Capacitance Decade boxes

[Specmaster]

This is a continuation of a topic that was started out of a conversation in the Test Equipment Anonymous (TEA) group therapy thread, and it seems to have sort spiralled into something a bit more substantial and so was suggested that it was moved to this section as it may have wider appeal than it was first thought.

It started when I was looking at the possibility of making a resistor decade box to use as an aid to calibrating multimeters, not to a lab standard but most certainly accurate enough for the average electronic enthusiast and I ordered what was advertised as a decimal thumbwheel switch from China. I already have the 1% and .1% resistors waiting for the arrival of said switch. It arrived and instead of being decimal, it was a BCD device which is not suitable, not without some special resistor values which I don't believe are readily available and if they are then they will almost certainly be expensive as I don't think they are preferred values.

However because of the way capacitors work, I think that they could be used in this context.

I have searched around and can find very little on capacitance decade boxes, it seems that they are almost as rare as hens teeth and I suspect that this is because most hobbyists etc are probably working on HF stuff and it appears that a decade box in this environment is a no no. As you may already have guessed, I have no knowledge in this field at all, preferring to work largely with audio and low frequencies.

So my question is this, is there a real need for such a device. bearing in mind that HF says no, but what about LF guys on here, do you think there is any merit in this beyond the use I had in mind, a test bed to test the accuracy of my varied collection of DMM's and LCR meters.

I welcome any thoughts / suggestions please.

[tggzzz]

Resistance and capacitance boxes used to be relatively common, typically used in the arms of Wheatstone bridges. They can often be found in sales, have lovely mahogany enclosures, and a certain steampunk appeal - and gravitas, since they are heavy and bulky.

Or, of course:  https://www.ebay.co.uk/sch/i.html?_from=R40&_sacat=0&_nkw=decade%20(capacitor%2Ccondensor%2Ccapacitance) etc

As for testing meters, it is worth considering what needs to be tested. Do you need to test linearity, or slope + offset?

Note that resistance boxes can be very sensitive to operator error, in the form of having too much current flow through low value resistors. Consider, for example, the permutations possible when changing from 1000.1ohms to 900.1ohms.

Even if circuits are "low frequency", whatever that might mean, long wires can have EMI/EMC consequences.

So overall I think their use is limited.

But having said that I've just bought a 7decade "potentiometer" in the form of a Kelvin-Varley Divider, just because. So far the worst non-linearity appears to be around a couple of ppm, but that could well be due to voltmeter non-linearity and/or substandard measurement techniques

[Specmaster]

Just maybe this whole thing needs a rethink.

As mentioned before, it was just a germ of an idea following on from my planned resistor decade box project which is based around a decimal thumbwheel switch and was mistakenly sent a bcd one and it was mentioned that the configuration is ideal for a capacitance decade box so I was just exploring that possibility as a way of making use of a switch that was not really suitable for its intended purpose. It is nothing more or less then just a feasibility and practicality study

[Dismounted]

Here in Australia, Altronics sells a resistor/capacitor decade box kit (http://www.altronics.com.au/p/k7520-resistance-capacitance-decade-box-kit/). I've got one of these, initially for the resistor part, and to be honest, haven't found a use for the capacitor substitution yet. I'ts sorta nice to just "have around" though.

These kits use BCD switches for the capacitor bank, just as you say.

One potential use may be quickly figuring out an approx cap for switch debouncing? Though the issue remains that for these "high frequency" signals, travelling over (relatively) long banana leads probably doesn't do much good...

[SparkyFX]

The capacitance decade might be fun with a 555 or other RC type setpoints. For applications with higher voltages and use cases sensitive to ESR value they are probably useless, as the requirements become very specific.

Otherwise decade boxes are a handy tool for breadboarding out designs, less waste and no "bent part" bin. Wouldn´t build one without access to cheap thumbwheel switches, although the commercial available decade boxes are horribly expensive.

[Conrad Hoffman]

General Radio Corp. made dozens of different types of decade capacitor boxes. They were common bench items used for all sorts of things. Typical accuracy was better than 1%. Silver-mica and polystyrene were most common. They also made some switchable standards boxes, not of the same accuracy as their fixed standards, but very handy. Also very expensive even today on the used market. What they did was a bit tricky. The GR 1417 used inductive voltage dividers, I believe a ratio transformer, to transform a single capacitance into various others. Same principle as their big standards lab bridges like the 1615. Not sure if it would be easy or hard to wind a suitable toroid for that.

[Specmaster]

The capacitance decade might be fun with a 555 or other RC type setpoints. For applications with higher voltages and use cases sensitive to ESR value they are probably useless, as the requirements become very specific.

Otherwise decade boxes are a handy tool for breadboarding out designs, less waste and no "bent part" bin. Wouldn´t build one without access to cheap thumbwheel switches, although the commercial available decade boxes are horribly expensive.

Well cheap thumbwheel switches is not a problem, depending on your finger size and type of resistor you plan to use they vary in sizes and cost, the smallest ones go for around £4 and the largest of the chinese ones go for around £8 for 10 switches. The second bank of switches arrived today (yes I ordered 2 banks from different sellers, a black and a grey one, (black is the bcd one and the seller didn't fill me with confidence as his description stated decimal but photo showed bcd, hence double ordering)) the grey one is ideal.

Here are 2 links to sellers on UK Ebay who are selling these, the grey one is the one that arrived today and the black one is another that I ordered yesterday and you see that this seller has included the 2 end pieces that are designed to clip the bank into whatever you're mounting them in.

https://www.ebay.co.uk/itm/10-x-Gray-Decimal-Single-Unit-Thumbwheel-Pushwheel-Switches-31mm-x-10mm-KM3/191621674657?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649

https://www.ebay.co.uk/itm/10pcs-One-Unit-Decimal-0-9-Digital-Pushwheel-Switch-Encoder-Thumbwheel-KSA-2/132399947213?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649

[SparkyFX]

The pieces i ordered are at the customs office, got to pick them up monday.

[Specmaster]

Good, I have just this minute finished the soldering up of my 7 decade resistor switch bank and in the process of verifying it all works OK. Looks like I've got an overall added impedance of 1.9 added on with the wheels set at 0000000 which given that the switch contacts are the weakest link, I think is pretty acceptable. Not after a lab grade instrument, just rough and ready but reasonably accurate value and as you step up in value so the 1.9 error gets significantly smaller.

To keep the errors as small as is reasonably possible without defeating the object on cost, I used 1% resistors up to 10K and .01% on the 100K and 1M and gives me a total of 9.984M which gives an accuracy at maximum resistance of just 0.16%. 

[Specmaster]

I just checked my Fluke 8505A using the decade box and it's off the wall above 4M, show the decade box has already proven its worth if you're into self calibration, I don't need Lab grade, accuracy to the 2 or 3rd decimal is good enough for the vast majority of cases and in my own case it is 100% of the time.

So thats another thing in the line up for some bench time then 

[SparkyFX]

Finished two 6-decade boxes this week (yeah, it took a while) and came across a similar problem:

I just checked my Fluke 8505A using the decade box and it's off the wall above 4M

checked both with my Siglent SDM3065X (6.5 Digit Bench Multimeter) in 2-Wire and 4-Wire mode, both decade boxes went off drastically above 5-6M. Didn´t think the brand new device has a problem, but measuring in the Megaohms range means leakage currents become relevant.
First i thought it was the isolation of the single Megaohms-Resistor crossing over the others from position 4 to position 5, bent it over to see if it changes readings... it did get closer, but only for a short amount of time. Because that grounded the circuit.
Then i grounded one of the multimeter leads and all readings were matching the set value quite close (from my expectations) up to 20MOhms with both in series.
Conclusion: Nothing wrong with it or my soldering skills, was just static.

[lordvader88]

I don't have any precision caps, but 1% resistors are a good start, I have lots of them. It's about time I make a Decade box. I just recently put some 10W load resistors on a perf-board. But I'm still using a ton of alligator clips/wires on it.

[Conrad Hoffman]

Just remembered that CDE (I think) made some cap boxes with BCD or similar switches. They made a big multi-section capacitor with the correct values for the switch scheme. The downside is that if any section goes bad, the whole thing is bad. You can quickly recognize them as they have dark red plastic cases and two rotary switches.

[ArthurDent]

There are different switching schemes to get the decade results you’re looking for. The use of ten equal components in series like the GR and ESI resistor decades use is perhaps the easiest to implement but there are others ways. You don’t actually need switches as my really old Welch resistor decade uses tapered shorting pegs with a 1,2,3,4 sequence. The CDE decade box use an encoded rotary switch with a 1,2,3,4 sequence of resistors and the Phipps & Bird uses individual slide switches with a 1,2,3,4 sequence of resistors. The IET uses thumbwheel switches and a 1,2,2,2,2 series of resistors and capacitors to go from 1 to 9 which is a little different. The high accuracy 6 decade ESI Dekabox I have uses coaxial stacked switches to give a wide range in a compact box.

The CDE capacitor decades are notorious for the capacitors going bad. The Aerovox I have uses silver micas and is still right on. The resistor decades generally last forever unless they are damaged. You can see on the photo of the IET decade box that one resistor had been damaged and I had to replace it.