Handling close-tolerance components

[ColinTheAmpMan1]

I have a Marconi TF 2700 Universal Bridge and the close-tolerance resistors to make the improvements to it suggested by Morgan Jones in "Building Valve Amplifiers". This entails replacing the 0.5% tolerance resistors on the main range switch with 0.1% tolerance components. I was wondering if there were any particular concerns when handling close-tolerance resistors. I am envisioning the use of clip-on heat-shunts and care in overheating the resistors, plus care in avoiding contamination of the outer casing of the resistors by finger-grease, etc.. Is there anything else I should be aware of?

The resistors came from Mouser in static-sensitive bags with the usual labels warning of the precautions to be taken when handling electrostatic-sensitive components. This is what has got me thinking.

It is a pity that Morgan Jones suggests that the bridge can be "easily improved", when access to the resistors is a little tricky and that he doesn't mention any special consideration that should be taken with close-tolerance resistors.

Colin.

[The Philpott]

If you have clip on heat sinks and wipe the resistors over with alcohol after fitment i don't see what other precautions could be taken other than ensuring that any bends in the tails should be a few millimetres away from the resistor.

[Radio Wrangler]

Yes, some precision resistors can be affected by static discharges.

The TF2700 is a very useful little instrument and more than adequate for testing general radio and audio components. A bit of improvement can be got from fitting better standards and the resistors on the top decade switch, but after that the pot and its scale makes it not worth going much further.

If you want to accurately measure resistors, then you may be better with a bench DVM with plenty of digits and accuracy - and 4-wire connections. I picked up a 6 digit Datron one some time ago at a radio rally for £20. I think it specs 0.01% on resistance which is substantially beyond my needs.

If you want to measure capacitance and inductance accurately, then the Wayne Kerr autobalance bridges are very good and use precision ratio transformers to (literally) transform a single capacitor as the reactance standard. Again in the 0.01% league for some models. The transformer turns ratios don't drift, so a single check of a known resistor and a single check of a known capacitor checks the whole calibration. These used to turn up for junk prices, but the nixie tube hunters lust after them like an audiophile spotting a 1930's radiogram. These also use 4-wir connections (Kelvin clamps) for good accuracy at low impedances

I've just (today) bought a Racal Dana 9143M databridge which is a 0.1% machine and fully automatic. Kelvin clamps again. At a radio Rally for £80

Better grade instruments are out there and offer fairly dependable specs that might be a more direct way to better accuracy than hot-rodding that TF2700.

Test gear is my hobby, otherwise i might have a hard job justifying anything better than a 1% LCR meter.

Just use a wrist strap with a 1 meg ohm safety resistor to ground when handling critical precision components.

Davi

[ColinTheAmpMan1]

Thanks for the ideas, guys. I will be quite happy with the TF2700 when I can get to upgrade these resistors (to be honest, it is good enough for my purposes as is, but if this simple upgrade gets me a bit of improvement, why not?). It is certainly a significant improvement over my Fluke 175 for low resistances.

I hadn't thought of wiping the resistors with alcohol (IPA, I assume, not ethanol, though I can lay my hands on absolute alcohol) after installation, nor of using a static-control mat and wrist-strap. I do possess a static-control soldering-iron, so I will use that, too.

Colin.

[G4_Pete]

I have two of these instruments and have found the correct setting of the D and Q controls to have far more effect on the balance points and accuracy than the resistors ever will. Providing all the resistors are in tolerance of course.

So I whilst I concede that closer tolerance resistors can improve accuracy, in practice on the L and C ranges I find it is very fiddly to get a good null on some components and hence obtain a good "accurate" reading that could be improved on with tighter tolerance components.

So I will leave mine as is.

Pete

[G4_Pete]

Thinking about this over a cup of tea in the the garden you could use your new precision resistors to use as standards and draw up a calibration chart for the bridge as is, you might be lucky and have a bridge where the resistors are already in good tolerance!
If not as you have them , as you say, you might as well fit them but you may have to sacrifice and snip the old ones to prevent too much damage to the wafer switch during removal.

Pete

[Radio Wrangler]

I like Pete's advice. Before fitting anything, why not use the bridge to measure the precision resistors just to find out how far it disagrees. If it's close enough then you can keep the resistors for use as calibration standards to check it again in the future and to check any other instruments.

Once you get to high precision components, they only make sense with 4-wire measurements and Kelvin clamps unless the resistance values are high.

David

[ColinTheAmpMan1]

I should probably clarify this a bit. The coarse balance switch components are 0.1% tolerance, but those on the main range switch are 0.5% tolerance. There are a total of eight here, but one is just a length or wire bent like a hair-pin and selected to be 0.1R. The next is 0.998R. The remainder are 10R, 100R, 1K, 10K, 100K and 1M. The originals are metal-film, but while Morgan Jones suggests replacing the five 100R-1M types with 0.1% metal-film components, he states that the 10R must be replaced with a 0.1% non-inductive wire-wound component. I have all of these and have discovered that access to the switch that they are mounted on is not as problematic as I thought it might be.

I have checked the 0.1% resistors that I intend to use and the TF 2700 actually gives the correct values. I don't think that this is particularly a justification for not replacing the 0.5% resistors.

Morgan Jones also advocates checking the fine balance variable resistor. He suggests that this can be done by using only the main range setting of 10M and checking a series of resistors from 100K to 900K in 100K steps. This is achieved by having five 0.1% tolerance resistors in series; one 100K and four 200K, such that when the 100K is at one end, values from 100K to 900K can be picked off. He then states that using a plot of measured value against known value will achieve a final error from +/- 1% to +/- 0.25%. That's good enough for me!

Colin.

[ColinTheAmpMan1]

To complete this tale, I have now replaced the 0.5% resistors with the new 0.1% items. I am continuing this in the "Success Stories" section.
Colin.