[kal]

A fellow Barco 800 owner and myself have been discussing offline how to clean up this +17VDC supply rail from the Switched Mode Power Supply (SMPS):


Here's a snapshot of the signal taken from my scope:
http://www.avsforum.com/avs-vb/attac...achmentid=4273


The 'noise' on this rail seems to be mostly:


- A low frequency (works out to 30Hz) rectified sine wave.

- A bunch of high frequency noise (0.8V peak to peak - seems like a lot for a +17VDC signal).


We'd like to add some extra filtering to get a cleaner image out of our BG800's.


What's the best design approach for getting rid of the 30Hz component? Add a bunch of large electrolytic caps in parallel to further flatten the AC component?


What about the high frequency component? Smaller ceramic disc caps in parallel? I have no idea.... this stuff's all a hazy memory from EE school...


Kal

 

[heathyoung]

The high frequency component will be hash from your switchmode power supply.


Low fequency component is half-wave rectified 60Hz, probably through the doubler (you are on 110V right?) I would have expected 50/60 hz on 220-240 (no doubler needed).


The way that the power supply works is that the 110v input is doubled, rectified (to give 312 or so volts) and this is put across a ferrite transformer at high frequency - 40-50khz or so. The pulse width is determined by the load, and the dips you are seing at 30 Hz are the supply caps draining as the voltage into them drops to zero on the swing of the sine wave. Bigger supply caps on primary (HT side) also mean these dips are smaller (smoother).


Bigger caps on the secondary (LT side) mean that the dips will be smaller as well. Be sure to use low esr caps for better performance as well.


Looks like your supply was under a reasonable load there, or your high voltage caps are drying out. Looking at your scope, you are 0.2V/division, so you have a 0.4v pp on the 30Hz waveform. Not *that* bad.


High frequency hash is a ba$tard to chase out of SMPS's. The usual trick is to use a coil in series, the coil will have a specific (high) impedence at increasing frequencies.


My suggestion - if it works, leave it alone, or else redo with LOW ESR capacitors of the same value. SMPS's can be very finicky beasts and unless you know exactly how it works, you can open up a major can-o-worms.


I have seen too many explosions from people who don't know what they are doing with SMPS's.


Higher rated caps (voltage, temperature) are a good thing. Capacitor technology has improved over the years, and if the unit you are fixing has high chassis hours, then there is a chance the caps are a bit dry anyways.


You mess with this at your own risk BTW.


Cheers

Heath Young

 

[mp20748]

I agee with Heath. Change the caps with the same value, but use high temp caps. After you change the caps you should see a lower noise level.


Change all of the caps...

 

[Curt Palme]

The 30 Hz is probably a ground loop that you've created with your scope. There should be no way that 30 Hz can get into the output of a SMPS. There is no half wave rectification in a BArco 800, it's full wave...


It's a Tek scope from the '70's right..;-)


Curt

 

[kal]

Thanks for all the hints guys!


I have to isolate a grounding problem I'm having first then I'll start on this... (another member's sending me a bunch of spare BG800 cards). There's another forum member here (Mikedd) that's wanting to clean up these supply lines as well so maybe he'll start off and comment... Any comments Mike?


I've actually already replaced all of the output filtering electrolytic caps on my SMPS with same value, higher voltage, 105C Panasonic FC series caps. The 17V line looked the same before and after. The only difference is that now my convergence is a lot more solid and drifts less (if at all) in the first hour of operation. I also changed another 50 or so odd 'lytic caps that had dried up to some degree (ESR was higher then normal) on various other cards a couple of weeks ago. Same result: Better convergence stability, but same amount of signal noise.


I had at one point also replaced the larger 400V caps on the front end to help smooth out the 300VDC signal right after the diode bridge. I actually made the picture noisier so I put back the old caps... I'll have to go back and try again and measure before/after. The originals caps actually measure just fine in terms of ESR, so they don't seem too worn. Though my unit is from 1990 with over 23000 hours on it.


What about using soft recovery diodes in the bridge rectifier? They're supposed to produce less noise/spikes when switching on/off as the AC voltage swings across them... They're big in the audiophile world... not sure how much they'd help here...


Heath: Yes, I'm at 110VAC. Re: inductors in series with the output to kill high frequency noise is a good point and made me look at the BG800 SMPS schematic a bit closer because I could have sworn there already were inductors in series with all low voltage output lines. Yep! There's a coil on each output line, but it's actually AFTER this +17VDC test point (Silly Barco! doh!) - so the photo above is BEFORE the coil. I'll have to get in and find a test point to 'scope after the coil for a real photo of the output.


Am I doing my math wrong to get 30Hz? Horizontal divisions on the photo above are 5ms. A full wavelength (two bumps since it's rectified) seems to take approx 6.5 divisions, so frequency is 1 / (.005 x 6.5) = roughly 30Hz ?


Curt: Yep - it's an old Tek 453 - one of the last 'scopes that still used Tubes/Nuvistors and then swiched over to Transistors in the last half of the builds. Mine's one of the later ones so it's solid state (a good thing!).


Mike - you going to try any of these ideas? Any thoughts?


I'll take another photo of the 17V line after the coil and after I've put in more filtering in the front end and we'll see how much the noise is reduced. Not sure why replacing the four (4) front end 400uF/400V caps actually ADDED more noise. I was using Panasonic TSHA 105C series caps...


Kal

 

[kal]

A follow-up question:


Some power supplies will use small tantalum or ceramic disc capacitors in parallel with the larger electrolytic filtering caps on the output supply rail lines. Why is this done?


Someone in a different thread mentioned adding small 0.1uF multi-layer ceramic disc caps in parallel with the large electrolytics and I'm just curious what this is supposed to do to to the signal....


Kal

 

[Curt Palme]

The small caps filter out high frequency noise, the larger ones are the brute force filtering.


This post is a good example how hard it actually is to improve on an item that a company has spent $1000's of dollars on R & D.


Don't get me wrong, I'm not raining on your parade, and I'm all for experimenting (see our local waste disposal area for the results of many of my 'experients'..;-), but sometimes even logical component replacement makes things worse than using the parts of the original manufacturer.


Even inserting small chokes/coils to smooth out ripple and hash can cause problems. Things like the additional resistance of the coil will result in a lower current flow, causing instability in circuits 'down the line'.


Have fun and experiment like heck..


Curt

 

[kal]

Good points! So far my local garbage dump is still fairly empty and I hope to keep it that way!


I'm very hesitant on doing any changes that aren't in fact direct part replacements with same (newer) versions of the same components.


And you're right: I'm finding that even doing this does not always result in something better.


I suppose this means that the warranty I had on this projector I bought from you is now null and void?


Kal

 

[Mikedd]

Guys,


I have changed out all the large 370V caps before the SMPS with higher voltage same filtering, higher ESR (Panasonic) This required some real modifications to get these to fit on the board itself. The noise floor came down a bit with much better stability. I would imagine that the originals were quite dry and out of spec by now. I think next I will go after the DC voltage rails in the SMPS and do the same thing. One thing I would like confirmed is that many of hese voltages like the +17V and +30V etc.. are regulated again at different points they travel to. I am willing to change the caps out on those as well since the visible noise could be introduced there as well.


Curt I agree it is next to impossible to improve on the real R&D these guys do, but my chassis has over 8K hours and many of those parts are good for only 3K maximum at rated temp. One thing for sure is that I got rid of convergence drift. I turn the thing on now from cold and its solid. Still got the visible picture noise/ripple though. It still l looks quite acceptable but that noise really bugs me. I was saying to Kal that if we could lower that noise floor the sharpness and depth would really stand out.


Mike

 

[kal]

Originally posted by Mikedd
I have changed out all the large 370V caps before the SMPS with higher voltage same filtering, higher ESR (Panasonic) This required some real modifications to get these to fit on the board itself.


Mike, which Panasonic's did you use for these caps?


I redid all the DC voltage rails with Panasonic FC series but this cap series doesn't go high enough to do the combined 385V/400uF required so I went with the Panasonic TSHA series (also 105C rated) but replacing these 4 caps (3 in the power input module, 1 in the SMPS) gave me more noise then before. I also had to do some real mods to get them to fit...

One thing I would like confirmed is that many of hese voltages like the +17V and +30V etc.. are regulated again at different points they travel to. I am willing to change the caps out on those as well since the visible noise could be introduced there as well.


Yes - there are often filter caps right at the card power inputs. Some of the signal boards like the RGB input/switcher also have voltage regulators on them with caps for filtering. These should also be changed. Ex: There are 3 voltage reg's on the RGB input board: -12V, +5V, +12V.

Curt I agree it is next to impossible to improve on the real R&D these guys do, but my chassis has over 8K hours and many of those parts are good for only 3K maximum at rated temp.


8K?? Try 23K!

One thing for sure is that I got rid of convergence drift. I turn the thing on now from cold and its solid. Still got the visible picture noise/ripple though.


I'm seeing the same after replacing the SMPS caps on the DC rails.


Kal

 

[Mikedd]

I redid all the DC voltage rails with Panasonic FC series but this cap series doesn't go high enough to do the combined 385V/400uF required so I went with the Panasonic TSHA series (also 105C rated) but replacing these 4 caps (3 in the power input module, 1 in the SMPS) gave me more noise then before. I also had to do some real mods to get them to fit...



Kal,


I used a 395 uF cap at 450 volts 105 degree. They were apain to fit. I made some jumper wires and soldered to the bottom of the PCB in order to use the alternate holes in the boards. At first it seemed to have no impact other than stability, but as I have run the set it is getting subjectively better. One thing I did forget to mention is that when looking into the tubes face with the contrast down the noise is quite visible as diagonal and rippling lines. The blue tube is the noisiest, red next then green. I took a peek at an Ehome and NEC and there was some noise but the ripple was missing, and they were plugged into the same outlets. I picked up an Extron synce stripper and am going to try that to see if I can subjectively separate what may be sync related issues from PS noise. I can evaluate how it all effects the picture. Too bad Chris Stephens isn't popping on here, as I'm sure he has dealt with this issue in his travels. On we go....


Mike

 

[kal]

Quote:

Originally posted by Mikedd I used a 395 uF cap at 450 volts 105 degree.

Mike, what SERIES Panasonic is this? Or better yet, what part number?


Kal

 

[Mikedd]

Kal,


Those boards are on the way as of tonight. The Panasonic caps I will look up and forward to you.


Mike

 

[kal]

Thanks Mike! Those caps have me intrigued since the Panasonic TSHA's I used had the opposite effect of what I was looking for. But since you mentioned "...as I have run the set it is getting subjectively better..." I may go install them again....


You're right about getting them to fit.. the ones I used required some creative retracing!


Kal

 

[kal]

Well, this is interesting.


I was speaking to Mike last night - he has also gone and replaced all of the 1000uF low-voltage electrolytic caps in his SMPS with Panasonic FC's (rated 105C) of the same microFarad value, but one step up in voltage.


His results were the same as mine: The picture/raster is much NOISER now then before!


So what's so special about the original German-made Siemens caps that Barco used back in the early 1990's? We've both been searching on the web for any info on these electrolytics but have come up blank.


Roland - I've noticed that you've been on the board more and more lately? Any comments?


Thanks!


Kal (and Mike)

 

[kal]

Quote:

Originally posted by heathyoung The way that the power supply works is that the 110v input is doubled, rectified (to give 312 or so volts) and this is put across a ferrite transformer at high frequency - 40-50khz or so. The pulse width is determined by the load, and the dips you are seing at 30 Hz are the supply caps draining as the voltage into them drops to zero on the swing of the sine wave. Bigger supply caps on primary (HT side) also mean these dips are smaller (smoother).

Heath,


What you wrote here makes perfect sense so I took a look at the AC component on the doubled/rectified spot (~300VDC) and saw a sawtooth waveform (more or less) with peak-to-valley dips of around 8VAC. What you see is the signal ride up on the original AC curve, then a linear decay down as the caps drain until the next rectified half cycle hits.


Makes sense. There are 4 caps. 2 are used on each swing. So I figured larger caps would result in a slower drain, thus supplying the latter stages of the SMPS with 300V supply with less ripple.


The existing caps still measure fine, and replacing them introduced noise so I figured I'd simply DOUBLE the caps.


So I added (4) 470uF/450V Panasonic TSHA 105'C caps to the existing (4) 400uf/385V ones. You don't want to see how I managed this physically.


The end result: ABSOLUTELY NO DIFFERENCE to the 300VDC supply rail. Same sawtooth waveform with the 8VAC peak-to-valley dip.


The extra supply is not helping keep the voltage up as the bridge rectifier flip-flops from one set of diodes to the other.


I give up!


Kal

 

[Mikedd]

Guys,


I reinstalled all the original caps and back to normal...well almost...some of the traces are just about beat. This was a bit of a frustrating experiment as I was not looking to "mod" or "improve" I had just wanted to remove some of the AC noise leaking into the DC rails. What I did notice was that the source of the noise increased with the 47uF caps and not really the 1000uF 40V. I'm wondering the same as Kal if these Seimens caps are Voodoo or something. I would still like to get to the bottom of the extra video noise and I'm convinced more than ever that the SMPS is at the root of it. I know people here have rebuilt those supplys so maybe someone could pipe up and let us know what might be up or what we are missing here. It is definitely ripple and is affected or more noticeable when running at certain scan rates. Perhaps Curt, Roland or Mike Parker can chime in here with some advice. As I said I'm not into hot rodding the thing, I just wanted to get the video noise down. The pre 96 Barcos seem to get noisier than my comparable Sonys or Electrohomes that I have owned. Hell at this point I would pay someone to rebuild it if it was reasonable and that way no one has to let the "Barco PS secret" out.



Mike

 

[kal]

Quote:

Originally posted by Mikedd What I did notice was that the source of the noise increased with the 47uF caps and not really the 1000uF 40V. I'm wondering the same as Kal if these Seimens caps are Voodoo or something.

I noticed the same thing Mike. My power supply is currently running with all new 1000uF/40V and 2200uF/16V caps (Panasonic FC series) but I had to leave the old 47uF/350V caps in place as the new ones (Panasonic EB series) made things worse.


The EB series is a much higher quality then the FC series. They're "High Ripple (High Frequency) Long Life High Voltage" 105'C caps who's main application is meant to be Electronic Ballasts & Power Supplies.


Kal

 

[Mikedd]

Kal,


You should have taken a picture of the PS after you doubled up the caps. Its pretty tight as it is.


Mike

 

[kal]

Quote:

Originally posted by Mikedd Kal, You should have taken a picture of the PS after you doubled up the caps. Its pretty tight as it is. Mike

Picture this: Four large caps tightly tie-wrapped together into a bundle and hanging from the Power Input Module. The Power Input Module has always had a screw holding it to the chassis so no worry about the extra weight pulling the board out.


My unit's running without the case and the ceiling mounted hushbox I use leaves a few extra inches for this to 'hang' in place.


Kal