CurtPalme.com Home Theater Forum

[racerxnet]

Register to remove this ad. It's free!

I have a little ghosting on my 8500 which always happens when I converge the Green in the service menu. I do so to get the grid square to the screen edges. If I delete the convergence the ghosting disappears again. Any Ideas on what I might try to eliminate it? I have the ultra boards on this PJ with firmware 4.2.

MAk

[Tim in Phoenix]

Hello

Some pics would be helpful.

[cmjohnson]

Not that it necessarily matters, but does firmware rev 4.2 even support Ultra features? I thought that was only version 5 and higher?

There's zero advantage to running Ultra boards if your firmware isn't high enough to support the extra functionality.

In fact, you might even have compatibility issues if the firmware on individual boards is too far mismatched from the CLM firmware.

[Tim in Phoenix]

Guys

The Ultra evolved from other specialized earlier versions. There was a Dome version, and a Hi-Scan version. Some Ultra functions were supported in v4.0. Full-tilt Ultras had firmware starting at v4.2c as early as 1998 in units from Christie Digital. They had decided that it was simpler to put all the features in all the projectors (8110 excluding) vs having specialized machines with long lead times.

[racerxnet]

Here are pictures of the left hand side with the vertical bands. I can see very faint horizontal banding, but is negligible compare to what is shown.

MAK

[cmjohnson]

That's ringing, it's in the deflection yoke system, and I have yet to see a Marquee that doesn't have it to some extent.

It's scan frequency dependent. The higher the scan rate, the more noticeable it'll be, and the space between the bright and dark areas will become smaller.

Is there a fix for it? I don't know. I think the ringing could be damped but the fix might only apply to a narrow range of scan frequencies.

[racerxnet]

[tschaeikaei]

Could this be damped with small caps (some pF) parallel to the coil windings?
Standing waves reflections and such things going on?
I don't have a coil at hand to look for caps already being there.
First guess only.
If someone wants to test it, there's trimmer caps (like on old radios) you could try.
Maybe it works.

I've seen this on nearly all projectors, the PG does it, the 801s here does it,
I'm not sure about my Marquees.

Impedance matching of the amp and coils could matter, too.
That's also frequenciy-dependant.

Regards, Julian

[cmjohnson]

Yes, you could probably do that. But that fix would probably apply only over a narrow range of frequencies.

You'd need to know the inductance of the yoke, the scan frequency, and calculate the required capacitance.

No, actually, you need to try to figure out the frequency of the ringing, which would be based off of an estimation
of how many pixels are between the dark areas, and knowing your resolution and scan rate, you can estimate the
fundamental frequency of the ringing, and damp it out with a properly designed trap. Which would be a tank circuit tuned
for HALF of the ringing signal's fundamental frequency, if my electronics instructor wasn't lying to me.

But the caps you use would have to be rated for the operating voltages involved. The HDM generates up to 1500
volts peak to peak. So you'd better use caps that are rated for that.

Pick your chosen resolution and refresh rate, get the scan rate, and calculate your LCR tank circuit values using what you already have.

But is this going to actually work?

[racerxnet]

That stuff is over my pay grade. I appreciate your time and answers.

Mak

[cmjohnson]

I gave this a little more thought and it occurred to me that it may be possible, without a LOT of engineering, to come up with a ringing fix which is related to what I just posted in my last reply, which might be applied on the HDM. Or it may need to be applied at the yoke itself.

There are two basic methods to damp the ringing. They can be used together.

One is to add a series resistance to the coil. This reduces the amplitude of the ringing and increases the decay rate,
by reducing system Q.

The other is to add a trap circuit that absorbs the resonant ringing frequency of the horizontal deflection yoke.

But be aware, the deflection yoke runs at up to 1500 volts peak to peak. Any components chosen to try to trap out this ringing
must be designed to withstand that voltage.

[barclay66]

Hi,

This could be tricky. The dampening trap circuit (usually some kind of snubber network made of a capacitor and a resistor) will increase the load on the horizontal drive circuit. Instead, I would recommend using an active variant. Maybe it could be possible modulating the H-drive power supply with a correction signal and opposite phase. The correction signal would either have to be generated by capturing the ringing itself (e.g. a sensing coil at the H-deflection coils) or by generating a specific pattern dependant on the H-frequency and flyback impulses...

Regards,
barclay66

[thewolfman]

I solve it by using only long retrace. Doing that makes it go away on its own. But if it's the right way I'm not so sure off..

Using short retrace one needs to go into convergence and move green only off to the left side to shift the ringing out.. and then press PIC and move the image back in again using phase. Your grid pattern will look strange after this, at least that is what happened to me.

I may have remembered it wrong but long retrace is the way for me.

[racerxnet]

[tschaeikaei]

I'd love to see an oszillogram of this. Maybe you could probe the horizontal deflection coil.
Be sure and use at least a 1:10 probe and an isolating transformer.

Or you simply try another HDM.

CM, i like your idea. But i wanted to avoid the math and therefore proposed a trimmer cap.
Oh and the standard formulas won't work with a sawtooth signal.

And: every signal consists of the fundamental and all those harmonics.
I mean, every low-pass will flatten the slopes, as they're very high frequency.
Could be possible using a narrow band filter (which only includes the unwanted ringing and not the slope rate frequency.
I'd scope and try instead of calculating it exactly.

No surprise long retrace works better.
I could also imagine adding horizontal porches could work (as it would blank out the problematic area.
But that's not solving the problem, it's only making it invisible.
Slope time increases and that reduces high frequency stuff being created by fast dI.
But Q is electrical charge. You meant G?

I alsp agree with Barclay, saing that everything you do to short (unwanted) high frequencies ..
will surely increase load on the amps and cause problems.
therefore i proposed to use just tiny caps that (hopefullly) won't matter too much.

Barclays idea of a inverted phase feedback signal for correction, wow, great!
But would it work if simply modulating the DC power supply rail?

I think the 180° phase shifted feedback could work the best (and at any h-frequency).
Maybe it could be easier to outcouple the signal using a small cap than using an additional winding.
Other stuff could also couple into the winding and ruin the feedback signal.
But that's not predictable, i could just imagine problems could occur that way.

I kind of hope my Marquees also do it
Sounds like fun to investigate.

Regards, Julian

[cmjohnson]

Q is the response curve of a filter. Higher Q, narrower bandwidth, lower Q, broader bandwidth.

A filter with a 10 db/octave slope is a high Q filter, a filter with a 3 db/octave slope is a low Q filter.


The trimmer cap idea is interesting but how easy are they to find in a 1500 volt (or more) rating?

I'd temporarily forgotten that triangular waveforms would be relevant in this case, which does complicate the filter design quite a bit.

I'd say that it's beyond my knowledge without cracking open a book to give myself a refresher course.

[racerxnet]

After setting retrace to long I can no longer fit the desktop to fit the screen. The icons can be seen on one side when moving the Phase, but then the opposite side is off the screen. Shrinking the image will not allow the full desktop to be shown either. The only way I have gotten the full desktop is to reduce the pixel count width. I'd rather see if there is an alternative and keep the 1920 x 1080 A/R. Would changing the timing resolve the issue with CRU? It has solved the ringing, but provided a new challenge.

Thanks,

Mak

[greg9518lc]

[quote="racerxnet"]After setting retrace to long I can no longer fit the desktop to fit the screen. The icons can be seen on one side when moving the Phase, but then the opposite side is off the screen. Shrinking the image will not allow the full desktop to be shown either. The only way I have gotten the full desktop is to reduce the pixel count width. I'd rather see if there is an alternative and keep the 1920 x 1080 A/R. Would changing the timing resolve the issue with CRU? It has solved the ringing, but provided a new challenge.

Thanks,

Mak[/quote]

I will send you my settings tonight after work for 207 mhz I think you will like what you see.

[racerxnet]

[barclay66]

BTW: Don't forget checking Your blanking settings...