DT-40 Pro + FB3 Kit having grounding issue

[VaThInK]

Hi guys,

Before I go on, I would like to say thank you very much to Bill for taking care of my scanners. He did a great job at tuning them and even modifying the scanners drivers, which I didn't actually expect at all and he didn't even ask for a penny . So if any of you is considering to get an FB3 kit, don't think twice and stop wasting your time. Get it as soon as you can! No other products in its class are comparable to FB3 when it comes to bang for your buck and in my case, Bill is kind enough to provide a service that he's not obliged to. Let alone providing support for his products. Last but not least, it only took around 2 days for the package to arrive in Australia from Pangolin. As Mr. Burns would say, EXCELLENT ! QM2000 would be hell of a lot better if $600 gets you this far !

Anyway, if you guys haven't notice it yet , I just bought a pair of DT-40 Pro and FB3 kit. I've played around with them for a while and I'm having what seems to be a grounding issue as informed by Bill. I have attached a number of pictures and video clips so you can visualize what I'm talking about.

If I drew a number of static dots on the same coordinate, it will be scanned as shown in Dots 1.jpg and Dots 2.jpg . The dots are oscillated from upper left to lower right very quickly. Have a look at the attached test pattern shots and clips under this effect (the ones starting with "Bad" on its filename). However, Bill have suggested me a temporary fix to this issue by shorting X-/Y- to common ground and the results are as shown in the pictures starting with "Good" on its filename. I did a quick geometric correction on all of those. Otherwise, they would be a bit off. Do those pictures after the fix look OK? Anyone else having this issue and hopefully have found a proper fix for it?

I believe this one is not an issue but more like a newbie question. I have also attached 2 extra clips. The one that's called Solid.avi is as the name suggest scanned the animation solidly while the other one is a little bit flickery and dimmer noticeably. Is this normal? I have a feeling that it flickers because of the bubbles in that animation, which stressed the scanners. Any comments on this one?

Any thoughts would be much appreciated. Thanks.

[The_Doctor]

It's odd that common-grounding the - side of the differential signal can help. I don't know whether either of your channels needed reversal to get the ILDA pattern right side up and with text reading right not reversed because I don't know the geometry and wiring of DT40's but if I assume that a a + on both X and Y sends the scan upwards and rightwards as expected, then your weird modulation isn't common mode, but affects one channel inversely to the other, though equally. Also, if your video is at right speed, that modulation is about 5 Hz, and it's very strong. That's not a very likely grounding problem, it looks more like something odd is getting in. Possibly a low frequency oscillation between both channels between the two halves of the scan amp?

I can't guess further, but that might give you ideas of where to look.

[buffo]

I've never seen the oscillation problems shown in your "bad" series of pictures and video. (Well, not to that magnitude anyway, and not nearly so slow either.) I can't begin to explain what's causing it - especially since grounding the negative output seems to get rid of it.

Sometimes you will have to do geometric corrections when you switch from single-ended to differential inputs (or vice versa). Though in my experience, it's a problem that only cropped up when I wired a differential signal to a single-ended input on the amp. Still, if you get reproducible results with the negative leads grounded and the corrections applied, then you should be fine.

As for the "good" series of pics - I'm hoping that the after-images of the dolphin are actually artifacts created by the CCD in your camera! If the show really looks like that in person, with all those after-images, then something is seriously wrong.

However, if there is in fact just the single image, then the slight flicker on the frame with the extra bubbles is not an issue. My alphalite used to display that same animation with a little flicker too. I can't remember if my QM-2000 displays it with flicker or not though. (Note to self - test that sometime this weekend!)

Adam

[DZ]

For some reason I can't view the videos, but just from what you were say I was just wondering if you have +x -x AND ground attached to the X amp, and +y -y AND ground attached to your Y amp. Attaching just +X and -X doesn't cut it. Aside from that, I'm not sure what else could cause it. Differential is suppose to eliminate induced noises, so it should not be a 60 cycle hum on the line.

[VaThInK]

The_Doctor: I did geometric correction not because of the image is flipped in X or Y directions. It's actually to fix unevenness. Without it, the image would look like a trapezium not a nice even square. This oscillation problem only occurs if I run FB3 as DAC and is connected to a PC. If it's isolated from PCs and runs off battery pack, this problem doesn't shows up just like when I join the "-" and ground together. Bill told me that he didn't have this problem when tuning my scanners but then he was using his PSU for the scanner amps and not the ones included. I'm going to try different PSU to see if it would makes any difference.

Buffo: I'm assuming the outputs of FB3 are single ended. When I received them, its outputs were connected to the positive and negative inputs on the scanner amps while the grounds are left floating. Now I have to short the negative to ground if it's connected to a PC. Yeah those artifacts are caused by my camera. I don't know why but my camera always act like that with bright objects. I'll try again this weekend with an SLR.

DZ: The videos are compressed with XviD so you gotta have the codec installed to view them. As I mentioned earlier, only the X+ and Y+ are connected. Ground are left floating. I just left it the way it is since I received it from Bill as I don't know how to connect single ended outputs to differential inputs. I assume the setup is correct.

Thanks for the help so far guys. I really appreciate it.

[The_Doctor]

DZ, a true differential should be fine with just + and - connected, that's how it's meant to be, the gnd is meant to be connected at only one end, and is there to further limit common mode error be reducing any error induced from outside. Connection at one end excludes the chance of ground loops.

It seems like Bill left it without the ground, and I think he probably knew what he was about.

VaThInK, do you have an oscilloscope? It could be worth prowling around that PC and seeing what its ground signals (and others) look like. That slow modulation might be a beat frequency between two very closely set high frequencies.

[DZ]

DZ, a true differential should be fine with just + and - connected, that's how it's meant to be, the gnd is meant to be connected at only one end, and is there to further limit common mode error be reducing any error induced from outside. Connection at one end excludes the chance of ground loops.

quoted from Pangolin's website:
Where to connect X and Y

There are two types of scanner amps. For this discussion we'll talk about the X signal; the Y signal works the same but of course uses different pins:

• Single-ended amps have two input lines per signal, such as "Ground" and "X". Examples include Catweazle and Laser Media. With these amps, connect the QM2000's Ground (pin 25) and the +X (pin 1) lines. Do not connect the -X (pin 14) line.
• Differential amps have three input lines per signal, such as "-X", "Ground" and "+X". Examples include Cambridge, GS MiniSax and TurboTrack. With these amps, simply connect the appropriate lines on the QM2000: -X (pin 14), Ground (pin 25) and +X (pin 1).

Quoted from LaserFX:
When compared to the ground level of the differential driver, the voltage level of the normal signal line shall be +5V and the voltage level of the inverted signal shall be -5V.

[buffo]

Ok - this is starting to make more sense now. The Flashback has single-ended outputs, but you've got them connected to the differential inputs on your scanner amps. This is a no-no, and is exactly the problem I described in my post above.

With a single-ended signal, you only connect to X+ and ground. If you connect to X+ and X- you will get the geometric distortion like you described. (I had this same problem at SELEM when I connected my projector to DZ's QM-2000 board.)

As for the differential signals, DZ is right. You need all three signals. And yeah, it's contrary to just about every other differential signalling protocol (LVD SCSI, for example, only uses 2 wires), but scanner amps are different; they're designed to use the ground in addition to the X+ and X- signals. (As to why this is, I haven't a clue. Maybe Bill Benner can shed some light on this.)

Bottom line: Re-wire your scanner amps to only use the X+ and the ground input, and let us know how that works. (Leave the X- unconnected and floating.)

Adam

[The_Doctor]

DZ, I suggest you take up the inquiry with Bill or VaThInK.

"As I mentioned earlier, only the X+ and Y+ are connected. Ground are left floating. I just left it the way it is since I received it from Bill as I don't know how to connect single ended outputs to differential inputs. I assume the setup is correct."
(From VaThInK)

Bill Benner is every bit as representive of Pangolin as any manual could ever be. As for comparison with the ground line, that can be just a way to say what you'll see IF the two differential lines are properly symmetrical about zero (if there is no DC offset error between two unconnected grounds).

Buffo, I did consider that maybe a ground might be needed to allow a DC value to work, but that wouldn't be so. So long as symmetry is ok the signals will be right, and the best way to avoid asymmetry is to leave the ground open, so long as neither end has a DC offset shifting its gnd relative to the other. That's the whole point of a differential input, that's how it reduces noise, by avoiding an absolute reference to any ground.

If there really IS a reason why a ground must be needed, then Bill might have to explain why he supplied a system set up without one. I still think he was right, and that it isn't needed unless you find a problem that needs you to forget differential and go with the standard single line with absolute ground reference, as he suggested as a temporary fix for VaThInK's problem.

Edit:
VaThInK, I'm not sure what you did for Bill's temporary fix. As well as connecting the V- inputs to local gnd at the scan driver input, you also need to connect the gnd from the FB3 to that input gnd/V- point, with the single end output to the V+ as you mentioned. I guess that is what you have though.

Buffo, I think leaving the V- floating would cause bother. At least, it would in a conventional differential, as it must anchored to something.

[VaThInK]

Unfortunately, I don't have an oscilloscope or access to use one .

I just tried to run my setup with 2 different PCs but ended up with the same oscillating problem. I also tried it on a laptop with its power adapter plugged in and it was less severe. On a scale of 1 to 10 with 1 being the worst. I would rate the severity when plugged to a PC at 1 and 9 when using a laptop. It's still there, but only very slightly. Still quite noticeable though.

Single-ended amps have two input lines per signal, such as "Ground" and "X". Examples include Catweazle and Laser Media. With these amps, connect the QM2000's Ground (pin 25) and the +X (pin 1) lines. Do not connect the -X (pin 14) line.

If the above statement is true, would it be valid if it's the other way around? Single ended controller on differential amps.

With a single-ended signal, you only connect to X+ and ground. If you connect to X+ and X- you will get the geometric distortion like you described.

I think it's not actually distorted. More like shifted back and forth from upper left to lower right constantly. Maybe the video clip aren't clear enough to show it. I have attached another picture to explain the oscillation issue better. The dots were shifting one by one quite fast. It will shows quite clearly in the video if you play them at slower speed. Try to watch "Bad ILDA Clip.avi" again in particular.

As I mentioned earlier, only the X+ and Y+ are connected

I was meant to say X+,Y+, X- and Y- not just X+ and Y+. Sorry, my bad .

Bill also mentioned that if I shorted the X- or Y- to ground it will create another issue. He said a minor one. Maybe geometry is the other issue. If it is, luckily it can be fixed easily with geometric correction on LiveQUICK.

Edit: Bill explicitly told me to short either X- or Y- to ground for the time being to fix my issue. Basically, X- and Y- are anchored to ground and ground is used as reference instead. Hopefully he will come up with a better solution. Check out the picture called "My Setup.JPG" to see how things are put together.