Resistor impedance vs frequency

I am interested in high frequency behavior of normal hole mounted metal film resistors. It would be nice to be able to prototype some high frequency digital circuits without SMD components.

***

General rule for SMD resistors from memory:

Low value resistors: impedance increases at high frequency (with 1 ohm resistor much earlier than 100 ohm resistor)

High value resistors: impedance decreases at high frequency (with 100 kohm resistor much earlier than with 1 kohm resistor)

The "turning point" is typically somewhere between 100 ohms to 1 kohm.

***

Philips (is it part of Yageo now?) has published impedance vs frequency graphs for their resistors, but if I remember correctly only for SMD types.

I can't find aywhere the same info for hole mounted resistors. SMD components have been in use relatively short period of time, so someone must have made these measurements.

It is quite strange that resistor manufacturers don't have this info available.

I found those equations too. But the problem is what to insert in them. Resistor datasheets give nothing.

I found impedance vs frequency plots for SMD resistors but HMD parts probably behave quite differently.

I checked some random metal film Vishay-Dale 36 Ohm, 2.11K, and 100K resistors on my HP 4274A LCRZ meter, which measures R and vector Z on frequencies from 100Hz to 100KHz, with different oscillator output (I tried 50mv and 5V). On all the frequencies with either output both R and Z were quite uniform at the nominal and the deviation was +/- couple /100ths across the whole bandwidth, which very well might be due to the instrument calibration.
I also have an HP 4800A, which gives impedance measurements from 5Hz to 500KHz. If you are interested I can put some resistors on that one.

Best, M  

In frequency range up to 100 kHz or even 1 MHz (depending on resistor value) HMD metal film resistors can be treated as ideal components. Now I am talking about impedance vs frequency, there are naturally other imperfections too.

The frequency range I am interested is something like 1-30 MHz.

Example of SMD resistors (0805 size):
100 ohm resistor works fine up to 1 GHz. Impedance of 1 Mohm resistor starts to change radically after 1 MHz. I guess there is a similar pattern with HMD resistors too but the change occurs earlier. But I have no practical information how much earlier.

huh?

I work in 1-3 ghz RF and I can tell you that nobody even considers resistor impedance change over frequency to be a factor in designs...  At least I've never seen or heard of a resistor changing impedance/resistance over frequency in school or in the work force..

SMD Resistor datasheets give nothing because there is nothing to give.  They are readily known to be fairly linear across frequency even though they might model as having L/R/C components.  You have to remember that these parasitics are extremely small in today's quality SMD parts and you would have to be working well into the 20ghz range to be affected by small SMD resistors.

Now I have seen the package parasitics play a role in impedance changes in leaded parts but only because you are forming filters with the parasitic L of the leads, the parasitic C of the package/proximity and the resistance of the part.  The parasitic filter changes your S-parameters and you'll get a wonky frequency response.

There are other things too like improper layout.  A lot of designers will have matched impedance traces but will forget to match the package of the part as well and then you have discontinuities which will also give you funky reflections and again your S-parameters will be all messed up.  This is especially true around series resistors because the signal will have the strength to drive through the resistor but the reflection will not and this might make the downstream side of your resistor look like the resistor is causing a frequency anomaly when it's simple phase problems.

Svart said:

I work in 1-3 ghz RF and I can tell you that nobody even considers resistor impedance change over frequency to be a factor in designs... At least I've never seen or heard of a resistor changing impedance/resistance over frequency in school or in the work force..

Click to expand...

From Philips/Yageo graphs (for 0805 SMD resistor):

1 Mohm resistor impedance at 10 MHz is about 200 kohm, at 100 MHz it is close to zero.

100 ohm resistor has the best behavior, about 20% variation of impedance up to 10 GHz. In RF applications resistor values are usually quite close to that. That is the reason you have no problem with your designs. Smaller packages naturally give even better high frequency performance.

If you increase or decrease your resistor value from approximately 100 ohms, the performance gets worse quite fast.

On a 10 Mohm/0.5 W I measured 120 fF at 100 kHz/5 Vrms. A 1 Ohm/0.5 W gives 18.6 nH. Measured with a HP 4274A.

Samuel

Again, it's important that you clarify that you are speaking of package parasitics vs resistance, not a real impedance change of the resistive element.  Your initial post did not discuss anything but resistance value vs. frequency which gives the false assumption to a reader that you are saying that the resistance changes over frequency without external influence, which is false.

I don't design anything RF using resistors larger than 0402 generally.  My next project will be in 0201!

Svart said:

Your initial post did not discuss anything but resistance value vs. frequency which gives the false assumption to a reader that you are saying that the resistance changes over frequency without external influence, which is false.

Click to expand...

My post was about resistor (name of a component) "impedance vs frequency". I haven't mentioned word "resistance" at all. As far as I know "resistance" is concept used for direct current only.


Back to the original subject: There must be impedance vs frequency graphs for hole mounted metal film resistors somewhere (similar to those graphs I have for SMD resistors). Most likely in old databooks since no one has used HMD resistors in RF designs for a long time because of the availability of much better SMD components.

Samuel Groner said:

On a 10 Mohm/0.5 W I measured 120 fF at 100 kHz/5 Vrms. A 1 Ohm/0.5 W gives 18.6 nH. Measured with a HP 4274A.

Click to expand...

Thanks. I think your values are a good starting point. I did simulations and got results quite similar to those Philips/Yageo graphs for SMD resistors. Except the bad behavior starts slightly earlier as expected.

I am going to do some practical measurements to find out if my simulations are too optimistic or not. I don't have an RF generator and level meter, but some of my friends must have.

By the way, what was the resistor wire length in your measurements?

By the way, what was the resistor wire length in your measurements?

Click to expand...

As short as possible--perhaps 2 mm free run from body end to test fixture on each side. Capacity increased to 200 fF when I lifted the resistor ~5 mm above the fixture. Didn't check that for inductance.

Samuel

> you are saying that the resistance changes over frequency without external influence, which is false.

No, real-resistance does change with frequency. On carbon-composition, stray capacitance bleeds from lead to carbon-lump, and the effective resistance tends to drop.

> As far as I know "resistance" is concept used for direct current only.

No, "resistance" has real meaning in AC. Svart likes to load in perfect 50 ohm resistors. Then the customer tells him the product is too small for a good antenna, and the best he can do is 73 ohms reactive plus 37 ohms pure resistance (often written as A+jB notation). In antennas, generally the pure resistance is "real work" (signal sent into space) and the reactance is in the way.

Kuehnel's Preamp book has some old-hole resistor frequency data.

> find out if my simulations are too optimistic or not

Hang a capacitor on EVERY node. Put an inductor in EVERY lead. If you don't, it is unrealistic.

In tube-amp work, I assume 10pFd-30pFd everywhere. Usually I know that other parts are loading 100+pFd already, and I don't need a sharp HF cutoff just a supersonic corner, so this 10%-30% error is "small". I should also add stray inductance, except I "know" that this does not matter in audio until you get down to loudspeaker impedance, and even then it is small.

Obviously in micro-SMD work, this parasitic capacitance is smaller due to smaller parts and traces, and inductance is less due to short traces.

> 100 ohm resistor has the best behavior

Inductance and capacitance are BOTH related to "size". Certain proportions lead to certain impedances. Co-axial cable "wants to be" 40-50 ohms. You can make it 75 ohms or 105 ohms only with very small center conductor and large outer conductor. Wide-space conductor pairs (old-old telephone lines) run 500-1K, but twisted-pair usually comes out 90-130 ohms. "TV twin-lead" was 300 ohms because it had a wide separation.

I think that nearly any "practical" part tends toward 50-100 ohms at something above GHz. That's just practical proportions and some universal constants for permittivity and permeability. Your real question is: at what frequency are your parts asymptotic to ~~100r? Then you can work backward and estimate where your "1Meg" resistor has an effective impedance significantly different than 1Meg.