Will do. I swapped the 1st & 2nd back & forth (both the originals) and didn't notice any change. The 3rd IF 6EJ7 was one of the bad ones that I replaced. I just now swapped it with the old one with the set still connected to the IF of the modulator and there are now two ghost images on the test pattern. That's a good clue!

It varies strongly with fine tuning. With the tuner connected, the "ghosting" goes away completely when I fine tune it into B&W, and I can exaggerate it by fine tuning it clockwise out of tune. That's what made me decide to try hooking the modulator directly to the IF cable and try to rule out the tuner. Maybe I'm using the wrong term when I say "ghosting" - if it's called overshoot, I want to make sure I get it right.

I don't have a way to inject baseband video.

I'm using a DTV converter box for the OTA content, and it has USB for the jpg images of the test patterns.

Thanks for the replies guys, more food for thought.

Any image that follows the main image on the right could loosely be called a ghost.

An overshoot is a following transient in the signal the opposite polarity of the main signal. It is possible for some circuits to produce a transient before the main signal, which is called a preshoot. Generally, if the transient is the same polarity as the main signal, it is called a smear.

TV sets generally use narrow overshoots and preshoots to "sharpen" or "peak" the picture. The one you are seeing is way too wide and too big to be pleasant.
Deliberate overshoots are relatively easy to generate in the video circuits. Peaking coils (in various places) and RC circuits involving small capacitors in the video amplifier cathode circuit may be used for this. A simple sharpness control may use a small capacitor and variable resistor in the video output cathode circuit.

Preshoots are harder to generate, because you can't generate negative time delay. You need a "non-minimum phase" circuit that delays the main signal a little bit. The IF alignment can do this, but it will not be adjustable by the user. TVs for many years used a fixed pre-shoot produced by the IF with a (possibly adjustable) overshoot generated in the video circuits.

Towards the end of analog design, video ICs became complex enough to generate symmetrical preshoots and overshoots in the best sets. Before that, only studio monitors had this capability. This was called "aperture correction" if it was in a studio camera, because it compensated for the shape of the scanning spot. In this case, the signals were produced by a tapped video delay line rather than an IF phase response. The aperture correction name stuck when it was incorporated in video monitors.

Here's Zenith's 1981 patent for a symmetrical peaking circuit (analog bipolar IC) that also included a noise-coring capability. Transient amplitude was detected from the first derivative signal across the resistor, and low level random transients would not be amplified.

https://pdfpiw.uspto.gov/.piw?PageNu...D4%2C296%2C435

That's over my head, yet fascinating. Thanks for the explanation, I hope it all makes sense at some point. At least now I understand what the overshoot looks like, and why I shouldn't use a crosshatch. I can see the overshoot now, even on a program fine tuned into B&W where it looked fine with the crosshatch. I should have some NOS 1st & 2nd IF 6EH7 tubes tomorrow and I'll see if they do the trick.

This is a good place for an analyst like a Sencore VA62 etc.

You can try lifting one end of the video detector diode & driving it with
your video source (the RCA plug). Remove the 3rd IF also ( 6EJ7 IIRC ). One volt P-P is pretty much standard. This will divide the chain in half & tell you
whether to look to the left or right of the diode. Divide & conquer :thmbsp:

BTW analysts are almost dirt cheap now. The sencore will do a LOT
more than most. If you get one be SURE it works, has all the cables,
& both the quick set up guide and full manual.

73 Zeno:smoke:
LFOD !

Thanks Zeno. I got the NOS 6EF7's and they didn't make any difference.

So do I pull the 6EJ7, and lift the anode side of the diode and drive the diode with my source? Or am I just lifting the diode and connecting my source to the IF cable of the set? Or is your suggestion in the context of me having an analyst? I'll search for one and get it if that's what I need to do. Sorry for not understanding 100% but I don't want to make a big mistake.

I'm pretty sure what he means is disconnect the detector diode on whichever side is easiest, hook composite video to the grid of the 1st video amp and see what happens...it's a good signal injection troubleshooting approach. If it gives you video then everything between the ant terminals and detector diode is suspect, but if it doesn't give you video then the video amp stages are suspect.
BTW good idea to not crank brightness too high when looking for video if video is weak you'll see it best with the raster slightly brighter than the minimum brightness needed to perceive that the screen is lit...many properly working sets with strong CRTs will totally wash out the video if brightness is set to max.
If you have video, and good IF voltages but no reception AGC might be improperly adjusted.

Correct on both counts. Keep in mind that it was common with cross hatch
& dot patterns to get a ghost to the right. Turning down the brite
eliminates most of & & FT can also. Also if the set has AFT or any other
auto color buttons turn them OFF til the set is repaired.

To inject to the 1st IF input you can make a tuner subber.
Take almost any solid state click type VHF tuner. Get Sams for the
set to get the B+ & AGC voltages. Make a supply for B+ & an
adjustable divider for AGC. Ground the AFT input.

73 Zeno:smoke:
LFOD !

Another thing to watch out for on late 60s Zeniths is an unsoldered tuner lead terminal...I've seen that factory miss on 2 Zeniths so far...

This has to be a safety feature. It prevents the HV from coming up if the set is turned on with the regulator tube removed.

I put this set away for a while and finally got back to it. After letting it run for hours and making some adjustments, it looks pretty good, so I think I might have been too picky about it. The convergence needs some work, but other than that, I think it's pretty good.