Marcel Duchamp wrote:
> First off, "video-frequency" ain't what it used to be.  It 'used' to be 
> about 5MHz; you're working with something faster; note those 1.5nS rise 
> times on the data sheet and the 400MHz bandwidth.

Very true. To be more specific, I'll need at least 640x480@60 resolution 
capability on the YPbPr input (27Mhz), and at least 1280x1024@75 on one 
of the RGB inputs (147Mhz). The latter will probably be using the direct 
input to the LMH6572. The remaining RGB input isn't as critical, but I'm 
hoping to get at least 1024x768@60 (70Mhz). For future proofing, 720p 
capability would be nice on the YPbPr inputs as well (82Mhz). 1080i is 
just a little more - 93Mhz. The datasheet quotes 70Mhz for the YPbPr 
inputs, but it's advertised as HD-capable, so I would expect it should 
at least be capable of 720p without too much trouble. We'll see.

> Ignore that.  It's either a complete mistake or it's for a special test 
> fixture at National.  Note figures 1 & 2 --- none of that stuff is shown 
> there; only in figure 3 labeled "Test Circuit"

Good. Grounds tied together, 75 ohms across the input then. Are the 
33ohm resistors and 220nF caps okay? Should I use a bias circuit like 
the one on the direct input to the LMH6572, or maybe the schottky diode 
system they mention in Figure 2? I'm not too sure what DC bias the 
LMH1251 wants on its inputs (if any).

> If the biasing is as you describe, you are right.  In any case, the caps 
> should not be reversed biased.

The LMH6572 op-amps are designed for a dual supply, or for a single 
supply referenced at 1/2Vcc. R16 and R20 are supposed to provide the 
virtual ground for the 6572. I think the ground connections on it really 
only connect to the gain setting resistors, so a simple resistor divider 
should work (as mentioned in the schematic). In fact, I've tested the 
second half of the circuit already. I've built it on a prototyping board 
and it works fine buffering a 1280x1024 VGA signal, though I do get some 
ringing (probably due to the missing 33ohm resistors and/or the crude 
wiring on the protoboard)

I have no clue what bias the LMH1251 uses though (maybe it biases the 
signal using internal resistors?). Pr and Pb are +/-350mV, so it can't 
be using GND as a reference for that. Actually, the datasheet mentions 
Vo BLK, the black DC output level, at around 2.1Vdc. White is 0.7V above 
that, or 2.8Vdc. Since the LMH6572 is referenced at 9V/2 = 4.5V, it 
should still be at a higher potential.

>> What about the width of traces? 
> 
> As the data sheet says: short.

*width*, not length :)

> The data sheet says 6pf of loading.  You can calculate this for the 
> board thickness and dielectric you plan to use.  With smd parts, you 
> should be able to get them fairly close, no?
I'm probably going to be using SMD 1206 for the passives. I can do 0805 
if you think it will help.

> By the way, did you note those 33Ohm series resistors?  Don't leave them 
> out.  And if your board shows any ringing on those lines, feel free to 
> experiment with them by trying both smaller or larger values.  They 
> absorb ring energy so may need some tweaking to get right.

I assume you mean the ones (that should be) between the LMH1251 and the 
LMH6572. I'll add them in.

> Good luck!
Thanks :)

For now, I want to at least get the component placement right. Swapping 
values is easy once I've built the PCB, but I want to avoid having to 
green-wire-patch it too much (especially considering the frequencies I'm 
dealing with).



-- 
Hector Martin (hector@marcansoft.com)
Public Key: http://www.marcansoft.com/marcan.asc

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist