Hello maybe someone knows something to mitigate this issue,

I made a PCB that is powered by trucks so its power is main source is from the battery or some socket made by the truck maker. Supply is always 12VDC to 24VDC and to power my PCB I use a DC-DC converter

We found out that the main power supply or socket where I power the PCB is also connected to very noisy devices as coils, motors or the alternator from the truck and they introduce in our PCB a positive signal going from few mV to even 5V in ground net, when it should be always 0V or close to 0V, the spike only lasts some miliseconds but it happening often therefore is affecting the stability of the whole PCB

Do you have any idea of a component to erase this voltage from ground network at the input power of the PCB before the the DC-DC?

Thank you,

 

How do you know that your ground net voltage varies? Varies relative to what? What is your reference point (your true zero)?

 

The short answer is you need a lower-impedance ground return path, eg. a fatter cable to the battery ground terminal. This would make it much harder to have any significant voltage appear at your board's "ground". I'm assuming you're measuring board ground against battery ground?

 

How do you know that your ground net voltage varies? Varies relative to what? What is your reference point (your true zero)?

I know that it varies because I measured it with a scopmeter and you can see the variation in miliseconds. The reference point the true zero

 

The reference point the true zero

True zero? Please define.

 

Please define.

= reference potential. Any voltmeter has two connections, since voltage = potential difference. One of those connections is a reference potential, which can be regarded as 'zero' for the purpose of the measurement. I'm curious to know where the TS's scope ground was connected when measuring the ground net voltage.

 

I'm curious to know where the TS's scope ground was connected when measuring the ground net voltage.

Right, that's exactly what I was asking, how the TS was defining "true zero".

 

= reference potential. Any voltmeter has two connections, since voltage = potential difference. One of those connections is a reference potential, which can be regarded as 'zero' for the purpose of the measurement. I'm curious to know where the TS's scope ground was connected when measuring the ground net voltage.

I checked within different pads of my PCB and other connections within the truck socket

 

That is not specific enough to answer the question.
Any ground conductor carrying significant current will experience 'ground bounce'. A 'star ground' system is probably required to overcome that, especially in an automotive environment. Failing that, depending on what your pcb circuit is doing, perhaps a converter with galvanic isolation between the input DC and the output DC might be a solution.

 

That is not specific enough to answer the question.
Any ground conductor carrying significant current will experience 'ground bounce'. A 'star ground' system is probably required to overcome that, especially in an automotive environment. Failing that, depending on what your pcb circuit is doing, perhaps a converter with galvanic isolation between the input DC and the output DC might be a solution.

If what you mean with star ground is to separate grounds, that has been made already. Also at the input of the DC-DC I use a common-mode-filter for positive and negative networks and a pi-filter just for the positive network. All of this just to get a power source as clean as possible but there’s something extra that I didn’t think about and for the moment I don’t know how to solve it

 

Do the DC-in and DC-out of the converter share a common ground, or are their respective ground points galvanically isolated?

 

Do the DC-in and DC-out of the converter share a common ground, or are their respective ground points galvanically isolated?

The DC-in (positive and ground) pass through X1 capacitor, then both pass through a common choke filter (I think both lines have 6-7 turns around the core, then X2 capacitor. The DC positive network goes to the DC-DC converter through a pi-filter and the DC negative line is then shared with the PCB in just at this single point. The DC-DC ground is shared with the whole PCB ground, but the noise is comming from main external power. I attach a very simplified schematic

 

If there is any current-carrying galvanic path between 'GND before filter' and 'PCB ground', then 'PCB ground' potential will be subject to noise as a result of that current. That path would effectively bypass the filter.

 

If there is any current-carrying galvanic path between 'GND before filter' and 'PCB ground', then 'PCB ground' potential will be subject to noise as a result of that current. That path would effectively bypass the filter.

Main power tracks are not close to any other tracks or even the poured ground cooper. There is a physical isolation space between the tracks of positive and negative tracks before the filter and the whole PCB

 

The problem is in the truck wiring, which does not go directly to the battery. Thus noise voltage is developed because of noise currents flowing in both the positive and negative sides of the circuit between the connection point and the battery terminals.Filtering will remove some of the noise, but probably not enough. a separate direct connection to the battery terminals would be the simple solution. An adequate filter will be big and heavy and probably much more expensive.

 

The DC-in (positive and ground) pass through X1 capacitor, then both pass through a common choke filter (I think both lines have 6-7 turns around the core, then X2 capacitor. The DC positive network goes to the DC-DC converter through a pi-filter and the DC negative line is then shared with the PCB in just at this single point. The DC-DC ground is shared with the whole PCB ground, but the noise is comming from main external power. I attach a very simplified schematic
View attachment 169138

Thanks for the drawing.
So, are you calling "true ground" the battery terminal, or where in that drawing?

 

The problem is in the truck wiring, which does not go directly to the battery. Thus noise voltage is developed because of noise currents flowing in both the positive and negative sides of the circuit between the connection point and the battery terminals.Filtering will remove some of the noise, but probably not enough. a separate direct connection to the battery terminals would be the simple solution. An adequate filter will be big and heavy and probably much more expensive.

When you talk about big and heavy filter do you refer just to the choke? Or what kind of filter do you mean? Big and heavy are relative terms

Thanks for the drawing.
So, are you calling "true ground" the battery terminal, or where in that drawing?

A true zero could be the a close connection to the battery, a connection not influence by the rest of the wiring, coils, etc

 

All of the parts would be fairly big and fairly heavy. A shunt capacitor, such as shown, needs to have an impedance at the noise frequency less than the source impedance at that frequency. That is why a filter also includes series inductors. But the series inductors must also be able to carry the load current without excess voltage drop or magnetic saturation. So after you calculate the values needed for the application then look up the parts and you would see what I mean. And of course all of it depends on how much power you need to pass through that filter.

 

Thanks for your answer Mister.

I will try to make the calculations, I am not expert on the field though. Power isn’t so much, the whole PCB won’t use much power, current won’t be more than 500mA so power through the filter will depend on the battery

Again thanks for your interest

 

OK, if the current drawn is only 500mA then the inductors will not be so very large, nor the capacitors. So the circuit that was shown will be suitable and a reasonable size. But there is a caution, which is that whatever is connected to that filter must not otherwise be connected to the vehicle ground, since that would bypass half of the filter and reduce the effectiveness. And you do need to have some estimate of the frequency spectrum of the noise that the filter is intended to block.