Capacitance from diodes

In the Waveshape, timbre, etc. (http://www.thereminworld.com/forum.asp?cmd=p&T=4206&F=3) topic, I pasted schematics and suggested using a standard power diode, such as a 1N4007, as a caricap diode.

I have been using these parts in non-critical sections of my design (HF VCF's etc) but am evaluating their use in some more areas, such as HF oscillators, where thermal drift is important. I am not too happy with the results for these applications - they pass - but only just.

Anyway - I thought I would share some of my test results which give capacitance vs reverse voltage.. Someone may find this useful. Please be aware that these tests were quickly done, and cannot be taken as any kind of 'gospel' .. Also, I have found parts vary hugely from one manufacturer to another - particularly if using parts in applications they are not specified for.

Test circuit and results (http://www.therasynth.com/html/diodecap.html)

EDIT -> I have now located data regarding the thermal influence on reverse biased diode capacitance.. It seems that a change of 1 degree C is equivalent to a voltage change of about 2mV..

If the above is true, it may explain why a higher control voltage range is required.. If the CV range is 1V, then there would be an equivalent 0.02% change in capacitance per degree.. with a CV of 20V, this reduces to 0.001%/C ....

Alas, even 0.001%/C is not good enough for direct use in an application like a high frequency VCO for music applications - The only way I can see to use diode capacitance for applications like this is by using an 'ovened' diode array (where all the diodes are on the same substrate, and one is used to monitor the substrate temperature, and this is used to drive a heater which keeps the array at a constant temperature.. I am using this method with a transistor array on my present design, and was hoping that using varicaps would reduce the complexity.. )-:

Fred wrote: [i]"I have now located data regarding the thermal influence on reverse biased diode capacitance.. It seems that a change of 1 degree C is equivalent to a voltage change of about 2mV.."[/i]

Fred, this is nothing new, it's not the capacitance which is depending directly on the temperature, but the forward voltage and following this, the reverse current and breakdown voltage and only following this, the junction capacitance. Simple Zener diodes are still worse, their reverse breakdown voltage has a temperature coefficient of -4mV/K. So one used to take one or two "normal" diodes with +2mV/K in series with a Zener diode in order to compensate such effects.

Perhaps it would help if you took the reverse idea, compensating your varicaps with Zener diodes in series... ;-)

Hi Thierry,

Oh, I had absolutely no thought that I may have discovered something new.. LOL

There are pages of complex (to me, anyway ;-) equations relating to the thermal behaviour of semiconductors, and even more data (and maths) relating to behavior at an atomic level..

What I wanted was a simple 'rule of thumb' regarding the temperature dependency.. and finding the 2mV "rule" helped.

I realise that capacitance, per se, is not "directly" affected by temperature - There are only a couple of variables which could affect Capacitance .. Main being the distance between plates (thickness / nature of dielectric) and area. When reverse biased, it is the depletion zone which acts as the dielectric.. The thickness of this zone is dependent on the reverse EMF.. But, obviously, electron 'infiltration' into this zone will be directly proportional to to temperature.. Therefore higher temperature = greater infiltration = narrower effective depletion zone = higher capacitance... I am not sure if the "forward voltage" is really the issue here - Forward voltage is a function of the 'width' of the depletion layer - a 'width' which needs to be 'overcome' when forward biasing the diode.. But you are right, they are effectively the same mechanism.. And this insight is helpful - When thinking about reverse biasing, I had forgoten this relationship - the 2mV/K for forward voltage is obviously equally valid when reverse biasing, as it relates (is almost entirely a function of) the width of the depletion zone.. Duh!


Regarding zener compensation - This would probably work well IF one could obtain Varicap diodes with a Zener built on the same substrate.. but as soon as one has seperate components acting to thermally compensate each other, one has problems with thermal coupling between these components.

Unfortunately I do not have the time to fully explore the schemes employed for thermal compensation of varicaps .. I know they exist, and suspect that Radio enthusiasts would laugh at my postings on this matter.. I have an oscillator (VCO) which works extremely well, using polystyrene film capacitors with slight NTC, and transistors held at constant temperature in a CA3083 NPN array.. (in fact, I use the 2mV/K from one of these transistors as a temperature sensor to control heating of the substrate) Thermal error on the resulting (post heterodyne) audio frequency from this is below what my instruments can measure..

I have been using diode capacitance for a number of waveshaping circuits, but have been doing this 'on the bench' without actually knowing how the parts behaved (what the capacitance spread, or thermal effects were) - Also, I used preset potentiometers extensively on these prototypes.. Presets I now wish to get rid off - So I thought it was time to do a little more examination of the diodes charactaristics.. And also to see if I could use them to simplify my HF-VCO..

And, as I was doing this, I thought I may as well make the examinations available to anyone who was interested..

One can get full data on the charactaristics of Varicap diodes.. But these are mostly available only as SMD, and are expensive. I have never seen data relating to the use of 1N4007's when used as a capacitance controlling element.

Thanks to Thierry getting me to think about it - I believe there may be a simple way to compensate for the thermal error..

A diode array (such as the CA3039) with one diode being used in forward biased mode, and the forward voltage from this diode (which will change at 2mV/K) is added** to any CV used to reverse bias the other diodes ..

seems too simple - But it should mean that lower CV's could be used*, therebye exploiting the zone where the greatest capacitance change occurs..

Now off to order some CA3039's (Anyone know of any other diode arrays? the more diodes the better - but not less than 3.. And hopefully not SMD.. And not difficult to get, as the CA3039 seems to be a part either discontinued or scheduled for termination..?)

*CV must always be greater than signal amplitude

**In fact, the Vf will decrease 2mV/K, so a small circuit may be needed to null for say 20/C and invert the -2mV/K to give +2mV/K which can be added to the CV.. All depends on what circuitry is producing the CV.. If this is an inverting summing opamp (as it most often is) then feeding the Vf into one unity-gain input will do the job.

Fred Mundell
Fundamental Designs Ltd.
Electronics Consultant.
<- See Profile Image for Email.
Designer of Theremins and other alternative electronic music controllers and instruments.

I have just confirmed the above..

Using a CA3083 NPN array, forward biasing one transistors b-e and using its Vbe as a reference, I was able to reverse bias the other transistors an use them as varicap diodes.. Simply inversely summing the Vbe of the first (transistor) diode with the control voltage applied to the other (reverse biased) transistors, I was not able to measure any effect as a result of temperature change.. Ok - This does NOT mean it is perfect.. I can only measure changes of 0.5pF directly (I will later connect the varicaps to one of my oscillators, which will give a much better idea about stability)

One slight difficulty using a CA3083 is that the transistors will only stand a reverse b-e voltage up to 5V before they do strange things..

Anyway - Thats all for now.. I have more important tasks to complete.. Modifications to my HF-VCO can wait, and I have enough data to do the required work on the waveshapers..

But I will add the results of my later tests to my page when I get a chance.