AVS Forum banner
Status
Not open for further replies.
1 - 20 of 53 Posts

·
Registered
Joined
·
1,049 Posts
Discussion Starter · #1 ·
Just wanted to thank Ken Hotte for some of his recent comments on upgrading electrolytic capacitors to improve projector scan line performance.


I've been having a problem with video noise on my eight year old Ampro 4200, and decided to swap out all 14 capacitors in the low voltage switch mode power supply.


While I haven't eliminated the noise, there has been a definite improvement in the sharpness and definition of the video image. The image I get from my DVDO Iscan isn't nearly as pleasing as it was, because the scan lines are standing out more. I find myself using the HTPC more to take advantage of the higher resolution 1280 x 720 image. I think the swapout has been the best single improvement I have made to date.


I followed Ken's advice about going to higher voltage rated capacitors and generally going up one size (say 600 uf -> 680 uf) for output filter capacitors. It's surprising that in all cases the new capacitors were physically smaller that the ones they replaced.


If you decide to try this, you might want to discharge each cap before you remove it, just to eliminate any surprises. The Digi-Key company has a nice selection of high grade Panasonic FC, EB, and HB ranges(all 105 degree C) to choose from. My total cost was under $50.


I had forgotten that electrolytics have a lifetime rated in hours. I'm not sure of the definition, but I think it suggests a 20% change in capacitance after rated hours. On my new selections I tried to pick 5000 hour rated units. I think a lot of the original equipment types were more like 1000 hour units. That's pretty easy to exceed on an old projector.


I notice some electrolytics on the horizontal output board. Does anyone have a feel as to whether they are worth going after, or where else I might look to eliminate video noise? I've tried everything on the rgb input side, such as short cables, common power outlet plugs, lifting ground plugs, and an rgb cable buffer/isolator unit.


Thanks,

Ray
 

·
Registered
Joined
·
6,660 Posts
There are other things you can go after to eliminate noise in the PJ. If it has a central board that handles sweep duties,and any of the input or line amplifiers, then go after any electrolytics there. If the unit has electrostatic focus and/or electronically derived focus control circuits, they are CRITICAL to image quality. Go after those, if any electrolytics are involved. Specifically any line drawing or horizontal/vertical control circuitry. The dividers/ clocking circuitry are VITAL for line placement and control which DIRECTLY equates to focus and control. So, re-doing the VIM or control board in a marquee creates a VERY stable and film like image, as well as doing the rest of the boards. in the marquee chassis i have, the appearance of the image, after doing the rest of the PJ, as that the noise floor and stability of the image took a HUGE jump forward after the mods to this main control board. SO, the VIM board on the Marquee, and any equivalent on a Ampro chassis. For me, it was a 'appearance' of about a 3-5db change in the video S/N ratio. That's absolutely HUGE. But then again, I had done every single part in the PJ before that, the control module was the last thing I did, and seemed to be vital to the image quality,as Chris Stephens said it would be. Funny, for some reason, I left it for last.
 

·
Registered
Joined
·
1,049 Posts
Discussion Starter · #3 ·
Thanks Ken,


That's great feedback. My unit does have the dynamic focus control. The Ampro use a swing up panel on which are mounted what are refered to as the "registration" control boards. I think this is the same functionality as the Marquee VIM main control board.


The registration control is the last place I would have thought to look at. I just opened up the unit, and sure enough there are a lot of electrolytics (some rather large) in addition to all the digital to analog converter ic's.


Wow, there are a lot of electrolytics spread throughout this chassis! I'll get started on the control board area and report back to you when I get some results.


Ray
 

·
Registered
Joined
·
139 Posts
With switchmode power supplies you have to be careful when replacing electrolytic caps. Sometimes the designer requires the ERS zero created by the capacitor in the circuit. This can effect the stability of the power supplies control loop if ample phase and gain margin aren't provided. To ensure stability you must do either open loop gain and phase margin testing or as a minimum do step load testing using active or passive loads.


Don
 

·
Registered
Joined
·
1,049 Posts
Discussion Starter · #5 ·
Thanks Don,


I wasn't totally clear, but I only went to larger capacitors on the downstream secondary side of the output transformer where there were ripple filters on the low voltage outputs.


If there was any doubt about the closed loop functionality of a capacitor within a circuit, I kept the same capacitance, but did go to a higher voltage and temperature rating.


Ray
 

·
Registered
Joined
·
18,507 Posts
As a rule, I have found that low value high voltage caps fail first...


Don't forget that every resistor should be a 1% metal film non inductive..;)


I'll be interested to read about what heppens when you 'up' the capacitance of many of the 'lytics..


Curt
 

·
Registered
Joined
·
21,464 Posts
Cap failure in all old (and even new) electronics is IMHO is the number one cause of product failure. High voltage caps are the most vulnerable to failure. Designers have a tendency to ru voltages too near the edge of a high voltage cap's working voltage. With a low voltage cap, the tendency is to put a little higher voltage cap in instead without raising the voltage it is carrying. It is cheap to do so and usually there is no benefit in raising that voltage. At high voltages (used for tubes (but not for tube filaments)) the design tendency is to up ze voltage. The tendency for low value high voltage failure is not surprising. Also by comparison the value of a high voltage cap ia always small compared to a low voltage cap. Cap value is expensive and size is very bulky at high voltage. Now for cap replacement in CRTs. Several purposes, replace older caps to prevent future failures. Most owners do not do this but boy should they have a good tech do this for critical caps. There are not that many. Replace leaky caps to improve performance and prevent failure. Increase the value of some caps for better performance. Increasing the voltage will only improve reliability. Should have no efffect on performance. Have to be careful re what cap values are increased. By passing some caps with high quality film caps. Replacing some caps with better quality caps. Quite a sport. Significant gains in performance and reliabilty are obtainable. Wish the experts would post specifics as to caps and brands, numbers etc. for Marquee boards. Particularly for upgrading the input coupling caps and the many small value decoupling caps..
 

·
Registered
Joined
·
596 Posts
When replacing electrolytic caps, I would recommend using the Panasonic HF series of caps as they have the lowest ESR of any caps I have seen. These are a little bigger than most, so you will need to take that into consideration. Actually, the best caps to replace would be the electrolytic caps that are at each semiconductor which provide filtering as close as possible to the active device. In parallel with each of these is usually a .1mf disc cap used as a bypass. This helps in the high frequency areas and I recommend these be replaced by metal film types of as large a capacity as will physically fit.

IMHO, the best application of cap replacement is to replace the electrolytic caps in the video signal path all the way to the crt. These can be recognized as they will be low voltage high value non-polar electrolytics. The effect will be dramatic if you replace them with polypropylene metal films. Since these are much, much larger, you will need to arrange them in such as way that they will still fit and probably hold them in place with silicon or a strap. There will probably be at least 2 or 3 of these in each of the three color circuits, and their purpose is to block DC from the power supply rail. Try this, and I think you will like the results. The down side is that these caps will probably cost about $20 or more each!

Thanks,

Bob
 

·
Registered
Joined
·
6,660 Posts
I have made the situation a bit murky. What I was speaking of was TWO different boards, but did not make that clear. I was speaking of the control board, which from looking from the back of a marquee chassis is on the left. It is the one with the diagnostic lights. Similar attacks should be mounted on the VIM, or input module, re the 15volt rails.


There is a entire run of circuitry on that board that runs at the 5 volt level. Lotsa logic circuitry. These chips need to have their .1uf ceramic caps removed, and swapped out with something like the following: Kemet brand, 47uf 6.3volt tantalum caps. The leads will have to be bent, and then cut, and then mounted on the solder pads. The cap is T350F476K006AS17301. There is also a run of OP-AMPS on this board as well. Some higher voltage rails are used. The buffer caps (two for each chip amp) of each rail should be swapped out to be Kemet brand 16 volt 22uf caps, or higher value. Maybe 33uf tantalum caps. They should be mounted in the same way. Same series of caps from Kemet, but 16 volt rated. Anything bigger and you raise the ghost of unreliability. Anything lower in value, and you lower the effectiveness of the modification.


I have suffered NO failures or odd behavior. This modification of the control board brings about a huge increase in the stability of the line drawing and drops the noise floor as a consequence, due to the fact that lines are being drawn in their proper spots, and therefore this equates to alignment, which is gun-to-gun alignment (overlay) which directly can be equated to maximum contrast range, or video S/N ratio. The effect this circuit has on damaging the rest of the PJ's dealing with the quality of power in the -/+ 15 volt rails is quite critical as well. When you run at 1920x1200 at 72hz like I do (simply a very high scanning number, well over a 2X multiple of the original DVD transfer, and nearly a perfect multiple of the original HD master the DVD transfer was created with), to get the BEST DVD overlay possible; these modifications, and the clairity they bring to the PJ are CRITICAL.


Fixing this at the chips involved, stops them from polluting the rest of the parts that use these rails, and gives them greater stability as well. So, the effect is doubled, in a way. Of course, those voltage rails would be corrected as well, (all other boards and circuitry) right at the parts involved, on their specific boards in question.


Video S/N ratio is what it's all about. Achieving that is what brings you closer to film. The PJ only has so much light output that it can dish out in a linear fashion. Dropping the noise floor has the same effect as increasing the overall output capacity of the PJ,whilst bringing every aspect of that Video S/N range and signal capacity to a more linear and perfectly represented level. This brings about tremendous stabiltiy to the image, and makes the experience considerably more 3-d like, and film like.


If you can't increase the output of the PJ, drop the noise floor, and linearize the behavoir of every gradation of a signal inbetween those two points. MASSIVE improvement in image quality is fully assured. A greater RANGE of signals can then be properly presented. Subjective analysis of content (DVD, HD, etc) can then take place. For instance, films like Dark City, look fantastic. You get the Video S/N ratio right where you need it, at the bottom of the range. (besides everything above that!)


Basically, HD mastering suites wish they had the image qualites that I have to suffer through enjoying. I don't care if they are using a G90, Barco Cimemax, or whatever, they ain't getting the image I am, period.


I did this and more to a ECP 4100 chassis, in an extreme way, about 1.5-2 years ago now. The ECP ROCKED. It TRASHED am mint condtion NEC GP-5000 in image fidelity. And the GP-5000 is a 9" tubed Liquid Coupled unit, with HD-10 lenses.
 

·
Registered
Joined
·
21,464 Posts
Bob. I am not sure I understand your post. The only caps directly in the video circuit of a Marque are blocking caps on the RGB inputs. There are many many power supply related caps but NONE to my knowledge block DC. The supplies are in fact DC supplies, not AC supplies. I think I am a little confused. Please understand this is not a bashing response and you have supplied valuable info to us all. I think it is just a case of perhaps the wrong semantics.
 

·
Registered
Joined
·
596 Posts
I am not familiar with the Marquee circuitry, so there might be only one set of non-polar caps in the video signal line. I'll restate my probably confusing post on DC as it relates to these caps. If you were to short these caps out, there would be a DC bias superimposed on the signal which would move the average voltage of the signal out of the linear operating range of the next stage. This capacitor serves as a single pole high pass filter into the next stage. It is sized based on the input impedence of the next stage device. It needs to be large enough in capacity to allow the lowest frequency signal that we want. I'm sure it was sized appropriately so you should not change the value. If you size it too large, there would be a chance to allow low frequency video noise to enter, but the purpose of this cap is to make sure that there is no DC voltage prevelent on the signal - Just the AC video signal. Because a relatively large value cap is needed, and since the voltage will fluctuate between positive and negative, it was necessary to use a non-polar electrolytic for reason of cost and physical size. IMHO, this is by far the most important cap changeout that can be done. The actual performance of a non-polar cap especially at the video frequencies we are dealing with is several orders of magnitude worse than a film type cap. I hope this makes sense!

Thanks,

Bob
 

·
Registered
Joined
·
6,833 Posts
The Marquee only has three coupling caps in the entire video chain. There is one for each color, it is right after the BNC connectors/relays (dc blocking) from there to the CRTs it is DIRECT COUPLED, yep, no other caps in the chain. There is no need for low frequency consideration, because this is VIDEO.


Because of the direct coupling design of the video circuit, I can't see where one could add caps in the video chain and improve on its image. However, I do agree with KBK on lowering the noise floor, because that would improve the signal to noise ratio, but that could only be accomplished by way of the DE-coupling circuits. The power rails in the Marquee are all high current rails, and because of the nature of video there is no transit demands for power like what happens in audio circuits when low frequencies (bass) is being amplified. The only time video pushes a demand is on bright scenes, and we do know that that places a demand on the power supply's (current), because video is in an envelope. Sometimes it seems like we are talking about audio, when the same tweaks that worked in audio will offer little benefit for video. Adding big, large, special, multiple caps is not new, however, it has become new for video. We are excited about what these caps are doing to these projectors. I have many schematic diagrams of video projectors... why did all of the engineers of these projectors forget to do this cap thing in their design of these projectors, did they not know about the benefits?
 

·
Registered
Joined
·
21,464 Posts
To many EE, a cap is only a cap and once you get past the value and working voltage. the cheaper the better. There are many many characteristics of caps that traditionally trained EE do not seem to recognize. See the Pooge thread for my additiobal comments. Those video engineers wre by no means infallible. We will try some of these cap mods Mike at my expense and your gain and I expect one of us will be eating his words. Do not bet against this old pooger.
 

·
Registered
Joined
·
596 Posts
Mike - I agree with what you are saying about the PS cap supplying a lot of current in low frequency audio. An electrolytic cap is an excellent device for doing this as it has a relatively low ESR (equivalent series resistance) at these low frequencies. Therein lies the problem. Video is a much, much higher frequency than audio, and the ESR at video frequencies of an electrolytic cap is very high. Therefore, when the cap is asked to supply current to the active circuitry at high frequencies, it causes a voltage drop so that the circuitry sees non-linear voltage depending on frequency. That is why there is almost always a small ceramic cap in parallel with the electrolytic. It's pupose is to help lower the ESR at higher frequencies. By changing the electrolytic to one with a lower ESR and using a film type bypass around it, we have reduced the ESR of the power supply that the active circuitry sees. Power supplies are supposed to be ideal voltage sources that have zero internal impedance so that at any load condition and at any frequency, it will supply current. With non-linear ESR with respect to frequency, the active circuit being used cannot have a linear response. Yes - power supplies have a frequency response! By changing the type and values of these capacitors, we can make it more linear, and therefore allow the active video circuit to be more linear.

Mike, as to the cap in the video "chain", I am not advocating installing an additional one, but changing the existing one to a much higher quality. You WILL notice the difference!

Thanks,

Bob
 

·
Registered
Joined
·
6,833 Posts
Bob,

I can follow very well what you are saying, and I agree with a lot of what you have said, however, this CAP thing has gotting a little out of control. One could read these threads and assume that by adding or replacing caps, that you could change a projector with a bandwidth of 60Mhz to a bandwidth of 200Mhz, just by replacing some caps. It's seems to always be an argument from the ESOTERIC audio fraternity about caps and their benefits in the VIDEO circuit. Audio is rather simple, but to achieve that very high bandwidth that these projectors are capable of the value, voltage, ESR, etc, are all very important elements, by just putting in a better or more expensive cap is just that, now, the question would more likely be: how did it effect the overall bandwidth (good or bad).


Donald Bennett gave the best answer on this, but his truth not only applied to a switch mode power supply, it is more important in a video circuit because of impedance and bandwidth, these caps are not just randomly chosen from your local Radio Shacks grab bag when the engineers designed the projectors, they are not cheap, maybe on a LOW bandwidth projector but not on any high bandwidth projector that i know of. Research needs to be done to find out who made these caps and why they were chosen. But to order a bag of expensive capacitors and to start swapping caps, no, I have better things to do. The only thing that i would agree with is the De-coupling that KBK mentioned, other than that, I can't technically see where adding caps would make that big of a difference. All of the testimonies have been great, but they were about AUDIO circuits, and I agree, but this is VIDEO. If the coupling cap on the Marquee is capable of handeling 130 Mhz bandwidth, how could this be a REGULAR cap?


I have in front of me a Marquee 130Mhz Video Input Module, there are a very few electrolytic caps in the video chain (about 5), and two of them are for the blue gamma circuit. The other caps are for De-coupling, the same applies on the neck board. the entire video chain is direct coupled. The caps on this board are all SPECIAL. You would not find cheap caps on a high bandwidth board, well, not on a Marquee.


Everybody mods the Marquee, why don't someone get a Kloss Novabeam and do a cap thing to it, that way we will clearly know the results of adding the caps. Modding a hi performance CRT projector is like fixing something that works very well...;)
 

·
Registered
Joined
·
6,660 Posts
The big problem with logic circuitry that EE's tend not to realize, is that.. as far as quality demands on the power supply, the WORST circuit to EVER deal within this regard is.. DIGITAL circuitry. Due to the square wave behavior of this type of circuit (basic design parameter), the slew 'range' demands on a power supply actually approach INFINITY, which is WAY beyond that of the demands of analog audio circuitry. So linearity within the slewing capacities of digital circutry power supplies is absolutely CRITICAL, and WAY beyond that of analog power supplies, which is the OPPOSITE of what you would think.


Think about it for a second.......DC, to lightspeed, and back.... just to replicate a square wave.


In the useage of digital cicuitry in the PJ... well! The 'bounce' back of energy demands into the power supply for these "transient demands' VIRTUALLY GUARANTEES that large amounts of timing errors and noise will be the EXACT resultant.


Largeish, SINGLE design, localized buffers are the correct approach. This is eactly what the Electrohome engineers have done. However, the caps put in are merely 'acceptable' and no where what WE want them to be for the BEST quality. So, swap them out with some tantalum caps(stay away from 'wound' capacitor designs here), and boom! much lower power supply noise (due to the fact that transient demands from the associated digital chps are handled by the LOCALIZED tantalum capacitor... remember, the transient current demand is nearly INFINITE!!!), and more precice behavior in the digital componentry, and....less damage to the power supply lines for the (power-supply-line coupled!!) analog circutry.


Much more stable 'clocking' behavior as well. Line drawing is MUCH more precise. Instability in the transient behavior of the power supply leads to line drawing 'jitter' so to speak. Fix it with wholesale cap changes at the power supply buffer circuitry. In this case, the local chip level buffers on the Marquee logic board. Remember, this is the MASTER board for ALL line drawing duties. Everything flows from it.
 

·
Registered
Joined
·
3,902 Posts
Oh geez...I'm beginning to get drawn into this thread and the ideas it presents.

I'd assume my NEC XG could also benefit from cap replacement.

Any idea how I might go about getting schematics so I can ID the caps that must be replaced? I'd also assume that even though NEC is no longer making CRT pj, these schematics are still proprietary???
 

·
Registered
Joined
·
596 Posts
Ken - You hit the nail right on the head! Digital signals (which are really analog in nature, but just interpreted digitally!) theoretically take a voltage from 0 to some value (5Volts) in zero time. I SAID THEORETICALLY! In actuality, just as you said, there is some slope to this line. In order to have a perfectly vertical wavefront with 0 to 5 in zero time, you would have to have a bandwidth capability of the circuitry and the power supply rail of infinity. That will never occur because of the construction of the IC and because no power supply will ever have zero internal impedance at an infinite frequency. Now lets say that the power available to the digital chip has a bandwidth limitation because of ESR at the risetime of the signal. When this happens, data is mis-interpreted or just lost. That is why it is imperative that the caps used at these locations can provide adequate bandwidth to the chip. Hence the use of very low ESR caps.

Mike - The engineers that designed and speced all of these projectors know what they are doing. There is no way that they can use the type and quality of caps I am suggesting, as they would no longer be able to market the projector because of it's cost. For example, on my GP-5000, I replaced six RGB caps that cost about a dollar each and could almost fit in a thimble with caps of the same value that cost $25 each and are about 3" long and 1-1/2" in diameter. Bypass electrolytics for power supply use are not nearly as expensive nor are the sizes much different.

If you are interested, I might be able to drag those old articles out of my attic and fax to you if I can find them. They were well written and provided documented readings of testing from various types of caps. Try it - I think you will like it! I envy Ken's work in what he has done and would love to see his image produced. I am just getting started, but will end up at the same place he is at. Just wish I had more time!!

Thanks,

Bob
 

·
Registered
Joined
·
21,464 Posts
Bob. your post re lowering PS impedence at high frequencies is right on the money. I had a long talk with MP before your post this morning and he will verify that I told him exactly the same thing. We also talked about Marsh's ultracap (essentially replacing one cap with 10 caps in parallel at 1/10 the value each). Boy does the ESR go down when you do this. The ultra cap contains essentially 10 caps in one. Every localized electrolytic should be bypassed by a .1uf and and .01uf. I still do not think MP is convinced. He is more stubborn than me and that is SCARY.


Ken. you still haven't answered my question as to the specific substitution blocking cap at the RGB input. If you consider this proprietary, I understand.
 
1 - 20 of 53 Posts
Status
Not open for further replies.
Top