Tuesday, January 22, 2013

I drove a couple towns over and picked up a Gates Studioette M-5381 yesterday. Let's have a look.

Solid logo, I haven't yet put a fine point on internal date codes, but the manual was printed in 1962.

Only 24 inches wide, seated in my work chair. I'm going to have to make a better place for it & will follow up with a better front panel shot later.

Recessed VU meter, quite modernized styling over the RCA BC-3C from the same era.

The front panel of this unit has escaped the rough hand of modification. There are some additional holes on the sides, but for once the face is left alone (which reminds me, I really need to post pictures of the basket case RCA).

This is where the serial would go these days, curious if there are identifying marks inside that would point to record with the manufacturer (not that it would do me any good).

GUTS. Hinged front is the control surface. Moving into the chassis, lower left is a pair of microphone preamplifiers, each with input and output transformer (albeit somewhat puny ones for the era, no need for 5 hz on AM radio, thanks) and room for one more. Shifting right we see a solitary UTC A-21 doing duty on line level inputs, keeping those voltage regulation tubes company. Continuing on to the right there is the program amplifier with three pentodes and a dual triode, the circular transformer being input and the larger rectangular transformer being output.

Shifting to the back row, at far left there is the monitor booster amplifier, 150 ohm balanced line input to 12AX7 pushing... a push pull 6BQ5 monitor amplifier which wisely has its output iron bolted to something more substantial than phenolic PC board, and on to the power supply.

Here's a closer look at the power supply, nothing unusual: mains transformer, rectifier tube, a couple of chokes. The output transformer for the amplifier stage is in close proximity, which will confine power hungry pulls on the B+ well away from high gain stages. Those two relays will probably go unused. I haven't seen a power switch on this unit yet, I will certainly be paying this section some scrutiny when update time rolls around.

Staring too long at this may make me delay diving right in. No! That's the wrong angle, this looks like meditative fun to sort and resort. Wind and dive.

Conductive plastic or mechanical fortitude? I know which direction my preferences lie.

I found the maker's mark in the Daven cap interesting enough to photograph, given the amount of scuffing (original as far as I can tell).

Apparently the #3 gap is standard from factory, designed in room for expansion. Who knows what will take place there in the future? Since I have the manual, and having roughly 52 hours notice of collecting this device spent some time fruitlessly searching for a digital copy (of which I could find none) I intend on scanning up a pdf and sharing that with the digital universe once I get a chance.

Here's a closer look at the preamps.

A truncated shot of the program amplifier (I think I'd term this summing section, but since it was named from the factory I'll show it a little respect). Observe the dirt build up from the right end..

Closer look at the monitor booster.

Yes, some clean up is in order.

...

The fan was added post factory, it would have been a good idea to put a filter on it. After seeing this I'm pretty sure the fan is getting the heave, I plan on undervolting the mains a couple percent, which should rein in the heat a bit. Honestly, having looked at the underside of the lid I don't think excess heat was ever really an issue.

Exposed tabs! SO much nicer than having to work through a punch hole in the chassis, this one's a keeper for sure.

Sunday, January 20, 2013

I was recapping my Roland SMX-880 a while back and got hung up on the fact that I hadn't observed the fact that the 10uf electrolytic at input jack was a bipolar. Quick test confirmed that this cap is situated in line with the signal, it is the first thing the signal hits.

I wasn't about to put some polarized cap there, so the unit was set aside to await a parts order. In the meantime I read a bit of opinion on the internet I found agreeable: "The best sounding capacitor is a piece of wire".

Brilliant. Upper right corner, I hadn't any plans to subject it to DC anyway.

Saturday, January 19, 2013

Where was I? Oh yes, I had polished the legs of all analog ICs in channel 8, reinstalled them, and then observed function. Success!

So naturally at this point I disinter all the patina laden ICs in channels 5 through 7, clean them up, reseat chips, reinstall channel cards and apply power.

Nothing. Well, unless you count the full meter deflection of 5-8; audio certainly isn't passing through any of these.

That includes channel 8, of course, which, by my own observation was just passing the first tier of functional test and literally has not been touched since.

Pull card 5 (again, if I was thinking clearly and weighing out all the evidence this step would have been reduced to one selection, but I was at this point a touch flustered). Power up, 6-8 pulse the VU. Power down.

Leave 5 out, pull 6 & power up; 7 & 8 pulse the VU, power down.

Pull 7 & power up, channel 8 shows a minor bump on the VU and now passes sound. Power down, reinstall 5.

Power up, channel 5 shows a minor bump on the VU whereas channel 8 does not deflect at all, both channels pass audio. Power down and install 6, which completes the pattern as expected.

Channel 7 contains a problem that was pulling down channels 5-8. Having my quarry fixed, I now begin to puzzle this out logically, and based on the power rail observation made previous I begin with measuring the rails for shorts. -15 volts to common measures 0.8 ohms.

Ah, the dogs, they have your scent. All integrated circuits are pulled, the short remains. Of course it wasn't going to be that easy. I pore over the schematic for all points tying directly to the negative rail, and choosing for now to ignore resistors narrow my focus to a dozen or so parts and begin to pull them out, one at a time to determine which part is failed.

Next time, I'm going to start with the caps. The dense two sided PCB has made me out for a bungling idiot, and before I actually pull any traces that are important (this was just an insert) I shift my method to cutting.

Which, while destructive, limits the destruction to parts at hand. I'm still exacting far more destruction than I feel I should on this simple task. This particular component was just fine before I came along, WOOF.

Having ruled out transistors I then focus on capacitors. C4 here is the problem, a Matsushita 1 uf @ 35 volts, it's marked as being polarized. Tantalum? Odd package, I can get those values in a similar dimension in ceramic, which is my plan.

Of course, I threaded up a reel of tape and learned my record levels have been killed (previously 3 & 4 had recorded well), now everything reads high to tape and plays back low, with the exception of previously recorded material. So now I chase a common point in all channel recording. I'm thinking bias as a starting point, have mercy.

Friday, January 11, 2013

Having temporarily ruled out a logic problem (which seems logical in retrospect, since I have observed the status LEDs appear to be driven from the state of the channel card itself and not the logic commanding the card), it was back to troubleshooting the machine as a complete unit.

While I had all the cards pulled I naturally brushed out cobwebs and hit the edge connectors with contact cleaner, so it was only logical to fully reassemble the unit and see if a good cleaning did the trick. No change.

Next step was to narrow down point of focus. For brevity sake I'm just going to observe the behaviour of the VU meter on power up and power down, since in consulting a friend who has a very functional version of the same machine I was informed that meter deflection of any sort in response to the on/off switch was unusual. So I swapped cards 4 & 5.

No change in 5-8 deflection at power up, though now #4 was pegging the channel 5 meter as a steady state while the natives along 6-8 returned to zero. Quickly power down & confirm card was seated correctly. I pull channel 4 card & channel 6 card to compare, they are, for all intents & purposes identical (beyond the fact that they carry factory stickers indicating their home channel).

I then install the channel 4 card into channel 6 slot, placing channel 6 card into channel 5 slot. 5-8 full deflection tap and now channel 6 is pegged. Power down.

While channels 1-4 have not been considered suspect, due to them actually functioning (and bolstered by the fact that channel 4 has in fact recorded to tape and passed the listening test, which I would like to think is pretty stringent). I then swap Channel 3 with #4 in 6. 5-8 full deflection tap and channel 6 remains pegged. Power down.

At this point I overlook a step that could have saved me a little time and effort. It should be obvious that I am not a factory trained technician, nor am I intimately familiar with the workings of this machine (getting there). I'm pretty much working in the dark with my empirical approach and a set of freely downloaded schematics.

Make no mistake, I am brimming over with a sense of gratitude toward the good people at AnalogRules.com, for without these highly useful documents I would quite simply be fucked. However, there came a point at which deep circuit analysis turned into a hide and seek sort of affair due to a lack of clarity in the scan. So yeah, I'm pretty much just making this process up as I go along.

So, my next step was to remove all the cards, set 1-4 aside as known functional and install 5-8 into the 1-4 slots. Anyone care to guess what happened?

Nothing, unless you count 1-4 meters deflecting fully on power up. I run signal test to confirm a complete lack of function.

Based on this observation (and brownie points to whoever catches my mistakes here) I shift focus on channel cards 5-8 themselves, and ignore for the moment the steady pegging of the low number cards in high number slots (another hint). I had deemed the probability of discrete failures on the four cards, right next to each other, as highly unlikely, but it's worth a look anyway.

All my analog ICs are looking a touched charred (this batch is among the cleaner looking). Granted, there's patina on the analog ICs in channel 1-4 too, but I'm not paying them much scrutiny, as they are not problematic. Incidentally, all the digital ICs are looking quite fresh & nickel plated, so it's a metallurgy thing, perhaps there's a high content of silver in the legs. Whatever.

So, channel 8 is my guinea pig, I've plucked the chips and go to work.

600 grit sandpaper and a little time, before and after shot here.

The ICs cleaned up pretty well, flushed the sockets with contact cleaner, reseated the ICs into channel 8 card and sink that channel into channel 8 slot - with no other cards with the exception of the logic control. Power the machine up. Minor deflection on channel 8. The card passes sound!

It is at this point that, flushed with a sense of progress and ready for a break I happily report to my magnificent and wonderfully patient wife that I've finally secured what appears to be a solid lead amidst a convoluted mess of red herrings. HAHAHAHA, oh the humanity.

Wednesday, January 09, 2013

So back to the Otari MX-5050 mkIII 8. Symptoms as follows:
On power up meters 5-8 deflect fully then return to zero.
On power down meters 1-4 deflect fully then return to zero.
Setting monitor to all input, audio passes through channels 1-4 while 5-8 has nothing.

So, my thinking is naturally focused on the split point between channels 4 & 5. With that in mind, and based on the observation that a lot of the CMOS gates used in logic direction are four per IC I started my analysis there.

As you can plainly see, that angle was not fruitful. There is not a single instance of component pairs straddling the 1-4/5-8 configuration. With that, I cracked the case and began to pull cards.

Here's the control card to go along with the (partial) schematic above. The fact of the matter is that this card does not have the same part number as the schematic I've been working from. I settle into the idea that I may have to resort to a trace by trace comparison to see if I can narrow down a culprit IC, but since I'm pulling cards now I may as well go for broke and thoroughly give everything a visual inspection.

So far all the cards & traces look good. I know, that just rules out the obvious...

...having been bit in the ass by the obvious after chasing the esoteric a few times too many that's where I like to begin.

While I have the motherboard exposed like this I do a continuity check on my pin 2/3 swap on the inputs and outputs of both channels. Everything measures up well.

Without really getting into photography it's difficult to demonstrate the look of backlit traces. Let's just say that evidence of pulls, shears or cracks do not exist. Everything passes on the visual level. One thing I do notice, is that the +/-15 volt supply rails (brightened in the picture) just span 1-4 and 5-8 (with a contiguous ground reference). This is the first point of a well defined 1-4 versus 5-8 I have observed on this machine, and it may prove useful later. But for now I simply apply power and confirm that the supply rails are delivering at all locations.

Next up, chasing a series of red herrings..