News:

;) This forum is the property of Proton software developers

Main Menu

Video buffering

Started by TimB, Mar 05, 2021, 06:33 PM

Previous topic - Next topic

TimB

Hi all

Part of new circuit I'm working on I analyze a pal video signal. That side is taken care of using the LM1881 and a comparator.

As part of the development, and in its use I need to overlay my own signal on a video signal to aid diagnosis.

The basics are, the Pic works out when line data starts etc using signals from the LM1881. The comparator triggers when the video signal is > x value. The pic uses that data to do calculations etc.

I need to overlay a signal on the video that an external monitor can see.

The issue is that, if I feed my own signal in to the video it will be seen by the comparator and trigger it. Making a feedback that will mask the signal I want to measure. So I added a video buffer. It's a dedicated device. However its not working.

The signal in from the camera is running at around 3.5v around a 1v pp
The signal out from the buffer is running at around 3v but the PP is more like 0.5v.
I have no idea on what I have wrong. Any ideas?

The device is a MAX4389 and my schematic is below.

Please note. The board is made, its built. Recommending another device is pointless.

Thanks Tim

Video buffer.PNG

tumbleweed

#1
Instead of trying to combine the signal with the output of U4, don't you just want a non-inverting opamp summing circuit (ie add both signals as inputs to U4)?

The way you have it right now,  the signal is being seen as feedback on the opamp output.

david

Hi Tim,
The buffer has the usual gain of 2 for video buffering and R8 is expecting a termination of 75R which results in a composite in to composite out gain of unity.  This is the first priority to check.
The output impedance of the buffer is very low - approaching zero, so trying to mix a signal in to this point via 330R is doomed.
I think your best option (least intrusive on an existing design) is to try mixing in after R8 and keeping R7 as high in value as possible.  I'm not sure if D1 is essential but we don't know what is connected to it.

Cheers,
David

tumbleweed

Quote from: david on Mar 05, 2021, 11:28 PMI think your best option (least intrusive on an existing design) is to try mixing in after R8 and keeping R7 as high in value as possible.
That will create a voltage divider between R7 and the 75 ohm termination,  and you'll lose most of the signal across R7.

U4 is just a high bandwidth opamp... I would use a standard summing circuit.

TimB


I will try and go through the comments one at a time.

tumbleweed
Are you suggesting connecting the user signal into the - input of the video amp?
Also at this time I'm more concerned with the video buffer not working

David
There is no 75ohm pull down on the schematic I posted but there is one on the board as the actual schematic used had one. However it was the wrong side of R8, I will add it in the right place and see what happens.


Now I'm very sceptical if its worth is as its not "just working"
The issue is that I do not want the signal I inject into the video line to effect the comparator signal I want to test.

To help understand what I'm doing think of it like this
I keep track of the video timings to know what scan line I'm on. I time from the start of the line to a set point then look for a signal from the comparator that indicates the video voltage has gone past the threshold. I count in a set loop the number of times that "high" signal is seen. I'm only interested in a small section of the scan line. This is where signal injection comes in, I will mark around the test area to ensure I get my timings right.

Now with further thought I should be able to ignore any feed back I introduce as I will turn it off before my loop tests.
I just need a few Us pulse so I can see where I am. On for a few Us just before the test and a few Us after.

I will probably drop the video buffer now and feed the signal directly to the Video line as video txt overlay POV systems do.

Mapo

Hi Tim,
why not try the MAX7456, it eliminates all overlay insertion problems and it is a chip designed for that function. The result is better

TimB

Mapo

From my first post

"Please note. The board is made, its built. Recommending another device is pointless."

david

Quote from: tumbleweed on Mar 05, 2021, 11:53 PMThat will create a voltage divider between R7 and the 75 ohm termination,  and you'll lose most of the signal across R7.

U4 is just a high bandwidth opamp... I would use a standard summing circuit.

Hi,
Yes R7 will drop a lot of level looking in to 75R termination but we don't know the source voltage feeding R7 - it sounds like it may be 5V logic and not a 1V p-p video signal so some attenuation is going to be needed.
Certainly the circuit must be measured when terminated in to 75R and not just measured with a x1 scope probe via R8. 

Cheers,
David

tumbleweed

Quote from: TimB on Mar 06, 2021, 09:07 AMAre you suggesting connecting the user signal into the - input of the video amp?
No, a standard opamp noninverting summing amplifier. Both signals go to the + input.

https://masteringelectronicsdesign.com/how-to-derive-the-summing-amplifier-transfer-function/

TimB

#9
tumbleweed
Thanks,
The summing amp is going to just replicate the issue I was trying to solve and that is feed back into the original video signal. That was the whole reason for the buffer chip, so I could isolate it.

David

Ok Thanks,  The source is a POV camera like those used in drones. They run from 5v to ~12v. I have it at on 5v. The video signal though is 1v p-p just offset from ground.
I did have a monitor connected and got no signal. I think now its a lost cause. I hoped it would be as simple as "change resistor X for another value".

Tim

david

Hi Tim,
I was curious about the drive for the overlay signal and was wondering if that was derived from a logic source at 5V.
If your camera is producing 1V p-p video but biased around 3.5V then your buffer (with a gain of 2x) will be well into clipping. It almost sounds like the camera is expecting to be AC coupled.  Going by the data on the buffer it seems that for a single supply of 5V they use a bias point of 1.5V for the buffer.
For academic purposes you could try adding a 56R resistor at the input of the buffer to ground to see if it biases nearer 1.5V and restores video output but it will also reduce video level.

Cheers,
David

OG

If my understanding is correct;
You want to superimpose the VideoOverlay signal on the opamp output.
OPAMP's output pin does not allow this. Because that pin is out.

TimB


Hi All

First off many many thanks for the help on this. It is appreciated.

Back to basics

Yes I want to overlay some data on a Pal signal. It's not hard You can do it just by applying a high to the Pal signal at the right point.

To reiterate I know there are loads of devices to do it. But I have the board, its built. Telling me to do it another way is not an option now. I need to fix the video buffer, that is all.

It looks like David has the issue in that the Video signal is being clipped by the buffer.

BUT as I also said. Now I'm not that bothered as I can do what I want and not make feedback over the siganl I was trying to test. Its a simple solution. Do not test while I'm injecting the signal.

Thanks

Tim

Below is a schematic of a shield that uses the same system

lm1881.PNG