noise on the supply on input of the LDO

[glias]

Hello all,
I have a problem with my board :
I have one analog preamp board which +/-6V supply from a power supply and another one which is connected to the preamp to control a mux which is on the preamp. The second board use a µC and control a LCD display. Thses boards are connected with a flat cable (awg28 about 1.5meter with 20 conductors) to supply the second board and also permit to control the second board.
On the second board, there is a LDO to provide a 5V to supply µC and the display (HCMS2903). The output of the regulator is quite clean, but I have a problem on the input of it : there is a very high AC ripple (about 95mV peak to peak) and as the voltage supply the op amp on my first board, I have this ripple on the preamp board too...

The total consumption of the second board (µC+ display+leds) is about 70mA (average value, I can't tell you the peak current).
I oberved that the display is the cause of this ripple because, when I have changed the program to disable it, the ripple became very low.
I tried to place a big cap on the supply of the display : it doesn't change anything on the input of the LDO, so I tried to place a big cap (1000µF) on the input of the LDO, It improves a little the ripple, but not enough...

What can I do ? place another big caps on the input of the regulator ? or I'm wrong and there is another thing to check and improve ?

I hope that you could help me

thanks

I forget to mention you that I have placed a tantalum cap on the supply of the display (it's not specified on the schematic) and one 100µF-16V electrolytic cap on the input of the LDO and another one on the output as specified in the datasheet.

Without 1000µF-16V electrolytic cap :



With 1000µF-16V electrolytic cap :

 

[philcorb]

Hi,

Just a couple of suggestions

1) can you power the display off a separate power supply (lab supply, battery, anything) to make sure it real is that causing the ripple

2) perhaps using RC decoupling onto the display power supply might help. That will help decouple the noise from the supply and onto ground. If you know the current consumption of the display, the IR drop of putting on a series resistor can be calculated.

 

[rhaynes]

Everything is working as designed. It will be impossible to eliminate that kind of ripple because the circuit draws current and there is resistance in the supply. Your problem is a little more complex than you may realize because there are probably ground loop problems also. Before I make any specific suggestions, is this what your basic power wiring looking like? The resistors represent the wire resistance.

Ray

 

[glias]

Hello
sorry for my late answer, see the diagram ; I think the same thing, I must be ground loop

My PCB board is FR4 4 layers with ground plane on bottom and inner layer.
I also put the gerbers to permit you to comment my layout.


regards

 

[glias]

Nobody to help me ?

 

[rhaynes]

I was planning on drawing a nice diagram explaining the issue but since I try and squeeze these posts in at work I don't have time right now. So let me just say a few things. Since all supplies have some output resistance when the load current varies the voltage will also. Draw a schematic showing all the resistances including wiring and traces. Guess at the resistance and using the actual current determine the voltage drop across each resistance. You'll see that things get messy. So we typically use regulators to isolate these variations. With the power being fed to brd#1 then on to brd#2 and since you don't have regulators on brd#1 and current changes on brd#2 are going to be seen as voltages changes on brd#1. There are several options:
1. Run separate connections from the +/-6V supply to brd#1 and brd#2 instead of running brd#2 power through brd#1.
2. Supply a higher voltage to brd#1 and regulate to +/-6V and send the unregulated to brd#2
3. Run brd#2 off a different supply than brd#1.
4. As phicorb mentioned above try and reduce the current noise at the source by adding an RC (large capacitance, small R) at the LCD.
5. Increase the power supply rejection ratio at brd#1.

Let me know if you need more detail on any of the above.

Ray

 

[glias]

Hello Rhaynes
I'm so sorry for my late answer, I'm going to prepare you a more detailled drawing of wiring and send you a picture (private way if I can) of my boards to see better the environement.

regards