Noise filter for DC

The best solution is to not be on the same power circuit as the welding machine.  You should be able to isolate it with a UPS.  If you want to build your own noise circuit, you should look for a low-pass filter design: http://www.allaboutcircuits.com/vol_2/chpt_8/2.html

I had try by ups, but no help.

It is that noise is generated ower the cables on secondar part of rectifier.

Did the UPS that you used include line conditioning?  There are line conditioners used for audio applications: http://www.amazon.com/Furman-Standard-Conditioning-Outlets-Spacing/dp/B0014598WQ

What's the source being monitored in that picture?
Possibly, shielding would be more efficient than filtering.

Or, if it's, e.g., warping a CRT display, changing to an LCD display would stop it.

UPS is for primar side (230V AC) of rectifier, ... there the signal is clear.
Problems are on the secondar DC side.

The sorce monitored is 24 V power for driving machine:
http://www.google.si/imgres?q=power+supply+siemens&um=1&hl=sl&sa=N&rlz=1T4ADFA_enSI455SI455&tbm=isch&tbnid=exhIHbhWp_wtrM:&imgrefurl=http://store.gibuys.com/power-supplies/siemens-6ep1434-2ba00-sitop-power-10-power-supply-used.html&docid=Mk_hN7iDmtDwKM&imgurl=http://store.gibuys.com/media/catalog/product/cache/1/image/9df78eab33525d08d6e5fb8d27136e95/d/s/dsc00840.jpg&w=640&h=480&ei=HvBeT9SlLcrxsgaqoIG8CQ&zoom=1&iact=hc&vpx=690&vpy=74&dur=1203&hovh=194&hovw=259&tx=152&ty=154&sig=115801785739580837645&page=4&tbnh=150&tbnw=221&start=42&ndsp=16&ved=1t:429,r:14,s:42&biw=1280&bih=538

You may want to read this paper on 24v noise sources and how to deal with them in Europe: http://literature.rockwellautomation.com/idc/groups/literature/documents/rm/gmc-rm001_-en-p.pdf

According to the data sheet I found for the 6EP1 434-2BA00, its input is supposed to be 340V-550V (400V-500V on the label) 3-phase, 47-63Hz.
How are you putting a 230V UPS on the primary side of the rectifier?

Anyway...  disconnect the load wires from the output terminals, connect your scope to the output terminals, and watch while the welds occur.
If there's no noise then, you know it's being induced into the load wires.
In that case, shield the load wires and ground only 1 end of the shield.

If there's still noise at the terminals without the load wires connected, add filtering at the output.

edit: for filtering, I was going to suggest something like a 50V 1000uF 10,000 hour @105c electrolytic (~$30 each) across the output... but I see that Rockwell paper linked by Callandor recommends just a ceramic 50V 1uF such as
http://www.newark.com/vishay-sprague/2c20x7r105k050b/capacitor-ceramic-1uf-50v-x7r-radial/dp/66C6721 (10%) or http://www.newark.com/kemet/c330c105j5r5ta/capacitor-ceramic-1uf-50v-x7r-radial/dp/72K6453 (5%) across the output (grounded common), or one of those from each side of the output to ground (ungrounded systems).

Ok, i will be more professional now:
I have used sheme with elements (pictures), but no help:
problem from the picture still persist.
What must i do?
IMG-3595.jpg
IMG-3596.jpg
IMG-3598.jpg
IMG-3599.jpg
IMG-3600.jpg

As I said, the 6EP1 434-2BA00 (the item you linked us to with that google search) is supposed to use 400V to 500V 3-phase as input, not 230V single phase.

So what power supply are you actually using?

The picturit  was just simbolic the real one  (see above (the last one: recifer) is:
6EP -1334-2BA00.  I did not expect to cause confiuson, i was yust looking for some
24 DC Siemens power suply.


Maybe more about procces: The welding machine uses to start welding  betwen two points high electric potencial diference (around 10000V). This tension diference produces spark
betwen two electrodes an then curent begin to flow.
This tension produces high magnetic field into machine cel.

We have to stop machine for more days to replace all cables. We well do it next month.

For the test i have used testing sistem by independent  recifer and two cables: one SHIELDED and another not. And i had put them together with "old" cables. I had made some measurments and the signal on the shilded cable was better. Still noise persist but much lower.

So i preume that shilded cables will resolve the problem.

Thank you.

It wouldn't hurt to put those 1uF 'snubber' filter caps on all inductive loads that turn on/off often (such as across control relay coils), too.

I would only use the ferrite bead inductors/chokes on sensitive components that are affected by those few-cycle spikes being induced by the welder flux, if there is any left once the wires are shielded.

Ok, thank You. I will let You know  the result.