So, the other night we sat down to watch some Dr. Who (final stretch of the Key To Time arc in the Tom Baker era) and lo and behold, the television, a Sony Bravia KDL-32M3000, was broken.

Neil nails it at 0:37. "No more telly.."

Actually, the symptoms were highly diminished brightness with a murky red hue, the images on the screen would essentially pulse: movement by an actor would be jerky, as if watching them through a strobe light with a sort of blur between frames. It was a lot like REALLY bad animation, running dark and murky reddish images at roughly 4 frames per second. It was, put simply, unacceptable.

Suspicious that it might be the source, I went to change inputs, only to observe the sluggish response with the menu windows as well, certainly not a source problem.

Since I'm not really super keen on the whole disposable culture thing, running out and buying a new TV isn't really high on the list. In searching the model family I learned this is a very common failure mode. I pretty much followed the following tutorial, and figure I may do someone a favor with clear and legible pictures.

I took to muting the sound, as there is no narration..

I had initially planned on testing voltages between these two points, but faced with the level of disassembly I decided to just go for it. The LCD was bought new in 2007, easing my apprehension a little.. sort of a nothing to lose now project.

All those circuit boards hanging off the back? They need to be removed. Disassembly down to the LCD section is a few layers deep, I took some pictures along the way in case I had to reference anything on reassembly. The entire process was not to bad, but I've certainly been inside simpler devices.

10 ohms, tied into wire runs each end and about to be encapsulated with yellow heat shrink.

The offending ribbon connector is the largest PITA of this process. After gingerly scuffing the insulation layer with some sandpaper I folded it up and secured it with some 3/4" blue masking tape so I could solder with impunity. A bit of liquid flux was added to the partially scuffed pad to assist soldering (I feared going deeper due to the incredibly small traces, so once I visually observed a bit of metal I stopped).

With the pre-tinned wire I essentially placed the wire over the pad and directed heat from the top side. Took the better part of a second at 600f. Once that contact was made it was returned to its relaxed position and secured with tape.

The actual voltage feed I want is on the left, but the pads are close & small enough that my tinning one incurred a little crosstalk to the other.

Not what I would call a structural solder joint, that's what the tape & shroud is for..

Overall view.

While I had it apart I took a moment to admire the elliptical full range drivers that are stock in this thing. Not that the driver struck me as magnificent, just that the shape is neat enough to imagine several dozen of these in a line array configuration. No expectations...

Rainbow artifacts aside, the picture looks as good as new.

It's been a while since I've pounced on an impulse buy, being that I'm essentially hovering in a state of critical mass I've pretty much stopped frequenting the places I run across this sort of thing. That said, I needed 100 feet of ethernet cable, so I made the trip to Next Step.

While I was there, I couldn't resist perusing the piles..

I struck out on the ethernet cable, but wound up scoring a heavy 2U modular video rack processor by Datatek. In poking about the internet I find reference to it being a D701 system, but no where on the device is a badge confirming this, and since eBay sellers are not the end all of cold hard fact I'll leave it somewhat open. Someone has marked KEVU 34 on the underside, call signs date to 1991-1997. I think I may have watched afternoon cartoons through this thing.

Anyway, let's have a closer peek at these modules. Left bank contains, top to bottom, left to right: D706 Leading Correction Generator, D709 Trailing Axis Corrector, D702 System Input & Power Supply, D703 Transmitter Delay Corrector. Right bank: D707 Trailing Correction Generator 1, D708 Trailing Correction Generator 2, D704 Receiver Delay Corrector, D705 Video Low Pass Filter.

D706. All the correction generators read similar to a graphic EQ, with a time based delay as opposed to a frequency. Each time base is variable on a +/- scale adjusted by a pot as opposed to steps of a switch. Without looking at this thing on a scope or finding operation manuals I can only speculate what function this serves. I'm guessing video sync alignment.

Gut shot of the D706 reveals a nice early potted integrated circuit. We're peeking into roughly 1974 here near as I can figure it.

D709. The trailing axis corrector strikes me as the module with the most potential. Perhaps it is the generous front panel patching. Perhaps it is the word "AXIS". I don't know, in skimming over the vast array of elements that comprise an NTSC analog signal, it could sink into numerous slots.

D709 guts. Fully discrete topology.

D702. Power section, with gain control! The power switch to the left is clearly missing the lamp cap, though it seems to actuate just fine. The bypass lit plunger was very loose and required a bit of tightening.

Not much more to say here, D702 gut shot. Does appear to be of more recent manufacture, but I neglected to pull date codes from it to compare.

D703.

D703 really doesn't appear to support anything other than set & forget operation.

D707. The trailing correction generators look similar in function to D706. D708 is also similar, but where D707 has +120, +240, +360, +480, +600 & +720, D708 carries on into +840, +960, +1080, +1200, +1320 & +1440/SPL.

Hey, this looks familiar. D708 is similar.

D704. Another set and forget module.

I really like the wandering line of the ground plane to unpopulated fiberglass. I'm also impressed by the flying wire jumper geometry.

Because no post would be complete without some borderline photography, here's the D705. Another set & forget (I think calibrate is the preferred nomenclature) though honestly this is one function I'd rather have hands on with.

33 adjustments does seem a bit much, how's about one big knob?

Here's a shot of the module back plane, relay bank & power supply. Foreground is the I/O. Speaking of big knobs, there's another that may do better on the front of this unit. Since my video processing is starting to populate my rightmost open rack I'll overlook the nuisance factor.

See, I remain fascinated with video synthesis and a device such as this seems a good "block" to have on hand. If it turns out it does nothing interesting I will have a short module frame enclosure for a future build.

Tropical fish capacitors! Tone zombies friend. These look dessicated, and to some extent falling apart. I'll file that into the "look for" list in the event of magic smoke, I really don't want to mess with any potential mojo prior to discovery. Speaking of mojo, I need some more nondescript milky white through hole parts on hand.

Fast forward 20 years and this assembler will be circuit bending Speak & Spells.

Pretty cryptic in terms of a part number. Not that I think I'd have a lot of luck finding a replacement even if I did know what I was looking for.

Clean heat sink & oxidized legs; oxidized heat sink & clean legs. What gives?

The AC mains fuse housing (5A!) was also very loose in the chassis, and the W/F Correction switch on the rear panel is in need of replacement. I'm hoping these are signs of simple nuisance/post decommission storage type failures as opposed to catastrophic BBQ. Remains to be seen, may need to freshen up those relay contacts.

From 4 to 16 ohms.

About as simple as rewiring gets.

For ease of installation I started to set my Weller WES50 soldering station on top of one of the speaker magnets. To my dismay I noticed the indicator LED was going bonkers, and having just recently lost a meter I was not optimistic when I pulled it out and saw a red indicator staring back at me. Temp was still up, so I pulled and switched the wires.

As it turns out, the magnetic field of the driver triggered the temperature lockout setting, which is of course reversible. Funny that feature has remain hidden to me for a decade.

I purchased a Sony MX-6S a while back & seeing as how I queuing up to play a somewhat stripped down minimalist set last Sunday I found it high time to bring it into the realm of usefulness.

The MX-6S is a passive resistance mixer originally bearing three RCA connectors feeding the AUX channels & three 1/8^th inch monophonic miniature jacks feeding the MIC channels. The center MIC/AUX channel is switchable to the destination of output channel 1 or 2, which took the form of ~2 foot pigtails terminating in 1/8^th inch plugs.

I can begrudgingly support both RCA & 1/8^th inch miniature in the studio; pigtails, however, have to go & since I've going to be in there already...

Removal of existing output wires is as simple as desoldering the points on this circuit board (upper grey wire shown here is channel 2 output, whereas channel 1 was connected to the fifth & third notch at the lower right of PCB) and pulling the wire out. Pulling the wire was further simplified through studied application of snips.

Rather than destroy the 3x RCA jacks to harvest the phenolic mounting board, I fabricated a new one out of mild steel to mount the 1/4 inch phone jacks, the original MIC connectors simply unscrewed to allow drilling the phenolic out to accept the larger jack. Further drilling was required at outputs. The MIC inputs are a rat's whisker away from an interference fit with the switch bracket, this probably wouldn't have been a problem if I hadn't run out of open frame jacks.

Output is further modified with a switching jack that will route channel 2 output to channel 1 if a cable is not plugged into channel 2.

Much better. The two input banks are different from one another: AUX is set up for line level with a 100K log as volume, MIC was originally terminated across a 560 ohm resistor feeding a 5K log as volume. This is equivalent to different gain structures when fed by something that is not loaded down on the MIC input (I opted to not reinstall the 560 ohm load resistor) in my very limited experimentation I find the higher impedance inputs far easier to control.

Worked like a champ, I expect it will see a lot more use now.

In the days since I updated last I have brought this beast on line.

It has not been without struggle mind you.

Phase 1: rebuilt bias dropping circuit to reflect values close to those originally published for the 7027A configuration of the amplifier (130K dropping with a 15K straddling the bias filtering caps, 10K trim on top of a 56.1K to ground). This is pretty much where I was at end of April. Netted a nominal 4 ma idle current on three of the tubes, whereas one read no current. I was busy checking voltages on the no current tube when my meter gave up the ghost (shunting enough voltage to ground to increase at the wall current consumption by a half amp). Meter no longer trusted for voltage work, everything is shelved in await for a replacement.

Phase 2: knowing that 4 ma idle current is indicative of an over biased situation, regardless of potential wiring error on my part, I remove and replace the 130K with a 150K per later revision. In doing this work the tubes are removed and replaced. All four tubes JJ E34L, one pair had been in a red plate runaway situation and was easily identifiable by the darkened paint on the envelope; darkened pair inner, non-abused pair outer - as I've always been installing them.

Phase 3: armed now with a new meter I ease the amp up and read nominal idle current in the 20 ma range, except at the one outer socket - which has switched sides......

Break out tube tester and confirm that one of the non-abused E34L is in fact useless. Dig through the stacks to come up with a handful of used EL34 & 6CA7.

Phase 4: the next value of 2 watt resistor I have is banded 170K and reads at 200K, all three 170K resistors read similar. Fire the amp up and read in the 50s for the E34L and lower 60s for a pair of Sovtek EL34WXT I have sitting in the outer sockets. No red plate at idle just yet, though I realize this is ranging them a bit hot. I plug my Tektronix RM503 into the extension cab socket (modified to be wired in parallel to the main output, not a switched series stack) to have a peek at waveform while the 4 ohm dummy load dissipates power and observe a tinge of orange on the Sovteks. Pull the meter out and read well over 100 ma. Shut down the amp and pull the Sovteks, which tested high on transconductance compared to the E34L I already have installed (which have served as the baseline changes to the amp over all).

Phase 5: find a 174K 1% of sufficient girth and swap that in to main bias dropping resistor slot. Test out the pile of tubes for something closer to the level of the E34L. Sovteks test high, the Sylvania 6CA7 is off the charts, a pair of Matsushita 6CA7 weigh in a bit lower. These are installed as outer pair. I've observed a roughly 7-10% travel in current adjustment with my 10K bias supply pots, enough to trim in an already similar quartet, but proving insufficient to balance out dissimilar tubes, which remains one of the goals.

Anyhow, with the 174K I can level the E34L out at 42 ma, while the Matsushita 6CA7 are stuck in the upper 20s. I'd rather err on the side of too cold for the moment and finish up some adjustments so I can house the amp and actually get to listening tests. This entails adjusting this phase inverter balance pot so I plumb the dummy load and o-scope into the speaker jacks and run a sine wave oscillator into the input, grab the knob and give it a twist while watching the trace on the screen.

This process is not incredibly helpful; as the adjustments only seem to have bearing on amplitude as opposed to balance of waveform, though to be fair the waveform could be out of balance and I'm just missing the baseline. I crank the volume dial up to a point where I can hear the frequency rattling the guts of the amp/dummy load (which is getting warm) and observe a clipped output. Though clipped the output still looks balanced, so I ramp the amplitude up to the point of ringing the waveform, adjust the balance pot to smooth that out - have a glance at the amp which is groaning a little under load to observe the quartet of outputs well into the orange.

Flick the power switch, which does nothing. Kill power at the variac, walk away and eat lunch.

Fire the amp up without adjusting anything, read low 40s & high 20s at idle. Plug 4 ohm speaker cab in and play a little while watching current draw. Normal use has draw ranging from the 50s to the 80s, though the amp is in no way being run wide open.

In the end the phase inverter balance was adjusted by ear (I reached for the loudest position, extreme swing on either side fell into cut-off) & the amp is buttoned up for observation while playing with standard sources. Being able to read current while playing will allow me to avoid running the amp into self destruction.

So, still not through the hazard territory, but well on the way. I need to bring B+ down to original values (my wall voltage is a touch higher than 1969s) which should push the point of runaway out even farther, and I think upscaling my adjustment pots to 20K might be worthwhile, but lower on the priority scale.