discrete resistor coupling and attenuation

I'm building a series of octave signal sources covering 25-400Mhz and the signal chain includes series combinations of couplers and attenuators. Conceptually, I would like to eliminate the coupler and use the shunt leg of a discrete resistor tee attenuator to sample the signal for input to a PLL, and to a power detector at the output. A reversed coupler would still be used at the output for measuring power reflected from the load. The off-the-shelf surface mount couplers and attenuators have published typical performance data that is quite good, but the specified performance is usually inferior to the typical performance data. I am considering using .1% thin film resistors in a 402 or 603 package to replace the SMT couplers and attenuators on .062" FR4 (nominal k=4.3). Although the discrete resistor approach has the benefit of greater freedom in selecting the degree of coupling and attenuation and a reduced overall footpriint, I am not sure how much impact the resistor parasitics will have, or if there are other problems in matching. Any comments or suggestions?

At up to 400 MHz I think there will be no problem with SMD resistors as attenuators or passive couplers. Only isolation can be an issue for > 20 dB.
I used e.g. SMD 100 Ohm trimmers as adjustable attenuators for up to 3 GHz with excellent results.

you can make an resistive unequal power splitter using only resistors. Maybe 1 dB thru line loss, and 10 db coupled loss. What coupling value do you want?

Thanks for the replies, I am coupling the output of a VCO to an ADF4002 with a mainline attenuation of 3-4dB and 10-11dB attenuation at the ADF4002 input. I'm also coupling the output of the signal source to an AD8361 using ~4dB of mainline attenuation and ~12dB of attenuation at the AD8361 input (see attached). Since I have an amplifier in the signal chain and I want to minimize phase distortion due to reflections, I am giving up gain to improve the match between components. The signal source modules will be connected to a tunable matching circuit at the input of a power FET which will in turn drive a resonant load that produces a magnetic field. The object of this project is to provide a clean sine wave to provide a magnetic pumping current at about 4A to investigate the Microwave Induced Magnetic Isotope Effect (MIMIE). Broadband noise and harmonics are less important than the phase coherence of the CW signal. Can you tell me which 100 ohm trimmer you used? As time goes by, I will be moving up in frequency and I will have to deal with parasitics. I appreciate hearing from people who have more experience in dealing with the realities of circuit performance than I have.