Isolated DC-DC converter excessive output noise !

Go To Last Post
18 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi!

I have designed an isolated DC-DC converter circuit based on the Texas Instruments DCR011205. I have created the PCB according to datasheet, and I have used the noise reduction ideas provided but TI tech support(see pdf attached). However now after I have soldered everything together, I have measured with an oscilloscope (PicoScope 2205) the output ripple of the DCR011205, and it is much higher than expected.

I have attached my schematic, and the measurement results with 68 Ohm load on output of the DCR011205.

The interesting thing is that if I add a 100nF capacitor between the two isolated ground ( input and output gnd ), then the ripple voltage drops to a fraction (5..10mV which is quite good). Even touching the output with hand ( human body acts kind of like gnd ) the noise drops significantly. This means to me, that the output is kind of "floating" , but I don't have any idea how to solve this problem. Adding capacitor on output between gnd and vcc did not help. Only thing that has helped, was that I have connected the input gnd and output gnd with a capacitor - but then the hole meaning of a 1000V isolated dc-dc converter is lost.

Please help me with this issue.

Thanks,

Miklos

.

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

First off, your schematic is wrong... Why L22 is there I wonder, and the concept of an isolated DC-DC converter prohibits connecting both grounds together. I would review the entire schematic if I was you.

The other thing is what is the purpose of this? Your output says RS232. Why are you not using a charge pump device like the MAX232?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

hugoboss wrote:
First off, your schematic is wrong... Why L22 is there I wonder, and the concept of an isolated DC-DC converter prohibits connecting both grounds together. I would review the entire schematic if I was you.

The other thing is what is the purpose of this? Your output says RS232. Why are you not using a charge pump device like the MAX232?

- My schematic follows the recommendation of Texas Instruments - see the pdf attached - so if it is wrong, then TI guys don't know what they are doing.

- Regarding L22 - I don't understand either, why is it necessary, but I also tried my circuit placing a 0 Ohm jumper instead of it, but it didn't make any difference.

- My RS232 IC is placed after the digital isolator circuit, so it needs an isolated supply - and this was my maybe confusing naming convention for isolated Vcc and GND used by the RS232 IC.

- And I know, that grounds should not be connected together, but connecting them via a capacitor seemed to be a kind of solution regarding the noise.
But I know it isn't a real solution, that is why I posted this question.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Bypass /ERR and SYNC pins to GND as TI recommends with low ESR ceramics.
Output ripple You observe could be a result of poor real ground on scope side. That's why extra cap between grounds works - primary side probably has better ground.

Edit: both L2* are needed - these are parts of symmetrical filter.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I will bet that your source power supply is line grounded, and so is your "scope" by means of the host computer. The output side of the supply is clearly NOT grounded (by intention). Thus, there will be significant currents flowing from the output side through the scope ground lead, giving a lot of apparent noise.

As a result, I will bet that if you have an isolated scope (think laptop running on batteries with no charger), you will see a very different noise reading, and it will be more realistic (that is, more like what is really there).

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Have you heard about this IC?

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

An old thread on converter noise Design for EMC (RF emmission)

Stan

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I don't think that this is a matter of excess converter noise. Instead, I believe that it is a matter of false display of apparent converter noise.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

'To measure is to know, but knowing how to measure' seems to apply here :)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The capacitor you added acts as an Y2 safety capacitor between line and ground. Any respected power supply has one two or three such capacitors. True, their values are around 2.2nF.
UL define the maximum current allowed to pass through these capacitors in normal conditions since are connected to chassis, so in case the ground connection is lost, to not become life threatening.
You have just discovered an EMI reduction solution to your power supply :)

George.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As Angelu suggests, the capacitor across the grounds does have an effect and it is a common inclusion in off-line power supplies. Why? There is capacitance between the windings of a transformer and especially when high frequencies are involved, this parasitic capacitance couples some energy across. Placing a capacitor gives a path for this energy to flow rather through a torturous path via your circuit. So, its not a bodge but common design practice.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MBedder wrote:
Have you heard about this IC?

Yes, I have heard about this IC, but here I rather use my solution for the reason, that if I isolate tx and rx signals plus the supply separately, then I can attach any serial module to my base pcb and not just RS232.

ka7ehk wrote:
I don't think that this is a matter of excess converter noise. Instead, I believe that it is a matter of false display of apparent converter noise.

Jim

Yes you are right. I have also asked TI support regarding this issue, and they have suggested, that I should remove the long (over 10cm ) ground lead from the oscilloscope probe, and to remove the clip as well, so shortening the track from the measured signal to the scope.

And it has actually worked, now I have a noise around 3mV peak-to-peak, which is what I have expected.

But finally a HUGE question arises in me: what is the reason, and the explanation behind this, that removing GND-lead and the clip from the scope probe had such a huge effect on the measurement ?!

I guessing that it has something to do with high input impedance of the probe, long scope probe leads acting like antenna, poor gnd connection.
But I can't connect the dots.

Can somebody explain this? ( Or point out which direction to search )

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:
As Angelu suggests, the capacitor across the grounds does have an effect and it is a common inclusion in off-line power supplies. Why? There is capacitance between the windings of a transformer and especially when high frequencies are involved, this parasitic capacitance couples some energy across. Placing a capacitor gives a path for this energy to flow rather through a torturous path via your circuit. So, its not a bodge but common design practice.

Meanwhile I googled the Y2 capacitor, and actually I found, that Linear Tech suggest to use capacitors with their isolated DC-DC converters. ( But so far I haven't seen it recommended by other manufacturers )

From the datasheet http://cds.linear.com/docs/Demo%... :

Quote:
EMI mitigation techniques used include the following:

1. Four layer PCB, allowing for isolated side to logic side
‘bridge’ capacitor. The bridge capacitor is formed between
an inner layer of fl oating copper which overlaps
the logic side and isolated side ground planes. This
structure creates two series capacitors, each with approximately
0.008" of insulation, supporting the full
dielectric withstand rating of 2500VRMS. The bridge
capacitor provides a low impedance return path for
injected currents due to parasitic capacitances of the
LTM2881’s signal and power isolating elements.

2. Discrete bridge capacitors (C3, C4) mounted between
GND2 and GND. The discrete capacitors provide additional
attenuation at frequencies below 400MHz.

Capacitors are safety rated type Y2, manufactured by
Murata, part # GA342QR7GF471KW01L

So it seems, that this solution could be used with TI isolated dc-dc converters as well. Any suggestion regarding this?

And still I would be also very thankful, if you could help me to understand why removing the gnd lead and the clip from scope probe has helped.

Miklos

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Your earth lead is an antenna. Try putting the earth clip on the probe tip, crank up the vertical sensitivity and move the loop around. See what you pick up.

Do you need the cap between grounds? Scope earth on the incoming gnd, probe on the outgoing gns. What can you see? Add a capacitor. Now what do you see? I've found some brands are better than others even when it is a generic dc/dc converter. I dare say transformer design plays a fair part.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:
Your earth lead is an antenna. Try putting the earth clip on the probe tip, crank up the vertical sensitivity and move the loop around. See what you pick up.

Do you need the cap between grounds? Scope earth on the incoming gnd, probe on the outgoing gns. What can you see? Add a capacitor. Now what do you see? I've found some brands are better than others even when it is a generic dc/dc converter. I dare say transformer design plays a fair part.

I knew that the gnd lead acts like antenna, and if I put it in the air, then I will see 50 Hz signal on it (from mains). But I haven't thought, that if I connect the scope probe gnd lead on the actual circuit gnd, and the scope probe pin (without clip) to the Vcc then I will see noise in MHz range.

So my guess is now: Does the isolated dc-dc converter emit such (high frequency) signals in air that can produce noise on the long (10cm) gnd lead, and so on the scope? That is the only explanation in my mind right now.

I have tried measuring the signal between the two GNDs. It is almost 9V (!) peak-to-peak, and 50Hz. See picture attached.
If I put a 100nF/50V X7R capacitor between the two GNDs, then the amplitude drops to a fraction, but still 300mV peak-to-peak.

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Welcome to the real world of switchmode supplies. Do they emit high frequencies? If you let them they do. The layout and design is critical. A product i was working on recently was emitting a nice signal at 100MHz! It failed CE emissions test miserably. As for the scope probe, Tektronix used to make a special pcb socket to allow proper measurement of signals that the probe plugged into and connected to the earth ring around the probe. So don't always believe what the scope is showing! A lot of the ringing you see on the scope with digital signals is due to the earth lead and if you do any rf work, you learn quickly that the earth lead picks up a lot of stuff.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here's some words that might explain better:

http://www.national.com/rap/Stor...

Robert Pease recently passed.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In a DC/DC converter you want to the change to the other state to be as quick as possible for efficiency reasons; you want the edges to be as steep as possible (high di/dt). Unfortunately, fast edges generate boatloads of harmonics. And that's why good layout practices are required to maintain EMC levels at an acceptable level. One the most important ones is to keep the loop area of high di/dt nodes as small as possible; usually the high di/dt node is the junction switch/diode/inductor. Depending on whether it's a boost or buck converter you need to pay extra attention to the input or output part.

You might like the information in this LT appnote.