JVC 30K Bandwidth Fix - Page 4 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #91 of 305 Old 12-20-2004, 07:03 AM
LJG
AVS Special Member
 
LJG's Avatar
 
Join Date: Jun 2000
Location: Brookville, NY
Posts: 4,017
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 18 Post(s)
Liked: 16
Quote:


Originally posted by mp20748
I. And with 6' of good quality RCA cables, I had a perfect HF test pattern at the other end of the cables. So I could not see a reason for a line driver. And I would never run component beyond 6', with 3' being the best way to go.


Hi MiKe:

What type of cabling would you run for over 3-6 ' ?

Lon
LJG is offline  
Sponsored Links
Advertisement
 
post #92 of 305 Old 12-20-2004, 01:22 PM
AVS Addicted Member
 
Gary Murrell's Avatar
 
Join Date: Jul 2003
Location: Kentucky
Posts: 10,927
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Mike would you ever be interested in looking at the new 5u from JVC??,

IT has a brand new rock solid transport that is much improved over the original model, should be very reliable, seems to be so far since release

only problem is that some folks have said the component out is well below the quality of the 30k and 40k

heck I would be willing to send my unit to you for a while to look at if you would be interested

I know you are a busy guy, but I just thought I would ask

-Gary
Gary Murrell is offline  
post #93 of 305 Old 12-20-2004, 03:22 PM
Member
 
moggy's Avatar
 
Join Date: Oct 2002
Location: Melbourne, Australia
Posts: 125
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally posted by mp20748
I don't think 600mhz is really needed (for 1080P). I think we'll need far less. But when using a 300mhz chip, we're no where near the best bandwidth performance for the signal. And because it's so hard to get these chips to perform at specs, you'll need a higher bandwidth chip to get the best headroom for the usable signal. And that's what puzzles me about these designs that use, say a 30mhz chip for a 30mhz signal.
For this circuit (30mhz), I would use a 100mhz chip. A 100mhz chip is very easy to tame (parasitics), and it'll provide excellent headroom for the usable signal.

Only 30MHz? How do you arrive at that? I calculated 66MHz. also, I thought your ATSC was about 50MHz or thereabouts anyway?
Our 1080i DVB is only 50Hz, so it's a bit less in MHz, but ATSC is about 55MHz , I thought. Correct me if I'm wrong.
Yes, I think a 100MHz chip(g=1) would be adequate but getting a 280MHz chip(g=1) is not hard.
eg. my favourite OPA3691 chip is 500MHz but at g=1 it is 280MHz with a flatness and -3dB point that are good enough for this puppy. It is dual supply but can be made to work with a single +5V supply. In fact the circuit has unused positions for resistors and caps that were most likely to be used for a dual supply chip version. Fitting a daughterboard with a 16pin SOP chip would be a pain though.

Quote:
Actually, I'm not using a chip at all there. I removed the BA7623, and I'm by-passing that circuit with a direct path to the RCA's. I thought of using a replacement for the BA7623, but after seeing that once it was removed from the circuit, the roll-off problem decreased. And with 6' of good quality RCA cables, I had a perfect HF test pattern at the other end of the cables. So I could not see a reason for a line driver. And I would never run component beyond 6', with 3' being the best way to go.

When I started out with this, I did not have a diagram. So I used the scope along the signal path. That chip was the first thing I changed, then I arrived at the low pass filters. [/b]

With good quality 75ohm cable there is no reason that you can't go past 2m in length. My projector is 12m in length away, so a short length is not an option for me if I were to drive the projector directly. If the initial driving circuit has a 75ohm output then all will be ok but I doubt that an emitter driver with 150ohm is an accurate 75ohm drive.
Yes, for your short lengths, I guess you could live without it but I would check your pulse reflections on screen or by using a good cro.
HF rolloff will occur with long cable lengths, yes, but I intend to use 5 runs of RG11 to my 1292 (RGBHV) .
This leads me to the next sentence by the previous circuitous route.

The other option of course is that if you bypass the opamp as you suggest and have a transcoder as close as possible to the RCA output pins then that would suffice I believe. I, in fact, have to use my transcoder that I built that is dedicated for this machine so this might well be the easiest option. I can put the transcoder with it's inbult buffereing and good 75ohm drive only about 50mm away from the back of the 30k. Voila. I'm starting to like this idea. I just need to load the 30k with the right resistor for 1Vpp out of the 30k(or change my transcoder to have a g=1). You say it looks great with just 1m so with only 50mm it should also look great.
For others that only use component(YPrPb) then simply putting an external buffer might be the most practical method if you need cable lengths longer than about 1m. It needs to be able to handle 2Vpp and have g=1. Not too difficult.

I look forward to seeing some photos.
To help with the previous conversation I've now also put a schematic bitmap of the back panel pcb containg this opamp.
www.mogford.com/audio/ntsc-panel.png

It's all sounding rather good. I'm glad someone told me about this thread, I'd never thought that the JVC was lacking. I can't wait to see the final results.

Regards,
moggy
moggy is offline  
post #94 of 305 Old 12-20-2004, 07:43 PM
Advanced Member
 
Briands's Avatar
 
Join Date: Apr 2002
Location: Louisville, KY
Posts: 935
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I like to pretend I understand a lot of this stuff, but sometimes... I think i know how Raster must feel.

There are SOME brilliant people here... and some ... well not so brilliant...

Then there are those of us in the middle who benifit from the scraps of wisdom that drip off the table...

Brian
Briands is offline  
post #95 of 305 Old 12-20-2004, 07:48 PM
AVS Addicted Member
 
Steve Bruzonsky's Avatar
 
Join Date: Jan 1999
Location: Gilbert, Arizona
Posts: 17,656
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 34 Post(s)
Liked: 23
Mike, did I miss something? Is it within reasonable possibility to replace the RCAs with BNCs for component out???????? Would this give a further picture improvement?

"Doug Winsor" used to troll at some AV Forums as "Steve Bruzonsky"! My home theater at:
http://www.avsforum.com/avs-vb/showthread.php?t=1158431
Steve Bruzonsky is online now  
post #96 of 305 Old 12-20-2004, 09:30 PM
Member
 
moggy's Avatar
 
Join Date: Oct 2002
Location: Melbourne, Australia
Posts: 125
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally posted by Steve Bruzonsky
Mike, did I miss something? Is it within reasonable possibility to replace the RCAs with BNCs for component out???????? Would this give a further picture improvement?

If you don't mind hacking the back panel then (75ohm) BNCs are always better than RCAs. Whether you'll pick the difference is questionable.
I calculated that the disturbance caused by (say) 5mm of an RCA connector of not quite 75ohm is equal to just a fraction of a pixel in time and won't be seen in any normal HD video response of (say) 100MHz.
This is why I believe that cable is the most important aspect. It needs to be true 75ohm cable with the usual +/-2% accuracy. Cables amplify errors over much longer times (ie depends on length of cable).
Many BNC mods claim to be the bees knees but there is very little to suggest that it is the cure-all.
Mind you, I still consider that using 75ohm BNC connectors is a good idea.
I have a personal hate of putting RCA connectors on RG59, RG6 or RG11.
If using BNC, make sure they are 75ohm and not generic (50ohm).

For the JVC, it depends how hard it is to change the connectors.
Since I have to hack into it anyway it might be worthwhile. Having imported mine into Australia, I have no warranty to worry about. If it's too difficult, I'm not going to bother, the other mods are more important.

On a similar topic, I also plan to use BNC connectors for my pc graphics card, they just seem a more robust method of connection. It creates a 75ohm impedance over a much longer path and eliminates any VGA connector pertubations. However, the main reason is to eliminate the LP filter just as in this 30k example. The second reason is that BNCs can use real cable instead of that RG179/miniature coax rubbish that has to be connected to a VGA connector.
You can tell I love RG59, RG6 & RG11, can't you?
PCs can clock at a much higher rate than normal 1080i HiDef and this stuff for a pc is important. My htpc is used at 1920x1080x75Hz. That's a 190+MHz pixel clock. egads! omigosh! I may have to down grade to 50Hz if the pc mod doesn't help with full pixel resolution.
Anyway, I have digressed.

Regards,
moggy
moggy is offline  
post #97 of 305 Old 12-20-2004, 09:49 PM
AVS Special Member
 
jcmccorm's Avatar
 
Join Date: Mar 2001
Location: Madison, AL, USA
Posts: 3,379
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 13
Moggy, the anti-aliasing filter for HD is set to 30MHz. The luminance sampling frequency is 74.25MHz so the filter is set to half (roughly).

Cary
jcmccorm is offline  
post #98 of 305 Old 12-20-2004, 10:48 PM
Member
 
moggy's Avatar
 
Join Date: Oct 2002
Location: Melbourne, Australia
Posts: 125
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally posted by jcmccorm
Moggy, the anti-aliasing filter for HD is set to 30MHz. The luminance sampling frequency is 74.25MHz so the filter is set to half (roughly).

Cary

Ok. Hmm.. I don't quite understand. What's the point of 1920 pixels if you only see half of it? I'll have to read up a bit more on your ATSC system.
I assume you're talking about ATSC offair HD?
This may be only a factor for analogue displays, not digital, if so then the analogue display fraternity is being dudded. I must read more to see what really happens.

If the 30k is truly 30MHz BW then 100Mhz is ample for our instance and I needn't worry about my projector showing 1920 pixels at full depth of resolution or changing the RCA connectors.
Unfortunately, the 30k manual says very little about such things.

See ya.

Regards,
moggy
moggy is offline  
post #99 of 305 Old 12-21-2004, 05:17 AM - Thread Starter
AVS Special Member
 
mp20748's Avatar
 
Join Date: Nov 2000
Location: Maryland
Posts: 6,339
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 13
Quote:
Originally posted by moggy

With good quality 75ohm cable there is no reason that you can't go past 2m in length. My projector is 12m in length away, so a short length is not an option for me if I were to drive the projector directly. If the initial driving circuit has a 75ohm output then all will be ok but I doubt that an emitter driver with 150ohm is an accurate 75ohm drive.
Yes, for your short lengths, I guess you could live without it but I would check your pulse reflections on screen or by using a good cro.
HF rolloff will occur with long cable lengths, yes, but I intend to use 5 runs of RG11 to my 1292 (RGBHV).

The other option of course is that if you bypass the opamp as you suggest and have a transcoder as close as possible to the RCA output pins then that would suffice I believe. I, in fact, have to use my transcoder that I built that is dedicated for this machine so this might well be the easiest option. I can put the transcoder with it's inbult buffereing and good 75ohm drive only about 50mm away from the back of the 30k. Voila. I'm starting to like this idea. I just need to load the 30k with the right resistor for 1Vpp out of the 30k(or change my transcoder to have a g=1). You say it looks great with just 1m so with only 50mm it should also look great.
For others that only use component(YPrPb) then simply putting an external buffer might be the most practical method if you need cable lengths longer than about 1m. It needs to be able to handle 2Vpp and have g=1. Not too difficult.

Because of this, I've designed a mini line driver circuit to retro replace the BA7623. It's a 1-1/2 x 2" mini board with 400 mhz bandwidth (-3db). I'll have some completed boards ready by the end of the week.

Very high quality cables are very important for best analog video performance. And the 30K deserves the absolute best. Once modded, it'll produce a superb video signal. Definitely the cleanest HD I've ever experienced from a consumer product. And that would probably explain why that emitter driver does so well. Actually, it should not do well at all driving a 75 ohm capacitive load. So that's why I've looked into adding the line driver circuit. Longer cable runs was not my main reason, I wanted a circuit that would produce and maintain true 75 ohms on any length of cable (>30"). I'll have five of the completed boards by next week.


Quote:
What type of cabling would you run for over 3-6 ' ?
Lon

I'm not sure what the experts would recommend here, but from my experience. Once you exceed 6' of cable run, it should be true 75 ohm cables. And I know of very few true 75 ohm cables on the market. Belden and Granite Audio makes the only TRUE mini 75 ohm cables. Other than those two brands, I would not go below RG59 or RG6 for any other brand named cables. My preference is Belden 1505 and 1694. And If not using the best mini, I would use RG59 or RG6 even for short cable runs. Skip right over the fancy named stuff. I could tell you some real stories with some of the designer named stuff.

Quote:
Mike would you ever be interested in looking at the new 5u from JVC??
-Gary

Sure. Someone sent me a HDSAT box to look at. So I'll take one more.

Quote:
There are SOME brilliant people here... and some ... well not so brilliant...
Briands

Actually, I'm one of those "in the middle", I've just had some good instructors on this particular subject. I've always loved improving the quality of the signal, but it was only until I became a member of this forum, that I was able to take my troubleshooting skills to another level on this. I'm not an engineer, so I've learned from some of the best out there.
Keep your eye's open for postings from these guys:

Tom Rosback

VideoGrabber (TIM)

Jvanderwalker (Jim)

jcmccorm (Cary)

When you see their post, they come across very humble, and almost as if they'll learning from others. But for real, these guys are humble Heavy Weights. My favorites are Tom Rosback and Cary (Tom is a master at understanding and lowering noise)...and it seems we'll soon be adding Moggy to the list.

Quote:
Mike, is it within reasonable possibility to replace the RCAs with BNCs for component out???????? Would this give a further picture improvement?
Steve

Not really, for short distances I can't see the need for BNC's over RCA's. Unless the goal is better grounding, and you'll always want the best grounding for un-balanced video. However, I'm thinking to add BNC's to my 30K, because I have BNC's on the MP-5. My modified 30K is the only consumer device with RCA's that could possibly benefit from BNC's.

Moggy, what do you think about a very clean 400mhz line driver?

It's all about the performance... Got Marquee!

 

High Performance Marquee Video chain modifications.  Now available!

mp20748 is offline  
post #100 of 305 Old 12-21-2004, 05:41 AM
AVS Special Member
 
Wireless's Avatar
 
Join Date: Feb 2001
Location: East Tennessee, USA
Posts: 2,130
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Briands, I'm with you.

Duh, I'm waiting on the purdy pictures before opening my 40k.
Wireless is offline  
post #101 of 305 Old 12-21-2004, 06:02 AM
Member
 
moggy's Avatar
 
Join Date: Oct 2002
Location: Melbourne, Australia
Posts: 125
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally posted by moggy
[b]Ok. Hmm.. I don't quite understand. What's the point of 1920 pixels if you only see half of it? I'll have to read up a bit more on your ATSC system.
I assume you're talking about ATSC offair HD?
This may be only a factor for analogue displays, not digital, if so then the analogue display fraternity is being dudded. I must read more to see what really happens.

Thanks for setting me straight. Yes, 30MHz BW. But wait, there's more:
Cr & Cb are probably only 15MHz. Y is half filtered but Cr & Cb should be quarter filtered giving only 15MHz BW. It's possibly only the Y that could benefit from a new opamp. The 3 circuits are almost identical so I'm not entirely sure if it's 30/30/30 or 30/15/15. It's a moot point. It's easier to treat all 3 the same.
I'd assumed a higher clock rate(not sure why I did) which when filtered would be closer to 74MHz(my approximation of 66MHz was out) pixel rate. The use of the pixel rate as the basic luma clock rate means that yes, it needs to be filtered. Still, 30MHz is a big drop in resolution.
A digital signal to a digital display is the only option to prevent this loss, I think.

Quote:
If the 30k is truly 30MHz BW then 100Mhz is ample for our instance and I needn't worry about my projector showing 1920 pixels at full depth of resolution or changing the RCA connectors.

I'll still do all this but the response increase on paper doesn't seem as dramatic as I originally thought. Having response out to only 30MHz is a bit of a downer. My hope is that the FL filters were their way of anti-aliasing or filtering and it's removal may actually be increasing this 30MHz to something much higher. Hoping......
Unfortunately I only have a 20MHz CRO so I can't measure any of this.

Mike, when you first saw this HF rolloff, do you know what the frequency was when it rolled off dramatically. I'm just interested.


Byee for now.

Regards,
moggy
moggy is offline  
post #102 of 305 Old 12-21-2004, 06:24 AM
AVS Addicted Member
 
Steve Bruzonsky's Avatar
 
Join Date: Jan 1999
Location: Gilbert, Arizona
Posts: 17,656
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 34 Post(s)
Liked: 23
Quote:
Originally posted by mp20748
Belden and Granite Audio makes the only TRUE mini 75 ohm cables.

Mike's only saying this cause Granite Audio has a special promotional deal (its not what you think) up through CES 2005. Click to find out. HA!

"Doug Winsor" used to troll at some AV Forums as "Steve Bruzonsky"! My home theater at:
http://www.avsforum.com/avs-vb/showthread.php?t=1158431
Steve Bruzonsky is online now  
post #103 of 305 Old 12-21-2004, 06:35 AM
Advanced Member
 
sandbagger's Avatar
 
Join Date: Dec 2001
Location: Detroit Rock City
Posts: 939
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 4 Post(s)
Liked: 21
Quote:
Originally posted by Steve Bruzonsky
Mike's only saying this cause Granite Audio has a special promotional deal (its not what you think) up through CES 2005. Click to find out. HA!


I could think of a few reasons her eyes are crossed

Kevin
----------
Motor City Custom Audio
Bringing you Custom subwoofer kits, with flames if you want
Onix and Melody Valve Hi Fi Audio Dealer serving the Midwest and Canada
sandbagger is offline  
post #104 of 305 Old 12-21-2004, 06:48 AM
Member
 
moggy's Avatar
 
Join Date: Oct 2002
Location: Melbourne, Australia
Posts: 125
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally posted by mp20748
Because of this, I've designed a mini line driver circuit to retro replace the BA7623. It's a 1-1/2 x 2" mini board with 400 mhz bandwidth (-3db). I'll have some completed boards ready by the end of the week.
Moggy, what do you think about a very clean 400mhz line driver?

I assume you are going to use that Intersil opamp. It seems very capable.
Your method is the best and the most flexible. From discussion elswhere 400MHz is a bit high but hey, why not.
I'll either fit a OPA3691 which I have a few of or simply use my external transcoder as the buffer(it uses OPA3691s). After the emitter driver I'll have to decide on values of the series cap and 75R if I use my transcoder. Hmm..

BTW, what do you make of the switched notch filter on the Y signal just after the FL? By previous reasoning it puts a notch in (or out) when in 1080 mode. Any idea what it's doing? A quick calc says it's a notch at 10MHz.


Quote:
and it seems we'll soon be adding Moggy to the list. ]

I'm educated as an electronics engineer but I've never made it my career. I used to work in broadcast television in the analogue days as a technician and have only learnt digital tv from reading.
I first saw analogue HD in 1985, that was something! It was the marriage of Di and Charles on a small widescreen hdtv.
The advent of surround sound and digital HDTV in Australia in 2001 has re-kindled my interest. Setting up the best HT setup is a challenge. I don't have the same advantages that you guys have in the US. HD here is offair only and that's slim pickings. My D-Theater is imported at great expense. I currently work as a pcb designer for R&D at Varian Inc. here in Melbourne. That's enough about me.

Quote:
My modified 30K is the only consumer device with RCA's that could possibly benefit from BNC's.]

ditto.

Regards,
moggy
moggy is offline  
post #105 of 305 Old 12-21-2004, 07:34 AM - Thread Starter
AVS Special Member
 
mp20748's Avatar
 
Join Date: Nov 2000
Location: Maryland
Posts: 6,339
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 13
Quote:
Originally posted by moggy
[b]Your method is the best and the most flexible. From discussion elswhere 400MHz is a bit high but hey, why not.

400mhz is a bit high, but it offers two advantages. One offering the best possible capacitive load driver. With the other dealing with the drop in bandwidth performance when operating with a single 5 volt supply.


Quote:
BTW, what do you make of the switched notch filter on the Y signal just after the FL? By previous reasoning it puts a notch in (or out) when in 1080 mode. Any idea what it's doing? A quick calc says it's a notch at 10MHz.

Yep, it appears to be a low pass filter for Standard (480I/8mhz) video.

Quote:
I first saw analogue HD in 1985, that was something! It was the marriage of Di and Charles on a small widescreen hdtv.

That's around the same time that I saw it at INFOCOMM at the DC Convention Center. The HDTV gear took up about as much space as a regular home refrigerator. It's amazing to see how much they've been able to miniaturize that technology.

It's all about the performance... Got Marquee!

 

High Performance Marquee Video chain modifications.  Now available!

mp20748 is offline  
post #106 of 305 Old 12-21-2004, 03:21 PM
Member
 
moggy's Avatar
 
Join Date: Oct 2002
Location: Melbourne, Australia
Posts: 125
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:


Originally posted by mp20748
400mhz is a bit high, but it offers two advantages. One offering the best possible capacitive load driver. With the other dealing with the drop in bandwidth performance when operating with a single 5 volt supply.


Fair enough, your chip specs would tell you all this I suppose.


Quote:


Yep, it appears to be a low pass filter for Standard (480I/8mhz) video.

It's a notch, not a LP filter. It's influence is only around 10MHz . Mabe there is a 10MHz clock when in 480p mode and the signal doesn't care at higher freqs??
The notch is switched by the same signal that goes to the FLs. Maybe it's not a 720/1080 switch but in fact a 480/HD switch. That makes sense now especially if the clock rate for 720p is the same as that for 1080i. I don't know for sure here.
What is the normal BW for 720p? is it the same 30MHz as 1080i?
Anyway, knowing that it is a 480/HD switch, I don't think we need be concerned with it, which is good.


Quote:


That's around the same time that I saw it at INFOCOMM at the DC Convention Center. The HDTV gear took up about as much space as a regular home refrigerator. It's amazing to see how much they've been able to miniaturize that technology.

I saw it in West Berlin at the Funkschau. It was all Philips cameras, vtr and monitors. I also saw 3D television there. It whet my appetite and have been anxiously waiting for HD ever since that date. I'm also into 3D tv as well. Young minds can be easily impressed.

Regards,
moggy
moggy is offline  
post #107 of 305 Old 12-21-2004, 03:37 PM
AVS Special Member
 
VideoGrabber's Avatar
 
Join Date: Oct 1999
Location: SW Michigan
Posts: 2,799
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
moggy asked:
> What is the normal BW for 720p? is it the same 30MHz as 1080i? <<br />
Yes. Higher line rate, with lower data density per scan line.

- Tim

- Tim
VideoGrabber is offline  
post #108 of 305 Old 12-21-2004, 03:55 PM - Thread Starter
AVS Special Member
 
mp20748's Avatar
 
Join Date: Nov 2000
Location: Maryland
Posts: 6,339
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 13
Quote:


Originally posted by moggy
It's a notch, not a LP filter. It's influence is only around 10MHz . Mabe there is a 10MHz clock when in 480p mode and the signal doesn't care at higher freqs??

Ok, that makes sense. I had to revisit the schematic to understand this better, and yet I still can't understand blocking frequencies around 10mhz

I was thinking it was a low pass for 480I, since 480I is an option on the 30K, and there may not be anything in the FL that would allow for 480I filtering.

Anyway, I'm glad you caught it, and it's nothing to be concerned about.


Now, can someone elaborate more on why a notch filter is needed there?

It's all about the performance... Got Marquee!

 

High Performance Marquee Video chain modifications.  Now available!

mp20748 is offline  
post #109 of 305 Old 12-21-2004, 04:38 PM
Advanced Member
 
mysphyt's Avatar
 
Join Date: Aug 2001
Location: Midland, MI, USA
Posts: 708
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Here are the pics and instructions for this mod. I haven't completed the mod, so no finished pics yet, just got all the parts. It's in PDF format, about 1.4MB.

JVC 30K Mod Doc
mysphyt is offline  
post #110 of 305 Old 12-21-2004, 06:53 PM
Member
 
moggy's Avatar
 
Join Date: Oct 2002
Location: Melbourne, Australia
Posts: 125
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:


Originally posted by VideoGrabber
moggy asked:
> What is the normal BW for 720p? is it the same 30MHz as 1080i? <<br />
Yes. Higher line rate, with lower data density per scan line.

- Tim

Thanks.
That's why they can use the same FL filter for 720p or 1080i. It's the same 74.17582MHz luma/pixel clock rate.
Clever.

It's a pity a 150MHz clock wasn't used so that the filter could be higher and thus give us closer to the 1920 pixels BW.
DVD players use 27MHz and 54MHz clocks almost as standard.


I've also found out that 480i pixel clock is 13.5MHz. So perhaps those 10MHz notch components are to filter this out.
Perhaps they used a switched notch filter since it would suck the 10MHz(or 13.5MHz?) out with sharper filtering rather than a series LP filter which would have only a 1st or 2nd order LP action.
This would give the 480i a really nice response rather than the usual 4.5MHz for offair NTSC.
I think SD is realising it's full potential by having a typical 8MHz (or whatever)BW to play with. The menu text looks half decent too. (The ideal would be 13.5MHz BW, I think)
Why couldn't they have done something just as clever at 74MHz for HD instead of chopping it back to 30MHz? But as Mike has said, the designing engineers are very conservative to prevent EMC problems.

As you might have guessed I don't use my 30k for offair or recording. It's ONLY used for D-Theater playback. I don't even have any blank tapes or NTSC playback tapes. I have about 45 D-Theater movies.
There is no chance in Hades of there being a Region4 version of the 30k being made.
BTW, all D-VHS machines on sale here are SD only and cost A$3000. I don't think they even sell them anymore now that DVD SD recorders have come on line.
The only way to record HD is offair onto the HD of the htpc.

Tomorrow, I'm off for 3&1/2 weeks over christmas. Hooray. Soaking up the sun up north in Queensland. Lovely 32deg every day. Bloody humid though.
I've digressed again. I must like chatting. Oh no!

Regards,
moggy
moggy is offline  
post #111 of 305 Old 12-21-2004, 07:22 PM
AVS Special Member
 
Mark_A_W's Avatar
 
Join Date: Dec 2002
Location: Melbourne, Australia
Posts: 8,110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Queensland???

When are you going to mod my video card then????



Have a nice holiday.

Loving my Electric Bike!!
Mark_A_W is offline  
post #112 of 305 Old 12-21-2004, 07:42 PM
AVS Special Member
 
jcmccorm's Avatar
 
Join Date: Mar 2001
Location: Madison, AL, USA
Posts: 3,379
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 13
Moggy, regarding the 74MHz vs 30MHz, it's that pesky Nyquist guy again. The sampling frequency (74.25MHz) is double the "frequency of interest" so they have to filter at half that to reconstruct the original signal.

You made an interesting point regarding analog vs digital. Are the luma and color difference data sent at 74.25MHz to the display device? (well, for DVI it's already converted to RGB, but what's the pixel clock?)

Which is not to say that engineers aren't still going to go overboard to fix or prevent EMI problems, which I'm guessing is what Mike has discovered here.

Cary
jcmccorm is offline  
post #113 of 305 Old 12-21-2004, 08:29 PM
Member
 
moggy's Avatar
 
Join Date: Oct 2002
Location: Melbourne, Australia
Posts: 125
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:


Originally posted by jcmccorm
[b]Moggy, regarding the 74MHz vs 30MHz, it's that pesky Nyquist guy again. The sampling frequency (74.25MHz) is double the "frequency of interest" so they have to filter at half that to reconstruct the original signal.

Yes. Unfortunately our frequency of interest or original pixel frequency is 74MHz, isn't it? They should sample at twice that but it appears they don't. It's like sampling audio at 22kHz to get 10kHz BW instead of 44kHz to get 20kHz.
Quote:


You made an interesting point regarding analog vs digital. Are the luma and color difference data sent at 74.25MHz to the display device? (well, for DVI it's already converted to RGB, but what's the pixel clock?)

For digital I don't know, I assume the maximum clock rate whatever that might be.
For analogue: I found a data sheet for such a (active) filter and it had 30Mhz for luma(Y) or Green and either 30 for Blue & Red or 15MHz for Cr & Cb. I assume this was to cater for either RGB(or DVI?)(30/30/30) or difference Component(30/15/15).
Since the 30k is difference component I hazard a guess that it should be 30/15/15.
But what is the 'actual' clock for Cr & Cb coming out of the big chip? I can't guess. Hopefully it's 74MHz because if it's 37MHz then we may not be filtering Cr/Cb enough with just a ferrite.
Also, each FL is identical AFAIK so it may be 74MHz with each 2nd pixel being repeated for Cr & Cb. That scenario would be just dandy.

I guess I can muse over these things till the cows come home but knowing that Mike has made it work is probably all we need to know.
Still, it's interesting.

Regards,
moggy
moggy is offline  
post #114 of 305 Old 12-21-2004, 11:42 PM
AVS Special Member
 
VideoGrabber's Avatar
 
Join Date: Oct 1999
Location: SW Michigan
Posts: 2,799
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
moggy indicated:
> Unfortunately our frequency of interest or original pixel frequency is 74MHz, isn't it? They should sample at twice that but it appears they don't. It's like sampling audio at 22kHz to get 10kHz BW instead of 44kHz to get 20kHz. <<br />
Actually, this isn't quite correct. You missed the point of Cary's comments about Nyquist. The sampling rate (clock) of 74 MHz is already twice the highest frequency required (37 MHz). A 37 MHz signal already contains all the information to reconstruct the full 1920 pixel scan line.

Unfortunately, due to the need for filtering (for anti-aliasing and anti-imaging purposes), we need to roll off the response before we hit that 37 MHz mark, and thus can never quite achieve that full 1920 horizontal resolution, in the analog realm (without some attenuation). But we can get close.

SMPTE specifies that the output response (NOT bandwidth, which is the -3dB point) should be maintained "virtually flat" out to 30 MHz. I.e., +/-0.05 dB. The rolloff should start AFTER that point, and be down at least 12 dB at 37 MHz, and continue to roll off at a steep rate (IIRC 6th order), so that by the next octave, when you hit the sampling frequency, you're down at least 40 dB.

What this means is that image integrity (full detail, with unimpaired MTF) is maintained out to 1560 horizontal (i.e., 30 MHz), and rolls off smoothly beyond that, with the bandwidth (-3dB point) being 34 MHz (equivalent to 1770 horizontal). There can still be content all the way out to 1920, but 1770 is about the effective limit of reproduction (if everything in the source chain is perfect).

Based on some response curves generated by Oliver Klohs' analysis, the stock filtering on the JVC 30K creates a bit of peaking (a small hump of about 1 dB from around 10-18 MHz), which then starts rolling off and is down 3 dB already by 25 MHz (or 1300 horizontal). By the time you hit 30 MHz (which should be flat), response is down a full 8 dB. And 34 MHz is down more than 12 dB. This noticably impairs sharpness and fine detail, especially in the highest quality display systems, with the best source material.

There was some controversy last year when it was initially thought that the 30K was down 25! dB at 30 MHz, but it was discovered that the original 1080i test tape (as documented by dr1394) had its own 17 dB of rolloff at 30 MHz. Only 8 dB was contributed by the JVC 30K (which is still a pretty serious hit).

What Mike's clever mod promises to do is push that 1300 horizontal resolution out substantially, to 1600 or beyond. The fact that it's possible to do that without any increase in spurious noise products is impressive. And it obviates the need to use firewire to an external DAC (like those in the SIR-T165 that DO extend out flat to 30 MHz), to extract the full PQ that the JVC 30K is capable of delivering. Kudos to Mike!

- Tim

- Tim
VideoGrabber is offline  
post #115 of 305 Old 12-22-2004, 12:07 AM
AVS Special Member
 
VideoGrabber's Avatar
 
Join Date: Oct 1999
Location: SW Michigan
Posts: 2,799
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
mysphyt wrote:
> Here are the pics and instructions for this mod. I haven't completed the mod, so no finished pics yet, just got all the parts. It's in PDF format, about 1.4MB. <<br />
Thanks, Chris! Great job on integrating everything into a nice document. And seeing how far down into the guts of the machine this mod is located, I'm even more impressed with Mike's signal tracing and analytical skills.

It would be nice if the FL modules could be bypassed from the back of the PCB, without having to pry off the glued on daughtercard. At least we know where everything is located now.

- Tim

- Tim
VideoGrabber is offline  
post #116 of 305 Old 12-22-2004, 12:39 AM
Advanced Member
 
spatz's Avatar
 
Join Date: Oct 1999
Location: Germany
Posts: 895
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
here a picture of the new D5 JVC, that works very reliable and looking at the picture does not have these filters you are bypassing. There are some caps and inductivities around this AKM Dac, but no FL parts.
I do not have a 30k any longer so I cannot compare the image quality of the new deck against the old.


Eidophor oil projector on a fork lifter connected to an original SONY BETAMAX, stunning picture, never seen anything better
spatz is offline  
post #117 of 305 Old 12-22-2004, 12:50 AM
Advanced Member
 
spatz's Avatar
 
Join Date: Oct 1999
Location: Germany
Posts: 895
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
btw, found one bug in this unit, seems that it outputs HDCP all the time, even if you playpack a non copy protected D-VHS HD tape ?
Anybody who can confirm this ?

Eidophor oil projector on a fork lifter connected to an original SONY BETAMAX, stunning picture, never seen anything better
spatz is offline  
post #118 of 305 Old 12-22-2004, 03:25 AM
Member
 
moggy's Avatar
 
Join Date: Oct 2002
Location: Melbourne, Australia
Posts: 125
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:


Originally posted by VideoGrabber
[b]moggy indicated:
Unfortunately our frequency of interest or original pixel frequency is 74MHz, isn't it? They should sample at twice that but it appears they don't. It's like sampling audio at 22kHz to get 10kHz BW instead of 44kHz to get 20kHz. <<br /> Actually, this isn't quite correct. You missed the point of Cary's comments about Nyquist. The sampling rate (clock) of 74 MHz is already twice the highest frequency required (37 MHz). A 37 MHz signal already contains all the information to reconstruct the full 1920 pixel scan line.

Er, sorry, no, I didn't miss the point.
I guess it comes down to what is the real spec for your HDTV. Is it 1920 pixels or 37MHz? 37MHz does not represent 1920 pixels. You say it does. How? Maybe I'm missing something here. 1920 pixels wide requires a clock every pixel to carry the information, yes or no? 2260x1088x29.97=74MHz.(approx.)
How can 37MHz be a representative of 1920 pixels?



The above is my first thoughts on the matter but I knew I must be missing something and I was, see below:

Well, I've been doing some reading and I think I have it now.
You are correct in stating that 37Mhz re-creates 1920 pixels, but not in the way I originally thought.
It seems that one white pixel followed by one black pixel is one cycle of the 37MHz.
This works if the definition of 1920 pixels is one white followed by one black pixel over the entire 1920 pixels wide.
That makes sense. This is how we have always defined resolution as I've rediscovered.

My original thinking was that we need to resolve individual white pixels followed by white pixels, ie. each pixel has information.
It's a matter of definition and I hadn't quite got it right.
Sample frequency is twice that required for 960 white pixels and 960 black pixels interleaved. This is what is defined as 1920 pixel resolution, not 1920 individual pixels.

I hope you've all been patient with me.


To view 1920 individual pixels with black between each individual pixel we would need to sample at 150MHz as a minimum. This has never been asked for AFAIK(except for silly me).
The only way to get near a "real" 1920 pixel res is to be entirely digital. This is where digital displays with HDMI or DVI will shine.

Ahh. I'm exhausted. I'm off to have a beer.

Regards,
moggy
moggy is offline  
post #119 of 305 Old 12-22-2004, 03:36 AM
AVS Special Member
 
VideoGrabber's Avatar
 
Join Date: Oct 1999
Location: SW Michigan
Posts: 2,799
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Spatz, nice closeup shot! but...
> looking at the picture does not have these filters you are bypassing... but no FL parts. <<br />
I could be wrong, but if you look in the upper right-hand corner at the tan-colored rectangular device labeled "04 97" that looks oddly similar to the FL units, with their 3 legs on one side, and 2 on the other.

- Tim

- Tim
VideoGrabber is offline  
post #120 of 305 Old 12-22-2004, 04:45 AM
Advanced Member
 
spatz's Avatar
 
Join Date: Oct 1999
Location: Germany
Posts: 895
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
yep, but they are filtering the signal that comes from the video head.

Eidophor oil projector on a fork lifter connected to an original SONY BETAMAX, stunning picture, never seen anything better
spatz is offline  
Reply CRT Projectors

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