capacitors drift and temperature characteristics.

[SArepairman]

Which capacitors have the best temperature coefficients?

I understand that polystyrene capacitors have a tempco of ~100ppm/deg C but they track in the same direction, so its possible to get pretty good matching.

What capacitors would you recommend for low drift/excellent matching for a timing application? Do they have matched tracking capacitors (like the matched resistors).

I read glass capacitors have almost no aging and only a tempco of ~60, but is it always in the same direction?

[Conrad Hoffman]

I think that polystyrene is also supposed to cancel out to a large degree the tempco of associated inductors in tuning circuits. For timing, I'd try to stick with NP0/C0G if possible. An old trick from the past is to combine silver mica and polystyrene in a certain proportion to cancel the tempco, but you might have to play with it a bit.

[SArepairman]

I think that polystyrene is also supposed to cancel out to a large degree the tempco of associated inductors in tuning circuits. For timing, I'd try to stick with NP0/C0G if possible. An old trick from the past is to combine silver mica and polystyrene in a certain proportion to cancel the tempco, but you might have to play with it a bit.

Well the ratio of the two capacitors matters the most, I would like them to track rather then to cancel anything out  AND minimize long term drift.
Right now I am thinking to try and match glass capacitors.

glass and silicon seem to be the only ones that dont drift with age.

[SeanB]

Depends on value, voltage and just how big you want them to be. Best is vacuum capacitors, but they are generally big, heavy and limited to a few hundred pF of capacitance, but can handle high voltage and high RF current easily. Small is NPO ceramic, available to a few nF and a voltage of up to 100V or so.  Higher values polystyrene is the best but is no longer made, and then you are left with polypropylene and polyester, which can go into the 100uF range up to 1000V, though not both at the same time.

[The Electrician]

Higher values polystyrene is the best but is no longer made.

I happened to notice on eBay the other day these beauties from Russia:

http://www.ebay.com/itm/Polystyrene-Capacitor-K71-4B-71-4-10-0uF-160V-2-1pc-or-more-/380791403351?pt=Vintage_Electronics_R2&hash=item58a8efb357

[T3sl4co1l]

C0G is available as large as 0.1uF (a bit more even, I think) thanks to MLCCs.  The days of massive honking discs are long over!  They're still relatively expensive though.

C0Gs are also very low loss, so they work well in resonant circuits.

N750s were the most popular for compensating coils, weren't they?  Still available I think, under a different type code.  SL comes to mind??

Tim

[SArepairman]

well, I need capacitor 10nF and 100pF. I would like them to be stable and have a matched drift.

How about air capacitors? Surely they must drift in the same direction due to mechanical expansion.

again, the absolute drift can be utterly shit, I just want the ratio to hold

[T3sl4co1l]

Seems like an odd ratio, might be a better way to do it?

Tim

[SArepairman]

Seems like an odd ratio, might be a better way to do it?

Tim

nmv, 100pf and 1nF

typo

[Conrad Hoffman]

Both should be available in NP0/C0G. Almost no drift with temperature and they generally don't change with age. There were some axial caps made by Corning Glass Works (marked CGW) with spectacularly low tempco and drift, but they're long gone. Glass caps aren't all that great, in spite of glass seemingly like an ideal dielectric. Almost anything else you do will probably be worse than the NP0 solution.

[SArepairman]

Both should be available in NP0/C0G. Almost no drift with temperature and they generally don't change with age. There were some axial caps made by Corning Glass Works (marked CGW) with spectacularly low tempco and drift, but they're long gone. Glass caps aren't all that great, in spite of glass seemingly like an ideal dielectric. Almost anything else you do will probably be worse than the NP0 solution.

Can yoou give me some drift numbers for the best C0G capacitors?

As I said the silicon capacitors have the right values, 15% tolerance (i don't care) and the one datasheet I found with a PPM number specified +62 PPM but the long term drift is practically unmeasurable.
Some typical C0G figures I see on the internet are ~30ppm.

WIkipedia says:
(ZnMg)TiO3    32    5
(ZrSn)TiO4    37    0 ppm
Ba2Ti9O20    40    2

Making the zirconcium tin titanium capacitor ideal....


Do you know where I can find (ZrSn)TiO4 capacitors?

or do you know the name of the corning capacitors so I can purchase some on ebay?

[SArepairman]

this is the best I found:

http://in.mouser.com/Passive-Components/Capacitors/Ceramic-Capacitors/_/N-5g8m?P=1yzrk0iZ1z0x7xvZ1z0wpzo

[Conrad Hoffman]

Yes, those CF series caps look really good. And you can buy them! The problem with exotic dielectrics is they often can't be had off the shelf. IMO, when you get into ultra high stability design, doing your own testing is the order of the day. I've seen the Corning caps on eBay, but there's no way to tell what's what. On line info on the various types is nearly non-existent. The few I have are, I think, 0.01 uF and show essentially zero TC on a GR 1615 bridge, maybe on the order of a couple PPM. I have no part number for them. If you can live with a significant but known TC, the quartz standards used in the Andeen-Hagerling bridges should have excellent long term stability. http://www.andeen-hagerling.com/ah2500a.htm

[vk6zgo]

Higher values polystyrene is the best but is no longer made.

I happened to notice on eBay the other day these beauties from Russia:

http://www.ebay.com/itm/Polystyrene-Capacitor-K71-4B-71-4-10-0uF-160V-2-1pc-or-more-/380791403351?pt=Vintage_Electronics_R2&hash=item58a8efb357

One problem you may find with high values of Polystyrene caps is high inductance,due to their construction.

Back in 1974,as part of modifications of an old Marconi TV Transmitter for PAL colour use,we replaced the  0.47uF Metallised Paper coupling caps in the "Sub-modulator"(which were looking a bit worse for wear),with nice new Polystyrenes.

When we swept the modified device,we were horrified,as the frequency response was "fish-shaped".
We re-fitted the old caps & all was OK,tried the new ones,& the "fish". was back!

We couldn't find anything else in the voltage & capacitance rating,other than waxed paper,so we used those.
Musn't have been too bad---lasted around 20 years!

I've seen plenty  of polystyrenes used in HF circuitry since then,but always lower values.
Maybe there is a change in construction with the higher values which causes the above problems.
I have noticed that the large values are usually bigger physically than equivalent values using other dielectrics---the Russki ones look quite large,too!

[SArepairman]

Hmm, the CGW capacitors that I found on ebay seem to have a data sheet temperature coefficent of ~60+-25PPM (all of them that I found are around the same). This seems to match up with the various datasheets that I found when I googled glass capacitor. Everything I found tells me that ceramic capacitors are better then glass capacitors in terms of tempco.

When you say old, do you mean the ones with the dots code? I only found one CGW capacitor with the dot code on ebay (i think its 180pF), I cannot find the datasheet for it...
I really would like to find what you are talking about

Those C0G are good but I think I might go with silicon because they keep beating the long term stability thing to death on the manufacturers website..... plus the inductance is lower, which actually might help my circuit (there will be less nonideality associated with frequency). Its not something I considered initially but the C0G and silicon seem very close. Those 15ppm C0G's might be better... now I need to determine the effect of the parasitic inductance on my circuit.

I am glad inductance was brought up because it might end up being a bigger problem then temperature coefficient!

[SeanB]

Poystyrene caps are wound differently depending on manufacturer and intended application. Most are wound with a single leadout tab at the end of the foil, and have high inductance, but are intended for timing applications or use at audio frequencies where the inductance is not going to have much effect on the response. Others have multiple tabs that are bonded together during assembly to a leadout wire, or have an extended tab at each end intended for metal spraying to make a capacitor intended for high current pulse applications. The latter types are generally also encapsulated in epoxy after they are heat sealed, as the leadout wires are very fragile and will break off with normal handling unless reinforced.

[Conrad Hoffman]

Maybe Corning just got lucky with a particular batch, but every one I've tested has been crazy good. It's a small greenish blue axial cap, 0.375" long and 0.127" diameter. The markings are: CGW, 103, K5L and 8416. If anybody has data on what they really are, I'd love to know. I have some other CGW parts that are similar, but with higher TC.

[SArepairman]

http://www.ebay.com/itm/10pk-1300pf-50V-Capacitors-Corning-CGW132K5L8232-/390533500249?pt=LH_DefaultDomain_0&hash=item5aed9c4d59

are those the ones? they seem to fit your description.

is there anything wrong with paralleling them?

[Conrad Hoffman]

Yes, that's them but with a different value. Certainly you can parallel them with no problem. I've built standards using them in series to get down to the standard 1000 pF reference value. The only problem is the guy only seems to have a couple left, and I have no idea of other values or batches will have the same performance.

The problem when you start working at the levels you describe is that data sheets are generally worthless. They tell you what not to use, but only testing will guide you further. Everything today is digital and crystal controlled, so the demand for low PPM capacitors is nil. Analog may not be dead, but the threads here on very precise voltage references and such are the exception, not the rule.

[SArepairman]

I have ordered some similarly colored glass capacitors and i will compare them.