Manufacturing Errors

[PJC58-Hythe]

I'm interested to know if anyone else has come across manufacturing errors in equipment.

I was working on a Murphy A482 this morning that had stopped working. After a fresh set of valves and the usual suspect capacitors and out of tolerance resistors were replaced I discovered it worked OK on AM but when VHF was selected V2 (first IF on VHF & Mixer/ Osc on AM) went into oscillation. After poking a around with a meter I went to measure the voltage on Pin 1 only to discover that C23 had never been fitted as there was definitely no sign of one having ever been soldered in. I fitted C23 and now the set works without any issues, I'm not quite sure how it worked in the first place but it clearly did as the customer had been using it for a couple of years.

[G6Tanuki]

One of the coil connections in the local oscillator of my Eddystone 840A was open circuit, though there was solder round the connection the coil winding 'tail' had not been stripped of the enamel. So no continuity, no oscillations, and one of the four wavebands could never have worked even when they were trying to align it at the factory.

Might explain why I got it so cheaply.

A few minutes work to take the coil out of the coil box and properly strip the enamel off the end, solder and re install the coil and all was working again.

Quality Control?? Don't make me laugh.

[GrimJosef]

Similar to G6Tanuki's - my AVO CT-160 had a wire passed through one of the valve socket tags, but the solder joint was dry. I suppose there might have been contact at some point - before age had oxidised the wire and/or the tag - in which case it could have passed a QC test before leaving the factory. It was one of the less commonly used sockets (a 4-pin one, I think) but it had me scratching my head when it failed a known good valve.

Cheers,

GJ

[Welsh Anorak]

I remember a Hitachi CRT TV that came straight back from being sold with it not going into standby. True, it didn't. Hitachi technical were baffled as we were! Eventually after a good while on the bench I discovered a resistor that was completely the wrong value - I mean completely the wrong part. Changing it provided a cure.

The TLO said that couldn't happen as it was auto-inserted from a bandolier and if it had jumped out of sync the whole chassis would have been wrong! But it did...

One reason I often buy graded goods is that they will have undergone much more stringent tests than done at the factory - and they're still guaranteed and cheaper.

[chriswood1900]

Dry joints are very common and I have found quite a few as described above over the years in AVO and similar equipment. Other faults have included capacitors reversed compared with the circuit diagram or PCB, I always check before removing polarised caps.
Also missing bits from circuit diagrams, links added after design and changes in value of resistors not updated in the documentation . Finally the wrong resistor loaded in the pick and place machine but close enough to work to get it out the door!

[Cobaltblue]

Its surprising the number you see in even pre-war sets although its probably less than 1 in 10,000 joints in reality.

I have come across a dozen or so in the last 45 years.

I suspect on pre-war sets there was often a good mechanical joint so the set probably worked its entire life without the joint being soldered.

An extra 70 years or so of oxidation not so much!

You wonder how many were picked up during normal servicing.

Cheers

Mike T

[Philips210]

A work colleague asked me to look at a fairly new Aiwa micro HiFi unit that had a difficult to describe distortion in one channel only. It had been previously looked at by the dealer that sold the new unit to the customer. They were unable to find the fault.

I ordered the service manual and started comparing the voltage conditions in the left/right channels and found an anomaly in the pre-amp stages of the left channel. It was looking like a faulty transistor but trying a replacement made no difference. When faced with a difficult fault I always have a good look at the circuit board on both the component and solder sides. Something seemed wrong.
I did some cold resistance checks in both left/right pre-amp stages and again there were differences. Carefully checking with the circuit diagram it seemed there must be an open circuit resistor. The open circuit resistor was more than that, open in the sense that it was missing altogether! It was never fitted during production. A new resistor put the fault to bed.

Always a relief to fix an awkward fault.

Regards,
Symon

[radiomobile]

Some years ago I was asked to look at a Bang and Olufsen Beolit transistor portable. The customer was not happy with the sound quality which was indeed lacking in bass for such a prestigious set. It had been returned to the manufacturers who declared it was fine. On examination I disccovered that a coupling capacitor which should have been 220nf was in fact 22nf, not immediately obvious as the caps used were Mullard "tropical fish" types and the orange and yellow stripes were rather similar. Fitting the correct 220nf improved the bass considerably. The set was around 10 years old by then so the customer had suffered a poor quality sound for all that time

[MotorBikeLes]

I had an Advance E2 RF sig gen that would not work. Probably 30 years old, but looked unused. One half of the double triode oscillator valve had no solder on one resistor to valve socket connection.
A high end Samsung Hi-Fi unit (SCM8500). PSU voltages all wrong, and some consequent damage. The mains transformer secondary was connected 20-20-0 (if it was 20v) instead of 20-0-20.
Then there was a Crown HiFi (?) with no CD play. Supply voltages all too high. The mains transformer secondaries feeding the + and - supplies to the CD should have been roughly 8-0-8v, but were 16-0-16.. It was a dealer return we had purchased, and the best solution was to remove the CD section, fit the CD front into the case with araldite, and flog it cheaply as a radio/record player.
A few months later, my son turned up with an identical model, no CD play. It was his pal;s mother, and I was asked to try to do something. I contacted the Crown importers (was Wendy there then), explained the problem, and they had a transformer for a later model which appeared the same. It was tested for me, correct 8-0-8 secondary, I fitted it, fixed the simple CD fault, and all was well.
There must have been others, but that is enough for now.
Les.

[kevinaston1]

Years ago, an expensive digital guitar processor was returned dead. On examination, the CPU socket had never been soldered on the dual sided PTH PCB; but the chassis had stickers from multiple operatives stating it had passed their test.

A photograph was sent to the manufacturer for their comments.

What came back was "well, it could have worked"

[G6Tanuki]

I remember a piece of kit that used a relay to short the output to ground if a set of safety critical constraints were violated. The relays were fired by a SCR which would latch and keep the relays in the output-shorted state so long as the equipment was powered.

Problem arose when there were two kinds of relay, with anti-back-EMF diodes across the coil.

Depending on if you specified the A or B version the diodes were wired in different directions. The thing was specified with version A relays but version B were sometimes installed at the factory or were supplied as official maintenance parts.

Meaning that the diode provided a total short circuit round the relay coil, blowing the SCR and failing to provide the intended fail safe.

Some of the devices came from the manufacturer with the SCRs exploded, having clearly been destroyed during acceptance testing......

A really nasty letter was sent to the equipment designer and his employer, copied to the Health & Safety executive..

[Tim]

A Ferrograph 422 I worked on some years ago had uneven levels on the input channels( user had to set the concentric knobs a good way from each other). It turned out the signal input of one channel was connected to the “ top” of the input level pot on one channel, ( with the output from the wiper) and the other channel had the signal input connected to the wiper, and output from the “top”. The soldering appeared original, and feeding both channels to the “top” of the pot restored proper operation.
The assembly operative/s may have a slight excuse as ISTR all the wires were the same colour.

I also remember a Defiant radio that used one of those output transformers with a small out of phase winding on the primary for hum bucking. Usually the HT from the rectifier and first smoothing capacitor is fed to the tap, the anode of the output valve fed from the “ proper” primary, while the rest of the set is fed from the end of the ( out of phase) hum cancelling winding. My set hummed quite badly and after trying all the usual tricks, it was found the HT was fed in at the “ top” of the winding, and the rest of the set fed from the tap. Correcting the wiring to the usual arrangement reduced the hum considerably.

[livewireless123]

A few months ago my brother passed me an Eddystone 670a from a neighbour of his who used to be a 'ham' and knew that I was interested in radio.
I soon got it going, the mains filter caps were leaky, but the uch42 seemed to be very noisy and intermittant. the slightest touch produced loud crackles.
I found what looked like a used one in my spares box but it was just the same.
in the end I found that the screen grid feed resistor was just touching the valve pin,no sign of any solder. Quite hard to get at as it's inside the coil box.
it was still noisy after that, the second uch42 turned out to be microphonic
putting in the original valve at last cured it.
I was amazed at the size and weight of the set, they must have had big passenger cabins on the queens.

[G6fylneil]

I remember a 28V linear PSU that blew its supply fuse as soon as power was supplied to the equipment it was mounted in. I dropped the PSU front panel down, and soon found that the earth terminal on the mains input was connected to the PSU front panel earth tag, and to the 110V tap on the mains transformer primary (instead of the adjacent transformer screen terminal). I assume the PSU was tested with the front panel dropped down (to access an internal preset perhaps) and so passed the insulation test. The other 20 or so units were OK.

[ben]

Once had to look at a 14" Sony TV (circa 1998 vintage) that worked fine until the owner wanted to use the SCART socket, no signal was received. I thought it would be some complicated switching IC at fault, but it turned out that some pins of the socket were bent over and never soldered to the PCB. Didn't say much for Sony's quality control!

Also had a 4 way extension socket with the earth and neutral reversed.

[TonyDuell]

I have a Trend Taxer, a piece of equipment used to check X21 data links. While figuring out how it worked I came across a bit of circuitry that made no sense whatsoever, inputs to CMOS logic left floating, outputs linked together, etc.

Turns out that there was a 40174 chip (6 bit latch) fitted where there should have been a 4017 (divde by 10 counter). All the soldering looked original. But as this bit of the circuit is only used for one fairly obscure test, the unit passed the internal self-tests and whatever final tests were done at the factory.

[G6Tanuki]

As a schoolboy in the early 70s I had given to me one of the typical mid50s 5valve superhets covering LW/MW/SW. It had strange instabilities at various points on the frequency ranges. I spent a week of evenings trying to sort it when I should have been revising for O levels.

And I found the cause was a wiring error there from the start. The radio used the common cheap trick of sharing the screen feed resistor and bypass capacitor between the frequency changer and the IF amp. The earthy end of the capacitor was connected not to earth but to the AGC line. There was a 3 way tagstrip, the kind with 2 insulated tags and the one in the middle being the fixing point to the chassis. The 2 insulated tags were used for the AGC line, the earthed one for several other connections but the screen decoupled was wired wrongly.

Moving the decoupled to the earth Tag removed the instability and I was able to listen to Luxembourg without accompanying whistles.

[Craig Sawyers]

I recently restored a boat anchor signal generator - a CT218. AM/FM, 85kHz to 30MHz.

Before I even turned it on, I recapped it - a shed load of iffy paper caps, and electrolytics that showed no signs of there being a capacitor between the terminals.

But while I was in there, I noticed a resistor that was soldered at one end, but although the other end had been formed to be soldered in, it was just stuck in the air. No sign of solder either on the floating end or the tag it was supposed to be soldered to.

Now this was the cathode resistor in part of the amplitude modulation circuitry. So it could never have worked. Perhaps it was simply used in CW, so no one noticed that AM did nowt,

So much for QC on a mid 50's mil spec signal generator!

Craig

[stevehertz]

Totally unsoldered joint on a Carlsbro Cobra PA amp. Immediately to the right of the grey wire in the photo. Worked fine for decades then obviously oxidisation got the better of it.

[G6Tanuki]

Another one I came across was a radio with a bit of a hum problem. Replacement smoothing capacitors made little improvement, a new rectifier improved things a bit more, problem eventually resolved by removing the connection between one end of the HT secondary winding and the valve socket and reconnecting it to the pin on the valve socket that actually had an anode behind it.

Poor thing had been running on half wave rectification from new!