We get 2 of this RLB,
I and JA2TY made a modification of Anritsu MR55A1 IF Return Loss Bridges. (60 ohm impedance)
Changed
a) The bridge resisters from 60 ohm to 50 ohm 1%
b) BNC connector from BNC-P to BNC-R as well.
Original: Anritsu MR55A1 IF RETURN LOSS BRIDGE changed Z=60 ohm to 50 ohm BNC-R
There is a shield plate in center, which separates input and output. Brown colored one in center is RFC.
3 bridge resisters are inside aluminum ingot for minimizing internal coupling.
left-hand: input from TG center: ground right-hand: output to Spectrum analyzer
A 270 ohm resister and a 5pF capacitor are used for balancing for 75 ohm impedance. Eventually, I removed those parts.
Result: Measure this RLB. left:JA2DJH, right:JA2TY
Characteristics of the modified Anritsu MR55A1.
Used for measuring: TR-4172; SPAN: 1GHz, Center frequency: 500MHz
Directivity: 38.9dB/JA2DJH, 30.7dB/JA2TY at 430MHz
The upper line is of reference.
Point of importance:
Carefully selected resisters checked by DMM down to 4 digit.
Select around 50.5 ohm in case used 51 ohm resister.
Measure resisters by DMM positioned at "4W ohm".
Termination resister is suitable for a reference resister. Use both with the same figure.
RFC is problematic. See the below professional RLB.
Glued tightly all components after fixed.
Comment of JA2TY:
If you built a 100MHz bridge, you could enjoy for one week for the fine tuning, for a full month to 1GHz bridge
and for a full year to 2GHz bridge. Hi Hi
2GHz Return Loss Bridge
2GHz is on way
JA2TY has built this RLB in March 2009
Modified based on OZ2CPU's circuit. Ferrite beads FB-801 used, inner coax is 2.9mm semirigid cable.
Directivity: 50dB at 430MHz and 35dB at 1GHz.(Span:1GHz)
RFC is wound by 5 turns on a ferrite core. Then semirigid cable with 5 Ferrite beads used.
Poor directivity is resulted in lower frequency in case RFC is removed(ferrite beads only).
Sep.,2009
Ferrite core used in double and shielded by a copper sleeve. Coax cable is 1.2mm semirigid.
Flat directivity obtained around 35dB! level.(Span:1.8GHz)
29, Sept,.2009
Used copper plate beside ferrite beads. Nice directivity: above 40dB lines.(Span:1.8GHz)
JA2DJH has built this RLB in July, 2009 in line with JA2TY's RLB.
2 RFCs are used FT23-61+FB801-43x5. Characteristics: Span;1.8GHz, Directivity:37dB at 1.2GHz usable to 2GHz
Utilized NEC's 800MHz/11dB directional coupler.Adapted original circuit; its strip line and terminator for return line.
I used HP 908A termination resister for DUT.
Reference was made from F6BON.
Copper tapes adapted for optimum circuit balance in improving directivity curve at high end, which is effective only in DUT side.
Ferrite core is wound by copper tape, which is also shielded between input and output by copper plate.
Return lines are of a bare copper wire, which is also shielded by ferrite beads.
Eventually,10dB directivity blow 500MHz was obtained.
As directivity is around 35dB, so I am considering to modify it to double ferrite cores.
August, 2009 Minor modification done: for double ferrite cores, used small and big ferrite in mix.
Used FB301 cores in a return line.
Close up view of the bridge components. See paralleled 100
ohm and right-hand is RFC.
Coax cable is of 2.9mm semirigid which is mounted with FB801
cores. 
Tried large size cores in tandem, but failed no chance to
apply lid.
Characteristics: inferior than previous one.September, 2009 Additional minor modification done.
1)For RFC, replaced enamel wire with 1.2mm coax cable.
2)Fold with aluminum foil over the semirigid cable.
Result: Up to 500MHz.October, 2009
Minor modification done.
RFC used dual ferrite beads.
13 two-hole beads are used, inner core used FB101. Up size of bridge.
Copper tapes used for circuit balance, 4 of tapes used. DUT to earth, Gen to DUT and RFC (hot and cold) to earth.
This is final characteristic.used TR-4172 SPAN: 2GHz, Center frequency: 1GHz at 1.8GHz Directivity: 40.9dB, upper line is reference.
Directivity line is smoothed. TG is 1.8GHz max.
September, 2009 Another RLB built using a Takachi case TD6-11-3 by JA2DJH.
Coaxial connectors and semirigid cables. Ferrite beads in dual use, outer beads are two holes ones.
Bridge circuit is based on the microstripline. Parallel of 100 ohm resisters are used for bridge circuit.
After touch up adjustment, all parts are to be glued.
Used ferrite beads FB-101 and outer beads are two holes ferrite ones.
One of popping beads (see small beads in center) is very effective for high frequency directivity enhancement.
Characteristics: Span: 1.8GHz CF: 900MHz Directivity: 35dB at 1.2GHz.
Near to flat directivity, however desirable additionally 10dB. Objective is 45dB directivity all through frequency range.
29, Sept., 2009 in line with JA2TY RLB.
Directivity at 2GHz: 42dBUse copper plate upside of ferrite beads. Characteristic:upper line is reference. Span: 2GHz Directivity is almost 45dB to 42dB!
I hope this is my final modify.
From DL forum, example thin coaxial cables are
1) RG174 d=1.56mm
2) RG316 d=2.01mm
3) RG58U d=3.30mm
4) Semirigid d=3.50mm
We can get semirigid cables and RG-58/U but cannot get RG174, RG316 etc.
There are another semirigid cables d=0.86, 1.19, 2.2mm.
I get 1.19mm semirigid cable in Japanese Ham fair 2009.
Inside view of Wiltron 60N50/ 5-2000MHz Directivity: 40dB
This is inside view of JA2TY's Wiltron 60N50 professional RLB, it's broken so directivity is only 20dB.
He said very difficult to open RLB so it's not use bolts and nuts.
signal pass: Left to Rightleft-hand:outer lid center:RLB block right-hand:inner lid Inside of bridge: used four cores for RFC. Downside is DUT.
He said difficult to built this cavity block.
Take photos inside of 60N50 used micro lens.
Close up of RFC parts, winding 6 turns on FT23? cores. Close up of bridge resisters 51 ohm
Used thin coax cable, outer shield pipes are very fragile so don't drop on floor.(tnx ja1ceb)
Close up of lid, small bolts makes capacitor of balancing circuit.Like resonance of micro wave circuit. Projection of aluminum is placed on bridge resisters.
I get Wiltron 60NF50 (5-2000MHz Directivity: 40dB) in Sep., 2009
Wirtron 60NF50 Directivity: 40dB minimums.
Characteristics of Wiltron 60NF50used TR-4172; SPAN: 1.8GHz, Center frequency: 900MHz Directivity: 44.9dB at 1GHz, upper line is reference.
Directivity is almost 45dB!! Reference also flatness, high frequency downed only 2dB.
On way this characteristic! It's very difficult for me.
HP 8721A(0.1-100MHz Directivity: 40dB)
Spec: Frequency Range 100 kHz to 100 MHz
Directivity: Greater than 40 dB, (1 to 110 MHz)
Load-Port Return Loss: Greater than 30 dB
Maximum Input Power: +20 dBm
Frequency response: +/- 0.5 dB
source: gen in, reflected: scope, load: DUT A tight lid so cannot open, I want see inside.
Characteristics of HP8721AUsed for measuring: TR-4172; Span:200MHz CF:100MHz
Directivity: 40dB at 110MHz.
anzac RB-3-50(3-1500MHz Directivity: 40dB) tnx JE1BMJ
inside view of JE1BMJ's anzac RB-3-50 professional RLB
Characteristic of RB-3-50; Ref: 0dB, CF:1000MHz, Span:2000MHzFlat characteristic to 1.5GHz.
JE1BMJ comment
Using 2 of Semirigid cables with FB, fixed by polyurethane farm.
Every ports have small bolts makes capacitor of balancing circuit.