Variable capacitor

I need to make a variable capacitor from 1000 pF to 19,000 pF (roughly). Using something like a 11 position rotary switch is fine. I can build something homegrown, but would like to know if there is something already available, so that I am not missing anything obvious. If I should build it, I will appreciate some design tips so that I do not miss anything obvious. Thanks.

Thanks. I need an assortment of caps, from 1000 pF to 10,000 pF. It would be best if they were of axial type, makes them easy to parallel and solder together. What kinds of caps would be best? Polyethylene film or ceramic or what? I do not need much from them in terms of accuracy or voltage.

Yes. There will be another, parallel non-switched 1,000 pF cap. This is for frequency regulation. With 1 megohm pot for duty cycle adjustment, and permanent 1,000 pF cap, I would have the following 11 values for frequency:

pF Hz

(that's for a low frequency switching circuit, hence very low frequency).

The exact values do not matter terribly too much. What I want is to find a cheap assortment of radial caps that would let me implement these capacitances easily.

Use a BCD or binary coded switch ( or make one from multiple poles). Thay way you can get 16 values from 4 capacitor values - see also :

Think "1-2-4-8-etc, binary" and a small array of miniature toggle switches.... BTDT, very useful.

That's interesting, but in my case, I want to select frequency with a rotary switch, not with binary bit toggle switches. Maybe one day I will sell this welding machine, and no one will understand binary bit logic except for those literate in copmputer logic and such, so no one will want to buy this welder.

Regarding selling it, it is not at all my current purpose. It would be uneconomical to spend so much time to make a few hundred bucks. I want to have a super nice modern welder for myself, made out of this cyber tig, hence my current project.

That's my plan. Since I do not know what the future will hold, I have to make sure that I end up with something sellable.

search for capacitor decade box, they are quite expensive, so a couple of rotary switches will do what you want.

martin

No problem, you can get rotary switches with a binary output. Google "rotary hex switch".

Given this low frequency, you could make a synthetic variable capacitor using 1 cap, a few resistors, a pot and a dual opamp. You'd need both + and - bias supplies for the opamp. Varying the pot varies the effective capacitance to ground.

Since this is not a sine wave oscillator, why not tune it with a variable resistor; like a nice 5 or 10 turn pot? You would only need 2 or 3 capacitor values to cover the range. A 74HC4046 should work well.

Another approach that I would look at is to cover the top octave, and use some binary counters to get the lower frequencies. Use the rotary switch to pick the octave.

There are also a number of ways that you could do it with a single tuning knob, and no band switch.

Tam

Tam

The cap is "grounded" by connecting to the summing junction (virtual ground) of an opamp with grounded noninverting input. A feedback resistor is connected from output to summing junction. Output voltage is -R * Icap. The cap will appear grounded to external cctry, e.g. a 555 or other timing or oscillator chip that uses a grounded cap.

Drive a bidirectional current source aka transconductance amplifier with the output of this amp. See an opamp book for how to make a bidirectional current source with an opamp and 5 resistors. Output of current source is connected to high side of cap. The effective value of the cap (current for given dV/dT) is then C*(1 + K1*K2) where K1 is the gain of the transZ amp (volts per amp) and K2 is the gain of the transconductance amp (amps per volt).

Using a pot for either the fb resistance in the trans-Z amp would vary system gain and hence the effective capacitance.

The alternative is 74HC132 device which do the same job as 555, except it simplier and more flexible. It can produce nice squarewave output or other duty cycle waveform depending on waveform and it work on 2.5V to

10-15 turn pot is availble cheaply and easier to tune freq than caps. You can switch three cap for x1, x10, x100 freq range and 10 turn pot for between 1 to 10.

Yes, that's my plan indeed, it will be approximately exponential scale, from 50 to 500 Hz.

I do intend to keep it. I would not undertake this project to make a few hundred bucks. In the time that I spend on this, I could have made thousands of $$s by computer programming.

I agree. There will be some drilled holes in the front panel, but not many. I could cover them with a picture of a naked woman or some such.

that's interesting, can you elaborate?

No. The external circuit still charges/discharges the capacitor. Current charging (or discharging) the capacitor is sensed, amplified, inverted and injected into the node charging the cap. It's inverted so it flows in the same direction in or out of the node as actual capacitor current.

Since this current is an inverted amplified replica of actual capacitor current, it appears to the circuit as another capacitor connected from actual cap to ground. If the gain is K amps/amp, then the external cctry sees a load (to ground) that looks like C + K*C or C*(1+K).

If this is for varying the frequency of your weld pulse thing, then I would suggest going for caps in the ratio 1,2,5,10,20,50,100,200,500 etc. or even just 1,3,10,30,100,300 etc. since that way you cover a large range in few steps. I doubt you'll see much difference between 180Hz and 200Hz or whatever it is in your welder, but it might be more interesting to explore a wider range.

Also forget about selling it, I reckon. If someone buys it then it will be more pain than it's worth, they'll keep calling you to fix it or they'll break it and accuse you of causing everything that ever went wrong in their life. Just build it to work the way you want, maybe you'll even feel like keeping it.

The only concession I would make with regard to selling it, is to make the modifications to the welder itself basically reversible so you can get back to what you started with.

Chris

Check out datasheet for XR2206:

the pot is used for duty cycle adjustment

I am not quite up to date on all the terms that you use, sorry. I am new to electronics.

If you can't get a binary encoded switch, you can use one with multiple decks to do the same job. Fir that matter, you can get the same range of values with less capacitors if you want.

I don't suppose I could interest you in a single chip solution, no weird stuff to glue together into a messy bundle? 14 pins, only Vcc, gnd, and all