strange ripple noise in DC power supplies ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello,

I have 2 independent DC power supplies sharing the same ground.

I am using a full wave rectifier using center tapped transformers for both power supplies.

Like this:

File:Fullwave.rectifier.en.svg - Wikipedia, the free encyclopedia

One of the power supplies outputs 1 amp, the other about 20mA.

Whenever they are both running at the same time I get a strange 100Hz ripple noise on the ground. It is about negative 40mV. I have attached a picture of what it looks like.

It only happens when they are both running at the same time. If they run independently this noise is not present.

There is some type of interaction going on that I cannot figure out.

Any ideas?
 

Attachments

  • noise.jpg
    noise.jpg
    204.7 KB · Views: 173
AX tech editor
Joined 2002
Paid Member
Hello,

I have 2 independent DC power supplies sharing the same ground.

I am using a full wave rectifier using center tapped transformers for both power supplies.

Like this:

File:Fullwave.rectifier.en.svg - Wikipedia, the free encyclopedia

One of the power supplies outputs 1 amp, the other about 20mA.

Whenever they are both running at the same time I get a strange 100Hz ripple noise on the ground. It is about negative 40mV. I have attached a picture of what it looks like.

It only happens when they are both running at the same time. If they run independently this noise is not present.

There is some type of interaction going on that I cannot figure out.

Any ideas?

This looks like the charge pulses for the reservoir capacitors causing a voltage across the ground wire.
Depends on the length and size of the ground connetion, and depending where you measure.
What is the level when you run either only the 1A or the 20mA separately?

jan didden
 
This looks like the charge pulses for the reservoir capacitors causing a voltage across the ground wire.
Depends on the length and size of the ground connetion, and depending where you measure.
What is the level when you run either only the 1A or the 20mA separately?

jan didden

Hi Jan,

I just checked it again. I was mistaken. When I run the 1 amp seperately I am seeing this signal on the ground. What I mean by that is that I have my scope ground hooked up to my output ground and the scope probe on the center tap of the output side of the transformer (which is ground in my set up). It is the 1 amp PS causing the issue. Very weird......
 
This looks like the charge pulses for the reservoir capacitors causing a voltage across the ground wire.
Depends on the length and size of the ground connetion, and depending where you measure.
What is the level when you run either only the 1A or the 20mA separately?

jan didden

Yes, I connected a wire from the center tap transformer input to the output ground and I still see the signal! But if I check ground at different places around the circuit the pulses are less or even non existent.

I do not understand "charge pulses for the resovoir caps causing a voltage across the ground wire". I do have 30,000 uF for the 1 amp circuit and 15,000 uF for the 20 mA circuit. Could the large amount of capacitance be causing the issue?

Sounds like that may be my problem. I do NOT have a ground plane for this -

?
 
You are seeing charging pulses in a 'ground'. True ground only exists at one and only one point in a circuit. All other points, even when connected to ground with a nice thick conductor, are not ground unless no currents flow along the conductor. You decide which point you call ground.

The bigger the caps, the worse will be the charging pulses so the worse will be the voltage drop along a 'ground' conductor. Treating the transformer secondary centre-tap as ground is a common mistake.
 
The charging circuit, from transformer through the rectifier to the smoothing caps, must not share any common wire or wiring with the load circuit that receives the smoothed DC supply.
These two separate circuits must be grounded at ONE POINT.

That fixed it! I was taking my "ground" to different components from different locations off of the ground wire. I changed it so all the grounds come from a central point (star grounding?) and problem is now gone.

OK, I have a question and I bet it is a can of worms... (ask 10 people and 5 will answer one way, 3 the other and 2 something totally different). If I lay this out on a board am I better to:

1) Have a ground plane

2) Set up a "star" grounding layout?

I realize one way done properly will always be better than the other done improperly, but for my setup of two independent single tap transformers with 2 independent DC power supplies sharing the same ground which way would you lean?


Thanks!
 
If there were only 2 power supplies in your circuit, then grounding was not an issue at all. But there must be some other circuits which can be disturbed by gnd noise. You should design the whole system, not the PSU alone!

1) or 2) ? These general rules/strategies are misleading. A designer always have to know that there are noise sources, and transfer paths to the useful signal (s). You have to analise and eliminate/reduce them below the acceptable level. First you have to identify the noise sources (in this case one of them is the charging current of the puffer capacitor) and entry points to the signal path (which is unknown by us, since we dont know what do you want to supply).

1) The star point grounding can be almost perfect or absolutely terrible depending on the details.

2) The grounding plane ensures some degree of noiselessness without considering any details, but to make it perfect attention is needed also.

3) You missed the chain style grounding. If the signal path is straightforward, and you can apply separate PSUs to every stage, then you have the chance to avoid every ground loop problems entirely! Be careful, the whole system have to be free from ground loops!

4) You can forget GND noise problems if you use differential signal paths.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.