CurtPalme.com Home Theater Forum

[Nashou66]

Right Drags, I was just looking at decoupling caps and I'll say it again, I really wish they Made SMD OsCons in smaller case sizes. I was thinking of stacking SMD Film chips as decouplers and then adding a larger size SMD oscon on top since now the other chips will raise the hight up off the board and possible clear any SMD IC's that might be too close together.

But with the EL5166 use a 4.7 uF decopiling cap and then try a 47uf OS CON radial cap across the pins of the amp.

Athanasios

[draganm]

don't mean to argumentative but i'm not a fan of stacking caps. My concern here is Oscillation by running so many different caps in parallel. My understand from the reading i've done is that by running them in parallel you get a generally flatter impedance curve over the time domain. However the problem I see from the research i've done is that the sharp response curve of MLCC's still leaves lump and bumps in the impedance curve. These deviations from constant impedance manifest themselves as Oscillations in the time domain.
OR, to put it in simpler terms your filtering is lumpy I don't need to to explain why Oscialltions are bad for the Op amp.
The other thing that bothers me about this it's so old school. I mean running all those caps in parallel is what they've been doing now since the invention of the micro-processor. It doesn't use any of the newer wideband cap technology out there now.
For instance, the plain-jane 470nf 50V MLCC used to de-couple the main chips in the VIM, C33,49,65. they now offer MLCC's with 3 node capacitive circuits with the 3rd node being a balanced reference to ground.
page 4
http://www.johansondielectrics.com/images/stories/surface-mount/x2y/JDI_X2Y_2010-06.pdf

not only that, but the design offers extremely low ESL which is important (if you want I have a Murata paper on the importance of low ESL in filtering).
This is just one example and doesn't even get int to the new low ESR / low ESL Tantalum Polymer caps. These are now offering very low ESL and low constant impedance up to about 10Mhz at which point it could hand off to the new MLCC. Maybe use 2 modern caps instead of 6 old ones?
So how about we try and come up with a new and unique filtering solution for the primary RGB opamps using technology from this Century, Who's up for it.

[Nashou66]

Drags, were you hacking my computer last night?

I was looking into EMI filters , here is an excellent paper on it.

http://www.syfer.com/category_docs/sm_filter_article.pdf

It explains what you were talking about and some other things.

I agree that with certain caps stacking can be detrimental. This is why I like Oscons, super low ESr and ESL as frequency increases.
The key is to do as MP does use caps and inductors to smooth out the filtering so you do not have the Lumps or Bumps.
But with the right combination of stacked caps you can avoid those bumps. This is why I think and others are doing this as well is to use
1nf,10nf,100nf, 1uf and a larger cap across the +/- pins for those times you need power overhear requirements.

The reason I like Oscons is, on the newest version of the CCM it has that really bad vertical band issue. I tried the stacking method and adding larger caps across the +/- pins of the associated chips and added inductors in line of the signal, it worked but also altered the wave form a bit. Only when I added a 100uf 20 volt Sanyo Oscon did all those bars go away. Also it improved even more when i removed the other smaller caps i added parallel to the 100nf caps there.

But i still think the EMI SMD chips might be the ticket here.

there are some other good papers on that site about filtering, especially one where they add those chips between the pins on connectors, like the ones on the Vim and CLm etc.

Athanasios

[tse]

You can get an idea of what different caps do by looking at a simple Spice program from Kemet. It can show the combined impedance of paralled caps. Low ESR up to 1000MHz is important for the opamps with the very high bandwidth.

http://www.kemet.com/kemet/web/homepage/kechome.nsf/weben/kemsoft

Scott

[Nashou66]

Scott I have a question for you. Why did electrohome choose not to use Voltage regulators on the boards them selves Like the NEC's do? have been looking at the Super regulators made by Belleson and others and they seem like much cleaner power sources.

It would have been nice to be able to just replace the on board regulators with these super regulators.
Bob Stephan is going that route with His Xg's and is having the owner of Belleson make special one to handle the highr frequencies a PJ uses.

http://www.belleson.com/

Athanasios

[tse]

[Boilermaker]

While I don't have a Marquee, I have followed this thread (and others that are similar) and really enjoyed watching the effort put forward by Drags, Nashou, Scott, MP and others to squeeze every last bit of performance that is possible from these projectors. It reminds me of doing the same thing to audio equipment many years ago and it was fun!

[draganm]

[Nashou66]

I agree, i tried using TI's spice program and got lost

Athanasios

[tse]

[Boilermaker]

Scott, thanks for the reply.
I had forgotten about the series capacitance effect of coils.

I have had a couple of conversations with Nash about the possible use of super-regulators as a direct replacement and what possible gains (if any) might be achieved. As a test platform, we have decided to use the Moome input card. Evidently, the input card picks off its power from the +5 volt rail on the Marquee. Nash is looking at the possibility of temporarily feeding it from the +15 volt rail and using a Belleson +5 volt regulator to serve the DAC. It appears that there is already a place on the Moome board for a 3-terminal regulator that appears to be jumpered. Or perhaps he could just use a wall-wart.

He could then easily measure any difference in output from the DAC when feeding a 1080P@72 test signal and also "see" any difference in noise at the 5 volt input to the DAC. It should be interesting!

Thanks,
Bob

[Nashou66]

Well I have not taken a look at the LVPS in the 2004 longbows yet. I was going to do some work on them with a few ides I had but may not now after looking at it.

Here are some pics, its a 2001 printed PCB. A few interesting things about these PS's.

1) The Filament add on PCB to fix the over voltage is now part of the main PCB.

2) Near one of the LM353N DC op Amps they added a 10uf cap and diode over a resistor. Not sure what its for.

3) They used much better caps all over. Nippon KME and a few Panasonic's in the 85 volt section, as well as Dragans fav Nichicons
in other places. and a few Jamicons.

The large red Wima cap is gone replaced by a smaller in size black film cap, not sure the brand.

Here are some pics, pay note to the Filament section and the added cap diode combo.












Athanasios

[dvh99]

i will add a little something, i replaced the diodes on the conv and vdm board.
when adjusting convergence the line now goes smoothly without any "hiccups" that most have seen while adjusting the conv.
the end of the lines from the raster dont show those funny ends anymore.

had those diodes lying here a while but i finally put them in, the conv is off a bit so it needs to be converged after replacing the diodes.

[Nashou66]

[Nashou66]

Well I wasn't going to change out the caps with some caps I had planed on using for the older Marquee's I had before I got the 2004 Ultra's.
I know many do not believe in boutique caps but I am one who has to try things out before I toss the thought out the window.
So i put in all the largest caps in the LVPS and upped the 85 volt rail from the 33uf to 100uf 160 volt and one 47uf 160 volt.
I wanted all 100uf but could not find anymore of these caps so the two in series
are the 100uf's.

Why could i not find them? Because they are no longer made..... Yes they are the famous Black Gate caps.

Man are some of these large compared to new caps. From researching these I guess they are larger due to materials
used and manufacturing to get the special characteristics.

There were some add's from the original Black Gate makers before Rubycon took over the production
where they used them in Televisions. So i figured I try them in the power supply.
Also Walter has done his Zenith 1200 years ago and kinda inspired me to try them.





Before and after. Notice how large some caps are compared to the replaced caps.

I still have the smaller caps to change out.






I only have done a small run in period on my test bed and no source but looking at the internal
Greyscale pattern looking for background noise I always see in the marquee.

There is a reduction compared to my other Ultra in the theater, not sure if its the
Black Gate's or just the increase in uf's on the 85 volt rail.

I guess I wont really know till I finish it up let the caps settle after about 50 hours or so and then compare
with video material.

I also have some Soft recover rectifiers I want to also replace. that will come at a later time after
i first try them in an older LVPS before I put them into the Black Gate Power supply.


Yeah I know I'm crazy or just bored.

Athanasios

[1031]

Hey.. thats quite expensive bunch of caps that you have Anyway, i´m interested to hear, if you notice differences whit those blackgates.

[mp20748]

[stefuel]

It's my experiance that most of the Marquee noise is coming from the people who work on them

[1031]

[Nashou66]

well one thing about the Black Gates I do not like is they only come in 85c temp. But from what i have read about them is
if they are used in fairly low temp areas they are fine. also they handle SMPS better than other 85c caps. That is one reason
i did not use them on the VNB's as it gets really hot with high resolutions. On those I used panasonic 100uf caps i think the EU series.

The 33uf caps replaced were panasonic 105c. So we'll see.

I always have a hard time scoping the power rails. i never seem to find anything. looking for anomalies on other wave forms
like focus and astig etc is easier to see.

Bob Stephan gave me some pointers, to lower the volts div way down and then use the AC couple selection on that V/D control.

I'll measure it from the MB harness is, if/When i do it. Might be some time from now. I'm getting some test
Super regulators Bob wants me to test for the 5 volt rails on the VIM.

Lots of experiments and little time and ambition as of late. Geeze it took me close to a year to try the Black Gates

And the Places i got them from are all sold out. A year ago when i was ordering them I bought the last of some of the Values from 3 different places.


Jarmo over all cost for two PS was just over 400 dollars. A couple bellow sales paid for them back then


Athansios