Ground Loop Isolation Wows New PSU Build

Thread Starter

Kelvinchill

Joined Sep 11, 2017
6
Hello to everyone, I wonder if anyone can offer some advice. I run a mobile disco and am frustrated with a ground loop problem. I have tried all the recommended solutions and some not recommended. I am about to build a custom power supply that will provide me with 19v 12v 9v 6v and 5v to power the numerous devices and peripherals I use.


After trying multiple different power supplies for laptops and peripherals some double insulated some not, connecting ground wires from each device and ground loop isolation transformers from the cheapest to the most expensive. I have failed to totally remove the hum. Grounding all the devices reduces the hum but does not eliminate it completely.


I did in a fit of desperation ground the negative on DC input to the laptop and this did reduce the hum even further, after thinking more I realised that was a really bad idea as any mains ground fault could end up in the laptop.


So this is my solution and what I would like some guidance with please. I have purchased a Meanwell 15a 24v PSU and will step down the voltages with 5 buck converters with appropriate ratings for the DC devices I intend to run from my new “Power Brick”


There are Three areas I would like advice on please.


Is it necessary, silly or advantageous to link the negative rail of all the buck converters?


Should I earth the buck converters on the DC side of the Meanwell or even ground it with a switch inline to create a ground lift?


I intend to install a fan in the “brick” enclosure that will house the 24vdc PSU and the buck converters. I would like to use a 240vac fan as I don’t want to run a potentially electrically noisy item anywhere on the DC part of the power brick. I would like if possible to control the speed of the fan. I am thinking of a 120mm 220-240v fan with a 0.1 amp draw. I did think of putting in a rotary switch with 4 of 5 positions and dropping through resistors but if I did my sums correctly the resistors will be huge (physically) So I then thought of using a capacitive dropper circuit for each position on the switch but in honesty I am struggling with this concept. Any advice would be very much appreciated.


Thank you
 

MSFTF

Joined Aug 11, 2017
33
Something I was considering mentioning earlier was how it could be ok to use small 12V fans. They should be quieter electrical noise-wise than in the past times when they used brushes.
 

Thread Starter

Kelvinchill

Joined Sep 11, 2017
6
Something I was considering mentioning earlier was how it could be ok to use small 12V fans. They should be quieter electrical noise-wise than in the past times when they used brushes.
Hi Thank you for the reply,

I did think about this and there is certainly a wider choice of 12v fans and speed controllers, Do you think a 12 or 24v dc fan powered from either the Mean Well 24v PSU or a buck converter set to 12v would produce less noise on the output of the other buck converters than a 240v fan on the incoming mains? If so then it's an easy decision.

Now you got me thinking given that I am already into this for a fair few pounds I may aswell chuck another fiver at it and get a small dedicated 12v internal transformer to power the fan. Then any back emf and noise would be mains side and at least as well filtered as the output of the Mean Well.
Cheers

Phil
 

Sensacell

Joined Jun 19, 2012
3,432
Trying to get rid of hum this way seems like an expensive and potentially frustrating exercise.
You might do all this work and have zero change - a total waste of time.
You also run the risk of damaging your gear in the process.

Better to actually figure out where it's coming from and eliminate it, maybe hire an audio guru to help you- it will be cheaper and faster than building all that stuff by far.
 

AnalogKid

Joined Aug 1, 2013
10,987
With no equipment descriptions, part numbers, datasheets, or a system wiring diagram it is impossible to say. But ...
Is it necessary, silly or advantageous to link the negative rail of all the buck converters?
Necessary (maybe).
Should I earth the buck converters on the DC side of the Meanwell?
Probably (maybe).
or even ground it with a switch inline to create a ground lift
If by ground you mean earth ground, *not* the power line neutral, then yes (maybe).
I would like to use a 240vac fan as I don’t want to run a potentially electrically noisy item anywhere on the DC part of the power brick. I would like if possible to control the speed of the fan. I am thinking of a 120mm 220-240v fan with a 0.1 amp draw. I did think of putting in a rotary switch with 4 of 5 positions and dropping through resistors but if I did my sums correctly the resistors will be huge (physically) So I then thought of using a capacitive dropper circuit for each position on the switch but in honesty I am struggling with this concept. Any advice would be very much appreciated.
For 50% fan voltage, a first-pass approximation of the resistor needed is 1200 ohms at 12 W. A 25 W resistor is not "huge" in absolute terms, but we don't know how much room is available.

ak
 

Thread Starter

Kelvinchill

Joined Sep 11, 2017
6
Hi Sensacell, it's a valid point and something I should look into. The hum is not the only motivator in this, I have two many low quality wall warts that are unsightly. The wire quality on them is always very poor, they are not intended for long term repeated insertion and moment. They are the bane of my life, different cable lengths, it would be nice to start with something I can trust, plus I thought it would be an interesting project. Sometimes it's nice to do something over the top just because one can if you know what I mean :)
 

Thread Starter

Kelvinchill

Joined Sep 11, 2017
6
With no equipment descriptions, part numbers, datasheets, or a system wiring diagram it is impossible to say. But ...



I will draw it up with the components and come back later on this. I mention linking the negatives so all the voltages would be referenced to the same "zero" point I don't know enough about this sort of this which is why I asked as I had a niggling feeling that if the output of the buck converters could "float around" it may be the case that the potential difference between two devices could be higher than anticipated??
Necessary (maybe).

Probably (maybe).

If by ground you mean earth ground, *not* the power line neutral, then yes (maybe).

OH yes Earth Ground not Power line neutral

For 50% fan voltage, a first-pass approximation of the resistor needed is 1200 ohms at 12 W. A 25 W resistor is not "huge" in absolute terms, but we don't know how much room is available.

That's what I got it to (12w) my concern was that it would be dissipating most of that as heat or is that assumption wrong?

ak
 

crutschow

Joined Mar 14, 2008
34,285
Killing a ground loop requires a knowledge of how all the grounds are connected.
You need a good diagram showing all the connections (signal, power, and ground) for all the devices.
Then you can determine how to connect everything to kill the loop(s).
 
Killing a ground loop requires a knowledge of how all the grounds are connected.
You need a good diagram showing all the connections (signal, power, and ground) for all the devices.
Then you can determine how to connect everything to kill the loop(s).

@crutschow nailed it.

You eliminate ground loops by design.

This is audio and I initially see the hammer and the nail problem. The audio devices may use 1/2 Vcc as the 0V reference, so all bets are off unless those power supplies are isolated. Common non-isolated converters won;t work.

The system ground is your starting point. There has to be one of these. If something is powered on the other side of the room, you could already have two. Neutral and ground are connected at one point for the premise and then there are tow different points because they are on different sides of the room.

You may loose fidelity with an audio isolation transformer. Professional stuff should be all XLR and balanced and they will get along just fine. Cabeling makes a difference. Single-ended vs differential makes a difference, Using twisted pair shielded cables with the ground connected at ONE END ONLY makes a difference.

Differential line drivers and recievers would be a good thing or single-ended to differential converters would help too.

Sometimes just adding a small resistor like 5.1 ohms in the ground side is all you need.

The general strategy is to have all high current grounds combine at one point and have all of the signal grounds combine at one point. At that one point they are connected together. I had one piece of instrumentation, that allowed me to externally move this reference. It was a really ice idea.

A scope would really help.

Assume you have an amp and a pre-amp (mixer) that don;t hum when connected together. So, you start connecting stuff (inputs) and look at the output reference with respect to the real reference with a voltmeter and/or scope.

==

I made a very nice universal wall wart adapter using these http://www.ebay.com/itm/like/321751991546?chn=ps&dispItem=1 supplies off of ebay. Hammond makes a nice case with a translucent color. So, if the LEDs are red, use a red case.

These can be set for both voltage and current limit and this can really replace just about any wall wart.

The input i made as the most common DC power coaxial power jacks, the 5.5/2.1 and 5.5/2.5 V. One of those, i can usually find. 12 v @ 1A is a common adapter.

For polarity reversal on the input, I have to move the connections on the terminal strip. This isn't a problem when you use wire ferrules

www.powerstream.com has some nice ways of getting from the common size DC barrel connectors to other sizes and they also have a polarity reversal adapter.

I needed a 6V adapter and i made one quickly. I've used it for a few years now. I wish the design had a way to turn off the display/ I even bought the 6V adapter that I needed too.

The output was a Radio-Shack adapt-a-plug pigtail which isn't a great product, but it offered lots of tips to fit just about anything.

If you use an outlet strip, that could also be bad.

Ground loops I can solve, but sometimes I miss one or six. I did just that.

We had a home made box that we made so it was easy to calibrate and manually control 6 mass flow controllers. the box has a +-15 V supply and a 5 V supply that controlled 6 MFC's. the setpoint used the ground at the MFC that was 15 foot away.

When it was time to use a computer to control these with 0-5V signals, the reference for each one was different, so the values interacted. Using current output would have worked, but i did not have that available. A difference amplifier had to be added for each channel to work properly.

The standard technique is to use pseudo differential as inputs to a computer and current or isolated current output for voltage outputs. the current is converted to a voltage at the instrument.
 

AnalogKid

Joined Aug 1, 2013
10,987
Common non-isolated converters won;t work.
Yes, they probably will. I've done this in several MIL sensor and radio systems from watts to hundreds of watts- one bulk power supply with full safety and performance ratings, followed by a group of various down converters.

Audio gear running on a single-ended wall wart pretty much by definition is not DC coupled throughout, so interconnecting multiple pieces of this stuff, powered by non-isolated regulators of any kind, is not a de facto problem.

Of more concern are the noise frequencies generated by multiple unsynchronized switching regulators and the mixed sub-harmonics in the audio band. Unless heat removal is a problem, I strongly recommend linear regulators where possible.

ak
 
I host Karaoke 6 nights a week for many years. (similar job to Dj'ing). I found my gnd hum problems
to be the video connections. You have control over everything you bring in, can be on the same outlet grounded or not grounded.
If you connect to a TV at the gig, it is the ground on their coax input (i use the AV) that introduces a different ground from the street.
If you connect to video, and hum starts with that connection, see if it's possible to remove their other inputs while using the display.
 
Top