Video Distribution Amplifiers for 10MHz Refs

[KungFuJosh]

I recently got a Dwight Cavendish VP-712 video distribution amplifier. It's a little weird, but also kinda cool. The one I received was defective, so I had to clean it up and replace a couple caps. It's a million years old, but came with a cute little toroidal transformer.

Does anybody have a schematic for this?

Anyway, this seems like it might be a good candidate for use with 10MHz refs after some modification.

One issue is the BW. I'm assuming the limiting factor there is the LF353N. Allegedly, the TLE2072 should be a higher bandwidth drop-in replacement.

One oddity I found was the series resistors for the outputs all measure about 39Ω. The input resistor was closer to the expected 75Ω.

One concern is that the outputs are not isolated. I don't know how important that is, but I saw a couple mentions of preferring desired outputs for a 10MHz distro amp, despite all the other conversion projects not being isolated (like Extron stuff).

Besides that, the output voltage range will need to be adjusted to make the trimmers more useful.

Now the question is this: should I convert this for higher bandwidth and 50Ω, or should I use the shell for one of the other distribution amp projects?

Thanks,
Josh

[KungFuJosh]

Here's a square wave going through the distro amp at 10MHz and 100kHz.

[Performa01]

The LF353 has a GBW of 3 MHz ... in hindsight it's amazing how low the requirements for analog video have been back in its day...

However, the TLE2072 is limited to 10 MHz GBW too. I would not want to build a dedicated 10 MHz distribution amp with an OpAmp that is so slow that it can barely handle the fundamental frequency - even if I only intend to use sine wave references.

A much better replacement would certainly be:

LM6172 (32V, 100 MHz, 105 mA, 3000 V/µs, PDIP)
AD8056A (12V, 125 MHz, 110 mA, 750 V/µs, PDIP)
THS4082I(D) (33V, 175 MH7, 100 mA, 230 V/µs, SOIC)
THS4012I (32V, 290 MHz, 150 mA, 310 V/µs, SOIC).

Regarding "drop-in replacement":

- all these dual OpAmps should have an identical pinout, LM6172 and AD8056A certainly do.

- The AD8056A can only handle a max. total supply voltage of 12 V, i.e. +/-5 V nominal. This might require an additional voltage regulator if the LF353 currently runs at higher voltages.

- Most likely, not all of the above mentioned ICs will be available as PDIP; for example, THS4082I can only be found on the used market - if at all - so you'd have to make do with the THS4082ID, which is SOIC. Whenever you have to use SOIC, I would strongly recommend to build the entire amplifier (OpAmp, feedback components, supply decoupling) on a dedicated daughterboard and connect it to the mainboard using coaxial 50 ohm cable.

- The LF353 might have been fairly high speed back in its days (1970s) and for some reason JFET amplifier were usually faster than their bipolar counterparts back then. So it's no surprise, that we find an LF353 in this application. The replacements that I've listed above do not have a low input bias current like the LF353 does, but for a 50 ohm AC application, this should really be of no concern.

Another aspect is the maximum output current, which also determines the maximum output level. If we use 50 instead of 75 ohms, the OpAmp must deliver 50 % more current for the same output level.

The question is, what maximum output level do you want? An LF353 (30 mA) can deliver 2.25 Vp at best - you usually don't want to run into the current limit, so 1.5 Vp (20 mA) would be more realistic for everyday use. Incidentally, the TLE2072 limits its output current at 45 mA (=50 % more), hence the output level could remain unchanged even with a 50 ohm load.

I've added the short circuit output currents to the OpAmps listed above, so you can easily determine, what is possible with more powerful devices.

Of course the max. output level is also limited by the max. slew rate. Fortunately, the requirements are not hard to meet for sine waves. 1.5 Vp at 10 MHz requires ~9.5 V/µs. The LF353 was quite fast in its days with 13 V/µs, so the large signal bandwidth was hardly a concern, given the very moderate small signal bandwidth. Naturally, a TLE2072 beats that quite a bit with 40 V/µs - about the same factor higher as the GBW.

In any case it gets important for square waves, consequently I've also added the maximum slew rates to the list of alternatives.

Another consideration is the BNC connectors. The 75 ohm types are not compatible with 50 ohm, even though you can squeeze a 50 ohm plug into a 75 ohm socket and for a fairly static configuration it might still be fine. The difference in characteristic impedance doesn't really matter at 10 MHz either. Yet for serious purposes, we should not mix connectors of different impedances.

[KungFuJosh]

Thanks for all that info!

The AD8056ANZ says it goes up to 300MHz in the datasheet, so that's probably good enough for my 10MHz needs.

I checked the supply voltage in the circuit, and it's 13V (-5V and +8V), so we're not that far off.

Is that the chip you'd choose? Or would you prefer the LM6172 if you were doing this?

Do you have an opinion on isolating the outputs? If isolated outputs are really needed, then I would scrap this and build 3 of the DIY projects.

Thanks,
Josh

[Rubo]

Don't have any comments on the op-amps, but having done couple of 75 to 50 Ohm distribution amplifier conversions, I can comment on isolation.

It is imperetive (in my opinion) to have good isolation between outputs, because adding a load to non-isolated outputs changes the phase of the signal on the rest of the outputs.
If you will always use it in a particular configuration, without plugging and unplugging other devices to the outputs, that's not very critical.
But if you plan to plug and unplug devices, I would scrap the idea.

Check out the schematics for SRS FS730/735 to get an idea of a good distibution amplifier design:
https://www.thinksrs.com/products/fs730-1.html

P.S. Since you have a couple of VNA's, try sweeping that DA with 75 Ohm outputs, and do the same after changing one of the ports to 50 Ohm.
I haven't done that myself, so not sure what to expect. After I finally finish setting up my lab, I'll do it with one of my conversions and post the results.

[KungFuJosh]

I'm worried more about safety than phase. Yes, it will generally be setup the same without frequent changes.

My goal is very simple and stupid: I like the displayed frequencies to agree across devices. Though I gotta say, I was impressed when my SNA with its internal ref said 2.499999MHz when the GPSDO ref'd SML03 was outputting 2.500000MHz. Not too shabby.

This is one of the DIY versions I was considering for the distro amp: https://ohwr.org/projects/pda-8ch-fda-8ch/

However, if there's no real safety concerns, I will try and make this work first. Yes, I will see if I can find some good replacement 50Ω jacks.

I think it's weird that the output resistors are only 39.5Ω for a 75Ω system. Why? if there's something else providing the difference, then I would need to replace them with lower values not higher values to convert the system. I wish I had a schematic, but probing can be fun too. 🤷 OTOH, there's not a lot going on, so I might make a schematic if it feels worthwhile.

Thanks,
Josh

[Rubo]

My goal is very simple and stupid: I like the displayed frequencies to agree across devices.

Trust me, I'm with you:



But even if your frequency counters will agree on the frequency, if you measure the phase, they will disagree on it.
And you will see it on a oscilloscope.

I'm just trying to prevent you from activating your OCD when you see it (happened to me  )

[edavid]

Another consideration is the BNC connectors. The 75 ohm types are not compatible with 50 ohm, even though you can squeeze a 50 ohm plug into a 75 ohm socket and for a fairly static configuration it might still be fine. The difference in characteristic impedance doesn't really matter at 10 MHz either. Yet for serious purposes, we should not mix connectors of different impedances.

This is incorrect, 50 ohm and 75 ohm BNC connectors use the same metal parts and are completely compatible mechanically.

(This may be a moot point, since judging from OP's photo this amp uses 50 ohm female connectors.)

I think it's weird that the output resistors are only 39.5Ω for a 75Ω system. Why? if there's something else providing the difference, then I would need to replace them with lower values not higher values to convert the system.

I think it's an attempt to compensate for the output impedance of the video amps.

[Martin72]

This is incorrect, 50 ohm and 75 ohm BNC connectors use the same metal parts and are completely compatible mechanically.

Mechanically, they are compatible, unfortunately.
However, they differ in design (dielectric), but as Performa01 also noted, at 10 MHz that’s not a big deal.

[KungFuJosh]

Yeah, it's a little weird. The connectors look more like 50Ω, right?

I think I'll leave them alone since it shouldn't matter for this application. I didn't bother changing them on the Extron either, and that was fine.

[Martin72]

The connectors look more like 50Ω, right?

Yeah, and it's not exactly robust.
But if it serves its purpose, I wouldn't change a thing—it must have worked for them at some point, and the developers must have had a reason for it.

[KungFuJosh]

Should I be lazy and use a zener diode, or throw in an LM7805 to drop the 8V to 5V for the new AD8056ANZ?

Or should I be even lazier, and just use the LM6172? This is more tempting.

Also interesting, the LM6172 datasheet shows 37Ω resistors in series after the op amp, and it only specifies the types of cables for different impedances. I would love to assume that means I don't need to change those 39Ω series resistors. Can I get away with that?

I'll need to replace the 75Ω input resistor(s) either way.

Thanks,
Josh

[edavid]

The LF353 is not in the video signal path.  As marked on the board, it's the DC clamp.  There is no reason to replace it.  (Although you might need to disable the clamp circuit for your non-video application.)

[KungFuJosh]

I could be wrong, but that doesn't make sense. The potentiometer adjusts the DC clamp, and you can watch the waveform shift on the scope. It's a dual op amp, so if that is a dedicated function, that could just be one side of the op amp, right? Besides that, what else has a bandwidth limiting factor?

[edavid]

I could be wrong, but that doesn't make sense. The potentiometer adjusts the DC clamp, and you can watch the waveform shift on the scope. It's a dual op amp, so if that is a dedicated function, that could just be one side of the op amp, right? Besides that, what else has a bandwidth limiting factor?

From the photo you posted, it looks like the signal path is all discrete transistors.

Do you see the video signal on any of the pins of the LF353?

What's the measured bandwidth?

[KungFuJosh]

Good questions. I need to trace this out a little bit.

The output section does look like it's fed from some BD135s, but I need to see what happens before then. My thought was that the op amp feeds those, but I haven't confirmed anything yet.