Ground loops / star ground / shielding theory

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!

Moderators: luketeaford, Joe., lisa, Kent

User avatar
L-1
Super Deluxe Wiggler
Posts: 2171
Joined: Wed Dec 14, 2011 8:20 am
Location: Belarus
Contact:

Post by L-1 » Mon Apr 28, 2014 7:42 am

daverj wrote:The ribbon doesn't get calculated in at a 1A load, since there are no modules that use 1A. Each ribbon drives an individual module that might range from 10ma to maybe 150ma.

:hyper: L-1 Vocoder has a bit less than 1A at V+.

User avatar
mskala
Super Deluxe Wiggler
Posts: 2614
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:

Post by mskala » Mon Apr 28, 2014 8:08 am

L-1 wrote:Graham, you used one heatsink for both regulators.
But MFOS recommended used separate.
I can't find any explanation about this in a datasheet and somewhere else.
The mounting tab on the regulator is connected to its input voltage. So if you use uninsulated hardware to attach it to a grounded chassis, or two of them with different input voltages to the same heat sink, you're in a world of hurt. But from the photo it looks like he's using insulated hardware to make a thermal but not electrical connection between the regulators and the heat sink and avoid this problem.

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3279
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Post by Graham Hinton » Mon Apr 28, 2014 8:53 am

L-1 wrote: Graham, you used one heatsink for both regulators.
But MFOS recommended used separate.
I can't find any explanation about this in a datasheet and somewhere else.

How it's important?
The middle pin on a TO-220 package is connected to the tab so there is a thermal pad under each regulator to electrically insulate it.

MFOS are obviously trying to lose the thermal resistance of the pad, but like most people use an inadequate thermal resistance heatsink. When the regulator gets too hot it goes into thermal shutdown and the voltage drifts slowly down. This is why you hear all the voodoo about "use at 60%" for most PSUs. Large heatsinks get very expensive so there is a tendency to try to save the money, but this ends up compromising the spec. There are other ways to get the thermal capacity and surface area necessary.

You need to lookup how to do thermal resistance calculations. It is just like a series resistors circuit, but Watts and degC.

User avatar
daverj
Vintage Video Wiggler
Posts: 8514
Joined: Fri Jun 19, 2009 11:09 am

Post by daverj » Mon Apr 28, 2014 1:30 pm

neil.johnson wrote:
daverj wrote:The ribbon doesn't get calculated in at a 1A load, since there are no modules that use 1A. Each ribbon drives an individual module that might range from 10ma to maybe 150ma.
...As long as you're not using one of those flying busboards.

Neil
True. But I was talking about the individual ribbons attached to each module. If using a uZeus style flying bus board then you would calculate the ribbon that is part of the bus board the same as you would the copper traces on a distro card, but still not count the individual module ribbons at full power supply load, since they don't carry the full power supply load.

except in the case of......
L-1 wrote:
daverj wrote:The ribbon doesn't get calculated in at a 1A load, since there are no modules that use 1A. Each ribbon drives an individual module that might range from 10ma to maybe 150ma.

:hyper: L-1 Vocoder has a bit less than 1A at V+.
:zombie: :doh: :zombie:

User avatar
Nantonos
Super Deluxe Wiggler
Posts: 1275
Joined: Sun Mar 03, 2013 1:00 pm
Location: Boston, MA

Re: Ground loops / star ground / shielding theory

Post by Nantonos » Tue May 06, 2014 5:19 am

Graham Hinton wrote:
Nantonos wrote: I haven't measured the voltage drops over 0V and -12V yet. Does 49mΩ seem a reasonable value for that setup?
It is in the right order, but you did not mention the lengths of anything. Compare the measured resistance with the theoretical value from the sum of the parts. 0.5m of 32/0.2 is about 8mΩ and two 0.2m of 28awg ribbon is about 20mΩ and I would expect every connection to be 5mΩ or more.
The ribbon is 260mm and is indeed marked 28awg (0.32mm). The PCB bussboard is 390mm long. The cable from PSU to bussboard is 550mm long and each rail (+12V, -12V, 0v) is carried by one, solid core, 1.5mm cable. So the ribbon would be about 26mΩ, two connections for 12V is 2.5mΩ at each end = 5mΩ, approximating the cable to 15awg an online calculator gives 10mΩ/m so 5.5mΩ. Two more connectors for spade terminals is another 10mΩ? Subtracting that from the total leaves none for the PCB track, so some of the approximated resistances must be too high.

I repeated the measurements on -12V as follows:
R = 1.0965k
V = -11.998V
so I = 10.9mA
Vdrop = 0.00005V (0.5mV)
so R = 45mΩ, comparable to the +12V measurement of 49.1mΩ

and then for 0V. Difference current is 88.8mA
Vdrop = 0.0019V (1.9mV)
R = 21mΩ

That seems reasonable as there are six conductors in the ribbon instead of two, and six connectors in parallel instead of two.
Graham Hinton wrote:At 1A load that will be a ~50mV common drop, which is too high.
It seems to me the common drop (on 0V) at 1A would be 21mV? Either way that is a pretty high noise floor compared to a 10Vp-p signal.

Looking at the figues, getting rid of ribbon cables would be the biggest improvement (26mΩ). Doubling up cables from PSU to bussboard would help a tiny bit (from 5mΩ to 2.5mΩ).

The contribution from the PCB busboard and spade connectors and euro connectors all together is around 18mΩ. What are the equivalent figures for your solid aluminium bussbars and replacement module cables?
Graham Hinton wrote:I would be interested to know if that PSU can actually deliver 1A continuously and what the ripple measurements are. Do you have a scope to view the ripple?
I do not as yet have a scope unfortunately. Nor (to pre-empt well meaning replies from people living in Silicon Valley) do I live somewhere where a secondhand Tek scope can be picked up for $40.

My meter does have a setting "AC+DC volts true RMS" but this is a sum, it does not give the AC and DC components separately. Testing for full 1A load would require something like 25W 12Ω resistors, I don't have those either. The total load of my present modules is around 500mA on +12V and 360mA on -12V.
ondes | current rack
I am afraid a firmware change will not be able to turn a rather expensive 16-bit DAC into a 16-bit ADC, and flip all those op-amps :)

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3279
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Re: Ground loops / star ground / shielding theory

Post by Graham Hinton » Wed May 07, 2014 6:22 am

Nantonos wrote:It seems to me the common drop (on 0V) at 1A would be 21mV? Either way that is a pretty high noise floor compared to a 10Vp-p signal.
Yes, if 1A is flowing, but that is not the same thing as the noise floor. A noisy 0V common will not help the snr. Every inverting op-amp or adder with its +ve input connected to 0V may be considered as a non-inverting amplifier with a gain of 2 or more acting on its +ve input. So anything on the 0V gets amplified and added in with all the thermal noise from resistors and all the other gains in the signal paths. OTA based VCAs and VCFs raise the noise floor quite a bit due to the attenuation and then amplification at each stage.

AC drops on the power rails will be less than the DC because they are caused by dynamic load changes. Working out the resistances at DC gives an indication of how they will effect the AC and where the best places are to reduce interaction between modules.
Looking at the figues, getting rid of ribbon cables would be the biggest improvement (26mΩ). Doubling up cables from PSU to bussboard would help a tiny bit (from 5mΩ to 2.5mΩ).
Each individual ribbon cable does not carry a lot of current whereas the PSU feed carries the total. So you should be comparing the I.R products for each section.
What are the equivalent figures for your solid aluminium bussbars and replacement module cables?
On a recent set of 900mm long bars I had 0.75mV drop from the PSU at 1A and 80µV from the centre to a load at the end. In practise the load would be distributed so that is worst case figures. At 2A the feed drop would double to 1.5mV, but with 1A loads in each half of the bars everything would be within 80µV on the bars and a further drop in the module cables depending on each load.
I do not as yet have a scope unfortunately. Nor (to pre-empt well meaning replies from people living in Silicon Valley) do I live somewhere where a secondhand Tek scope can be picked up for $40.
eBay.fr? More on eBay.de and other EU countries for <100€.
Testing for full 1A load would require something like 25W 12&#937; resistors, I don't have those either.
The resistors are the easy part, you can buy them from most electronics distributors for between £1 to £2 each. They will need to be bolted to a decent large heatsink.

User avatar
thresholdpeople
Veteran Wiggler
Posts: 574
Joined: Sat Dec 21, 2013 6:17 pm
Location: NYC

Re: Ground loops / star ground / shielding theory

Post by thresholdpeople » Thu Sep 15, 2016 7:21 pm

Graham Hinton wrote: This is the type of distribution I am making now:

Image

With a very short heavy cable feeding from a PSU this makes the Rds as low as they can possibly be. I can make the busbars any length and number of taps, for wooden cases they may be screwed directly to the sides. Rc is limited by whatever gauge of wire fits the modules' mating connector, but the individual cables may be longer and it is better to have longer runs there than in the common cable.
Hi Graham. I'm curious what alloy of aluminum you used for the busbars. Did you opt for the 6101, or did you find the resistance was low enough using whatever aluminum at hand.

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3279
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Re: Ground loops / star ground / shielding theory

Post by Graham Hinton » Fri Sep 16, 2016 5:01 am

thresholdpeople wrote:I'm curious what alloy of aluminum you used for the busbars. Did you opt for the 6101, or did you find the resistance was low enough using whatever aluminum at hand.
The resistivity of aluminium alloys is higher than pure aluminium, but it is still a lot lower than most other metals. Brass, which is often used for connecting blocks, is quite a bit higher.

If you want a lower resistance, just use a thicker piece. The bars in this picture were 15x7mm, I use 15x10mm as standard now and have used 20x20mm for the 0V inside long cases.

User avatar
thresholdpeople
Veteran Wiggler
Posts: 574
Joined: Sat Dec 21, 2013 6:17 pm
Location: NYC

Post by thresholdpeople » Fri Sep 16, 2016 7:21 pm

Thanks Graham. It's easy for me to get caught up trying to pick the best spec, so the insight into your thought process is a great help.

I think I'm going to do something similar for the synth I'm working on. Chassis made out of aluminum, with a PTFE plate that holds the bus bars.

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1374
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Sat Sep 17, 2016 4:36 am

thresholdpeople wrote:Thanks Graham. It's easy for me to get caught up trying to pick the best spec, so the insight into your thought process is a great help.

I think I'm going to do something similar for the synth I'm working on. Chassis made out of aluminum, with a PTFE plate that holds the bus bars.
Be careful machining PTFE - you don't want to get it hot.

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
Poldenstein
Wiggling with Experience
Posts: 446
Joined: Sun Aug 04, 2013 10:01 pm

Post by Poldenstein » Sat Sep 17, 2016 5:54 am

amdagan wrote:Back to those busbars... has anyone done a cheapo diy version of Graham's setup?
Interesting thread.
At module side I went for a solution very similar to Graham's. I used discrete 24 awg cable crimped to duponts connectors (used a 5 bucks Chinese crimper wihout problems).
At PSU side I bypassed bus bars and went with an old school harness design, twisting all the copper wires together and using one BIG terminal to crimp them. The downside is that one cannot modulate the number of modules to feed.
Here's the thread
viewtopic.php?t=134925&highlight=

User avatar
thresholdpeople
Veteran Wiggler
Posts: 574
Joined: Sat Dec 21, 2013 6:17 pm
Location: NYC

Post by thresholdpeople » Sat Sep 17, 2016 7:25 am

neil.johnson wrote: Be careful machining PTFE - you don't want to get it hot.

Neil
Well, after exploring pricing options for PTFE, I may use ABS instead, so that'll be much easier to machine :)

User avatar
diablojoy
Super Deluxe Wiggler
Posts: 1617
Joined: Fri Jan 22, 2010 9:03 pm
Location: melbourne

Post by diablojoy » Sat Sep 17, 2016 9:04 pm

Well, after exploring pricing options for PTFE, I may use ABS instead, so that'll be much easier to machine smile
Hmm perhaps nylon ?
I found a good source is cheap kitchen chopping blocks from $2:00 shops about 10mm thick cut out what you need , I use it for quite a few things.
it might work ok for this as well.
It probably wont happen today but if it does it definitely wont go smoothly.

User avatar
thresholdpeople
Veteran Wiggler
Posts: 574
Joined: Sat Dec 21, 2013 6:17 pm
Location: NYC

Post by thresholdpeople » Sat Sep 17, 2016 9:33 pm

Thanks for the tip about cheap nylon sources. Nylon is prone absorbing moisture, so I'm going to avoid it. It's pretty humid in my area, and I don't air condition my studio.

At the moment I'm reading about GPO-3, which is a polyester fiberglass laminate material. It's UL listed and used for bus bar applications in industrial settings. It's about twice as heavy as PTFE or ABS, but so much cheaper than PTFE, and a bit cheaper than ABS.

The benefit I see to using it, or ABS, over PTFE is the tensile strength of the material, so I can use the insulation material to also brace and support the mid section of my chassis, and so get away with using thinner aluminum.

I'm already planning on ordering bulk aluminum for the chassis and the same supplier sells a large variety of plastics, so I get to be as choosy as I want at the moment. Due to budget constraints, I only get one shot to pick the right thing... luckily I'm an obsessive researcher. :hihi:

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1374
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Mon Sep 19, 2016 4:15 am

Nylon should be fine in this application - you're only dealing with 12V not 100s or 1000s of volts :woah:

The issue with machining PTFE is called Polymer fume fever

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
mskala
Super Deluxe Wiggler
Posts: 2614
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:

Post by mskala » Mon Sep 19, 2016 4:43 am

neil.johnson wrote:The issue with machining PTFE is called Polymer fume fever
Also that even if you manage to machine it without poisoning yourself, it's not good at holding an exact shape. Depending on the application, that may be no problem or it may be a dealbreaker.

User avatar
taichber
Common Wiggler
Posts: 149
Joined: Mon Jul 06, 2015 11:34 am
Location: Munich

Re: Ground loops / star ground / shielding theory

Post by taichber » Fri May 05, 2017 4:28 pm

Graham Hinton wrote:
This is the type of connector I supply for Eurorack now:
Image
It takes 7/0.2 or 16/0.2 wire and, ironically, is the type intended to mate with unshrouded headers.
You can tell me what kind of connector this is or where I can buy the connector only? I want to DIY a few cables.
Thanks!

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3279
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Re: Ground loops / star ground / shielding theory

Post by Graham Hinton » Sat May 06, 2017 4:49 am

taichber wrote:You can tell me what kind of connector this is or where I can buy the connector only? I want to DIY a few cables.
You can buy only the shells and contacts, but they won't be much use without the special £300 crimp tool.

User avatar
taichber
Common Wiggler
Posts: 149
Joined: Mon Jul 06, 2015 11:34 am
Location: Munich

Re: Ground loops / star ground / shielding theory

Post by taichber » Sat May 06, 2017 5:23 am

Graham Hinton wrote:
taichber wrote:You can tell me what kind of connector this is or where I can buy the connector only? I want to DIY a few cables.
You can buy only the shells and contacts, but they won't be much use without the special £300 crimp tool.
Where? Which crimptool? I like DIY :hihi:

Thanks for any help!

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3279
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Re: Ground loops / star ground / shielding theory

Post by Graham Hinton » Sat May 06, 2017 3:36 pm

taichber wrote: Where? Which crimptool?
There are several different makes of similar "board to wire" connectors, but you have to use the tool for the contact you choose. They are not interchangeable.

If you can buy from Farnell in Germany check their Harwin M20 range, otherwise search for <board to wire 0.1" 2 row>.

If you need any power cables made up, we have the correct tooling for all these Power Cables

User avatar
taichber
Common Wiggler
Posts: 149
Joined: Mon Jul 06, 2015 11:34 am
Location: Munich

Re: Ground loops / star ground / shielding theory

Post by taichber » Sun May 07, 2017 3:22 pm

Graham Hinton wrote:
taichber wrote: Where? Which crimptool?
There are several different makes of similar "board to wire" connectors, but you have to use the tool for the contact you choose. They are not interchangeable.

If you can buy from Farnell in Germany check their Harwin M20 range, otherwise search for <board to wire 0.1" 2 row>.

If you need any power cables made up, we have the correct tooling for all these Power Cables
Many thanks. This was helpful. Now I know how to search.

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Mon May 08, 2017 9:27 am

Just going to leave this here, Keith Armstrong over the course of several PDFs has left a big impression on how to properly tackle EMC issues.

This document includes advice about equipotential bonding over the "single point" technique, but there is *a lot* of advice in other PDF documents on the site, that are all informed from extensive work in the field.

http://www.cherryclough.com/PCB-design- ... es-for-EMC

For my products I actually ended up contradicting the single point ground advice as given by AES48 and deliberately bonded the chassis of my device to the PCB 0V. The reason was that once you think about the parasitic inductance, where the 0V line is tied through minimum 10s of meters cabling to the earth bond (this path estimated at say, 1mH of inductance). This means that for a low inductance 0V you need an RF reference which means a piece of metal such as a chassis or an unbroken copper plane).

User avatar
taichber
Common Wiggler
Posts: 149
Joined: Mon Jul 06, 2015 11:34 am
Location: Munich

Aluminium busbar

Post by taichber » Tue May 09, 2017 11:24 am

question for Graham:

What kind of aluminium do you use for your busbar?
Most of the aluminium rails (e.g. 20mm x 8mm) are AlMgSi0,5.

The best conductivity has AL99,5. But this is quite expensive.

Is AlMgSi0,5 ok with e.g. 20x8 size?
What screws do you use for the busbar? Steel?

Thanks!

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3279
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Re: Aluminium busbar

Post by Graham Hinton » Tue May 09, 2017 2:37 pm

taichber wrote: What kind of aluminium do you use for your busbar?
Most of the aluminium rails (e.g. 20mm x 8mm) are AlMgSi0,5.
Yes, usually EN-AW 6060 which is the same alloy.
Is AlMgSi0,5 ok with e.g. 20x8 size?
Don't get confused comparing resistivity. The final end resistance is all that matters, i.e. after all the holes are tapped. If it isn't low enough use a larger piece. I usually get around 50 to 60 microohms on a 17" bar. I made a 2.7m bar once with 40x10mm and got 570 microohms.

To measure this put it in series with a 12 ohm power resistor and put a 12V PSU across both and measure the voltage drop.

The short linking cables to the PSU have a much higher resistance, I try to keep them below 10 milliohms. This needs more attention than the bars.
What screws do you use for the busbar? Steel?
Stainless steel with Copaslip on the threads to prevent seizing. You also need a zinc jointing compound on the terminals against the aluminium face.

User avatar
taichber
Common Wiggler
Posts: 149
Joined: Mon Jul 06, 2015 11:34 am
Location: Munich

Re: Aluminium busbar

Post by taichber » Tue May 09, 2017 4:02 pm

Graham Hinton wrote:
taichber wrote: What kind of aluminium do you use for your busbar?
Most of the aluminium rails (e.g. 20mm x 8mm) are AlMgSi0,5.
Yes, usually EN-AW 6060 which is the same alloy.
Is AlMgSi0,5 ok with e.g. 20x8 size?
Don't get confused comparing resistivity. The final end resistance is all that matters, i.e. after all the holes are tapped. If it isn't low enough use a larger piece. I usually get around 50 to 60 microohms on a 17" bar. I made a 2.7m bar once with 40x10mm and got 570 microohms.

To measure this put it in series with a 12 ohm power resistor and put a 12V PSU across both and measure the voltage drop.

The short linking cables to the PSU have a much higher resistance, I try to keep them below 10 milliohms. This needs more attention than the bars.
What screws do you use for the busbar? Steel?
Stainless steel with Copaslip on the threads to prevent seizing. You also need a zinc jointing compound on the terminals against the aluminium face.
Many thanks Graham. Very useful information!

Post Reply

Return to “Music Tech DIY”