Film-Tech Forum ARCHIVE: Chasing hum bars?

This is topic Chasing hum bars? in forum Digital Cinema Forum at Film-Tech Forum ARCHIVE.

To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=16;t=000377

Posted by John Hawkinson (Member # 1135) on 05-28-2009, 09:19 PM:

Background: Today I hung around and watched while a bunch of electricians, audio engineers, designers, consultants and technicians tried to troubleshoot a persistant problem with hum bars in a 400-seat auditorium that they recently installed here (no film, sadly). They failed.

I'm mostly curious what some tips and tricks might be for dealing with this.

To be clear, when I say "hum bar," I mean a horizontal bar of luma (lighter or darker than the image) that either scrolls up or down continuously, or just stays in a fixed position, in a video image.

This room has a particularly complicated setup. It's got video (and audio) inputs scattered around the room, both composite and VGA. Generally speaking they go into Extron distribution amps (or VGA extenders) and then run on 5-BNC cables up to the booth, where they and a bunch of other video sources go into an Extron switcher. Outputs from those switchers go to three seperate scalers, each of which drives the 3 DLP projectors in the room.

Some of the VGA inputs are in floor-mount boxes, several of them are in a rack at the front of the room with a Dell in it, etc.

The hum bars show up on all 3 projectors, and they show up differently. Sometimes they scroll up, sometimes they scroll down. Sometimes they scroll fast (2 or 3 seconds of period), usually medium (5 seconds), sometimes slow (12 seconds), and sometimes fixed. Of course, it is far worse with some sources than with others (especially laptop variation).

The per-projector scalers are all outputting 60Hz refresh signals to the projectors. One at 1280x1024, the other two at 1024x768.

AC power for all of these things are on isolated ground outlets that run to a dedicated panel for the A/V equipment in the room, though of course they don't all take the same runs to reach that point.

Most equipment in racks is not grounded with star washers.

Electricians have put power quality monitors on the 120VAC circuits and don't think there are any power quality issues.

Neutral/ground voltage is sometimes as high as 3V. Most circuits are on the same phase (not that it is clear that has an impact).

We obtained limited success troubleshooting these hum bars by running all three video projectors off of the same outlet strip, and plugging that outlet strip into an outlet adjacent the rack with switchers & scalers. Though this improved the hum bars, it did not completely eliminate them.

Ground-lifting the computer at the front of the room, along with it's DA and its local monitor, dramatically reduced the hum from that video source. But of course, that's not an acceptable thing to do beyond troubleshooting.

How do you fix a hum bar? In audio, you'd install transformers and forget about it. But what do you do about unbalanced video signals? Converting to HD-SDI or fiber is a really pricey proposition. Jensen sells a $600 5-BNC isolation transformer for video...installing boatloads of those might not be terrible, but is it one of the right answers?

How do you see a hum bar? We stared at (composite) video signal traces on some oscilloscopes for a while, and no one was able to see the hum bar on the scope, even though it was quite visible on-screen. Of course, no one was precisely sure what we were looking for in the scope trace, but nothing seemed to move at the same (or similar) rate that the visible hum bars moved at.

What are clever tricks to try? We began at the sinks (projectors) and started ground-lifting and isolating various devices, putting them all on the same outlet strip, bypassing components, etc. This was not very fruitful. In most cases, we could make small changes that would reduce the severity of hum bars, but not eliminate them entirely. And many of the changes were not practical from a permanent perspective ones, like "put everything all on the same circuit."

Trying to ground everything in a single-point fashion seems unmanageable in a large-scale installation like this, with lots of interconnected devices and many places where chassis grounds touch parts of the building, and 120VAC circuits for things at the front of the room don't run through the booth the way the video (and audio) signals must.
Thanks for any thoughts.

--jhawk

Posted by Steve Guttag (Member # 268) on 05-28-2009, 09:50 PM:

Believe it or not John, it is the same to troubleshoot video hum as it is for audio hum...they are caused by the same thing...ground loops.

With coax cable...you have so many ground points and as you extend that out into the rooms, you extend your ground contamination.

The isolated ground plugs have caused more problems than they ever helped...they are often stupid, in fact and should never be used in a equipment rack. You WANT the equipment rack, the equipment inside and the ground of the power that is powering the equipment to all be on the same ground/ground potential. The only value in isolated ground outlets are if they are in remote locations and you want to keep the outlet (neutral/ground and hot leg) all powered from the "blessed" system. Keeping things truely isolated is VERY tricky with video since the shield of the coax cable is "grounded" everywhere.

You've got Extron (any brand would be the same problem)...how are they mounted? If their face plate comes into contact with the JBOX they are mounted to, you are going to get a ground path and set up a loop there. Grounding the Interface with your video rack ground would go a ways to combating it. Even if you try to float it (which will likely cause other issues with performance), the thing that plugs into it (the source) can then contaminate the ground.

Had you all not gone the "isolated" route, odds are you would have far less hum though. Isolation is a an all or nothing proposition...if you don't completely isolate, you have problems. Better to bond EVERYTHING than to try to isolate most.

Steve

Posted by John Hawkinson (Member # 1135) on 05-28-2009, 09:59 PM:

Thanks!

I should stress, by the way, that this isn't my room/project and I'm mostly an interested onlooker, and not really a Video Guy. But that said:

Steve, I buy your argument for why IG outlets may not help. But why do they hurt?

Generally, speaking, most of the Extron extenders/DAs are mounted into the metal frame of the junction boxes and the various racks they are in. You say ground the interface with your video rack ground -- how? Run a seperate ground 100 feet from the box to the booth in the same conduit as the video signal wires? Why should that be any better than the shield of the BNC connectors?

I'm not sure how you bond everything throughout a large room-wide installation any better than, say, the chassis bonding that you already get through EMT, etc. (Perhaps that's less relevant if nothing is actually touching the non-IG ground of the various outlets?)

Privately, someone asked me about video humbuckers. I'm told they were tried and didn't help. Though I'm not really sure what they are. Chokes I guess? Though are they available for RGBHV (5-BNC) signals?

In re troubleshooting, maybe it's the same, but in video, you can't tell if you have a hum without a valid signal present, which makes it a fair bit more annoying than in audio, where you can unplug and dead short cables and boxes and just listen. And then add to that that 5xBNCs are a fair bit more effort than an XLR or a TRS to disconnect, and it quickly becomes quite time-consuming. Maybe they should focus on composite video to start with...

--jhawk

Posted by Steve Guttag (Member # 268) on 05-28-2009, 10:13 PM:

IG outlets are typically wired by the worst technician in the room...the electician. With ground being critical to noise free operation, at best you are increasing the liklihood of a bad ground and are potential partially floating things plugged into it. They serve absolutely ZERO purpose within a rack...what are you trying to isolate from? You can isolate the rack itself from building ground but not the rack from its outlets.

As to running a separate ground to the device (100-feet) ABSOLUTELY...do you really think the conduit, with all of its connections and all of its bonding to various parts of the building won't change its potential from the isolated-ground reference rack? Run an 16awg wire or larger and you will guarantee the box is at the same potential...there should be zero curret flow on the ground wire and thus no potential difference...larger wires would be better, of course. Don't ever use the signal's electrostatic shield as your ground reference...it is busy shielding...it is not that it doesn't normally work but it definately is not the best method.

Video hum buckers are available for 5-wire systems and no they are not cheap (good things rarely are). Extron GLI 2000

Steve

Posted by Demetris Thoupis (Member # 833) on 05-29-2009, 12:23 AM:

The first thing you should have tried is plugged the video source DIRECTLY on each projector to see if the problem still existed. Then work your way BACKWARDS!
demetris

Posted by John Hawkinson (Member # 1135) on 05-29-2009, 12:40 AM:

Yes, there was a lot of that. Of course, with sourecs directly connected, there's no problem. And as you add pieces back, problems stack up. And oftentimes some things appear to fix problems, but the problems come back on their own an hour later without any changes. etc.

No tips on seeing hum bars on scopes? How about a waveform monitor? Is that the right tool? [I don't have one, but could probably find one...]

--jhawk

Posted by Mark Gulbrandsen (Member # 72) on 05-29-2009, 09:10 AM:

Yes, You can use a waveform monitor to see hum on video signals or you can use a scope but best to have one with TV trigger. Set either to "frame" sweep rate and you'll see it for sure. If it's just slight you can turn up the gain and then it should be plainly visible. If you look for it at "line" sweep rate it's likely to be invisible since you're just looking at one scan line.

Mark

Posted by John Hawkinson (Member # 1135) on 05-29-2009, 09:32 AM:

So, I didn't use a waveform monitor, but I spent a while with a 100MHz scope. On a composite NTSC video signal, viewing the entire 60 Hz window, triggering in Video mode (Tek TDS2000 scope), the triggering was fine and stable. But we couldn't see any evidence of hum bars. But they were clearly visible on screen and we were T'd off the signal to the projector.

So we concluded we weren't looking for the right thing, or it was too subtle to show up on the scope? But they were PLENTY visible on-screen.

--jhawk

Posted by Mark Gulbrandsen (Member # 72) on 05-29-2009, 11:15 AM:

Did you try cranking up the scope gain and looking carefully at the 7.5 ire black level? Any hum present on the video should be visible traveling through there.... You might also try using line lock on the scope... you may have triggered to the hum itself.

Mark

Posted by John Hawkinson (Member # 1135) on 05-29-2009, 11:41 AM:

I assume you mean the 7.5IRE black in SMPTE color bars, not some setup area or sync pulse on any old signal? I definitely didn't zoom in and stare at that section.

Ironically, of course, on the projection screen, bars are most visible in WHITE areas, not black areas.

I don't understand your triggering comment -- in a composite video signal, if I was triggering on the hum, then I would see the entire rest of the signal wandering around with the period of the hum, right? If I see a stable waveform, I can't possibly be triggering on the hum, since the hum is a component of that waveform.

line lock? Like triggering on the AC Line? Or do you mean triggering on a particular NTSC Video Line number?

Thanks again. I wish there were some good references for debugging video. I'm sure there must be books and magazine articles, maybe I should spend some time at the library, the Internet is not-so-helpful for this stuff.

--jhawk

Posted by Paul Mayer (Member # 355) on 05-29-2009, 01:18 PM:

Line triggering on an O-Scope means you see all lines in a field superimposed together. The trigger is the H-Sync pulse. Field triggering lets you see all lines in a field side-by-side, using the V-Sync pulse as the trigger. Field triggering is the display that will most easily let you see 60 Hz hum.

There are scopes that let you trigger on individual lines - just select the line number you want. These are great for looking at stuff in the vertical blanking interval - things like VITS, VIR, closed captioning, time code, and the like.

But a video monitor can work great as a hum detector too - in your case it showed you the hum just fine whereas the scope did not.

Posted by Bruce Hansen (Member # 281) on 05-29-2009, 04:22 PM:

Putting everything on the same power phase is the WRONG way to go. That is why your neutral voltage is so high, and that is the major cause of your hum problem. Balance the phases as much as possible, so neutral current is as low as possible. Hum buckers will usually correct the problem, but you may need to experment with where to put them.

Posted by John Hawkinson (Member # 1135) on 05-29-2009, 04:43 PM:

Interesting point, Bruce. Apparently this problem has been going on for months, (I only heard about it recently), and originally the circuits were more evenly distributed, and somebody went and changed them around to try to help solve the problem (of course it did not help).

(There's definitely a lot of belief in "on the same phase" out there; I can't say I've seen a good solid debunking of it...)

One of the electricians did show me his phase balance measurements and I can't remember exactly what they were, but I think they were something like 30, 17, and 9A.

That said, is the neutral/ground voltage that significant a problem? I guess it depends on how much leakage there is from neutral to ground in the various devices?

--jhawk

Posted by Louis Bornwasser (Member # 3063) on 05-29-2009, 07:06 PM:

A small grounding wire (or common) can actually increase the hum, just like in audio. In broadcast, grounding is accomplished with a 6" copper strap, bonded to the station ground.

I use a good, heavy battery booster cable to temporarily get a good, low resistance ground. Then I replace it with a heavy (#2 or larger permanent cable. In one instance, I used #000 to get rid of the last hum. Louis

Posted by Gordon McLeod (Member # 33) on 05-29-2009, 09:51 PM:

I have never put any concern into balanceing loads and neutral current
Theatrical loads are dynamic and constantly changing and often the neutral current will be changing radically as poppers kick in and out amps have more load demand. Also many cases for noise issues certain loads can not be on the same phase as other loads
I aways spec that the neutral must be able to handle on a continous basis the entire phase load with as much as 50% harmonic content and all transformers must be rated for this as well
Electricians live by the balancing rule to reduce the amount of copper not the real world system

Posted by Paul Mayer (Member # 355) on 05-30-2009, 12:16 PM:

I think balancing phases only matters when you're dealing with really big loads. When we used to install temporary TV lighting in hotel showrooms or convention areas (like for the Jerry Lewis national Telethon) we'd get loads around 600A per phase with everything lit. We'd try to distribute the dimmer racks to keep the phases within 50A of each other.

The sound guys used to insist on a separate "clean" drop (i.e. no noisy reactive or switching loads) just for them, usually around 200A capacity.

Posted by Fred Tucker (Member # 4482) on 05-30-2009, 07:04 PM:

The only time I would ever attempt down sizing a neutral is when wiring a house & then downsize it according to the 220V loads. For instance a 200A service would get a 1/0 neutral instead of 2/0 since things 220V appliances like water heaters, furnaces, a/c & stoves use very little 120V power and have little use for a neutral.

I've upsized neutrals in a office building where the electrical contractor tried to save a buck & hum was present on the computer monitors. Only the service wires everything else was #12.

Posted by Bruce Hansen (Member # 281) on 05-30-2009, 08:12 PM:

Also, make sure that someone has not connected the neutral and the ground together somewhere (other than at the entrance panel.

I have had a number of calls form people who have done the wrong thing of putting everything on the same leg, and have bad hum problems. I tell them to try to get close to balancing the legs, and bingo, the hum goes away.

Posted by Austin Pelnar (Member # 5043) on 05-31-2009, 08:19 PM:

Given its an auditorium My guess would be dimmer/SCR noise getting into the video cabling or possibly a bad power supply somewhere I say this because... I spent half a day looking for a bad switching power supply. It was blowing thru the 1694A/RG-6 SD that we use and was out of band you couldnt see it on a waveform monitor or a video monitor but once it was converted to serial digital it was a mess. Looked long and hard, space heaters, printers, lights, dimmers, sniffed around the building with a spectrum analyzer there was powerful noise everywhere starting at ~6Mhz. Gave up at 5 PM stuck a low pass filter on it and called it a night. 2 hours later was called back in for a dead camera, bad power supply replaced it and for the heck of checked and the noise was gone.
Myself I wouldnt think ground loop lest somebody has done something horribly wrong. The shields of analog video cables are supposed to be isolated from the chassis. Patchbays are made from phenolic, feed thrus and equipment mounted connectors should have plastic bushings on them. NOWHERE read NOWHERE should the shield be tied to equipment/safety ground. Grounding to clear up the problem is dangerous at best I once found a video cable with 7 amps of ground loop on it... the hard way.
Austin

Posted by Steve Guttag (Member # 268) on 05-31-2009, 09:51 PM:

quote: Austin Pelnar
The shields of analog video cables are supposed to be isolated from the chassis.

Where did you read this fantasy? I can't think of the last piece of equipment I've seen where the BNC connector wasn't right on the metal chassis. Once that goes into an equipment rack that is on ground...whola...you've grounded, somewhat poorly, the sheild of the cable.

Anyway...you all believe what voodoo you want on what does and does not cause hum and hum bars.

Steve

Posted by Mark Gulbrandsen (Member # 72) on 06-01-2009, 08:55 AM:

Steve is right!

Posted by Austin Pelnar (Member # 5043) on 06-01-2009, 08:55 AM:

Yes it can be grounded on the source and terminating end altho isolated is preferred. Grounding is a very complicated subject with some basic rules and is way beyond a fourm discussion. Do you ground the shield on both ends of a ballanced audio line or just the terminating end. In what situation would you ground on both ends, what situation would you not ground it at all.
Check the connectors on equipment from broadcast equipment mfgrs LEI, Grass Valley, Evertz, Leitch... The connectors are isolated. Patch bays isolated, proper feed thrus isolated, DA's isolated, analog side of a converter isolated. Isolated for this very reason if you got noise and the shields are grounded everywhere lots of luck to you.

Posted by Steve Guttag (Member # 268) on 06-01-2009, 09:07 AM:

You'd have quite a bit of problem if they were not grounded as proper RF shielding requires grounding at BOTH ENDS.

As for audio grounding...a PROPER system actually does ground at both ends. The problem is, most are not proper and grounding at one end avoid one problem while causing another...the lesser of two evils.

Steve

Posted by Bruce Hansen (Member # 281) on 06-01-2009, 08:58 PM:

Sorry to disagree, but for balanced audio, the shield should only be grounded on one end. BTW, the shield does not go through a "normaled" audio patch, unless you "make" it go through. This can be a problem if someone has wired mike inputs to an audio board through a patch panel. Phantom power will not make it to the mikes, you will need to put patches into the patch pannel to have the shields connect.

I have seen a number of pieces of video gear where the BNC input can be treated as sort-of a balanced input, where the center pin goes to the + input of an op-amp, and the shield goes to the - side of the op-amp. There is usually a jumper on the board you can use to ground the shield, if you want to.

Posted by Steve Guttag (Member # 268) on 06-01-2009, 11:03 PM:

You can disagree all you want...the fact remains, the proper way is to bond the shield to the case at both ends...if it causes a problem, the cover-up is to lift one end. The solution is to correct the potential difference.

For some fun reading: http://www.rane.com/note151.html

As for patch bays...it all depends on how one sets them up and what patch bay they use. I use "Sleeve Normaled" type such that all three conductors are normaled or disconnected if a patch is inserted. It is a little more costly but it is a better/safer way to go.

Again, do what you want.

Steve

Posted by Gordon McLeod (Member # 33) on 06-01-2009, 11:24 PM:

I agree with steve on the issue of shields a properly designed and grounded system will have better hum and sn with the shield connected at both ends
ALTEC even had data sheets long before the quoted rane one on the issues of grounding and shields
one needs to get back to basics first

Posted by Monte L Fullmer (Member # 2797) on 06-02-2009, 02:27 AM:

What is the drain wire and what function does this wire do? I've seen connections where one side has 4 runs and the other side only has the three w/drain that isn't connected to anything...

thx-Monte

Posted by John Hawkinson (Member # 1135) on 06-02-2009, 07:46 AM:

OKAY GUYS: Can we please take the general discussion of audio hum problems and proper grounding practices to another thread? Both because those are useful to people who don't care about video and spend their whole life in the FHF, and also because I'd like this thread to be useful to people debugging hum problems in the future.

In case you're wondering, I've passed some of the very helpful comments here (thank you!) on to the folks responsible for the auditorium in question, and they're going to try to see what to do...

--jhawk

Posted by Steve Guttag (Member # 268) on 06-02-2009, 10:11 PM:

Actually troubleshooting audio and video hum is identical and normally caused by the same mistakes/problems.

Steve

Posted by John Hawkinson (Member # 1135) on 06-03-2009, 10:01 AM:

Yeah, I hear you; but few are going to find the good suggestions here because they'll be searching in the Film Handler's Forum rather than the Straight to Video forum.

On the other hand, no one is recommending disconnecting the
shields from BNC video connectors at one end, right? [Roll Eyes]

--jhawk

Posted by Robert Minichino (Member # 3495) on 06-03-2009, 10:42 AM:

In RF land, a shield that isn't tied to ground at both ends is an antenna, whether it's an audio or video signal (or some other type of data). So any mucking with the shield connections (usually used to work around audio equipment design flaws) is definitely not for all applications. I've assembled plenty of complex audio systems, and I've never needed to lift a shield unless I was working around a problem in the equipment.

The hum is usually caused by ground loops, where current flowing through the grounding system is causing multiple ground points to be at different potentials, and those differences are superimposed on the low level signals that are referenced to ground.

You can fix this by either removing the dependency on the ground reference, or by tying every ground point together with very low resistance connections so that they're all at the same potential (this gets hard at RF).

To remove the dependency on ground referencing, you'd use differential transmission (twisted-pair Ethernet is differential, and so is pro audio), or isolation transformers to provide galvanic isolation. This isolation isn't perfect, but the induced noise is usually at a much lower level than the signal to start with, so the extra noise rejection the isolation adds brings the noise down to an acceptable level.

Another cause of hum is inductive or capacitive coupling of noise into the signal lines. Because the impedance of the interconnect cable is not zero, a noise voltage can be imposed on the signal voltage. The shielding of a cable mostly helps with capacitive coupling (electrical charge), but inductive coupling (current-induced magnetic fields) remains an issue. Differential transmission helps greatly eliminate noise that's inductively coupled into a signal line by canceling out signal added to both lines of a pair in the receiver. You can use a balun to take single-ended signals and transmit them differentially over, for example, Cat. 5. If your VGA cables have a long run in close proximity to AC current they will pick up noise regardless of any shielding. A little bit of physical separation helps greatly.

Posted by Jonathan Bodge (Member # 4072) on 06-08-2009, 11:03 AM:

Whenever we have hum bar issues on video shoots (usually when a engineering truck shows up and there is a mix of AC between truck & location) I knock on the door and ask for a humbucker. A small heavy passive box with a BNC in/out. This has always fixed it in the past but we know gremlins will always try to change it up in the future!

Posted by Jack Ondracek (Member # 1466) on 06-08-2009, 11:59 PM:

I've had RF problems in cases where several audio sources, some fed from different power circuits, terminated in a common mixer. The different wiring lengths from the breaker panel to the source equipment caused all kinds of problems before we figured that one out.