Disaster with a Marconi TF893A Output Power Meter

[Mark R]

Al,
Manual added to the archive, obliged.

As the enthralling content of the thread has gone into the mathematics of fixing the problem it has been stuck.

cheers Mark

[Skywave]

I realise that what follows is making a point that has already been adequately discussed and I would have made this post much earlier but for the simple reasons that (a) up to now my test bench has been totally cluttered with other work and (b) finding the time to do what follows has simply not been available. However, better late that never - and I hope what follows is of some value to this thread.

From the bottom of my test equipment storage rack, I dug out my Marconi TF 893A A.F. power meter and my Advance J2 A.F. sig. gen. For measurement of various a.c. voltages, I used my Tektronix 475 'scope. To measure resistance, I used my Fluke 77.
The J2 has an output that is stated as being of 5 Ω resistance: I decided to check this first. I set the J2 frequency to 440 Hz and by connecting a non-inductive variable resistance across the output of the J2 and monitoring the voltage across it with the 'scope, I determined that the J2's internal resistance was of the order of 4 Ω.
I set the 893A power meter to a Rin value of 5 Ω and trimmed the output of the J2 so that 100 mW was indicated on the power meter. (Just out of interest, when the Rin value of the power meter was changed to 4 Ω, the change in the power meter was virtually undetectable). The voltage across the power meter's input terminals (Rin) was 2v p/p = 0.7 v r.m.s.
I now introduced a variable, non-inductive resistance between the J2 and the power meter. When the voltage across the power meter's input terminals (Rin) was now 1v p/p = 0.35 v r.m.s, (i.e. half the magnitude of the previous measurement), the power meter indicated 24 mW (which is what we would expect of course: half the voltage in → quarter of power out to meter; O.K., 25 mW to be exact!).
I then removed and measured the value that variable resistance: it was 9 Ω.

And so to the arithmetic.

1. First part: prior to introducing the variable resistance.
We have: Vs / (4 + 5) = Vin / 5,
so: Vs / 9 = 0.7 / 5,
so: Vs = 1.26 v r.m.s.

2. Second part: having introduced the variable resistance.
We have: Vs / (9 + R) = 0.35 / 5, where R is the unknown variable resistance.
Since Vs is now known, we get:
1.26 / (9 + R) = 0.072,
and this yields a value of R = 8.5 Ω.
From the above, the measured value was 9 Ω.

Conclusion: measurement method and theory thus vindicated by a practical example. [The difference of 0.5 Ω (5.5%) is put down to intrinsic measurement error, tolerances of the measurement equipment and approximations in the arithmetic].

Al. / Aug. 25th., 2012 //

[AlanBeckett]

I'm back on the job. I hadn't realised it had been so long.

Lawrence kindly sent me some spare wafers which look the biz. So, I decided to practice with one of them. Step one is extracting the rivet which holds a contact in place. In the usual manner these are hollow. First I tried the usual method of drilling. The snag with this method is that the rivet almost always spins before it's released, and sure enough it did. Next I tried filing it with a needle file. This was pretty successful apart from chucking a lot of brass filings about and requiring a lot of pressure. Finally I tried my (pretend) Dremmel with a small stone. That worked best but was quite difficult to control. After removing one side of the rivet it was easy to push out the rest.

Now for a trial refit. For this I used a brass 8BA nut and bolt and a couple of washers. It's easy to tighten it up sufficiently, and I expect a dab of Araldite, Loctite or solder will stop it coming undone.

So, Next the real thing, which will be more interesting. Tomorrow is SWMBO's birthday so there'll be a short delay.

The Eagle-eyed will notice that I broke the trial wafer. Still, it's better than breaking the real one. Also, I think my fingers have got bigger and my eyesight a lot worse since I last did this sort of thing.

Alan

           

[Joe]

My favourite way of removing rivets is with a hammer and expendable wood chisel.

Rather you than me with that fiddly job, never mind re-connecting it all.

- Joe

[pwdrive]

They look cleaner than when they left my place Alan, to stop the rivet spinning when drilling I usually solder the rivet to the switch contact, just enough to stop it spinning. You'll soon have it sorted, if you forget the wifes birthday you would probably have all week or even months to fix it.

Lawrence.

[AlanBeckett]

Progress, or even success!
I removed the old rivet using Lawrence's 'solder' method - worked fine, although I was a bit nervous about putting pressure on the wafer.
Next fit a new contact. I tried to do it without slackening the switch but couldn't manage it. So I removed the knob, nut and washer and eased the switch backwards, then slackened the nuts on the switch to allow a bit more access. After about 15 goes, including destroying one contact, I managed it. SWMBO's old eyebrow tweezers proved invaluable. My tweezers are in there somewhere I'm sure. I put a dab of nailvarnish on the threads to lock it. Then it was just a matter of reassembling and reconnecting the wire. The wires are pretty thick so a bit of careful dressing is called for to reduce the load on the contacts.
And it works. I'm a bit dubious about putting 10W through the contacts, but Marconi thought it was OK.
I'll have to do some more tests in a bit, but I've got a few other things to do first.

Alan

           

[Skywave]

Yes, well done, Alan - yet another illustration that Patience, Persistence and Perseverance will always win the day.
Incidentally, I do like the look of your Advance sig. gen. Type 81: to my eyes it has that certain "I mean business!" look about it. Probably just another case of "eye, beauty, beholder".

Al.

[pwdrive]

Well done that man.

Lawrence.

[AlanBeckett]

Thanks Al.
It is indeed satisying to do something beyond the usual 'Check the valves, change the leaky caps and high resistors and switch on'.
The 81A certainy looks the biz, and I don't know of any others which look similar, so I would guess it was made for a special purpose.
I'm working on your list, but I'll report on that in the appropriate Thread.

Alan

[AlanBeckett]

Last lap, at least for now.

First I set it up connected to the 81A, and dialled up 600R and 1mW.
The scope says that the output is within 10%. Having said that I'm relying on an uncallibrated scope and a thick line on the 81A.

Next I connected it to the J-2 on the 600R output. Apart from a bit of hum - must look at the smoothing - the results were much the same as above.

Finally I connected it to the 5R output of the J-2 set to 100mW, and dialled up 5R on the Marconi - that's the part that didn't work before. The scope says it's 2V p-p, which is correct.

A couple of comments:-

The measurements were at 1kHz.

I could/should have got my super DVM out but I couldn't be bothered because it would have taken a long time to warm up.

The Marconi Manual does admit that the Impedance is mainly Resistive, but doesn't define the other part, although I have to think that it's Inductive. It's measured using a Bridge, and the fourth arm includes a Capacitor - RTFM - which no doubt explains the odd readings I got before.

Alan