Video Buffers

I think they are both voltage feedback devices, the first one being an AD828, with the two resistors, and the second an EL2244 using on the one feedback resistor.

Not sure that the means to the solution, but I know I ought to have included that info in the original post :-)

Thanks,

John

Joerg,

Thanks for the explanation. The buffer with the two resistors is a 2:1 amp, where both resistors are 1K, so that on the final stage, in series with each output line, is the required 75 ohm resistor you mentioned.

Where can I learn how to know what the value of the single feedback resistor ought to be on the output stage that uses the follwer, to arrive at the required impedance?

The circuit I am looking at works, but I would prefer to understand why the values are what they are, as opposed to just going through life always using a 680 ohm feedback resistor, and one day, having a problem and not being able to understand it :-)

This probably shows that I am not an expert in OpAmp theory

The amp in question is an EL2244, if that makes any difference.

Much obliged for you sharing your wisdom.

John

Thank you. I read the data sheets, and anything else I can find, but many data sheets seem to be generally written for people who are already fairly knowledgable, and although I can get along, I am not in the same class as a lot of folks here, so I don't always glean a lot of good from them.

I search for app notes, but this is so basic that what I have found so far talks way above it, and doesn't really offer the basics, which I am just learning as I go.

Often I have seen such a basic buffer without the resistor, just a short back to the - input as you made mention of, so it is device dependent, along with a little experience of the designer, I suppose. When I get the board built, I will experiment with values and see what the results are, but at least I now have a bit of an idea why it is there, and so I can go ahead know that it will probably be okay.

Much obliged, as what you said makes sense, and is just one more thing for me to learn and consider.

John

is a follower without voltage gain, mostly used to provide an output drive with a low enough impedance.

source termination, and they use that kind of termination a lot. IOW, when there is a 75 Ohm resistor in series to an output to make sure there will be no reflections on a coax cable even if the termination at the far end isn't so stellar. This 75Ohm will then form a voltage divider with the coax, assuming it also has 75Ohm impedance. This causes the signal to drop by 6dB in amplitude (to half its voltage) which can be made up by an amp that offers 6dB gain.

Thanks for the clear explanation. I have worked with video circuits for years, and it seems that the questions to this answer come up at every preliminary design review. With your permission, I will print this out ready to hand out the next time the appropriate question comes up.

Hello John,

The first example has a gain defined by the resistor ratio. The second one is a follower without voltage gain, mostly used to provide an output drive with a low enough impedance.

In the video world you normally need voltage gain whenever there is a source termination, and they use that kind of termination a lot. IOW, when there is a

75 Ohm resistor in series to an output to make sure there will be no reflections on a coax cable even if the termination at the far end isn't so stellar. This 75Ohm will then form a voltage divider with the coax, assuming it also has 75Ohm impedance. This causes the signal to drop by 6dB in amplitude (to half its voltage) which can be made up by an amp that offers 6dB gain.

Regards, Joerg

It could well be that the Amplifier is a current feedback type.

Rene

--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net

The single feedback resistor doesn't set the impedance. It may affect the offset voltage error or the frequency characteristics of the feedback (when you consider that the input has some parasitic capacitance). In classic op-amp theory, when you are dealing with ideal op-amps, the single feedback resistor does nothing and could be replaced with a short circuit.

My best guess in this case is that it somehow provides balance for the DC offset currents flowing in or out of the inverting and non-inverting terminals, thereby minimising offset voltage. Hopefully someone will provide you with a better answer. ;-)

Usually, I look to the datasheets and app notes for this kind of information but I didn't see any for this particular part.

--Mac

Hi Richard,

You are welcome. Just mind that this is not intended as advice since every situation is different. Especially with low gain active stages built around opamps your engineers have to check out and really understand stuff like phase margin, stability at low gain and so on. I had to redesign a few video circuits because they would occasionally become unstable. To tell you the truth, I kicked the opamps out of almost all of them and replaced them with transistors of the 10 cent category. At the end I just clamped the DC drift out, other times I didn't because the video gear is supposed to do that if it operates according to the standards (which it often doesn't...).

Regards, Joerg

Hello John,

I don't use opamps in video designs much, mainly for cost reasons. So I don't know the EL2244 but as far as I remember it is a voltage feedback and unity gain stable device. Probably it'll be happy with a plain trace from output to negative input. But it would be best to discuss this with an application engineer at Intersil.

Sometimes people use feedback resistors in a prototype to be able to 'give it some gain' later when there ain't enough gain. For example, if they don't know yet whether or not they'll need some kind of filter which typically will also come at a 6dB penalty.

BTW, there usually is a Spice model for these types of amps. So you could simulate a design with a program such as LTSpice (this one is free) by Linear Technology. LTC also makes lots of great opamps and their models should plug right into LTSpice.

Regards, Joerg

This application goes beyond opamp theory. The EL2244 boasts extremely good differential gain/phase performance, a measure of transfer function deviation with DC offset, and it is for this reason that the RC time constants at both inputs must be matched at least throughout a 5MHz video bandwidth. Use the 680 ohm if that's what Intersil recommends. Also, the 75 ohm series termination is a convenient way of eliminating the scourge of wideband linear cable drivers, indeterminate capacitive loading, in addition to providing attenuation/elimination of reflections. You know that the series termination->cable->terminated load is a net gain of 1/2, so it is no leap to conclude you need a gain of x2 prior to or at this stage, you could make the driver itself a gain of two also. Your circuit description suggests that Intersil is using the low impedance load driver configurations with the least gain, x1, to maximize the feedback and thereby minimize resulting waveform distortion where the amplifier is under the most stress, reserving the x2 gain elements for high impedance constant and determinate loads like the line buffer input. This explains the x1 buffer splitter that drives across a large board ( into a termination?), followed by the x2 and into the line buffer again at x1.

This is very interesting. How many transitors and in what configuration? Do you rely on relatively high voltage rails (like +/- 12) or can it be done easily with, say, +/- 5?

--Mac

Hello Mac,

Being a consultant I usually took whatever voltages the client's system offered. +/-12V is nice, so is a +/-5V situation. But there were occasions when I had to make do with +5V alone. That requires a lot of quiescent current to be able to properly drive a 75Ohm cable with source termination. On +5V only systems there was plenty of power available though, thanks to the widespread use of guzzlers such as PALs, Thevenin terminators etc.

Gain was achieved with the usual simple two-transistor stages and I used AC coupling a lot. You just have to make sure that the 3dB limit at the low end is low enough for the allowed droop at the end of a horizontal line which usually will be in the millivolt range. The same goes for any clamping but here FETs such as the BSS84 and BSS123 come to the rescue which can be had for around five cents a pop. These would also make good video buffers but, being a penny pincher, that was delegated to bipolar versions at half the cost.

If you need lots of gain, for example after elaborate filtering, there is the uA733 (Texas) which is a blazingly fast amp for its day and age. That was always my friend because of its great performance and low cost. The strange thing was that hardly anybody knew about it. I don't know how long it is going to be around considering that this chip design dates back to the early 70's. But it was transferred to SMT which surprised me. Then again, for under 40 cents it is a true bargain.

Regards, Joerg

Thanks! You seem to be the king of cheap discrete solutions. ;-)

--Mac

Hi Mac,

Nah, I think the real kings in that domain are in Asia. In Korea I once thought I had a rock-bottom cost solution for a $10 circuit. Then an engineer beat me to the punch and sliced more than 50 cents out of it. They had a wall that looked like modern art but when stepping closer it was reems of computer paper with part prices on there. Engineers would do an hourly pilgrimage to that wall. I wish they'd teach that at schools here.

Mostly I try to use jelly bean parts not so much for cost but to make Purchasing's life easier. After all, these are the folks who get those nice chocolates and bonbons from distributors in December and then it is good when they like you... Also, it'll be better for circuit boards that are farmed out to assembly places because that will be cheaper when you use their 'preferred parts' as much as possible.

Regards, Joerg