some industrial process GFI's.  Ferrite beads can help swamp out the RF stuff on the power feeding the actual GFI circuit.

150 ms is what they say here, that would be a minimum response time, actual trip time will probably be better than that.

here is a spec sheet for a GFI and the circuit that it uses showing the RC circuits that slow the trip time down>

 

Attachments

  • gf.jpg
    gf.jpg
    240.3 KB · Views: 8
CJ said:
here is a spec sheet for a GFI and the circuit that it uses showing the RC circuits that slow the trip time down>
Electronically, almost anything is possible, it could even be possible to implement RMS detection of the fault current; however, with purely electromagnetic devices, implementing a time-constant is not that easy. In essence, it takes only one half-cycle to trigger the mechanism; hysteresis and stiction (a newly-learnt word) would take care of spurious. I suspect there maybe some kind of damper providing the time-constant...
 

Latest posts

Back
Top