“New” AF126 duff

[FiveBobRepairs]

As documented in a different thread, I am experiencing the delights of a restoring Roberts R707. This not only has AF11* transistors, but they are in an enclosed module, necessitating a huge amount of disassembly.

So imagine my disappointment, after reassembly with all new AF12*s, to find FM was poor and AM was not working at all, just noise, unaffected by tuning. No beat from the oscillator when tested with another radio, so it all had to come out again. Turns out the new AF126 that I used to replace the original AF115, that serves as IF amp on FM, mixer/oscillator on AM, is open circuit at emitter.

Annoying thing is of course, whilst a different AF115 was originally shorted collector to case, that one was actually fine, no whisker symptoms. I only replaced all three so as to avoid the nightmare of taking it apart again. And then I had to, anyway. Best laid plans of mice and men.

Anyway, just wondering how common this, a duff “new” AF126? It certainly looked unused, all leads straight. Could it be my own fault, maybe bending leads or whatever? I’ll certainly be applying a quick component test in future, when installing in such hard to reach areas.

[paulsherwin]

The AF12x series isn't known for unreliability, but it's certainly not impossible to get a bad one. These transistors will be over 40 years old now and there's no knowing how they've been stored or if they were sourced from a rejected batch.

Always do a basic hfe test on any transistor before fitting it, and check there is an open circuit between any metal case and the cbe wires.

[Restoration73]

I have come across TO-72 package transistors where the B and E were transposed, so
always perform a test to confirm.

[FiveBobRepairs]

I do not possess a specific transistor tester, and lack the technology to do a gain check other than by lashing up a test circuit.

But I am feeling suitably ashamed for failing to spot even such a simple parameter, that was revealed just on meter check.

This particular specimen was purchased as bundle of five and is the last, the other four having already demonstrated themselves to be functional.

[paulsherwin]

You don't need a dedicated transistor tester. Almost all DMMs made today will perform a basic hfe transistor test, even the very cheapest DT830 ones.

[FiveBobRepairs]

Originally Posted by paulsherwin

You don't need a dedicated transistor tester. Almost all DMMs made today will perform a basic hfe transistor test, even the very cheapest ones.

I was about to say “mine doesn’t” but wow, you’re right, it does. I think I’ve been blotting it out as the transistor socket is a bit too cumbersome for convenience any time I have tried, but lesson learned.

[paulsherwin]

The hfe figure should be taken with a pinch of salt for Ge devices, but it's adequate for a basic pass/fail test. You can also use it to confirm the leadout.

[Argus25]

Originally Posted by FiveBobRepairs

I’ll certainly be applying a quick component test in future, when installing in such hard to reach areas.

Hi,

I think it pays to go further than this.

I have a simple test jig with a common emitter amplifier circuit that I plug the germanium transistor into with about 0.75mA collector current (with a socket) and a signal generator drive and scope on the output.

For RF transistors like this I check their output at 30MHz for radios like the EC-10. But if it's just for an AM broadcast band radio a test at 2MHz is OK.

But if it's a shortwave or FM radio you need to test the transistor at its operating frequency, at least. This way you can be sure the transistor is OK for the job and you can compare a pile of transistors quickly.

It soon screens out the duds, which in a lot of 20 to 50 transistors, even NOS, there might be one total dud and a couple of sub-optimal ones that would be better avoided.

As noted on past threads the AF178 is a great replacement for AF11x, they are bigger and the lead wires longer and it's an easier fit. Of about 30 or 40 of these I tested and graded for gain, I think I found one NOS dud as I recall. The AF125/6/7 is a very good transistor too, but it still pays to test them first.

[FiveBobRepairs]

Thanks for the tip re AF178, although not sure they would be an option in this case as headroom between pcb and the lid of the module is extremely limited. And whereas the originals are mounted very close to the pcb to keep the profile down, the holes are designed for the original pinout, more room is needed if the leads need to be rearranged.

AF12* do fit comfortable and allow the leads to be formed without strain, but anything bigger might get challenging.

Unless anybody cares to talk me out of it, I’m probably going to replace it (originally AF115) with an AF124, as I have one available. It serves as AM mixer/oscillator, and FM first IF. I lack the kit to do any elaborate measurements but have now tested it using my cheap DMM which claims to have established a an hFE of 110, though I don’t know what collector current is used, and I guess it’s just measuring dc? Still, I reckon the fact that it looks like two diodes on resistance check, no leakage to screen, and provides some dc gain, means it probably is a functioning transistor. Fingers crossed.

[paulsherwin]

That is a typical reading for a Ge transistor of this type.

[Argus25]

Originally Posted by FiveBobRepairs

Unless anybody cares to talk me out of it, I’m probably going to replace it (originally AF115) with an AF124, as I have one available.

Af124 is good, but..

Every AF114/5/6/7/8 and others of the same construction are affected by tin whiskers. This rule holds whether they are seemingly working or not. So its never worth replacing with the same type.

In many many cases the length and position of the whiskers inside them are simply not long enough or in a position where they have yet caused inter-electrode shorts, or shorts to the case..yet.

I have opened scores of these transistors working and failed and found whisker formation in every case, sometimes it is gross and you can see it without reading glasses, other times you can just make them out under a binocular microscope.

So if you have a radio with these transistors they should be replaced. Sure, you could wait for the radio to suddenly stop one day (that is how the faults present with whiskers, suddenly) while you are relaxing & listening to some nice music with a glass of chardonnay. Then you will have a repair job on your hands.

AF124/5/6/7 never have this problem and since you are short of space in this case, I would go for one of those as the replacement, rather than the AF178. So go for the AF124.

I forgot to mention that one other thing about doing a dynamic signal test on the transistors, with a generator & scope (for those who have this test gear) it identifies some transistors which are very noisy and have collector current irregularities, that the meter test and hfe tester don't notice and again better to avoid these ones too.

[FiveBobRepairs]

Thanks for those words of encouragement.

It’ll definitely be an AF12*, my slight uncertainty relates to whether it is worth the extra effort to obtain an AF125 rather than ‘124, since the original was AF115. Ie, to replace one ending in ‘5’ with another ending in ‘5’.

I do reallise, from reading these forums, as well as my own research that there appears to be no difference whatsoever between AF124 & AF125, other than the intended application.

My own best reference source is a relic from my childhood, “Mullard data book, 1970”. It contains a section on “comparables” wherein lies suggested remedies for AF11* that fail to oscillate, or that become unstable. Fair to assume same advice would apply to AF12*. I want to minimise the probability of needing such remedies, as access to the module requires copious desoldering of the module itself and other parts. Tempting therefor to wait a little longer while I obtain an AF125, just in case it is closer to the original, after all?

[G6Tanuki]

I think it's inevitable that there will be occasional failures in what is now 40-year-old "New-old stock" parts. The classic 'bathtub' curve for reliability doesn't always require the component to be doing anything other than just sitting there.

I've had a few AC127/128/153/176 with 'whisker-to-case' shorts, and also a couple of AF139/239 UHF RF-amp transistors that have gone noisy/low-gain.

And I recently fitted a BFY51 from a reputable source that I'd had in-stock for 30+ years but which failed to show any sensible signs of amplifying. So it's not just Ge transistors that have their issues.

[paulsherwin]

Originally Posted by FiveBobRepairs

I do reallise, from reading these forums, as well as my own research that there appears to be no difference whatsoever between AF124 & AF125, other than the intended application.

I think the AF114-7s were so named for marketing reasons, as manufacturers had become used to using an OC44 and two OC45s in their designs. The AF124-7 was designed as a replacement in a more modern encapsulation and followed the same conventions. There is no discernable difference between them, though Mullard listed different parameters for each type in their data books to give a different impression.

It would be helpful if somebody who worked for Philips/Mullard in the 60s could confirm this, but in the absence of that it's a reasonable assumption.

[FiveBobRepairs]

Thanks Paul. I think if I can find time to put it all together this weekend, I’ll risk the ‘124. Otherwise it’ll need to wait a few days as I’ll be busy with other things, in which I case I might use the the time to order a ‘125.

G6Tanuki.. Indeed I have taken on board the message that it was rash of me to assume the part would just work, without testing. Lesson learned.

Up until recent resurrection of my childhood interests, the only AF11*, or any other transistors that I ever replaced would have been 45-50 years ago, and probably straight off the production line. That explains why I don’t remember many being defective when purchased.

I do vaguely recall, from repairing for friends & family in early 70s, that AF11* was a fairly common failure. I don’t recall whether shorts to can were usually present, I would not have reallised the significance in those days.

[Nuvistor]

My experience in the 70’s was shorts to the can with these transistors, of course there were replaced with the same type.

[Argus25]

I find it a bit of a shame that the only transistor in the world which had about the same sized case and the linear pin array matching the AF11X transistors and the same electrical properties and was resin filled and free from the whisker effect was the Tesla OC169 & OC170. (photo attached on right hand side)

When these appeared for sale, from an ebay seller in Europe, a few years ago, they sold out fairly fast. I bought a supply of them for future radio restorations because I could see they were ideal, but I should have got more of them. The very odd thing is there appears to be no more.

There probably is a large number of them in Europe, hiding somewhere. It is hard to believe the factory made just one batch or something. Perhaps there is someone with links to part suppliers in Europe who can research this and find some more of these.

[paulsherwin]

They may have been produced for military use in relatively small quantities. Tesla was a major supplier of semiconductors to Warsaw Pact countries, though not to the USSR who insisted on their own supply channels.

[G0HZU_JMR]

On this old thread here:

https://www.vintage-radio.net/forum/...d.php?t=144591

I posted up a few test results for some classic Germanium transistors that I tested on my VNA. This showed that the AF117 and the AF178 and AF127(6) were fairly similar devices although this assumes the devices I obtained were genuine.

I took s2p data of several Ge devices at various Vce and Ic operating points and uploaded the results here on my qsl webpage:

http://www.qsl.net/g/g0hzu//RF%20Tra...ata/Germanium/

There are s parameter files for several classic Ge transistors at various operating points. However, I'm not sure if anyone will ever make use of these data files as they are not easy to find on my qsl page. I did wonder about donating them to the Radiomuseum website as they seem to hold lots of datasheet and schematic info. But their website/forum looks a bit cliquey so I'm not sure how to even ask them.

Of course, anyone is welcome to copy/use/distribute these s2p data files from my qsl website and they could be stored on this website too if the data is thought to have any value. To anyone not familiar with s2p data files the data will look like a load of meaningless numbers. But these data files can be used for basic amplifier design/checking using something like RFSIM99 or QUCS. It only takes a minute or so to design/check an amplifier using these freebie simulators and my s2p files should be accurate although I can't guarantee I haven't made an error in one or two of them. There are too many to check although I hope/expect they are all good!

Most manufacturers of modern RF transistors will provide similar s2p data files for many of their devices and my attempt to measure these classic Ge transistors should allow people to use modern design/checking methods with these classic old Ge BJTs

[Argus25]

Originally Posted by paulsherwin

They may have been produced for military use in relatively small quantities. Tesla was a major supplier of semiconductors to Warsaw Pact countries, though not to the USSR who insisted on their own supply channels.

Paul,

That is good to know. When I have been searching for vintage TTL IC's often quite a few Tesla ones come up for sale. Probably then, if they were for mil applications they would be of great quality & reliability. So I might get some and try them out.

It is interesting about how many NOS semiconductors one might expect to be faulty in a batch of say 100. Probably less faulty transistors as a % than IC's because of the multiple transistor junctions in IC's. I read a remark from a manufacturer in the mid 1970's who used 100's of 74 series TTL IC's and they said about 1/100 were faulty new.

Generally I have never ordered enough of any type of transistor to get any comparable figures. Though for old transistors the figure appears very high. For example out of 10 or so very vintage 2N94 nos transistors I bought, 3 were faulty and one too noisy to use. So age has a lot to do with it. On a pcb I built with 66 TTL IC's, on 3 occasions so that is nearly 200 IC's total, not a single one was faulty