Marquee Modifications and Performance Enhancement 3 - Page 4 - AVS Forum
First ... 2  3  4 5  6  ... Last
CRT Projectors > Marquee Modifications and Performance Enhancement 3
aspec2's Avatar aspec2 09:59 PM 06-23-2004

Capacitors don't truly block DC until they are charged.

Capacitors pass AC. The rate at which they pass current is determined by the resistance, size of the cap, and the frequency. Note in the equation that Robert posted. Remember it is the rate of change of voltage with time.
Capacitors resist voltage change, inductors resist current change.


mark haflich's Avatar mark haflich 10:11 PM 06-23-2004
True. They really do not pass anything until they are charged. They fill up, then discharge. It is not like a brick wall. I know all that. There is plumbing in and plumbing out so to speak. It is a 1st order filter, not a wall. So . . . .?
aspec2's Avatar aspec2 10:23 PM 06-23-2004

In this application, it is not a filter. It time shifts the voltage to the op amp, and (dare I say) acts as a DC safety stop.

Robert Zuch's Avatar Robert Zuch 10:52 PM 06-23-2004
Thanks to everyone for the very positive comments; they are very much appreciated!

>An RC filter blocks DC and AC up to the pole

Mark, the RC filter blocks DC but does not block AC. It attenuates AC signals at the rate of -6db/octave below the cutoff frequency (if you halve the frequency then the filter gain drops by 6db). The frequency where the roll off occurs is not determined only by the cap, but instead by the RC product. Doubling the resistance will have the same effect as doubling the capacitance in terms of filter response. So all AC signals pass through the filter but lower freq. signals are attenuated.

Since we don't expect video signals to contain any low frequency information, the amplitude response of the filter is not important. The RC product was selected so that the filter's cutoff frequency is far away from the frequency of the video signal. This assures that the filter's phase shift is the same for faster changing and slower changing video signals (or a constant group delay).

Other than the possibility of 60/120Hz hum, you should not see any "low frequency contaminants" on the video signal. Spurious noise travels at high frequencies, much the same way as an AM or FM carrier transmits. If you have a 10Hz unwanted signal on your video, then you have something broken in your system. Better to fix the system than try to filter the problem.

>As this relates to these projectors, my questions:
>1. How many or what mods that improve a circuit, potentially cause strain or magnify a weakness in a down stream circuit?
>2. Is it possible to make improvements and be forever chasing gremlins?
>3. Is this why some may say that they see visual improvement with a mod that may not be theoretically correct?

Glen, the answer to question 2 is absolutely! We spent quite a bit of time trying modifications that do not work. They may improve one area but have a negative effect elsewhere. Then there are the mod's that have very small effects on PQ. These can be difficult to judge because after investing many hours of work we tend to see improvements whether they exist or not! When the evaluation is subjective we tend to shade our opinions to justify our efforts. So our philosophy on mod's is that we do not include a mod unless we can see the difference after a 24hr. break. If it requires an A-B test switching back and forth to discern the change then it is probably not significant enough to matter. If on the other hand we can turn on the projector the next day and see a better picture than we were getting before the mod, we have something. Of course you need to use source material that you are very familiar with.

As to straining or magnifying a weakness in another area... sure that is always a risk. Understanding what the modification does and how it interacts with all of the circuits alleviates a lot of this risk. Thorough testing takes care of any remaining concerns. In the Marquee you need to be very careful with the safety circuits and anything that effects beam current.

>Many the hour was spent plotting exponential curves for inductors and caps USING A SLIDE RULE. THANKS, Robert for reminding me how long it's >been since the olden days.

Walt, I'm guessing that your slide rule is quite a bit faster than our abacus!

Bob Z.
Reference Imaging
mark haflich's Avatar mark haflich 11:45 PM 06-23-2004
Let's see. It attenuates the AC at 6db/octave from the pole point! Let's see duh, that's a first order filter. My gosh. But I knew that. Whoopsie, I said it is not a brick wall, I said it's a first order filter. Thanks for setting me straight.

Its not a filter. Please, it is a filter. It may serve other purposes with respect to down stream components, but it is still a filter. The video spectrum starts at 0 hz and goes up. Any filter regardless of where placed near the low end of the video spectrum will attenuate part of that spectrum. Then attenuation will be below the pole at 6db per octave for a first order filter. It is an RC product. You can raise one and lower the other to maintain the same result. Robert. I taught physics during the late 60s at college level to help put myself through law school. Your posts are great and I respect your work, but . . . . I already know all this.

Bottom line. Every bit of the spectrum passes through a series cap and pollutes it to some extent. That is a fact. Some caps are better at not polluting what passes than others. You know this damn well. The quality of a series cap is critical. Good night all.
bblue's Avatar bblue 05:34 AM 06-24-2004
It is a filter of course, Mark. I think what Bob is saying is that even though it's a filter you don't relate to the 'frequency response' in the same way with video as you would with say, audio, because of the nature of the waveform. Even though the RC circuit feeding the non-inverting op-amp looks the same as one that might process audio, the filter actually is doing something different for video and it and the first stage become a differentiator producing a pass voltage based on the rate of change rather than the amplitude of the signal. You are still getting rid of LF component that really isn't needed, but also you are choosing this filter characteristic such that all of the significant range is completely out of the filter range so the resulting delay is linear at all frequencies and the rate of change is at the correct amplitude. Bob, correct me if I'm misinterpreting something.

So extrapolating from that, any capacitor change of the same value but different characteristics may change the phase relationship of the part of signal which needs the constant group delay, so the character of the capacitor matters.

Also, for sure you could not go lower in capacity for that component since it would move the pole closer to usable signal and may affect phase relationships and would, in effect, lower the signal because of a lower rate of change.

But what of raising the capacitance? In doing that, you're lowering the pole moving it further away from usable signal, and presumably are increasing the rate of change from the filter which would look like additional gain. What I don't quite grasp is how the rate of change could ever be higher than the amplitude of the source signal? Or is that really a non-issue, since there is actually a linear voltage loss through RC and the entire passband which is made up for by gain in subsequent stages, and that gain wouldn't be needed if the RC network wasn't there in the first place?

Hopefully I have that right.

As I mentioned earlier I've been using 1mf caps for some time. If you did that and the electronics had enough headroom, and your color tracking and overall luminance was calibrated after that, I think you would end up with pretty much the same results as with the .47 and a setup to match.. Yes or no?

Last night I spent some time trying different capacitors in that network, from .47 to 2Mf, compared to my 1mf 'standard'. By comparison, the .47 was duller with not enough contrast and too much color, like a high gamma. The 2mf made the picture just jump out at you -- very punchy with very rich colors and exceptionally good low level detail, yet there didn't seem to be significant increases in black or full white level (a subjective view). It did have an awesome character about it.

But if all this is is a result of voltage gain, you would think that the same thing could be achieved with contrast and brightness adjustments, yet attempts to produce the same results with C & B were not successful. What else is happening?

The last two hours I spent comparing 1mf caps of different types. One was a stacked metal film, one was some sort of small mylar and the other was a higher voltage polypropelene. They each had a slightly different character (this is subjective, by viewing same HD source, not scientific). The mylar had sort of a muted look -- everything was there, but the colors were drab. The other two were quite close, but I found my self repeatedly preferring the stacked film for an over-all natural 'being there' look and revealing the biggest difference in HD sources.

I'd like to repeat this again with some recommendations on .47mf's to try, and my projector recalibrated with .47mf's in place beforehand (if Bob confirms what I think he will). Any suggestions on types to try?

aspec2's Avatar aspec2 06:01 AM 06-24-2004

Sorry. When one is hauling a load of rum and browsing the forum late at night, one must remember to keep one's shut or one runs the risk of making an A$$ of one's self.

mark haflich's Avatar mark haflich 07:01 AM 06-24-2004
Thanks everyone. No problem, Walt and Robert. Everyone here is trying to help each other.

Very nice post Bill. I especially appreciate your patience and demeanor with me.

I am not sure about how capacitor parameters other than value or size affects group delay. Maybe they do, I just do not know. From my audio days, I know about ESR and especially DA. No one here has discussed these parameters and their effect on the video signal passing through a series cap. Does ESR contribute to RC group delay? Do you think the circuit design engineers actually took things like ESR into account? Clearly, they choose not to use electrolytics but beyond that?

Also isn't spectrum, spectrum. That is both audio and video are AC signals l but video compared to audio goes a lot lot higher in frequency. Please all I am asking are questions, I am not not trying to irk or embarrass anyone. I really would like to know from a pure knowledge perspective and for our mutual quest to improve our beasts.

To my friend MP, the danger in trying a value change in a cap somewhere and then making an observation that something in PQ deterriorated, may have nothing to do with the value change but everything to do with the quality or vice versa. I think experiment by Bill and other will lead to some value change from .47 and clearly will result in a a better cap as well and less deterioration due to the series.

Another thing. Our projectors do not have changeable gammas (no gamma adjustment). Perhaps a cap value increase for the RGB caps in effect changes the gamma curve?
mcpherv's Avatar mcpherv 04:52 PM 06-24-2004
Everyone, the new thread is here:
(see below - link is on page 6)

Thanks to eveyone for their support. The discussion from page 2 onward in this thread will be copied over to there. I have added a bunch of new information to the new thread.

EDIT : That was the plan, but things kinda didn't work out. If you are reading this, the next few posts might seem a little weird because some of them were originally posted in the new thread.
techman707's Avatar techman707 05:04 PM 06-24-2004
WOW !!!! Do you still have tips on your fingers?
mcpherv's Avatar mcpherv 05:04 PM 06-24-2004
They are getting a bit worn :)
aspec2's Avatar aspec2 08:38 PM 06-24-2004

Now this IS providing a service. Great job.

GlenC's Avatar GlenC 10:54 PM 06-24-2004
This thread, as with many others, is really great, I have the feeling of being able to converse with "Rocket Scientists". Contact with such knowledgeable people here really enhance and bring the learning experience to new levels.

I may be way out of my league here, but asking questions is how I learn.

I have a question on the coupling caps, .47uf vs. 1uf. I hope I am correctly understanding the effect of using one over the other.

The Marquee has three identical video circuits, R, G & B, however, with the Green C-element, the outputs are not identical. The filter is for color correction only. With the 9500LC having the Green C-element, the Drive gets set much higher to balance the light output with that of the non filtered tubes to properly set the gray scale and white balance.

1. Would it make any sense to use the 1uf cap on the Green (and Red, if a Red filter is used) and stay with the .47uf with the non-filtered tubes?

2. Could this possibly reduce the Drive (edited) setting on the Green closer to that of the others?

It's my understanding that the G2, drive, brightness and contrast are somewhat dependent on each other and that an overall increased contrast setting can reduce image detail.

In an attempt to improve the Marquee performance for Home Theater use, is there any reason that the modifications to the R, G & B need to be identical?
jcmccorm's Avatar jcmccorm 11:03 PM 06-24-2004
Well, as soon as I have the time to read through all of this, I definitely will :)

In the mean time, please accept my thanks for putting it all together.


jcmccorm's Avatar jcmccorm 11:06 PM 06-24-2004
Glen, the AC coupling capacitor is not related to the filtering (C-element). Although I don't have a colored C-element, I think I can safely say that you'll be doing a color calibration again when you add it, but it doesn't have anything to do with the video chain on the VIM and neck boards.

GlenC's Avatar GlenC 01:58 AM 06-25-2004
Originally posted by jcmccorm
Glen, the AC coupling capacitor is not related to the filtering (C-element). Although I don't have a colored C-element, I think I can safely say that you'll be doing a color calibration again when you add it, but it doesn't have anything to do with the video chain on the VIM and neck boards.

Cary, My reference to the C-element/filter is that it is the only major difference in the final operation of the three video chains. It reduces light transmission creating a need to increase the green drive (I originally said G2).

My 9500 G2 is 66 R&G and drive is 25,90 R&G. If anyone is using a Red C-element, what are your Red G2 & Drive settings?

I would guess that the Green C-element is a major contributor to the majority of wear on the green tubes.
mcpherv's Avatar mcpherv 02:46 AM 06-25-2004
Walt, Glen and Cary - you're welcome, and thanks for the encouragment.

Glen - As far as I can reason from the other thread, the 1uf cap is amplifying the video signal to a greater extent (according to the equations given by Robert, where Vout is proportional to Capacitance - note I'm working from memory) than the 0.47. In theory, this is basically acting in the same way as contrast modulation (which also multiplies the video signal, but roughly constant over the frequency range, while the "differentiator" circuit depends on frequency a bit as well), in that it also effectively increases the amplitude of the signal downstream at the crts - therefore, you could probably change just the one value, and make adjustments for it to keep the video looking correct, but you would be wearing the tubes just the same.

Guys, I forgot to add this to my welcome message (although I have stated this in the other thread) - the contents from pages 2 onward in the previous thread ( will be copied to here soon, I think (I would imagine sometime today). I spoke with Alan on wednesday, and he didn't have a problem with it.
bblue's Avatar bblue 02:52 AM 06-25-2004
I don't think you should change the capacitance value of just one channel to compensate for anything. Also, it's beginning to look like you should not deviate from the original .47 value at all since the gain structure of downstream circuits depend on it.

My G2/Drive settings for these 'new-glass' rebuilt tubes and red c-element are G2 r62, g59, b61, DRIVE r32, g89, b18. That combination and the 1mf caps gives me a 25-100 IRE temperature of 6500 +- 100 degrees.

I'm chainging out the caps to .47mf and recalibrating tonight to examine the difference.

madclammer's Avatar madclammer 07:26 AM 06-25-2004
Awesome thread. The best thing on the net for me these days. Im not up to doing any of this work myself yet but hopefully will learn to do it soon. I would like to ask Tim or mp if they have tried and or reccomend jeehong lee's hvps resistor mod. I am still planning to purchase mp modded boards from etech but wonder if this hvps mod might be worth doing as well ? hope im not breaking any "rules" in asking this. thanks. so glad i went with a marquee !
mcpherv's Avatar mcpherv 08:12 AM 06-25-2004
I finally got a chance to look at a VIM and a Neck board in person today, which is nice, so that I can visualize what we're trying to do, finally (remember I don't yet own a Marquee - I just received these boards in the mail from Paul Bellemare - Gn2). The one problem I've found is that the MC34084 opamps are wide-body soics, whereas the OPA4134s are narrow. I am going to try an come up with a solution here (we could prbably use an adapter - I am going to look for one -I don't like doing this, but it actually might make things easier for placing our filter on the contrast lines). Thanks everyone!
jcmccorm's Avatar jcmccorm 08:30 AM 06-25-2004
It's easy to place a SOIC8 narrow in place of a wide body. Usually there's still some overlap of the pins onto the pad. If there's not, just bridge it with solder.

When the part you're trying to place is a wide-body but the pads are for narrow, it gets tougher. :)

mark haflich's Avatar mark haflich 09:17 AM 06-25-2004
Just remember sports fans, that many of these changes may end up making no noticeable improvement and some may end up damaging the machine or degrading picture quality. Somethings else not so far listed but perhaps being done in proprietary mods may cause a greater PQ improvement than these suggestions. I am just cautioning all thats all. I am sure there are things in here not being done by the proprietary boys that may yield additional substantial improvements in PQ.

I know Vic knows all this. Thank you for your awesome work and posts, I am confident that there are some true gems in your suggestions that have not yet be considered but some of us veterans.

An observation re RGB caps and Bill's experiments. I believe Bill is using some version but perhaps not the very latest of MP's modified VIM. If so, his observations as to cap value and type etc etc may end up being relevant to a MP modified VIM. However, since I am also using a MP modified VIM I eagerly await his conclusions and recommendation. Whatever it is, I hope he sends me 3 of those caps.

Though I do not have a schematic in front of me, the VIM I think contains a gamma correction circuit for the blue tube. I know the Marquee has blue tube gamma correction and since each neck card is the same, then if there is blue gamma correction it must be done on the VIM. I see the point re changing cap value dependent on whether the tube is R, G, or B. Changing the cap value could indeed result in something again to a gamma change at least with respect to the x axis crossing point.
mcpherv's Avatar mcpherv 09:31 AM 06-25-2004
Thanks Cary, that helps out. The ones we are dealing with are SO-14s - but I can see how this would still apply here. :) I will remove the notes I put on about this.

Mark, that is indeed a fair point about cautioning people. I am going to add in a disclaimer at the top saying exactly this. Also, you have a fair point about the fact that you guys are using the MP VIM and therefore things might be different with regards to the caps. Hopefully, though, things are still relvent to a VIM that has not had the MP treatment.


EDIT: Disclaimer added.
Robert Zuch's Avatar Robert Zuch 10:00 AM 06-25-2004
Originally posted by GlenC
1. Would it make any sense to use the 1uf cap on the Green (and Red, if a Red filter is used) and stay with the .47uf with the non-filtered tubes?
Glen, since phasing is important (as well as amplitude), I'd recommend keeping the RC product the same for all three colors.

Congrats Vic on an excellent thread!

Bob Z.
Reference Imaging
mcpherv's Avatar mcpherv 10:45 AM 06-25-2004
Thanks Bob, I have put a lot of effort into this, and I am pretty happy with the result.
Everyone, we are experiencing technical difficulties right now :D . Give us some time to get things in order.
Lightningman's Avatar Lightningman 11:06 AM 06-25-2004
Hi Bill,

Originally posted by bblue

Last night I spent some time trying different capacitors in that network, from .47 to 2Mf, compared to my 1mf 'standard'. By comparison, the .47 was duller with not enough contrast and too much color, like a high gamma. The 2mf made the picture just jump out at you -- very punchy with very rich colors and exceptionally good low level detail, yet there didn't seem to be significant increases in black or full white level (a subjective view). It did have an awesome character about it.

But if all this is is a result of voltage gain, you would think that the same thing could be achieved with contrast and brightness adjustments, yet attempts to produce the same results with C & B were not successful. What else is happening?
Thanks for doing the experiments. Is there anything you DIDN'T like
about the 2µF substitution? From what your description sounds like,
it seems like you increased the PJ's ability to display a wider "contrast
range" w/o changing the endmarks (black & white). This sounds very
interesting indeed and was what I was hoping would change.

As to the cap change vs. C&B adjustment: I'm NOT sure this would
be the same thing. By feeding a "hotter" input, you did something to
the signal itself, whereas C&B don't directly connect to the video signal,
but rather to an amplification stage. If you change C&B you would also
shift the 0 & 100 IRE points, but (at least from what I can see in your
description), this doesn't happen with the cap change. From the way I
understand this portion of the circuit, by changing the cap values, you
won't change the "background" brightness of the picture, which you
would at least do by increasing the brightness control of the PJ.

mcpherv's Avatar mcpherv 12:03 PM 06-25-2004
JAEHONG: I am VERY sorry about what happened to this thread...... If theres any way to repair it a bit (like you create a new thread again, and paste what you orginally posted, and we can try to patch it together up to my large post...) please say so!

The merge was not quite successful. Alan tried several times, and went to lengths to try to help, but it just isn't possible. Thanks Alan!
The new (again) thread is here:

Thanks, and sorry for any confusion this caused.
bblue's Avatar bblue 12:23 PM 06-25-2004
I'm pretty sure that as far as adjustments are concerned, increasing the capacitance is identical in net function to increasing the contrast adjustment (despite my earlier observation that it isn't). The difference is that the first CLC449 stage is run at a bit higher level and the contrast multiplier (AD834 or AD835 depending on your board) is run in a different (lower) range as a result. Black level would be unaffected.

My VIM is an MP modified -03 with the AD834 multiplier and it may be that there's more headroom to be had there than on the -02 with the AD835 multiplier. I just don't know yet.

bblue's Avatar bblue 12:29 PM 06-25-2004
I've put the original .47mfd metal stacked film caps back in my 9501 and spent some time with grey scale and tracking adjustments, and comparing some pre and post cal settings.

First, this is an -03 version VIM (my favorite with the AD834 contrast multiplier) and regular neckboards, all MP modified. Mark, I don't know if this -03 is the latest rendition of VIM mods or not. Probably not, but certainly representative of a well tuned board. The VNB's are the latest revs. Based on what I've seen so far, this particular board (at least) has sufficient headroom and can easily benefit and make use of a hotter signal by virtue of the 1mfd cap replacement. I could swear there is finer detail, especially in backgrounds, with the 1mfd cap installed, but I haven't gone the route of measuring it yet.

I used two white level reference points to compare the relative outputs of the two caps. With the 1mfd caps at my standard settings the light output at the screen was 12.5f/L at 100 IRE, and 6.85f/L at 80 IRE. With the .47's it was 11.9f/L for 100 IRE and 6.6f/L for 80 IRE.

Starting from scratch with .47's and C & B at 50 and using Drive on the tubes to push out as much as possible on the high end (limited by the green drive) I got 8.8f/L at 100 IRE, a gamma of 2.5 and a color temperature range of 6503 +100/-200 from 0 to 100 IRE with an input reference .7v p-p.

To achieve higher light output I needed to set contrast at 53 to get 10.54f/L, gamma of 2.5 and temperature range of +/-100 of 6503 degrees.

Higher light output again, contrast to 57 to get 12.1f/L, gamma of 2.5 and temperature range of +150/-250 of 6503 degrees. The extra low range was due to the blue tube running out of steam. The final tube settings for all of this were Drives R34, G93, B23, and G2s R63, G61, B62.

I'm using Colorfacts for calibration which uses by default, the desktop for its pattern and level generation. So the next step was to compare the desktop to overlay, since overlay and VMRx would be used in HTPC's. The desktop is considerably hotter in output than overlay is! You cannot increase the level of overlay past its default without white crushing, Fortunately, the NVidia drivers have individual level tailoring for each, and if you set desktop at 91%, it will match the overlay output.

bblue's Avatar bblue 12:29 PM 06-25-2004
This is looking great! Keep up the good work.

First ... 2  3  4 5  6  ... Last

Mobile  Desktop