From: John Larkin
Subject: Re: Modelling of Chip Resistors at 2.4GHz
Date: Mon, 14 Oct 2002 09:08:57 -0700
Organization: Posted via Supernews, http://www.supernews.com
X-Newsreader: Forte Agent 1.91/32.564
On Sun, 13 Oct 2002 02:01:37 GMT, firstname.lastname@example.org (carltons) wrote:
>In article , John Larkin
>> On Sat, 12 Oct 2002 09:05:39 GMT, email@example.com (carltons) wrote:
>> >Does anyone know of a source of information about the model for a standard
>> >chip resistor such as a 1206 or 2512 resistor at a value of around 100 ohm
>> >at 2.4GHz?
>> >I'm not talking about the microwave type, but just the plain jane Koa,
>> >Panasonic, etc. resistor that we normally use.
>> >Thanks for any help.
>> >Steve WB4CZR
>> Rough guess for a 1206: 1 nH + 0.2 pf. So it's mainly R+L at a couple
>> of GHz, and actually a pretty good resistor.
>> One other issue is the problem of making connections. If a circuit
>> board trace is just the width of the resistor, and maybe the other
>> side of the resistor is a big ground sheet (say, for a termination)
>> things are pretty good. But if a wide trace ends in a skinny part, the
>> inductance goes up. A decent termination for a 0.1" wide, 50 ohm trace
>> is three 150-ohm 0603 resistors in parallel, spread across the trace
>> width. So paralleled 1206s might be better than a single 2512. Or use
>> a single 1206; soldered to wide traces, it can stand well above its
>> rated power dissipation.
>Thanks guys. I would have guessed about 0.7 nH John, but we're both close
>to one another on this. I've been having this argument going with a
>fellow engineer involving the 100 ohm resistor (1206) used in one of our
>splitter/combiners. I contend that we use a 2512 since it would quarter
>the shunt C of the part and lower the inductance at the same time. I also
>have seen some data on the web (I don't remember where but it was from one
>of the big vendors) that the effective resistance of the standard resistor
>at 2.4 GHz is about 60% of its low frequency value due to shunt C. I
>can't really see a big increase in R due to skin effect even at 2.4 GHz
>since most of the resistors are so thin to begin with and have the correct
>shape to compensate anyhow. BTW, the 2512 would also be about the same
>width of the microstrip now in the combiner's Zo lines which should also
>be desirable. I have also come up with a configuration change which could
>make all of this less sensitive to the parasitics of the resistor, but I
>would still like to know what to use as a resistor model for any freq up
>Problem: We have old equipment and it's hard to measure S11 with
>reliability. We don't have any fancy design tools like Sonnet or
>Eagleware. What we both agree on is the fact that the layout does affect
>the performance of anything we put in there. I wish I had a HP8753D and a
>decent fixture, not to mention the cal kit. (I'm dreaming again)
Hey, there's a pile of equipment in the lab waiting to sent out for
cal, and there's a Boonton 72 c-meter near the top of the heap. A
1-meg 1206 resistor measured about 0.1 pF, or maybe a bit less. This
is tricky to measure in free-air, as the leads themselves have
capacitance. I just stuck one end of the r on one terminal of the
meter, and brought a lead from the other terminal pretty close to the
other end of the r, zeroed the meter, then touched the resistor.
I guess I could TDR the resistor sideways along a piece of coplanar
waveguide on a PCB; that would be fair, I guess. Nah, too much work.
A 450-ohm 0805 resistor soldered onto the end of a 50-ohm hardline
makes a damned fine 10:1 probe, well into the GHz.