Restoration/repair of a 1947 RCA Victor TV set.

[N2IXK]

Just got started on the restoration of an RCA Victor model 721TS television set. The "721" designation means 1947 model,  21 tube chassis.  Thought I would write the project up here to show my methods for bringing vintage electronics back to life, without damaging often unobtanium parts.  The schematic and service literature I am using is available as a .pdf here:

http://www.earlytelevision.org/pdf/RCA-721TS-Rider-TV1.pdf

Now the WORST thing you can do to 60+ year old electronic gear is to just plug it in and let it rip. Many components (mainly capacitors) have become degraded over the years (materials technology being a LOT cruder than modern era). Electrolytic caps need to be "reformed" at the VERY least, and the typical wax paper coupling capacitors are almost unanimously leaky. Bringing equipment to life slowly and in stages helps reduce the risk of burning stuff up.

First thing I decided to do was just to give the chassis a good general inspection.  Looking for signs of previous repairs, particularly sloppy soldering or "anachronistic" components, that don't match the vintage of the rest.  Found some crudely cut and respliced wiring (see picture #4) connected to the speaker (which has a nice hole punched in the cone which will require repair), as well as a capacitor (picture #5) which was added across a portion of the horizontal output (flyback) transformer winding. This was an old-time repairman's trick to cheat a little extra width out of an aging set.  Clipped out the added cap, and removed the speaker carefully and set aside to prevent further damage.

Gave everything a thorough dusting/wiping and looked for degraded wiring, missing parts, etc. other than the butchered speaker and the added cap, everything else cleaned up well and looked to be fairly unmolested. I pulled all the tubes at this point, and ran them through a tube tester.  This set had a mix of some RCA Victor branded originals (most of which tested perfectly good after all these years), plus a mixture of other brands that had been installed as replacements over the life of the set. A handful of leaky or weak tubes were found, and replaced as needed.  The 10" CRT was checked and found to still have excellent emission.

This set uses a large power transformer and full wave tube rectifier (type 5U4) to supply various voltages , ranging from +280V to -85V. The same transformer also has low voltage windings to supply tube heaters.  This makes the set safe from the problems inherent to "hot chassis" designs which gradually took over as TVs began to be built down to a price. I found a pair of molded paper "micamold" capacitors, C143 and C144 (.01uF/400V) from the incoming AC line to chassis ground. I went ahead and clipped them out, as these capacitors have been known to fail explosively when hit with line voltage after sitting dormant for decades. These caps will be replaced with modern X/Y rated AC capacitors.

At this point, I took an ohmmeter, and looked for continuity across the power cord prongs when the set was turned on and off.  This checks the primary of the power transformer for continuity, as well as the line cord, fuse, switch, etc. Checked for shorts to chassis ground, as well. Everything looked good at this point, so I put all the tubes EXCEPT for the 5U4 rectifier back into their sockets, and connected the power to a Variac.  With the rectifier tube out, there will be no B+ supply generated, and nothing should burn up.  Turned on the set and slowly ran the Variac up while listening for hum and looking for smoke or other unpleasantness. Nothing but a nice orange glow from the tube heaters.  Setting the Variac for 117V, I measured the voltage between pins 4 and 6 of the missing rectifier tube, and found ~750VAC. This is a great sign, and shows that the power transformer is also not shorted.

I let the set cook like this for a while, to make sure nothing was overheating or arcing. So the tubes were glowing happily for the first time in decades, and the project is off to a good start!  Next up, a few more quick checks, and replacing the electrolytic caps....

[Vgkid]

This is pretty cool, I do look forward to this.
Ironically We have one in our garage(I believe thats what the manual says, wherever it is).

[T3sl4co1l]

Cool!

Not many grids on that CRT, is that one of those magnetic focus types?

Yeah those Micamolds are a gotcha... the true silvered mica ones are generally okay (give or take possible formation of silver whiskers), but those are only the small values, if any... the bigger ones are oiled paper or something like that, and leaky!  Needless to say, all the waxed paper tubulars are getting the nix?

Tim

[calzap]

Wow, massive amount of point to point wiring.  Made me research history of the PCB.  It was  invented and used commercially to a limited extent prior to WWII with the main inventor being Paul Eisler in Austria although there were others.  Eisler fled to England to escape the Nazis, and his expertise became available to the Allies.  Eisler's ideas were refined by Harry Rubinstein of Centralab to make products more robust.  PCB production began big time on the Allied side during the war to make proximity fuses for large ordinance because the connections could tolerate higher G forces than point to point, and the fuses could be made more efficiently.  PCBs were used in other war-related electronics too.  Most of the expertise and manufacturing effort were used for defense purposes during the war, but afterward, PCBs began to be used more and more in consumer and industrial electronics.  Apparently, not in RCA TV's in the late 40's though.  The 50's were the decade when PCBs finally replaced most point to point wiring.

Mike in California

[Tandy]

Found this fascinating retro video about manufacturing RCA TVs

[rsjsouza]

Nice work and report, N2IXK! I look forward for the rest.

I started the valve bug as a kid when my dad restored his father's Philips BX462. A similar procedure was followed, but only with the schematics in hand (something harder to obtain in the 80's) he was able to find that some "technician" had removed a couple caps for apparently no reason.

One thing that you may want may to look for is the silk wire coils - I heard years ago they can accumulate moist and change their inductance. I am not sure if this is entirely true, but the BX462 lost sensitivity over the years and we could not figure out the reason other than this (we replaced the two heptode/triode valves to no avail).

When I was a kid I disassembled a few valve-based TV sets and, as you mentioned, all of them had the 300mA "P" series heaters (PL36, PY88, PCL82, PCL85, etc). This could give you a quite nice jolt if you were not careful, but what we really stayed away was from the "hat" of the horizontal rectifier, either a beautiful DY87 or a DY802 that had the X-rays warning (I still have those).

[N2IXK]

Cool!

Not many grids on that CRT, is that one of those magnetic focus types?

Yep.  Uses external electromagnetic coils for focus and the ion trap. Focus coil does double duty as one of the power supply filter chokes, along with the field coil on the speaker.

The 10BP4 was the first mass-produced television CRT.  It was introduced in 1946 in the 630TS chassis, and was the most common CRT of the early postwar era. Datasheet available here:

http://frank.pocnet.net/sheets/049/1/10BP4.pdf

Yeah those Micamolds are a gotcha... the true silvered mica ones are generally okay (give or take possible formation of silver whiskers), but those are only the small values, if any... the bigger ones are oiled paper or something like that, and leaky!  Needless to say, all the waxed paper tubulars are getting the nix?

Tim

Silver mica and ceramic caps generally won't need replacement, as their dielectric materials age much better than the paper types did.  All the paper tubulars are going to be replaced, along with all of the electrolytics.


All wiring in this set (and others of the era) was point-to-point. PC boards didn't come into general use in consumer electronics for almost a decade after this set was made.  And they didn't really become as reliable as point-to-point wiring until tubes were replaced with transistors, and epoxy/fiberglass materials started to replace the phenolic/paper PC board substrates. The heat from the tubes and the large heavy components tended to cause cracked solder joints and delaminated copper foil on early boards.

[Alex Eisenhut]

Fascinating stuff. I was under the impression that they used 60Hz high voltage supplies back then, I didn't realize the horizontal flyback idea was a "thing" already.

[N2IXK]

60Hz HV supplies were standard in prewar sets. Postwar, the flyback HV system was pretty much universal on magnetic deflection sets. The low cost small screen electrostatic sets either retained the 60Hz "brute force" HV supplies (similar to an oscilloscope of the day), or had a separate RF oscillator/air core transformer/rectifier setup to develop the accelerating potential for the CRT.

[N2IXK]

Back to the work...

The next thing I did was to make a couple resistance checks to rule out an open flyback transformer or vertical output transformer. The flyback is simple to check, with a simple resistance measurement between the plate cap leads of the 6BG6 horizontal output tube and the 1B3 HV rectifier tube. This portion of the winding measured spot on the 240 ohms called out on the schematic, and the other windings checked out as well. Same for the vertical output transformer, and the deflection yoke windings. I like to check these major components (and the CRT) at this point before putting a lot of time and new components into a set, just to discover a "showstopper" in the form of a defective part which is not easily available.  For this particular set, I have a fairly complete parts chassis kicking around, so even a bad flyback wouldn't be that big of a problem.

Another common failure point on RCA-type designs of this era are the large power resistors used in the LV power supply voltage divider network.  These resistors are mounted inside a sheet metal box on the rear chassis apron, with openings at top and bottom to promote convection cooling. Opening the resistor box found everything physically pristine, but both sections of the large wirewound resistor R153 were found to be open.  I suspect that this is a corrosion induced type of failure from decades in storage, rather than a burnout from excessive dissipation, as no shorts were seen on any of the power supply rails. I will save the failed resistor and do a teardown/failure analysis later. This resistor had already been grabbed from my parts donor chassis for a previous project, so modern replacements were called for. Because the 2 sections of the resistor were both odd values (1125 ohms and 610 ohms), I used 25W adjustable resistors from Ohmite (210 series). The units I installed were 1250 and 750 ohms total, and I adjusted the metal slider bands to get the precise values I needed.  Both new resistors stacked nicely end to end on the original mounting bolt, using the new washers that came with them. I put some heatshrink over the unused resistor lugs to prevent shorts. Ended up with a very clean installation.  Before and after pics below. Fortunately, the lower "candohm" resistor (R154) in the box checked good on all sections. These are a bit more difficult to replace when defective. I usually use metal jacketed chassis mount types here when needed.

Next on the agenda were the electrolytic caps.  Normally, I might try installing a rectifier and bringing the power up very slowly in hopes of "reforming" the electrolytics at least long enough to see if the set shows any signs of life. But a close look at the caps in this set showed definite signs of crusty electrolyte leakage from a few of them, so I just dove in and began replacing them all. There are 13 electrolytic caps in the set in total, with 12 of them housed inside 5 "twist lock" chassis mounted aluminum cans.  Each can houses 2 or 3 sections, with the common negative terminals connected to the metal can itself. In most sets, the can is connected directly to the chassis for a ground connection. In this set, however, none of the negative terminals are actually at chassis ground potential, with most being at -85V with respect to the chassis. They are mounted on phenolic insulating plates, and covered with cardboard insulating sleeves to prevent a shock hazard.

You have several options when replacing twist lock capacitors in vintage gear.  A very few values are still being manufactured, mostly ones used in vintage tube stereo gear. They cost $$$, though, and weren't available in the values I needed here. Another option is to disconnect the old cans completely, and tack in newer radial electrolytics under the chassis as needed. I tend to avoid this as it looks messy, and can get quite cramped in some areas.

The standard technique is to remove the old can capacitor, uncrimp the base, pull out the old guts, and install the new capacitors inside the old can before recrimping and reinstalling it. This gives an original appearance when the cans are exposed above the chassis, but is very labor intensive.

Because the aluminum cans are covered by the cardboard sleeves on this set, It allows me to save a LOT of time by not uncrimping and recrimping the cans and pulling out the old capacitor sections, but simply cutting them open above the base crimp, and discarding the upper portion of the can and the old contents. The new capacitors get installed in the old can bases, with the leads threaded through holes drilled in the base wafers, and soldered to the lugs under the chassis. The cardboard sleeves can then be reinstalled over the new caps, hiding them completely.

I begin by unsoldering all the connections under the chassis, making notes of where all the wires go. then the mounting lugs can be straightened out and the cap carefully wiggled free of the chassis, being careful not to break the phenolic insulator. Once the cap is loose, I roll it firmly with my palm on a hard surface, to break the glue bond (some kind of varnish) between the cardboard sleeve and the aluminum can. The sleeve will then easily slide off in one undamaged piece, and gets set aside for reinstallation later.

I use a Dremel tool and a fine toothed circular saw blade to cut all the way around the aluminum can just above the base crimp. Then the upper section of the can can be pried away from the base, and the internal connections cut off flush with the inside of the base.  I clean out any loose residue inside the base at this point, then drill a 1/16" hole next to each lug and an additional hole next to one of the mounting ears for the ground connection.

I use Nichicon or Rubycon 105C rated radial electrolytics to replace the old caps. No need for ultra low ESR caps in old gear, but sometimes they come in tall skinny form factors which can help fit multiple caps inside the old cans. I connect all the negative terminals together to a single lead that passes through the base wafer, along with all the positive leads, which connect to the corresponding lugs on the old base.

The base/capacitor assembly can then be reinstalled like a new twistlock. Make sure that the leads coming from the new caps get soldered to the lugs along with the rest of the wiring.  If the wiring under the chassis is very hard to access, it is possible to use this technique without removing the can completely from the chassis, if you have access to cut the can off above the chassis and drill the lead holes.

The pictures show what the rebuilt capacitors look like before the cardboard sleeves get put back on. There is one axial electrolytic (C160) under the chassis that also gets replaced at this point.

Next up, applying power and seeing if we have any signs of life from the set at this point....

[Seekonk]

A friend of mine was head of bio medical engineering at a hospital in the early 70's.  He would collect old tv's that had electrostatic deflection (like a scope) and feed a little bit of video into the horizontal.  That would make video of blood vessels or anything else just seem to pop out of the screen. That was image processing before the age of digital!

[Yago]

Thanks for posting, interesting reading

Are there tricks to removing a cap can intact as you mentioned?

[N2IXK]

No real "tricks" to removing twistlock cans intact. The biggest thing to look out for if to make sure that ALL the bits of wire are removed from the mounting lugs, along with as much solder as can be removed with desoldering braid or a vacuum desoldering tool. The carefully twist the lugs to straighten them out, and start working the capacitor out of the chassis. If it is tight coming out, reheat the solder residue on the mounting lugs while wiggling. Work slowly, and they will come out.

One other potential problem is when caps are mounted directly on a metal chassis, with the mounting lugs soldered down to the chassis.  You will need a SERIOUS soldering iron to unsolder these, because the chassis will soak away a lot of heat. This is where the old 200W soldering guns are great to have around.

[Yago]

Sorry, I was not clear with my question.

What I was asking is how to remove the can on a large old electrolytic, in order to preserve the can, not the capacitor, any tricks of the trade?

I have an old guitar amp that needs work.
Whilst I am completely unfamiliar with valve stuff, I could replace caps, taking care to keep the look of the original can. (leave the dodgy stuff to someone who knows or can teach).

Trying to avoid the "forth capacitor-can destroyed learning the process of removing it"!

Thanks again for taking time, wonderful stuff are old televisions

[N2IXK]

Phil Nelson has a great writeup on the procedure (with lots of pictures) here:

http://www.antiqueradio.org/DuMontRA-103Television.htm#restuffelectro

[Yago]

Thank you kindly sir

I'll be checking your posts now for more of this lovely old television projects and info.

[Alex Eisenhut]

Oh if you need tubes (except picture!) I inherited a TV repairman's tube box years ago. Mostly stuff from the 60s and I never did an inventory. But I can look if you have a p/n in mind.

[N2IXK]

Got back on the set for a bit today.  Soldered a standard choke across the disconnected speaker field coil connections to give some filtering. Stuck a solid state rectifier plug (pair of 1N5408s) into the 5U4 socket so I get rectification at low voltages without having to warm up a filament. All other tubes out of the set. Connected voltmeter from rectifier cathode to chassis ground. Connect to variac and apply power.

Start slowly running the line voltage up, until I get about 50V or so out of the rectifier. Take a couple quick readings on the +225, +150 and -85V buses, and verify that voltages are coming up OK, and there are no obvious smells or signs of anything overheating. Looks good so far.  Slowly raise the input voltage until I get close to the specified 280V at the rectifier, and check all the buses again. Things look good, The voltages are a bit high, but practically no current is being drawn at this point. None of the new caps are getting hot or bulging, so I guess I got them all in OK.

Power down, replace the solid state diodes with a good old 5U4GB from the junkbox, and apply power from wherever the variac was set. Voltages come up quite a bit lower, as to be expected with the much higher voltage drop of a vacuum tube vs a silicon. Goose the variac up to reestablish the 280V at the 5U4 cathode.  Let the set cook for a little bit and then power down and give one last check for overheating parts or weird smells. 

Next step I put in all the rest of the tubes into their sockets EXCEPT for the 5V4 damper, 6BG6 horizontal output, and 1B3 HV rectifier.

Apply power again while monitoring the +225 rail. Voltage initially goes to 250V or so (5U4 has quick heating filament), then gets sucked down below 200Vas the other tubes come up to temperature and start drawing plate current. Quick check on the -85V rail shows about -60V, which is fine as the set is still pretty lightly loaded at this point, and the voltage divider network is kind of unbalanced without full load on all rails. Goose up variac again to +280 at the 5U4 socket, and the input voltage is at 105VAC or so.  Looking good. Sticking a screwdriver into the focus coil shows a decent magnetic field, which adjusts with the focus pot.

Powered the set down, and installed the 5V4 damper tube. Applied power again and checked the DC voltage at the disconnected 6BG6 plate lead. Had about 260VDC, which seems fine.

Next up, I will attempt to fire up the horizontal sweep and HV stages and pop a CRT into the set as a quick evaluation before replacing all the waxed paper caps under the chassis. Not much in the way of pictures today, mostly a lot of meter readings...

[N2IXK]

The power-up continues...

 Checked the screen, grid and cathode voltages on the 6BG6 socket, and everything looked "safe". Hooked my scope to the grid, and got a reasonable-looking but low 30Vpp drive signal (service data says should be ~40 Vpp).  Things are looking pretty good, considering all the original paper caps are still in!

Shut down power, installed a 6BG6, and hung my DMM on the +225V rail. Made sure the 1B3 lead was in the clear, and applied power. Took a neon test probe (NE-2 bulb taped to a plastic stick), and held it near the 6BG6 plate cap. Bulb lit fairly strongly. Waving it near the 1B3 plate lead, it got much brighter.  Lots of high voltage RF being made.  No signs of arcing/corona/smoke from flyback.  No glowing plate or funny colors inside the 6BG6. Looking even better...

Stuck in a 1B3, and connected HV meter (Fluke 87 and 80K-40 HV probe) to CRT anode clip. Applied power and....whole lot of nothing.  About 4V if the meter can be trusted that low (.004 kV was the reading). Bringing back of hand near end of anode wire gave no static to hairs, and bringing a grounded screwdriver near it gave nothing at all. Bringing the screwdriver (now ungrounded) near the 1B3 plate cap (briefly) drew a 1/2" long RF arc. We have HV AC, but no DC.   

The 1B3 I have in it checks good (on a TV-10D/u) but it has a internal insulating coating that obscures the filament, so I can't see the filament glow.  Will have to dig up another one to try, I'm sure I have a few of them kicking around. 

Only other possibilities I see would be an open resistor under the 1B3 socket (R187 or R167) (somewhat common), or a shorted HV filter cap (C142, but these mica or ceramic "doorknob"caps are usually pretty reliable).

At this point, I'm going to go ahead and swap the 1B3 to rule it out, and then forge ahead on the recapping. Have all the needed caps ready to go in.   Would have liked to get some light on a CRT, but probably pushing my luck running this thing much at this point. The 1B3 socket resistors and the HV cap all require at least some disassembly of the HV cage to get at, so will investigate this area after doing all the underchassis work.

[T3sl4co1l]

Thinking the 1B3 isn't getting heat.

Tim

[N2IXK]

Swapped in a NOS 1B3GT tube from my stash, but still no DC at the CRT anode connector.

After a bit of thinking, I realized that I could check both suspect resistors without disassembly.   Checking resistance between pins 2 and 7 of the 1B3 socket with the tube out shows about 2M ohms, where there should be about 3.3 ohms. R187 *must* be open (or a wire is broken at the socket), as the filament winding on the flyback is just a single loop of wire and not at all likely to burn out. An open resistor here will kill filament power to the 1B3, and prevent any HV output. Checking the resistance from the CRT anode lead to pin 7 on the 1B3 socket shows about 1.8M, so R167 has drifted high, as well. Both will be replaced before the next power-up attempt.  I wonder if this failure is what brought about the retirement of this set years ago?

Anyway, on with the recapping....

[MickM]

Hi;
 Bob Andersen does a lot of 1940's TV restoration.
I have been watching him for a long time.
here is his channel.
https://www.youtube.com/user/bandersentv

He does a lot of capacitor rebuids.

Mick M

[N2IXK]

Another subscriber to BandersenTV here! 

In fact, he has a video up about the same model I'm working on here. He only did the first installment as of this point, though...

[N2IXK]

Spent a few hours replacing all the waxed paper caps under the chassis with axial lead film caps. Nothing complicated here, just methodical one at a time desolder, clean up the terminals, then install the new caps. This set is old enough to use values like .05 and .002 uF capacitors, which were simply replaced with modern caps of .047 or .0022 as needed. All the caps installed were rated for 630VDC, except for a couple in the horizontal output/damper circuit, which needed a 1000V rating. The RF bypass caps on the AC line input were replaced with 250VAC rated line bypass type ceramics. I tried to duplicate the original placement and lead dress as closely as possible, and added insulating tubing over the caps leads where needed. The new caps were all much more compact than the old ones, and the underside is a bit less crowded with all the old caps replaced.

Turning my attention topside, I unmounted the flyback and width coil, and removed the HV cage assembly.  Then I removed the 1B3 socket assembly from the chassis to get at the 2 resistors under the socket.  As suspected, R187 checked open, until i wiggled the leads a bit, then it would make contact intermittently. The molded case was cracked around one of the leads, allowing it to move. Both resistors under the socket were replaced with 1W metal film resistors. Care needs to be taken when working in this area to make only smooth, rounded solder connections to prevent corona discharge. This entire area operates at around 8500V above chassis ground. I carefully cleaned away the dust and flux residue, and gave the area a light spray of insulating lacquer before reassembly.

I powered the chassis up again, and got almost 9kVDC at the anode connector. A bit higher than the specified 8.5 kV, but there is no CRT installed to load the supply down at this point.

Let the set play for a bit and verified that nothing was overheating or acting strangely.  Now it is time to see what we get in way of an image....

Many older sets (including this one) support the CRT from the front of the cabinet. Operating the chassis on the bench with the 10BP4 installed can be done, but it requires careful balancing of the fragile tube and a serious risk of breaking the tube if it slips from its temporary mounting. It is also very difficult to probe around UNDER the chassis with the regular tube installed.  Fortunately, there were special ruggedized test CRTs made for service shop use that would work in a wide range of sets, and allow the customer's CRT to be left with the cabinet in the home, and pull only the chassis into the shop for service. The tube I am using here is a Sylvania 5AXP4, which is a universal test CRT for early 50 degree deflection angle sets. One of these is a GREAT timesaver if you work on old TVs.

Because the 5AXP4 is self-focusing and doesn't need an ion trap magnet, the ion trap and focus coil were carefully dismounted and set aside. The tube was installed, and held in place with a large nylon cable tie around the neck.  Applied power and got a raster of sorts.    A bit of adjustment to the height and width controls, and I was able to get full deflection and a good bright raster.  High voltage tests at about 8.2 kV. at full brightness. Tried tuning in an image from a signal generator, but nothing comes through. working the channel selector produces no flashes on the screen, indicating that no signals are coming through the video IF stages or video detector.  I still have no speaker connected, so no idea about sound at this point.   But the power supply, sweep circuits and CRT bias circuits all seem to be OK. Time to take a rest for the day and plot strategy from this point forward. May try injecting a composite video signal after the detector, to see if the video amps and sync circuits are all functional. Stay tuned for the next installment...

[N2IXK]

OK, got back on the 721TS again after a short break.  Since I couldn't yet get any kind of signal through the tuner and IF amps, I decided to inject a video signal right at the detector output, to see if the video amps and sync circuits were functional.  Lifted the detector end of C103, and connected the free end to a standard composite video signal (1Vpp into 75 ohms).  Firing the set up , I got some video modulation visible on the screen, and vertical sync seemed to lock in well. Horizontal would simply not lock, and frequency appeared to be a bit too high, even at the extreme end of the horizontal frequency and hold adjustments.  Sure enough, although the waveform itself looks good, the frequency checks at over 18 kHz, rather than the 15.734 kHz it should be. Tried a couple different 6SN7s, with no change.  Just for a quick check, I temporarily hung an additional 150 pF cap across the horizontal freq. trimmer (C136C), which dropped the oscillator frequency enough to get a stable image.  Guess it's time to start going through the horizontal oscillator and synchrolock circuits looking for components that have drifted out of tolerance. Could be a carbon comp resistor or a leaky mica cap shifting the frequency on me.

Picture looks ok (linearity needs adjustment) with the added cap, but width seems a touch low. Remembering the added cap that was installed across the width control, there may be something else at play here, as well.  Time to start checking resistors and other parts at this point...