DT-40 Pro + FB3 Kit having grounding issue

[buffo]

If your suggestion is safe enough, I'm willing to try it out.

It's safe all right. It's how the DT-40 pro's are supposed to be connected to a single-ended DAC. If you received the users manual for your scanners (actually it's not a manual at all, but rather a single sheet of paper) you'll see a diagram that looks like the one I posted above.

In any case, you aren't going to damage your galvos or your amps even if you conect the input signals incorrectly. (Well, let me qualify that - so long as you aren't overdriving them, you won't damage them.) You might get screwed-up images, but you won't break them.

Is 2(1/tan(opposite/adjacent)) the right formula to find the angle?

Have a look at this page on the Pangolin site; it gives you several different ways to calculate scan angle.

I have no idea how to measure the Kpps .

It's a setting in your software. You don't need to measure it, you just select a speed and you're done.

Adam

[cruch]

Is there an program for calculate this??
Maby some programmers in here can make it?

[The_Doctor]

Don't trust a program. Not that they're untrustworthy, they aren't if they're done right, but your experience and intuition together will guide you better once you've made sure you understand what can affect that scan angle.

The main problem is that noise mod, it's LARGE. Unless you root that out and dismiss it at source, you will risk damaging something with that alone unless you keep gains down for now.

I agree, the signal that can get through a leaky cable isn't really relevant here, unless it's being boosted in some kind of feedback loop. That's why I mentioned squegging, it's possible your scan amps themselves are providing the gain.

One cautious experiment you could try is to change the scan angle electronically (by changing signal level (or frame size or whatever) at source, in data or by attenuation before input to scan amps, and see if the noise mod changes too, and if so, in what proportion. As far as I know, it doesn't change, I think one of your tests was a dot redrawn in the centre.

Anyway, assuming you're using a moderate scan angle, and your signal is a few volts, that noise mod is also a few volts. Try an analog voltmeter, or a digital one with bargraph, this is something so slow and so big you shouldn't need exotic test gear to see it. I'm not sure which connections to examine though, I got confused about something, or maybe there's not yet enough info.

[The_Doctor]

Separate post for this as I think it's that important. (I mentioned it before but it got waylaid).

Did you have to reverse the polarity of ONE channel, X or Y, in software OR hardware, to flip that ILDA pattern so it appears the right way round and the right way up?

If the answer is NO, then that weird modulation is inverted on one channel, which might mean something's going in one channel somewhere and out the other, instead of going to ground as maybe intended. What's important about that is it would be on the analog side, not digital. So maybe look for a broken wire that doesn't look broken.

[Pangolin]

Hi Vathink,

Sorry for not being able to reply to this sooner, and thus having everyone speculate on what the problem is...

Here is a description of the problem. You actually have several problems which are interplaying.

First, the power supply really isn't that great. It is the intention of the power supply to offer "isolation" from the incoming AC. But it doesn't do a very good job... As a result, you have some "compling" effects, which lead to what's called a common-mode signal being applied to the -24V, Ground, and +24V. Another guy named Dan sent in his kit to be modified by us at Pangolin, and we saw the same problem there. What we see is that there is around 20 volts AC "coupling" -- some people might call it "noise" because it certainly is not a desired signal -- riding on the power supply voltages.

Second, we wired your amps and the DB25 for differential going into the amps. That was done VERY PURPOSEFULLY. If you run those amps single-ended, as they come from Jian, you will get a different kind of coupling -- coupling from the X to the Y through the common ground. You will have a ground loop, although that phrase is thrown around so often that I hate to use it.

Now, ordinarily when you go into an amp differentially, and there is some common mode noise (normally the noise is on the incoming signal, but in this case it is on the power supply lines, but it is a matter of tomato tomauto...) the amp SHOULD be able to subtract the noise out. BUT, here is where we get into Chinese gear... The input resistors on the differential receiver are not hand-matched, so the common mode rejection of that amp is just not very good...

So, here is the solution. In fact, the solution can take several forms, but this is an easy form. If you take a look at the pwoer supply, you will see that it has a LINE input, and an NEUTRAL input and a "FG" which stands for "Frame Ground". Outside of the US, the LINE and NEUTRAL might not actually have much sense. But hopefully, you can tie the FRAME GROUND to the middle pin (or outside ring or whatever it is in your country) so that the power supply frame itself is grounded. Next, you can run a small wire (heck, better make it a low-current fuse, just incase...) from the FG lead to the "ground" lead on the power supply output going to the scanners. That will solve your problem.

Another way to solve the problem is to run a "third wire" from the "ground" pin on the power supply to your computer.

Basically, you just want to calm down that coupling. If you can get it to some relatively grounded potential, that will solve your problem.

Bill

[Banthai]

God damn bill ... you are way off


You say Tom-Mate-toes ... we say Tom-mart-toes

but you say Po-tate-toes and we also say Po-tate-toes ...

so that song is not a good example


but seriously what you are saying is so correct, i work with equipment that is installed all over the world and we have issues with grounding in different countries, because of the way its done, for example copper ground stake.... or grounded to neutral, .... or as you say when it used both sometimes you get the crappy ground loop , and in some countries ( like india ) gound just does not exist ... and i was getting buzzed by our gear all day long ....

just to say here that
line = live
neutral = neutral
Ground = earth

also we call FG.... field ground

we also drive on the correct side of the road


all the best .... Karl

[carmangary]

You crazy brits. Best thing that ever came out of England was Americans. hehe.

[dave]

line = live

also active here,

hehe, at least we don't use the black wire for active/line/live here

[The_Doctor]

VaThInK, how much different was the speed of your video from realtime? If it looked like a 50 or 60 Hz modulation noise I'd have stayed with the obvious idea of ground trouble (which I didn't exactly ignore..) but it looked a lot slower to me, around 5 Hz.

Another question, for Bill, or anyone else who knows. Does ILDA have to use current loops on its differential lines? You wouldn't need such precision in matched resistors if you did it the way most people do, you'd just use symmetrical voltage around zero, high impedance inputs, and a decent cable screen grounded at one end.

[VaThInK]

Pangolin: The problem is that the amps PSU supplied by Jian only have AC, AC and FG marked on the input terminals. In here we call them Active, Neutral and Earth. So I don't have any clue which one is Active and Neutral. The FG is always connected to Earth the moment I notice my grounding issue, but then it doesn't seem to fix anything. I've just tried shorting the FG to ground of the PSU output and the result is the same just like the quick fix that you've suggested earlier.

Now I'm curious about the differences between the 2 solutions. Correct me if I'm wrong but I think both are pretty much the same. Since the X- and Y- of the FB3 are actually the same as ground (on the same level as R-,G-,B-, etc), shorting either X- or Y- to amps ground is basically the same as shorting FG to ground of the PSU output in my current setup. I have attached a picture so you can see clearly what I'm trying to say.

The_Doctor: The multi meters that I have are all digital and they don't have bar graph feature unfortunately. With or without the fix, all the scanned images are in proper orientation but at 100% size they don't look nice and square so that's the reason I did geometric correction. I think those clips show the speed in real time. The variation in oscillation speed depends on the size of the projected image and the complexity of the image. The oscillation speed will be slower if the image is bigger and more complex.

I have also tried Buffo's suggestion and it works without any issues, well, 1 issue actually. The image was scanned at 1/4 of the original size (or reduced by 1/2 in X and Y plane). Now if carmangary is right, that means right now I can only get a maximum of 1/4 of the actual size with FB3 as it doesn't have differential outputs, am I right? So if I wired my setup just like Buffo suggested I would get 1/16 of the actual size.