Maker Pro
Maker Pro

NTC caps, where are you?

J

John Larkin

Jan 1, 1970
0
We'd like to temperature-compensate a 50 MHz LC oscillator. The
inductor is generally the bad guy, typically in the +100 to maybe +150
PPM/K range. So if we use an NPO cap, a piston cap trimmer, and an NTC
cap all in parallel, we could maybe get the TC down around, say, 20
PPM in production. So I asked one of my guys to get us an 0805 or 0603
sample kit, low-PF range, TC's like N750 or something. Duh: there's
apparently no such thing, and Kyocera only makes NTCs to order in 100K
and up batches.

So, does anybody know of a source for NTC surface-mount ceramic caps?

Thanks,

John
 
T

Tim Wescott

Jan 1, 1970
0
John said:
We'd like to temperature-compensate a 50 MHz LC oscillator. The
inductor is generally the bad guy, typically in the +100 to maybe +150
PPM/K range. So if we use an NPO cap, a piston cap trimmer, and an NTC
cap all in parallel, we could maybe get the TC down around, say, 20
PPM in production. So I asked one of my guys to get us an 0805 or 0603
sample kit, low-PF range, TC's like N750 or something. Duh: there's
apparently no such thing, and Kyocera only makes NTCs to order in 100K
and up batches.

So, does anybody know of a source for NTC surface-mount ceramic caps?

Thanks,

John

Yea, you're supposed to think "PLL" instead of "NTC" these days.

Can you just use a through-hole part? With the right lead forming you
could surface mount it :).
 
J

Joerg

Jan 1, 1970
0
Hi John,

Now this one is really kludgy but if it absolutely has to be all SMT: How
about using a varicap diode controlled by an NTC or PTC resistor?

I have used that once in a remote temp sense app before the advent of UWB,
a VCXO in a narrow ISM band.

Regards, Joerg
 
J

John Larkin

Jan 1, 1970
0
Yea, you're supposed to think "PLL" instead of "NTC" these days.

Normally that, or even better DDS, but not in this case. This is an
instant-starting oscillator, sort of a weird application.
Can you just use a through-hole part? With the right lead forming you
could surface mount it :).

Ack, last resort. We want this to be small. The alternative is a temp
sensor+opamp+varicap, which is all surface mount, easier to tweak, but
not terribly appealing.

John
 
T

Tim Wescott

Jan 1, 1970
0
John said:
On Thu, 22 Apr 2004 15:47:59 -0700, Tim Wescott

Ack, last resort. We want this to be small. The alternative is a temp
sensor+opamp+varicap, which is all surface mount, easier to tweak, but
not terribly appealing.

John

I was going to suggest that myself, but "not terribly appealing" pretty
much sums it up. You may be able to get by with a varicap and a
forward-biased diode or two, but you're still heavily into "not
appealing" territory.
 
J

John Larkin

Jan 1, 1970
0
Hi John,

Now this one is really kludgy but if it absolutely has to be all SMT: How
about using a varicap diode controlled by an NTC or PTC resistor?

I have used that once in a remote temp sense app before the advent of UWB,
a VCXO in a narrow ISM band.

Regards, Joerg

Right, we may resort to doing that. At least it would be easy to trim
the TC... just change a resistor somewhere instead of juggling
capacitor ratios. Maybe an LM45 sensor and an opamp to drive the
varicap. The varicaps themselves have rotten TCs, but that will all
wash out in the end. I'd really like to keep the net TC below 20 PPM,
better if possible, without temperature cycling and adjusting each
production unit.

John
 
J

Joerg

Jan 1, 1970
0
Hi John,

If it's not much range you need to correct for TC maybe it could even be done
without an opamp. Something like a voltage divider controlling the diode
reverse voltage, with the divider so that the reverse voltage rises with temp.
Of course, yes, you'd have to add in the TC share of the diode. Unfortunately
since VR changes here, its TC isn't linear. That could be compensated again by
adding a forward based regular diode on the DC side but then it is getting
really ugly.

The only upside I see here is that if the competition ever disects your circuit
they'll have a hard time figuring out how the heck this works.

Regards, Joerg
 
F

Fred Bartoli

Jan 1, 1970
0
John Larkin said:
We'd like to temperature-compensate a 50 MHz LC oscillator. The
inductor is generally the bad guy, typically in the +100 to maybe +150
PPM/K range. So if we use an NPO cap, a piston cap trimmer, and an NTC
cap all in parallel, we could maybe get the TC down around, say, 20
PPM in production. So I asked one of my guys to get us an 0805 or 0603
sample kit, low-PF range, TC's like N750 or something. Duh: there's
apparently no such thing, and Kyocera only makes NTCs to order in 100K
and up batches.

So, does anybody know of a source for NTC surface-mount ceramic caps?

Thanks,


Do you, by chance have a uc on board ?

http://www.maxim-ic.com/quick_view2.cfm?qv_pk=2450

Fred.
 
J

John Larkin

Jan 1, 1970
0
John,

Check this one out. Look at table A at the end of the documnet. I am not
sure but maybe they have some suitable values in negative tempco:

http://www.murata.com/catalog/c02/es0014.pdf

Large tempco tolerances though.

Regards, Joerg

Hmmm, interesting. The -470 +-60 is the tightest, and that (if I
hacked the math sorta right) compensates a -120 PPM inductor within
about +-15 PPM, assuming I can get the capacitor ratio exactly right.
We'll try to get some.

Thanks.

John
 
B

Bill Sloman

Jan 1, 1970
0
John Larkin said:
We'd like to temperature-compensate a 50 MHz LC oscillator. The
inductor is generally the bad guy, typically in the +100 to maybe +150
PPM/K range. So if we use an NPO cap, a piston cap trimmer, and an NTC
cap all in parallel, we could maybe get the TC down around, say, 20
PPM in production. So I asked one of my guys to get us an 0805 or 0603
sample kit, low-PF range, TC's like N750 or something. Duh: there's
apparently no such thing, and Kyocera only makes NTCs to order in 100K
and up batches.

So, does anybody know of a source for NTC surface-mount ceramic caps?

IIRR the telephone industry solution to this problem was inductors on
nickel-zinc ferrite cores that more or less compensated polystyrene
dielectric capacitors, if for frequencies around 5MHz rather than
50MHz. Never used the approach myself, but I did see it mentioned in a
Philips ferrite data book.
 
Top