Quad 303 thundering (in a bad way)

[Leon Crampin]

At around the last time I repaired a Quad 303, I was working in the automotive electronics business. I was therefore dismayed to see that Quad had specified small signal devices from a manufacturer whose products I had just rejected on test for automotive applications. The problem was poor hermeticity of the encapsulating plastic giving rise to moisture ingress on accelerated testing. In an amplifier this could well give rise to severe noise problems. In an engine controller (or worse still, an engine mounted voltage regulator) the consequences would be dire.

The output triples need TO-3 hometaxial devices (preferably steel-cased RCA devices with an "L" stamped on the flange, indicting Liege production) in order to give a chance of unconditionally stable operation. For automotive applications, we used thousands of these devices which were of high quality with an excellent SOA (safe operating area). Any epitaxial device will be potentially unstable in this circuit. RCA in-house numbers at that time were 5-digit, no letters, eg. 16025.

If Quad had taken reasonable care in the testing and selection of their components, the long term reliability of their equipment would have been hugely enhanced. An observance of Ohm's Law in the selection of some of their resistor wattage ratings (eg. 405 op-amp feeds) would also have been a considerable help.

With the 303, I would suggest replacing all of the plastic encapsulated small signal devices with new, preferably metal can equivalents. The small signal diodes, whilst all glass, are from the same maker and are probably best replaced by now. Then reset the quiescent current, being mindful of the poor quality adjusting pot and the rather dubious bias stability.

Leon.

[Boulevardier]

Originally Posted by Leon Crampin

The output triples need TO-3 hometaxial devices (preferably steel-cased RCA devices with an "L" stamped on the flange, indicting Liege production) in order to give a chance of unconditionally stable operation. For automotive applications, we used thousands of these devices which were of high quality with an excellent SOA (safe operating area). Any epitaxial device will be potentially unstable in this circuit. RCA in-house numbers at that time were 5-digit, no letters, eg. 16025.

Leon.

Leon - on the subject of the triples, do you know if BDY20 (Ft = 1MHz, and still available at Cricklewood Elect.) would be a reasonable replacement for the output transistors?

Mike

[Craig Sawyers]

In a letter to Douglas Self in 1995, Peter Baxandall says:

"I was instrumental in introducing triples to Quad - we had used them for other purposes in the circuits division of RSRE. Though I had much to do with the 303 design, the input part of the circuitry departed from my original scheme and was due to Mike Albinson, who had just joined Quad from Murphy Radio, which had closed. The use of a regulated power supply was also my contribution, and had nothing to do with any notion that it might be "sonically superior"! It was done purely because in 1967 we were all scared of busting output transistors!"

So although it is easy to be critical of Quad's semiconductor choice in the 303, we are after all talking of a circuit design that was done 53 years ago, and a mere 13 years after the first silicon transistor was made at Bell labs.

Craig

[Leon Crampin]

Mike, the difficulty is that many devices changed over the years but masqueraded under an unchanged type number. The 2N3055 was perhaps the most affected victim of this crime, and I suspect the near equivalent BDY20 may be similarly afflicted. As a pro-electron coded device, probably Philips in origin, it probably wasn't hometaxial in structure - otherwise, at the time, I would have known about it and tested it.

We were always concerned at the time that the RCA devices were hard to second-source, but I was able to locate other suitable devices from Thomson CSF (as they were) and Toshiba. We did in fact take some Thomson devices from time to time.

Craig, my criticism of Quad stems more from their obvious lack of in-house component testing than of poor circuit design - briefly neglecting overrun components in some of their designs. Having finalised and fully toleranced (most essential) an automotive design, the next step for my organisation was to generate a list of approved components for it. This was a most revealing exercise, with many "military approved" components failing miserably. The key is to design your own tests which are relevant to the application, and not be sidetracked by existing test standards.

So, to the horror of the sales reps, I would put TO-92 devices in the dishwasher and string wire ended diodes up from the ceiling with a weight on the bottom whilst cycling them at full current, one minute on and one minute off.

The most revealing tests were for water ingress, intermittency at the maximum allowable lead pull as above and degradation of thermal resistance, chip to case after 1000 full power cycles for large devices such as TO-3s and press fit alternator diodes.

There was a lot of rubbish on the market, and I take credit for not allowing Rifa capacitors to be used by my employer. They failed my cycling tests. The best makers, generally speaking, at the time were Philips, RCA, Toshiba, Matsushita and Lucas.

Leon.

[Boulevardier]

Originally Posted by Leon Crampin

Mike, the difficulty is that many devices changed over the years but masqueraded under an unchanged type number. The 2N3055 was perhaps the most affected victim of this crime, and I suspect the near equivalent BDY20 may be similarly afflicted. As a pro-electron coded device, probably Philips in origin, it probably wasn't hometaxial in structure - otherwise, at the time, I would have known about it and tested it.

Leon.

Thanks Leon
I’ve always thought it was appalling that the 2N3055 wasn’t given a new type number when it was upgraded to epitaxial and its Ft upped from 800kc/s to 2.5 Mc/s. But then I suppose this would have been seen as a harmless bonus - for anyone except 303 repairers! What makes me a bit optimistic about the BDY20 is that it is still today specced at 1Mc/s, whereas the 2N3055’s specs were long ago revised to 2.5Mc/s on all the data sheets.

I suppose the only way to know is to try it.

Mike

PS - you must have been a nightmare customer for those sales reps, with your product-testing "gymnastics"!

[nutteronthebus]

I to have had problems with the front end of Quad 303's after replacing a driver board because of print damage ( gas ring and poker job ) . You can get working new PCB's and a kit of parts . I fitted new transistors and 10 turn pots and it was night /day with the new side a lot brighter . I then found that a lot of epoxy transistors had gone low gain with HFE as low as 30 ( new ones 300+) . As Leon says "poor hermeticity of the encapsulating plastic giving rise to moisture ingress " I now replace the plastic transistors on all 303's I service

Dave

[Craig Sawyers]

Originally Posted by Leon Crampin

Mike, the difficulty is that many devices changed over the years but masqueraded under an unchanged type number. The 2N3055 was perhaps the most affected victim of this crime, and I suspect the near equivalent BDY20 may be similarly afflicted. As a pro-electron coded device, probably Philips in origin, it probably wasn't hometaxial in structure - otherwise, at the time, I would have known about it and tested it.

We were always concerned at the time that the RCA devices were hard to second-source, but I was able to locate other suitable devices from Thomson CSF (as they were) and Toshiba. We did in fact take some Thomson devices from time to time.

Craig, my criticism of Quad stems more from their obvious lack of in-house component testing than of poor circuit design - briefly neglecting overrun components in some of their designs. Having finalised and fully toleranced (most essential) an automotive design, the next step for my organisation was to generate a list of approved components for it. This was a most revealing exercise, with many "military approved" components failing miserably. The key is to design your own tests which are relevant to the application, and not be sidetracked by existing test standards.

So, to the horror of the sales reps, I would put TO-92 devices in the dishwasher and string wire ended diodes up from the ceiling with a weight on the bottom whilst cycling them at full current, one minute on and one minute off.

The most revealing tests were for water ingress, intermittency at the maximum allowable lead pull as above and degradation of thermal resistance, chip to case after 1000 full power cycles for large devices such as TO-3s and press fit alternator diodes.

There was a lot of rubbish on the market, and I take credit for not allowing Rifa capacitors to be used by my employer. They failed my cycling tests. The best makers, generally speaking, at the time were Philips, RCA, Toshiba, Matsushita and Lucas.

Leon.

Well yes I do know something of reliability engineering, having been International Project Manager for MIXS (Mercury Imaging X-Ray Spectrometer) currently en-route to Mercury. Arriving in 6 years time. I suggest that instrument is rather more highly specified than an automotive application.

https://sci.esa.int/web/bepicolombo
https://www.youtube.com/watch?v=B4pRe8yrHmI

My point however stands - the 303 is a design that is well over half a century old. My particular example, connected to a 33, an FM4, and the optical output from the kitchen TV via a cheapo DAC. It feeds ceiling speakers and has worked flawlessly ever since I bought it from the original owner ten years ago. It has not been messed about with, and is used ever single day.

Craig

[Leon Crampin]

I hope you have at least evicted the Rifa from the 33, otherwise the ventilation may have to work overtime.

The automotive industry places severe constraints in terms of maximising reliability to cost ratio - and in quantity. Specialised applications are generally less cost critical. The British, and to a lesser extent the US and European components industry were partly responsible for the demise of the UK consumer electronics business, producing some shoddy rubbish. It took the Japanese to show us how it should be done. My experience indicated that taking the time and trouble to seek decent components in fact added very little to the final cost. Basic requirements such as good lead solderability could often actually reduce production costs when the component supplier got it right.

Leon.

[Radio Wrangler]

Originally Posted by Craig Sawyers

Well yes I do know something of reliability engineering, having been International Project Manager for MIXS (Mercury Imaging X-Ray Spectrometer) currently en-route to Mercury. Arriving in 6 years time. I suggest that instrument is rather more highly specified than an automotive application.

Not sure about that. Millions of miles away from the nearest driver sounds like a big advantage to me

David

[Ted Kendall]

A formative experience - during my initial BBC training, I spent some time at the late lamented Equipment Department. One of the more memorable mornings was spent with the Standards Engineer, Brian Ammon, known generally as "the terrier" for the persistence with which he sought out the right components for BBC plant and held erring manufacturers to account. He described explaining to the Hirschmann rep why the connectors they had supplied for OB use were not up to the job and him suddenly exclaiming "what you want is a tank connector" - to which the answer was "precisely!", two guys from Panasonic beseeching his forgiveness for shortcomings in their crystal can relays and other jolly tales. He was, however, emphatic in his praise for Japanese component manufacturers in general. ED had not long changed resistor supplier from Philips/Mullard to Panasonic, Ammon citing a failure rate on delivery for Mullard of about 0.1% as opposed to Panasonic's parts per million throughout the manufacturing process. In practice, they didn't get dud Panasonic resistors.

[SteveP02]

A quick update - all still good, but I have new idea about the original problem.
We had a leak in the conservatory where my speakers sit. I didn't notice until later when I noticed one speaker cabinet all slightly swollen and wrecked along the sides and bottom *sigh*.
I thought it was cosmetic damage as the binding posts were the lowest electrical items on the speaker and the whole house would have been ankle deep to get near them. However, opening the speaker up yesterday check the damage, I realised I was forgetting that the upgraded crossovers are too big to fit in the usual position halfway up the box, so instead sit on the base of the cabinets. I don't know if one got damp but it looks possible.
Many thanks the pointers and the interesting discussion.

[lesmw0sec]

Had exactly the same problem some years back, which turned out eventually to be one of the low-level transistors (can't recall which though...)

[barrymagrec]

The new inspector at a place I once worked introduced the idea of a dab of red varnish on every solder joint to show it had been inspected. After finding a joint which had been suitably varnished but not soldered at all I told him to stop the practice and just check the joints properley.

[SteveP02]

Update

IT'S BACK...

Still intermittent, but every now again then there's a low rumble, on one channel of one amp as far as I can tell.

I tested the values as laid out here by Dada Electronics:

1.Adjust PSU board (not shown) for 67v
2.Adjust RV100 to set 33.5v between pins 5 and 9 on the line of connections coming onto the board
3. Adjust RV101 to measure ~10mV between pins 4 and 6

The last step showed instability in the reading with the mV randomly spiking up from 7mV to ~20mV (reading was rock solid on the other board).

Any advice would be much appreciated.

[SteveP02]

Circuit diagram for driver board attached.

[Petedox]

Originally Posted by SteveP02

Update


3. Adjust RV101 to measure ~10mV between pins 4 and 6

The last step showed instability in the reading with the mV randomly spiking up from 7mV to ~20mV (reading was rock solid on the other board).

Any advice would be much appreciated.

Was this done during quiescent conditions after a good warm up? TBH I have never found this adjustment to 'rock solid' - there's normally some fluctuation. Also some would say that 10mV is pushing it a bit.

Which channel is affected? The rock solid, or the random spiking?

[Leon Crampin]

Steve, I would respectfully suggest evicting any Texas small signal devices from your amplifier unless you have already done this. The symptoms you describe suggest moisture ingress, which causes very noisy operation. Whilst you're in there, change all the lousy skeleton presets for ceramic based cermets with solid carbon wipers.

Unlike Quad, my employer was very concerned about product reliability and the provision of goods of lasting value to customers. Accordingly, as I have stated above, designs were carefully evaluated and toleranced (no over-dissipating resistors anywhere) and components thoroughly tested under real conditions of product use - that was part of my job.

It's great to keep these old amplifiers running properly, but in order to do so, the design errors need to be corrected and the rubbish components evicted. There's no other way.

Leon.

[SteveP02]

Thanks guys, the wavering channel is the one giving the noise, tested after a 30 minute warm up with only the power lead connected to the amp.

FWIW I've upgraded all the caps, trim pots and some of the resistors with the DaDa kit:

https://www.dadaelectronics.eu/uploa...oards-V2.0.pdf

Could you please point me in the direction of the Texas parts should I be evicting? I'm not sure whether my (cheap) multimeter has any damping in it but the readings on the "good" driver board and the two in my other amp don't show any fluctuation although they are incredibly sensitive to the trim pot, so a pain to set.

[Petedox]

Have you reverted the amps back to their 'non bi amped' state as previously suggested? You won't be able to fault find properly until you can isolate which channel of which amp is faulty.

[SteveP02]

Yes, the amp is standalone, LH channel seems to be the one giving problems.