Roberts R600: Fades

[Mike Watterson]

I checked the headphone socket.
Checking at the IF out is my next thing, but it's not like any whisker fault.
The lockfits are unsuitable for the PCB and very badly mounted. In my experience 100% of lockfit failures are bad mounting stressing a lead. A replacement 2N3904 or equivalent TO92 BC??? cheaper than a replacement capacitor, costs nearly nothing. I'd not replace lockfits on PCBs properly designed for them.
However that would only be preventative, it's not a lockfit fault. I've never seen noisy lockfits in over 30 years, maybe 40 years. Either working or not. That was the case on one of my Hackers. No doubt noisy lockfits can happen, but it was originally the same chip as in the TO18 (or whatever the little metal can BC107, BC108 etc used), just a different package.

The absolute most gentle turning on and simply waiting is 100% repeatable if the set has been off long enough. Initial second of correct volume, fade to nothing and then after a couple of minutes it comes back, distorted at first and then perfect. I fooled myself earlier about the switches etc as obviously it returns while you are clicking the switches.

It seemed like a capacitor fault after I realised it wasn't the switches or socket. Except it obviously isn't. Once it's working it stays working, even if turned off for a minute or two.

[Mark Hennessy]

I'd say that nearly all Lockfit failures in my experience have been noise. Probably 95% or more. Yes, complete failures happen, but I can't recall many. I've lost count of the number of transistor sets (mostly Hacker and Roberts) that I repaired for a local friend when he was selling sets, but it was high hundreds, perhaps more (he was very successful!), and if we assume that each of those radios had perhaps 10 or more Lockfits, that's a lot of meaningful data gathered. While it was good to gain that experience, it left me with no energy to restore any of my own collection - that's hopefully something for retirement, which is still a long way away. But that's why I was careful to point out that things might be different in other equipment, like TV sets, for example. My experience of transistor radio repair is extensive, but it is just that: my experience.

And as I said, when it comes to noise, it depends where they are in the circuit. In many cases, you just wouldn't know about it. And in other places, it's ruinous.

The classic location being the first transistor in Hacker amplifiers. It's so common that I wrote about it specifically here: https://www.markhennessy.co.uk/articles/...tm#lockfit - that includes a discussion (and typical Hacker schematic) about why circuit location matters.

I replace BC148s with BC108s in the Hacker amplifiers because that's what the service manual lists, but BC548s contain the same silicon as far as I can tell, and therefore are equally good electrically if you don't care about how things look. Obviously, there are many, many possible alternatives.

For many years now I've been promising myself I'll build a test jig to test for noise. Doesn't need to be anything complicated, and I have bread-boarded a couple of potential ideas that were really quite promising. It's just a case of getting around to it.

Next time I'm up in the attic I'll dig out an R600 to see what you mean by "very badly mounted".

Anyway, I'm not in a position to speculate whether it's a Lockfit problem or not based on what you've reported so far, so my reporting of my experience is just general, and intended to be useful to people less expert than yourself. For those folk, perhaps the most useful things to pick out of my previous post are 1. current limited bench power supply, and 2. start with 'scope or powered speaker at the volume control.

Oh, I've just remembered a reasonably common fault that happens with the Hacker amps, and perhaps could happen with this one too.

The VAS in the Hacker circuit - an AC128 (TR4) - can develop intermittent tin whiskers that cause this fading effect. Sometimes you hear cracking, which can be alarmingly loud, but often you don't. The fault changes the circuit conditions in a way that looks like heavy collector-base leakage (which can also be caused by C7 going leaky). What's happening is intermittent bursts of current leaking between collector and base via a high-resistance whisker (or not-so-high if you hear cracking), and this gradually shifts the DC conditions - the time constant being formed by the large capacitors in the circuit having a peak-hold effect. If present at all, the programme audio comes and goes while all this is going on.

Looking at the Roberts circuit (Trader 1852), the VAS (T3) is an AC128, so it could be behaving in a similar way. The equivalent to C7 is C22, but I'd suspect T3 over that.

Tin whiskers in TO1 packages seem to be much more frequent than they used to be. For years, folk on the vintage forums didn't believe me when I told them it happens, but it seems to be accepted now. I've also written about that in that article on my website.

Anyway, first/next step is to determine if it's before or after the volume control...

[Mike Watterson]

Thankfully the IF audio out is perfect. Those modules are not pleasant to remove.

The fading is ONLY straight after power on, ONLY if it's been off a while. Once it's going you can turn it off maybe for 10 minutes. Once it's going it's fine and has never then faded.

Next is to set up meters on all the likely points on the amplifier, there is maybe only just over a minute before it comes back, and then you'd want to leave it for an hour.

The Lockfits should click into the holes without the pins being bent. Clue in the name. The ends of the leads should not be bent over. The lockfits should be sitting straight and not bent or stressed.

Any of the soldered can transistors can grow whiskers if the inside has too much tin. I've a couple of sets with original AF117 transistors that are probably different production because they are fine. I'll leave AF117s alone if they are working today, because if they were going to grow whiskers they've had nearly 50 years to do it. I compared a Russian military part with a shorted AF117, I photographed the whiskers. Rarer on the OC171 or whatever similar part and on the VHF IF version. I'm sure it was a production issue, too much tin. The Russian part isn't a soldered can, more like the older top hats, but internally it looks more like a Silicon planar chip, though Germanium.

I put in 2 x BC547 (= BC107, BC147) which is simply a higher voltage part, maybe even by selection on the basis that the BC148 might fail eventually, being in something more suited to a BC108 etc. I think the Hackers are better made, but the PCB on my two Hackers weren't properly designed for a BC148 (which was dead). The other model had a BC108 and some other small differences, though the amps seem to be interchangeable on the RP25A and RP18. I forget which had the BC148.

[Mike Watterson]

I've never tested or replaced transistors for noise. But I've only repaired a few HiFis and usually dried out caps. I'd not have thought of checking a portable radio unless it seemed unusually noisy. Most of my commercial service work was on AV monitors / Multistandard TVs, Industrial / Medical / security and budget studio camera as well as various VTRs/VCRs such as Umatic, EIAJ, N1500, N1700 and early Betamax and VHS. Then it was Apple, RM, BBC, Pet, IBM etc in my own company, though mostly the ex Chubb guy did servicing and I was designing Z80 controllers and analogue controllers; it was the last time I did hand layout. From the next job it was all CAD, from 1983.

[Mike Watterson]

At power on, after it's been off for a while there is 0V at junction of TR5 and TR6 (to the + rail, which is 0V). That rises to the 4.45 V eventually.
So I took out TR3, the AC128. It shows as a P-JFET on the Chinese tester. Does nothing sensible on the simple Hfe tester. On diode test both junctions are about 165mV.
But on Ohms, with + to emitter, base to emitter and -Ve lead to the collector it reads about 150 Ohms. This gradually rises to about 4K and then it goes much higher. It then stays "working" for a while. Eventually it "resets" and the characteristic is repeated.

The construction does look like a soldered cap on metal ring base. I can't imagine how this relates to a tin whisker, but I shall use the fine cutting disk on the Dremmel tool with can well heatsinked and examine the insides.

Somewhere I have an AC128, but meanwhile I found a pnp Germanium "T7" in an OC71 style black case that has similar to AC128 gain. The radio worked perfectly.

I'll be curious to see inside the can, though not tonight. I've been opening alarm clocks, valves and transistors for a long time.

[Mike Watterson]

I replaced the o/p capacitor 640 uF C23 as its leakage is high. Only 6.5V rated and was getting over a minute of 8V to 9V at power on every time.
I readjusted RV3, quiescent current, to 3.5mA. It's very sensitive. The RV2 needed a larger adjustment with replacing TR3, AC128. It's not so sensitive due to R14 in series. Replacing the lockfit TR2 BC148 with the BC547 didn't require any change as the Vbe and hFE was virtually identical.

I've no idea which version of IF module is fitted as only the Roberts label warning not to service it is on it.

Is the R600 really later than the R700?

Anyway it's working well. I think it's the same model as the last radio my Grandfather had and he had it till the 1990s when he died. My mum threw it out.

[Murphyv310]

Good result Michael.
You stuck with it and you got a cure.

[Mike Watterson]

Certainly a strange fault I'd never heard of before that seemed like a capacitor issue. It was behaving like an old series chain valve radio with the really slow rectifiers due to extra cathode - heater insulation.

[Mark Hennessy]

As I said, when I've had this, it's usually been tin whiskers - the intermittent nature of the leakage being the clue. You might get audible (often very loud) cracks on the audio if the path through the whiskers is a fairly low resistance one, but often that is not the case, and the DC conditions just drift about slowly - as you've found, it's quite confusing when you first meet it. Listening to the output while tapping the case can often confirm the whiskers, but not always.

Sometimes, the whiskers can be measured from case to electrodes (has to be more than one, as the can isn't in electrical contact with anything in this position) - a simple voltage measurement of the can relative to earth can also be useful. But the heat and handling from de-soldering can be enough to temporarily clear the whiskers, meaning the transistor works perfectly upon test, but of course a replacement cures the fault. That's also confusing, but you can sometimes find light leakage (>100k) from case to one electrode, which confirms the presence of whiskers in some form.

Zapping the whiskers is a cure that'll last for decades if done properly. Of course, some won't want to do that; that's fine, no need to get into that here. But even if you don't re-use the transistor, it does at least prove that whiskers were the root cause...

I don't know how many times I've seen this exact fault with the Hacker circuit, but at least a dozen. But I've had a couple of instances where it wasn't whiskers - it was a genuine fault with the junction leaking very heavily. Not surprisingly, the leakage amount was highly temperature dependant, and it didn't change at all after zapping. In circuit, it worked fine when very cold, or when blasted with freezer. The Hacker circuit runs this transistor at about 4 times the power level than the Roberts - they even fit a small heat sink - so the time constant with that fault was pretty short. It's possible that the variations you're seeing in the static test are temperature related - that's easy to investigate should you be curious.

I've found that the Peak Atlas transistor tester is a lot better with germanium than the Chinese testers. Although occasionally - perhaps one time in 20 - the Chinese tester gives a sensible result when the Peak Atlas did not. I'm glad I have both, but use the Peak most. I got mine from Maplin for only £20.

Taking all the TO1 and TO7 transistors together, I'd say that actual junction failure is very rare indeed - certainly in my experience of transistor radios. Ignoring cases of incorrect battery polarity killing output transistors and similar abuse, I can probably count on one hand the number of dead junctions. I've even had germanium transistors with broken encapsulations that were working just fine. But I have had countless apparent junction failures that went away completely when the whiskers were treated - even though the measured resistance of the whiskers to might have been hundreds of kΩ, which could easily have been dismissed as insignificant. I've "zapped" hundreds of them now, and never damaged one in the process, and as I haven't had a reoccurrence ever, I'm confident in giving a 5 year warranty on whisker treatments (I give the customer the choice, and most of the time they prefer originality, though I might steer them towards replacement if there's lots of labour involved in getting to them - e.g. Roberts R707). Some time ago it got to the point where if I have any RF/IF fault at all in a radio using TO7 transistors, I remove and zap before doing anything else. Apart from a quick check to leakage to case, I don't bother testing before or after - I just zap and put them back in. Offhand, I can only think of one occasion where that caught me out because of a "real" junction fault - there must have been some others, but I really can't think of any. Luckily, that was in a Hacker where the transistor was very easy to access.

Yes, the R700 came first, as per RM.org and various brochures, etc. You can tell it's an older design - all germanium, for a start. It's perhaps the only Roberts portable to have Hacker-style separate AM and FM IFs stages. AM IF and FM RF end are both courtesy of modules - presumably Mullard, but it's been a while since I've looked in one - but happily the service data contains details of those. In the examples I've seen, the caps were not Philips axials - they were the usual "suspect on sight" types, like TCC Elkomold, etc. It was fairly quickly replaced by the R707. The R600 carried on in parallel with the R707 for a while, but was replaced by the R606 and later R606MB. Early R600s had the mesh grille, but most IME had the flat sheet type, with the black stripes painted on (though that paint isn't very resilient in my experience, so I'm always very cautious when cleaning them). Both grille types dent very easily. If you want to see a proper grille design, look to the extruded aluminium used by Hacker in the '70s - I dread to think how expensive that must have been!

I've got 3 examples of the R700 in storage, and should be able to make 1 decent one from them. Let me know if you need any parts from an R700 if/when you tackle one.

[Murphyv310]

Hi.
I've never encountered tin whiskers on any of the AC range of germanium transistors from the 126 to the 188. Most of the failures have been caused by incorrect polarity or actual physical damage, AC187/188 failure was often caused by a failure in correct biasing or cap failure. I was under the impression that on the AF11* range that the tin whiskers were to the can that is grounded and in some cases disconnecting the screen effects a cure albeit not very elegant, the AC range the can is not at ground unless the transistor is in a heatsink bolted to ground.