Low Cost High Frequency Chip Resistors

[trentender]

Hi-does anyone know a source of low cost 0603/0805 style chip resistors which have very good rf performance up to about 5GHz.I know thin-film is better than thick film for rf performance but everything I can find seems to be specified up to 40GHz+ which is way over the top for my application plus the parts become very expensive and are soldered usually face down rather than by end connections.

[T3sl4co1l]

Normal parts are usable at that frequency. There's not much that can be improved by design, that isn't dominated by body size already (0402 would be preferable, particularly for values very different from Zo), and particulars can be resolved with modeling or adjustment.

Tim

[Hamelec]

may be Vishay FC Series is an option

[PartialDischarge]

If you are working up to 5GHz I can't believe resistors are your cost problem.... What about your substrate and actives?

[tszaboo]

If you are working up to 5GHz I can't believe resistors are your cost problem.... What about your substrate and actives?

You would think. But an LNA or a SAW filter could be 50 cents, your standard FR4 with controlled impedance stackup can easily handle 3-4 GHz. And that just means they use the same substrate not some random one. 5 GHz wifi chip is also going to cost peanuts, and 2.4G will be below 1 dollar.
Meanwhile the FC series is 4-6 dollar in single quantity, and not cheap in large. And regular resistor is more inductive than resistive even at 5 GHz.

[TimFox]

What value of resistance do you need?
Any critical values besides 50 ohms?

[Hamelec]

4-6 dollar? More in the range between 1-2€.. single quantity

[tszaboo]

4-6 dollar? More in the range between 1-2€.. single quantity

I must be seeing things then 
https://www.digikey.nl/en/products/detail/vishay-dale-thin-film/FC0402E50R0BST0/7102843
UNIT PRICE   EXT PRICE
€6.00000   €6.00

[babysitter]

If its a one-shot, go for smallest size, consider mounting them upside-down, so the resistance layer is closer to trace plane?

[trentender]

Good advice, I will try out fitting an upside down but quality 0402 chip resistor (smaller meaning less parasitics) and see hows things pan out.Rather pay £0.01 for each chip resistor than £1.00+

[tszaboo]

Unfortunately, it's not that simple. The issue is with the internal capacitances of the resistor. Flipping it may reduce the inductance (in exchange for lower power handling) but it doesn't do much with the capacitance.

[T3sl4co1l]

Namely the end metallization is wraparound, so it gets a nice big chunk (relatively speaking) of parallel capacitance from the Al2O3 body; the special ones are a "flip chip" style I think, solder pads (DFN style) and element on the same side.  Field divides through thickness and length of the substrate, capacitance is lower.

Adjusting substrate thickness, and possibly material (thus, K), are other variables they might play with.  No idea if anything other than Al2O3 is ever used, actually.

Have seen a few precision resistors in the style, but being 0.1% or better, the cost might not be what you're looking for, anyway.

Other tricks can be played, like ground relief under the pads or body.  You really need a simulation model to do stuff like that, and if you're asking about specialty resistors at these frequencies in the first place, you really need those tools regardless.

Tim

[tszaboo]

What is strange for me, is that you can buy SMD attenuators for the fraction of the price than a single RF resistor. And they also work to 10GHz lot of the time. I even see 0dB attenuators. Probably to fill up the gap if you designed one in. Though, makes me think, these are just a Pi network of resistors, and what would happen if you mis-use them in circuit. Unfortunately, they never seem to disclose the internals, only the dBs, and in a Pi network, you obviously never have exactly 50 Ohm.
But obviously, they know you only need resistors to build test equipment, and that's the price hike.   

[mtwieg]

For just 5GHz I don't think fancy flipchip/ceramic resistors are appropriate. Is your bandwidth also near 5GHz? If bandwidth is much lower then you can definitely work with jellybean resistors, though you will have to compensate for their parasitics. Not necessarily easy, but definitely doable with sufficient modelling.

Also I would avoid vishay resistors like the plague for now... seems they're having massive supply chain issues, lots of parts at 99 weeks... Makes the price irrelevant IMO...