Sensor detachment from prism.

[Bob Hart]

This commentary is pretty much useless today but I thought an observation might interest somebody.

I have aquired three JVC GY-HD cams for recovery of oral histories from about 60 tapes. All the cams are faulty in some way but a common is detachment of the sensors from the prism. The tape transports are fine

It seems that the adhesive used degrades with time. The extremely low buying cost for these now obsolete cameras enables me guilt-free examination of the entrails whilst I do checks on the tape transports.

Firstly I am impressed with the fine detail in the castwork. Unlike the Sony HDV cams of the same period which wrap plastic bodies around a cast chassis, the JVC castwork comprises the entire camera body which doubles as supports for circuit boards and fittings.

My sense is that the Sony cams with plastic outer body layers may be slightly tougher when it comes to the casual light knocks and bumps of everyday use. Plastic has a dull response to light sharp knocks or vibrations. Alloy metal as used by the JVC is more "live" when knocked or vibrated.

My first thought relating to detached sensors was sensitivity to vibration transmitted by the metal body direct to the prism and to the sensor. Each sensor is glued to the prism by four small spots of white adhesive about 1mm in diameter.

To me, given the mass of the PCBs attached to the sensors and constant forces applied by the ribbon cables in their bent position, this is a big ask of four small 1mm glue spots. There is no other mechanical support other than the adhesive.

Then I observed a miniature fan which circulates air within an otherwise sealed enclosure and my thought moved to degradation of adhesive from heat. The adhesive material seems to be not brittle hard but a softer material.

The degraded material cleans easily off the sensor and the prism so my thought then turned to heat affecting the adhesive over a longer term combining with the constant draw of the ribbon cables.

The monkey wrench in my assumptions is that in all instances so far I have run into, it has been the blue sensor which has detached. It defies logic because the blue sensor is on top and thus assisted by gravity to stay put. The red and green sensors are actually hanging from their glue attachment.

Maybe in it highest position it receives more heat from the other two lower sensors.

Then I observed that the ribbon cable attaching the blue sensor to the main PCB is barely long enough and there is considerable tension caused by the soft fold in the ribbon cable.

If anyone is concerned with conserving their cameras against the likelyhood of the blue sensor coming loose, it may be helpful to add a spacer of about 3mm to 4mm between the SD slot assembly above the ribbon cable and the upper surface of the ribbon cable to force the cable downwards a little and convert the tension imposed upon the sensor by the ribbon cable into a slight pressing force to avoid a tensile load on the adhesive.

As the cameras are by and large feather dusters in light of today's technological advances, salvage work on them probably represents futile lost productivity for most folk.

[Andrew Smith]

Interesting! It's not often that we get to take apart one of our prized cameras.

Andrew

[John Nantz]

Hey Bob, another "like taking things apart" guy here. Having sold my GY-HD7 dearly loved cam (new $1,600) for something like $150 after about 10 years this post was an interesting read. To go with the HD7 I bought my wife an inexpensive JVC cam for her point-'n-shoot opportunities when I wasn't around. Had a Mic-In so was able to use the Røde video mic for better audio with it. It was nice little cam and great for my use as a B-roll cam or those semi-candid shots.

Eventually, the little JVC gave up the ghost and I identified the problem as the ribbon cable that went from the flip-out LCD screen to the cam. Found a replacement ribbon cable on eBay for really cheap, say ~$15 but never bought it or fixed the cam. The days go by and got the Sony AX100 and a little Sony CX380 off Craigs List for her. Gave the little JVC 380 away to a teenager when our neighborhood had a garage sale. The father didn't want to take it but I said that kids need to know what is inside and he can fun taking it apart so since it was free, and given my convincing sales pitch, they took it.

Having grown up in the days of wind-up clocks and tube radios and TVs, taking things apart was what we did. Doing our own car repairs, taking engines and transmissions apart, all a piece of growing up. Changing a throwout bearing on a stick-shift transmission ... no problem. Break jobs, honing wheel cylinders, rebuild carbs, etc. Taking things apart was what a lot of us did.

Now, if we can only figure out how to fix the little glue spots and align the sensors ...

[Bob Hart]

It is interesting what JVC have done. There are aspects of fine engineering, the split processing which did introduce a bad artifact in the initial batch of cameras. But a pliant glue alone to hold the sensors to the prism?

Why did they go there and why no mechanical reinforcement? Probably one of those "because we can not because we should" sort of things. Maybe there was a thermal issue. Maybe a mechanical attachment was affected and moved the sensors ever so slightly out of registration as a result of thermal cycling. There is a miniature fan tucked in there near the sensors so heating is an issue.

The older JVC KY-F50 SD camera heads had the sensors mechanically captive to the prism but maybe the precision required for 1280 x 720 1/3" sensors with smaller pixels with a mechanical system may have been impossible to achieve. Whilst the camera heads were notionally 2/3" an optical reduction was used to enable 1/3" sensors to be used.

Attached to each sensor PCB is a redundant strip with an elbow at one end with threaded screwhole and a wedge shape at the other end which is not elbowed. It seems that these strips are intended to fit into some sort of alignment jig and the sensors are energised to provide an image or signal to a scope or special instrument.

My imagining is that once the sensor images are coincident in the jig, the entire assembly is moved away from the prism, the little dops of glue added and then the sensors offered ujp again to the prism and kept in place until the glue cures. My thought was that they would have used a UV cure optical cement which is the very devil to remove from lenses but is practically indestructable.

However, the glue spots cleaned off the prism and the sensor easily and seemed to be of the same consistency as water cleanable white bathroom sealer. Maybe the material had broken down into the cleanable state.

It should be possible to make an alignment jig to realign the blue channel sensor live in the camera. It would require and A-B axis adjustment plus a rotational adjustment. I suspect that the soldering of the sensors to their PCBs does not provide sufficient accuracy for such a fine optical tolerance.

However applying a replacement adhesive is a killer of a task. Removing just one sensor from the prism without the other two also removed and aligned in the jig to apply glue then offer back to the prism is impossible.

This would be why JVC says the whole prism block assembly must be replaced. That of course then may require the downstream processing to also be adjusted. I swapped a functioning but troubled intact assembly from one camera to an identical model. It showed as a split screen colour channel artifact.

The ribbon connection to one half of the green sensor on the intact prism block seems kaput. That artifact appeared indentically in the substituted camera and the registration of the sensors appeared to be maintained so I am not sure how touchy the arrangement is.

I will have a good look at the ribbon cable attachment to its plug which attaches to the main PCB. It it has gone open on the PCB attached to the sensor then the game is over.

All other aspects aside, the miniature engineering work is amazing.