# Testing a video transmitter for video bandwidth

I need to test the bandwidth of a variety of video transmitters to figure out the maximum resolution I can safely output without losing detail. To this end, I came up with a solution using a TV pattern generator to generate a proper PAL or NTSC video signal with colour bars or with just blackness, and a 10 MHz signal generator, to generate a pixel clock. However, the 10 MHz signal generator is too expensive for a student like me - about £300 here. The TV pattern generator is only £75, which I can afford. I need a way to generate an adjustable square wave signal from 5 MHz to about 12 MHz, synchronised with a video signal. The frequency doesn't have to be precise, because I can verify it with an oscilloscope. I just need to see the point at which a typical video TX can handle no more pixels, because I have the opportunity to make a high resolution on screen display - but it doesn't matter if no-one can see the pixels!

• I recommend the $0.5M 1THz Agilent VNA. £1~$1070, right? :D – tyblu Dec 28 '10 at 11:27
• @tyblu - What a bargain! – Thomas O Dec 28 '10 at 11:53
• If you only have access to a sine wave source, run it through an op-amp with way too much gain. It will clip the signal to the point where it is a square wave. Bear in mind, though, that it will generate harmonics like crazy, probably out to 200 MHz or more. (I don't know if it's even-order or odd-order, or both. It's been a while.) You may want to run the signal through a low-pass filter to eliminate some of these harmonics. It'll distort your square wave a bit, but what's important -- the waveform or the spectrum? – Jesse Dec 28 '10 at 15:21
• So to simplify your question, you just want a variable frequency square-wave generator, 5-12MHz? Also, have you considered renting equipment? – W5VO Dec 28 '10 at 16:10
• @Jesse, many op-amps do not recover quickly from saturation, and at 10 MHz may not get a chance. This is not always spec'd in the op-amp datasheet. (An op-amp that recovers from overload quickly is called a "comparator.") For this purpose a Schmitt gate like the 74HC14 works and costs pennies. – markrages Dec 28 '10 at 17:07

It's not hard to make a oscillator of a few MHz. In fact, I've done it by accident many times.

Use the VCO from a 74HC4046. Supply the chip with power, and setup a pot to feed the VCO input.

If the VCO isn't a strong enough square wave, feed the output through a few stages of 74HC14 to square it up.

If you really need a phase lock with the video signal, you can phase lock this against the sync output of an LM1881. Use phase detector II in the '4046 and add your own dividers according to how many pixels you want in a line.

With all this said, surely the vendor of the video transmitter has a spec for the video bandwidth it will support? Also, NTSC kinda defines its own bandwidth requirements. From Wikipedia:

• That's broadcast NTSC. I suspect most VTX's will be limited to 3-4 MHz, but I'm unsure. If you're dealing with composite video, I think there is no actual limitation on bandwidth, although 100 MHz is usually more than enough. – Thomas O Dec 28 '10 at 18:01
• Thanks for the idea on the 4046, I've heard of that chip. I'll have to get some for myself at some point. – Thomas O Dec 28 '10 at 18:01
• Note that the original CD4046 is only good to about 1 MHz. The 74HC one is faster. The part is discussed a little bit in Don Lancaster's "CMOS Cookbook" (If you don't have it, buy an autographed copy from Don on eBay.) A fuller treatment of the loop filter is found in "PLLs: principles and practice" but it's just another control system, more or less. – markrages Dec 28 '10 at 21:19

Almost any FPGA evaluation board of the last 6 years vintage will give you an ability to generate clock signals of varied frequency using PLL multipliers and dividers; they probably have too much jitter for radio transmission carrier frequencies, but are fine for modulation. And you can make actual test pictures using the fpga logic. A resistive ladder DAC, or just a few resistors from various outputs to give you two video levels and a sync pulse can be used. You will want to take the input impedance of the video transmitter (probably 75 ohms, but you can test it by making a resistive divider with it as the lower leg) into account when designing these.

Since you have a scope available, you may do better to use that to measure the voltage for various modulation frequencies and plot a transfer function rather than to rely on a test that also depends on the bandwidth of your display (which if LCD is going to be sampled and then pixelated itself).

You may well have a video audio gain control to contend with, so you might want to make a circuit which generates a few cycles of a reference wave at low frequency (say 500 KHz or 1 MHz) and then skips up to a test frequency you control with DIP switches on the board, and outputs that for a half line of video or so, and then goes back to the low frequency. What you will do is measure the amplitude of the high frequency signal in comparison to that of the low frequency reference, which is there to set the AGC level. Doing this with square waves you will probably see a rounded off leading edge of the pulses, but the low frequency one will get to a flat top. The high frequency one will at some frequency never get beyond the rounded portion, and then start not even completing the rise.

If your scope has good measurement cursors, you might also simply be able to measure the rise time of the low frequency square wave until it reaches its full excursion, and calculate the bandwidth from that.