White Level Clipping - DC Coupled 480i YPbPr Output - AVS Forum
Forum Jump: 
 
Thread Tools
post #1 of 9 Old 01-08-2012, 11:25 PM - Thread Starter
Member
 
LordSte's Avatar
 
Join Date: Jul 2004
Location: Chicago, IL
Posts: 36
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I was told to ask any video design questions in the Video Processors forum, so here it goes. I've read some good info in this thread, but I'd still like more information on this topic.

I'm designing a circuit with a 480i YPbPr output. It's running off a single supply so there is a known DC offset added to the outputs which sets the blanking level at +0.8 volts. I've been reading all I can find about DC/AC coupling video inputs/outputs, among which are:

http://www.maxim-ic.com/app-notes/index.mvp/id/3303
http://www.maxim-ic.com/app-notes/index.mvp/id/3768
http://www.fairchildsemi.com/product...eo_filters.pdf

What I gather from these documents is that most equipment AC couples their video inputs and then uses a clamp or DC restoration circuit to set its own optimal DC level. Thus, anything connected into these inputs can have any DC level, since it is removed by the input circuitry. This is confirmed in Video Demystified (4th ed.) in Chapter 5 under SDTV YPbPr Interface: "For receivers, video inputs should be AC-coupled and have a 75-Ω ±5% output impedance." "DC offsets up to ±1V may be present." I have attached pictures of the quotes in case someone wants to look at them in their entirety.

According to this, everything should be fine if I leave my outputs DC coupled. However, when I connect my YPbPr outputs to a TV, the peak white level of the image appears to be clipping on every TV that I try. If I AC couple the Y output, by adding a 220µF series capacitor, the white level is normal. Adding the capacitor to the color difference outputs doesn't seem to change the resulting image. My guess is that the DC (average) value of the Y signal before considering the additional +0.8 volts, is greater than +0.2 volts. This would push the total DC value past the ±1V limit as stated in Video Demystified. Can anyone confirm this? If I readjust my circuit to lower the blanking level to +0.4 volts, then the white level looks fine even without the coupling capacitor.

What causes the actual clipping though? Figure 6a of that 2nd Maxim link seems to imply that it can handle any DC offset. Don't modern input stages include AC coupling followed by a clamp that forces the bottom of the sync pulse to a reference voltage? That shouldn't be affected by different DC offsets...or am I missing something?

This circuit I'm working on needs to be so small, that adding the coupling capacitor would result in me scrapping the project completely. Avoiding the field tilt issue is also desirable. I can redesign my entire circuit to use the minimum DC bias allowed to keep my signals within the linear operating range of my amplifiers. Hence, the signals will have the same voltage levels as shown in Figure 5.6 of Video Demystified (4th ed.). I've attached a picture of that figure as well.

My main question is...would this be sufficient in a compatibility or "good practice" sense? Or should I really be putting that coupling capacitor in there no matter what? Or am I understanding this whole AC coupling DC coupling thing completely wrong? Please set me straight!

Thanks for any help. Oh, and sorry about all the random questions scattered throughout the post. Answers to any of them will be most appreciated.
LL
LL
LordSte is offline  
Sponsored Links
Advertisement
 
post #2 of 9 Old 01-09-2012, 11:10 AM
AVS Special Member
 
Glimmie's Avatar
 
Join Date: Sep 2000
Location: Los Angeles, CA
Posts: 7,816
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 53 Post(s)
Liked: 176
The reason for clipping on some DC coupled input stages us due to the DC on the source upsetting the bias of the first stage, thus forcing the transistor into a non linear part of it's curve.

As you have found, it takes a rather large capacitor to properly couple a video signal into a 75ohm load. Too small and you get LF rolloff which causes tilt. If you want to use a smaller capacitor, it must be after the termination resistor.

There is no free lunch. Professional analog video circuits run form biPolar power supplies for these reasons.

Glimmie's HT Page
Being redone - comming soon!

Glimmie is offline  
post #3 of 9 Old 01-16-2012, 11:09 PM - Thread Starter
Member
 
LordSte's Avatar
 
Join Date: Jul 2004
Location: Chicago, IL
Posts: 36
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Thanks for the reply. Are you saying that the input stages, of all the TVs I tried, look more like Figure 6c of this Maxim app-note (except with an NPN emitter follower instead of the PNP)? I'm assuming that's why Mr. Jack suggests that the DC offsets going into these don't exceed ±1V? +1V for when there's an NPN, and -1V when there's a PNP?

I calculated the DC value of the Y signal for a pure white horizontal line (63.56µs) with blanking level at 0v, sync tip at -0.3v, and peak white level at 0.7v. It turns out to be 0.56v. If you add the extra offset of 0.8v that my circuit has, then that makes it 1.36v which is well above the 1.0v limit.

If I readjust my circuit to achieve the levels of Figure 5.6 in Jack's book, the max DC of the Y should be 0.86v. This is below the +1.0v limit, so I should be fine, right?
LordSte is offline  
post #4 of 9 Old 01-17-2012, 03:32 PM
Senior Member
 
Speed Daemon's Avatar
 
Join Date: Jan 2010
Location: Location, Location
Posts: 419
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 22
Quote:
Originally Posted by LordSte View Post

What causes the actual clipping though?

Ideally it would be an engineer watching a waveform monitor.

The only real reason that video clipping is necessary is to prepare the composite video signal for RF modulation. A modulated NTSC signal is inverted so the 100 IRE level is 0 power/modulation, and the tip of sync pulses are no more than 100% power/modulation. Because it's impossible to have less than zero modulation with AM and its variants (including VSB, naturally), it's vital that the video fed to the exciter never exceeds 100 IRE.

If you're using baseband, and don't have to worry about RF modulation, you can get away with exceeding the 100 IRE white limit (whiter than white) and/or 7.5 IRE black limit (blacker than black). Most video monitors will accept the "illegal" levels within reason. Still, the best practice is to keep the video portion of the signal between 7.5 and 100 IRE. Instead of hard clipping, more gradual gain control that operates much like audio limiters can be used to maximize video fidelity within the prescribed limits.

Quote:


My main question is...would this be sufficient in a compatibility or "good practice" sense? Or should I really be putting that coupling capacitor in there no matter what? Or am I understanding this whole AC coupling DC coupling thing completely wrong? Please set me straight!

NTSC video is supposed to contain relatively large amounts of DC, and it's not confined to the blanking level. AC coupling is always a kludge of some sort. IMHO you should always design to the standard, and leave it to the designers of downstream components to meet that same standard.

In the studio it used to be common to send luma, chroma, H and V sync separately. This allowed relatively crude circuits to reject ground loop distortion (a.k.a. "hum bars") while maintaining the DC levels where they count. More recently it was common to include sync with the green channel of RGB component, or a similar arrangement.

These days it's feasible to regenerate a usable baseband signal even if the DC reference levels are rendered useless by color under recording or "copy protection" distortion. Needless to say, with YPbPr color space the -Y information carried by the chroma offset channels can be used to good effect to regenerate the various reference DC levels post AC coupling.

This isn't really my area of expertise, but IIRC a purpose-built gyrator circuit might be a better solution than the relatively crude isolation capacitor. Just food for thought.

You get what you pay for.  For professional advice, pay the professional rate.
Speed Daemon is offline  
post #5 of 9 Old 01-17-2012, 05:37 PM
AVS Special Member
 
Glimmie's Avatar
 
Join Date: Sep 2000
Location: Los Angeles, CA
Posts: 7,816
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 53 Post(s)
Liked: 176
Quote:
Originally Posted by Speed Daemon View Post

Ideally it would be an engineer watching a waveform monitor.

The only real reason that video clipping is necessary is to prepare the composite video signal for RF modulation. A modulated NTSC signal is inverted so the 100 IRE level is 0 power/modulation, and the tip of sync pulses are no more than 100% power/modulation. Because it's impossible to have less than zero modulation with AM and its variants (including VSB, naturally), it's vital that the video fed to the exciter never exceeds 100 IRE.
If you're using baseband, and don't have to worry about RF modulation, you can get away with exceeding the 100 IRE white limit (whiter than white) and/or 7.5 IRE black limit (blacker than black). Most video monitors will accept the "illegal" levels within reason. Still, the best practice is to keep the video portion of the signal between 7.5 and 100 IRE. Instead of hard clipping, more gradual gain control that operates much like audio limiters can be used to maximize video fidelity within the prescribed limits.

We are not talking about clipping that is already present in the signal. We are talking about clipping that occurs in the video amplifier circuits that are not designed properly or other circumstances.

Quote:
NTSC video is supposed to contain relatively large amounts of DC, and it's not confined to the blanking level. AC coupling is always a kludge of some sort. IMHO you should always design to the standard, and leave it to the designers of downstream components to meet that same standard.

Analog video is supposed to have a stable reference point. That is not the same as having several volts of DC offset with the AC video riding on it. That can be allowed within a device but not on video that leaves a via standard interface. Video distribution is expected to have the blanking level at zero volts and most importantly stay at zero volts regardless of the changing video voltage levels. Capacitor coupling was/is quite common as most equipment re-establishes the DC reference through clamp circuits..

Quote:
In the studio it used to be common to send luma, chroma, H and V sync separately. This allowed relatively crude circuits to reject ground loop distortion (a.k.a. "hum bars") while maintaining the DC levels where they count. More recently it was common to include sync with the green channel of RGB component, or a similar arrangement.

I don't know where you got this information? NTSC analog video in a broadcast or production plant has been a full composite signal since the early 1970s. Only some early 1950/60s equipment used noncomposite video and added sync later. Separate chroma distribution was never used in commercial broadcast plants. Svideo was a consumer format. Component video, that is RGB or Y,R-Y,B-Y was used in island areas where high quality was important. Like graphics stations. After an initial period of composite digital in the early 1990s, broadcast video is now mostly component digital per SMPTE 259 for SD and SMPTE 292 for HD. Clamping DAs and processing amplifiers work quite well with full analog composite signals. In fact the sync is often used as the clamp voltage reference source.

Now in computer applications, RGBHV was indeed the standard with Sync On Green a later developement. But not in the analog broadcast and mastering facilities where composite video was the standard. Note too there were initially countless proprietary computer video standards in the 1970s thru 1990s none of which were compatible with NTSC or PAL broadcast video standards.

Quote:
These days it's feasible to regenerate a usable baseband signal even if the DC reference levels are rendered useless by color under recording or "copy protection" distortion. Needless to say, with YPbPr color space the -Y information carried by the chroma offset channels can be used to good effect to regenerate the various reference DC levels post AC coupling.

This isn't really my area of expertise, but IIRC a purpose-built gyrator circuit might be a better solution than the relatively crude isolation capacitor. Just food for thought.

I don't follow you at all here. A gyrator circuit is a capacitance multiplier circuit used to substitute for a choke. This has no application for AC coupling of video signals. AC coupling of analog video is quite commen.

Glimmie's HT Page
Being redone - comming soon!

Glimmie is offline  
post #6 of 9 Old 01-17-2012, 05:45 PM
AVS Special Member
 
Glimmie's Avatar
 
Join Date: Sep 2000
Location: Los Angeles, CA
Posts: 7,816
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 53 Post(s)
Liked: 176
Quote:
Originally Posted by LordSte View Post

Thanks for the reply. Are you saying that the input stages, of all the TVs I tried, look more like Figure 6c of this Maxim app-note (except with an NPN emitter follower instead of the PNP)? I'm assuming that's why Mr. Jack suggests that the DC offsets going into these don't exceed ±1V? +1V for when there's an NPN, and -1V when there's a PNP?

I calculated the DC value of the Y signal for a pure white horizontal line (63.56µs) with blanking level at 0v, sync tip at -0.3v, and peak white level at 0.7v. It turns out to be 0.56v. If you add the extra offset of 0.8v that my circuit has, then that makes it 1.36v which is well above the 1.0v limit.

If I readjust my circuit to achieve the levels of Figure 5.6 in Jack's book, the max DC of the Y should be 0.86v. This is below the +1.0v limit, so I should be fine, right?

In theory yes. Personally I don't like to see any DC on signals, video or audio. But we have come quite far in semiconductor design where a small amount of DC on a signal line is an acceptable trade off in a cost sensitive application using a single power supply rail.

Glimmie's HT Page
Being redone - comming soon!

Glimmie is offline  
post #7 of 9 Old 01-22-2012, 10:17 PM - Thread Starter
Member
 
LordSte's Avatar
 
Join Date: Jul 2004
Location: Chicago, IL
Posts: 36
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Cool. I think with enough tweaking I can make it so there's only an additional 0.3v on the Y and 0.35v on each of the color difference signals. This is the best I can get while keeping my amps in their linear output range.

Quote:
Originally Posted by Glimmie View Post

In theory yes. Personally I don't like to see any DC on signals, video or audio. But we have come quite far in semiconductor design where a small amount of DC on a signal line is an acceptable trade off in a cost sensitive application using a single power supply rail.

Are you saying that even in a dual-supply scenario you would still AC couple the output to completely remove all DC from the signal?
LordSte is offline  
post #8 of 9 Old 01-23-2012, 11:54 AM
AVS Special Member
 
Glimmie's Avatar
 
Join Date: Sep 2000
Location: Los Angeles, CA
Posts: 7,816
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 53 Post(s)
Liked: 176
Quote:
Originally Posted by LordSte View Post

Are you saying that even in a dual-supply scenario you would still AC couple the output to completely remove all DC from the signal?

No, with a bipolar supply you typically DC couple the output and the whole circuit for that matter.

DC audio power amplifiers have still AC INPUT coupling for two reasons.

1) There is no DC restoration issue with audio.

2) A small amount of DC offset into a high power amplifier would be awesomely destructive to the speakers.

Glimmie's HT Page
Being redone - comming soon!

Glimmie is offline  
post #9 of 9 Old 01-24-2012, 11:26 PM - Thread Starter
Member
 
LordSte's Avatar
 
Join Date: Jul 2004
Location: Chicago, IL
Posts: 36
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Glimmie View Post

No, with a bipolar supply you typically DC couple the output and the whole circuit for that matter.

DC audio power amplifiers have still AC INPUT coupling for two reasons.

1) There is no DC restoration issue with audio.

2) A small amount of DC offset into a high power amplifier would be awesomely destructive to the speakers.

Right. I was just confused because you said "I don't like to see any DC". But for video, I'm assuming you meant "any extra DC". Thanks for the help.
LordSte is offline  
Reply Video Processors

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off