Click here to go to our main page on microwave terminations (new for September 2020)
Click here to go to our main page on test equipment
Click here to go to out page on vector network analyzers
Click here to go to our main page on calibrating network analyzer measurements
New for January 2019. This page was contributed by Rob, who designs electronics for radio astronomy at a scientific research organization in Oz. Many thanks!
Editors note: I could not resist the temptation to include Big Long Slidin' Thing by Dynah Washington on this page.... she is singing about her favotite band member, who happens to be a trombone player. Recorded in 1954.
Big Long Slidin' Thing by Dinah Washington
Introduction
In a short-open-load-through (SOLT) calibration kit, the load standard is assumed to be a perfect match to 50 ohms from DC to the upper frequency limit of the cal kit - 18 or 26.5 or 40 or 50 GHz depending on the connector type of the cal kit. The manufacturer can go close to achieving ideal performance, but close isn't good enough. The load standard's reflection coefficient has a direct bearing on the corrected directivity error of the network analyser.
Rather than try to achieve ideal performance with a fixed load, which would be horrendously expensive and incredibly fragile, the manufacturers adopted a different approach: Why not develop a load of small (but not infinitesimal) reflection coefficient magnitude and variable phase? The variable reflection coefficient phase allows the network analyser to take several measurements of the load as the reflection phase is varied, and work out where 50 ohms should be by fitting a circle to the measurement data. Or, as the hardcore network analyser heads say, "the circumcentre of the fitted circle is the raw directivity error."
Thus was born the sliding load. It consists of a length of precision, low-loss, air-dielectric coaxial line. An absorptive (thus lossy) dielectric element fits between the inner and outer conductors, and can slide axially up and down the line. A knurled ring on the body of the sliding load positions the absorptive element. The element doesn't make direct electrical contact with the coaxial line's conductors, both minimising wear and avoiding a world of pain with sliding contacts.
Sliding loads allow very accurate measurements of low-VSWR DUTs to be made, but are still horrendously expensive and incredibly fragile. Because the sliding load is an air-dielectric coax, its centre conductor is located in position only by the test port's (or test cable's) mating connector.
Beware! Sliding loads are equipped with a mechanism to extend the centre conductor so it can be mated to the network analyser's test port connector, then retract as the connector nut is tightened. HP sliding loads have a toggle clamp at the rear, which breaks if you're too rough with it. Maury sliding loads have a locknut which is loosened to allow the centre conductor to extend, and then tightened after the connector nut is tightened.
A sliding load can be damaged beyond repair just by making one connection wrongly. You have been warned.
How to Use a Sliding Load
Clean it, inspect it, gauge it.
Clean the mating faces of the connector and the threads in the approved manner.
Inspect the connector carefully for dirt, lint, metal particles, and damage. 3.5 mm female connectors have precision slotless inner conductors, with six spring fingers inside the hole. Use a x10 hand lens, a headband magnifier, or a stereo microscope for inspection. If the centre conductor is missing fingers, or is bent, scratched or dinged from mis-mated connections, the sliding load is basically good for the bin.
The connector gauges are those cool-looking dial-indicator gauges that nobody ever takes out of the calibration kit. Find the gauge appropriate for the gender of the sliding load and its "master" standard. Clean the mating faces of the gauge and the standard. Mate the two, using the correct torque wrench. Set the connector gauge to read zero. Remate with the master to confirm that the gauge's reading is zero. Now you can gauge the sliding load. Get out with the little plastic insert beads supplied with the cal kit in a tiny screw-top jar and slip one down between the inner and outer conductor of the sliding load. It's easiest to extend the centre conductor to do this. Retract the centre conductor and connect the sliding load to the connector gauge while holding the two in alignment. The connector or gauge nut should always turn freely by hand while you do this. Binding is a sign of trouble. Torque the nut to specification. Compare the gauge's reading to the sliding load's specification. If the connector is within tolerance, you may now use the sliding load in earnest. Just remember to remove the centring bead from the sliding load before you start a calibration.
Clean, inspect and gauge the test port (or test cable) connector.
As above. The test port or test cable connector deserves the same care as the sliding load's connector.
Calibrate with all the fixed standards first. In HP/Agilent/Keysight network analysers, the sliding load is the last standard in the sequence.
Ready? Take a deep breath . . .
Extend the sliding load's centre conductor. Engage the centre conductor with the test cable connector's centre pin, but don't push it home. If you're over 35 years old, you're going to need a headband magnifier and a strong light to see properly. Trust me on this. Holding the test cable connector and sliding load in proper alignment, start the nut onto the opposite thread. The nut should always turn freely with your fingers. Binding means trouble - misalignment or mis-mating. Once the connector nut is well-started, there may be a little resistance as the two centre conductors are drawn into position. Never rotate the sliding load relative to the test cable connector. The connector nut is the only thing which should turn. Tighten the nut with a torque wrench while holding the opposite part with a spanner. Engage the toggle clamp (HP) or tighten the locking nut until the detent "clicks" (Maury). Lay the sliding load on the bench. Do not let it "hang" off the test cable. The network analyser requires at least five measurements of the sliding load. HP sliding loads have grooves machined into the body, into which the ring "clicks". Start with the ring on the position closest to the rear of the sliding load. Press "measure" in the cal menu, wait for the sweep to finish, advance the ring to the next "click", press "measure", wait for the sweep to finish, and so on until all positions are measured. Then press "done".
Unscrew the connector nut, withdraw the sliding load, re-install the protective centring bead and dust cap. Position the sliding ring to the rearmost position and stow the sliding load back in the cal kit with the long slot uppermost. This protects the centre conductor, lest dirt and polyurethane foam particles should get inside.
How Do I Verify My Sliding Load?
If you have inherited a cal kit which has been used, you'll most likely need to determine whether the sliding loads are fit for further use. Sure, you can send the cal kit back to the manufacturer for recalibration, which is fine as long as you have the money and the standards meet specification. But if they don't, you've blown your money, and your standards still won't meet spec.
Inspection and Adjustment
Take the dust cap and centring bead out of the sliding load. While holding the load horizontally, rotate it. The centre conductor should rotate (off-centre) with respect to the outer conductor. But if the centre conductor describes a little circle as you rotate the sliding load, then the centre conductor is possibly bent from misuse. If you hold the sliding load vertically, the centre conductor should be more or less concentric with the outer.
Next, inspect the sliding load as in "clean it, inspect it, gauge it" above. If the centre conductor protrusion is out of spec, it can be adjusted back into spec. Maury sliding loads have a knurled stainless nut that the locknut tightens against. This large nut adjusts the centre conductor protrusion. Turn the nut (carefully!) while gauging the connector to achieve the right setting. HP sliding loads have a backstop screw hiding in the end of the body, beneath the toggle clamp. The smaller grubscrew is the one to turn (carefully!).
Measurement
Since you are taking the time to verify a sliding load, you are assumed to have a network analyser and a known good cal kit of the same connector type. The kit can be the one that the sliding load is a member of, as long as the kit has a known good broadband load, and you don't use the sliding load under test in the calibration. Perform a one-port calibration of the network analyser over the full frequency range of the cal kit, with "set frequency low pass" if you have the time domain option, maximum source power, low IF bandwidth and lots of measurement points. Connect the sliding load under test to the test cable.
The HP 911 D/E sliding loads have a nominal 30 dB return loss from 3 to 26.5 GHz.
The Maury 8034 A/B sliding loads have a nominal 40 dB return loss from 3 to 26.5 GHz.
The return loss should stay approximately constant (within 3 dB) as the slide position is varied. In polar format, the trace should describe circles of approximately constant radius: 0.03 for the HP 911 D/E; 0.01 for the Maury 8034 A/B.
The best diagnosis is with either unwrapped phase or group delay.
Unwrapped Phase: A good sliding load will exhibit reflection phase linear with frequency above its minimum frequency, with the phase slope proportional to the slide position.
Unwrapped phase of a good HP 911E, slide set to the first position.
Unwrapped phase of a good HP 911E, slide set to the sixth position.
A bad sliding load will show pronounced ripple or 180-degree phase jumps.
Unwrapped phase, bad sliding load
Group Delay: Set the scale to 100 ps/div, reference 0 sec, ref pos 0 div. A good sliding load will show essentially flat group delay over its specified frequency range, with some ripple.
Reflection group delay of a good HP 911E, slide set to the first position.
Reflection group delay of a good HP 911E, slide set to sixth position.
The HP 911 D/E sliding loads' reflection group delay should vary from 150 ps to 400 ps approx as the slide is moved over its full travel. The Maury 8034 A/B sliding loads’ reflection group delay should vary from 100 ps to 600 ps approx. over the full travel of the slide.
If your network analyser has time domain, so much the better. You can then distinguish the connector mismatch from that of the airline and load. Set the transform to "low-pass step" and the format to "real". Before measuring the sliding load, sanity-check the measurement setup with an open and a short. The open should give a step from 0.0 to 1.0 at 0 sec, and the short should give a step from 0.0 to -1.0 at 0 sec. Now connect the sliding load. Set the display scale to something like
0.01 units per division. The connector mating plane will be at 0 sec. The connector mismatch for a good sliding load will be almost imperceptible, thanks to expensive, precision manufacture.
TDR trace of a good HP 911E, slide set to first position. The sharp rise at t=0.7 ns is the load itself.
The Effects of Connector Damage
Careless handling damages the connector interface, and causes reflections. Female connectors are much more fragile than their male counterparts, yet because most coax components have female connectors, the female cal standards are used more often. In any well-used cal kit, the female standards are most likely damaged from misuse. A damaged connector means death to a sliding load. If there is connector mismatch, the sliding load will still draw circles around a certain point on the Smith chart, but that point won't be the origin. In bad cases, the origin will lie outside the circle circumscribed by the load, giving rise to 180-degree reflection phase jumps. A calibration with a bad sliding load will be worse than one with the sliding load omitted.
Appendix 1: I've Broken the Toggle on my HP Sliding Load. What Can I Do?
If the sliding load is in good enough electrical and mechanical condition (apart from the broken toggle), you can repair it. As long as you still have all the parts, that is. The toggle assembly is made up of the toggle itself, a smaller over-centre link, a plunger, a coil spring, and a 1 mm diameter lock pin. Ease the toggle off both sides of the pin in the slot, then unclip the smaller over-centre link from the body. If the toggle has broken where the lock pin goes through the toggle, cyanoacrylate- glue the two pieces of the toggle together. When the glue is dry, drill four 0.95 mm holes through the thick part of the toggle, two on each side of the break, 3 or so mm away from it. Make two flat- bottomed U shapes out of 0.9 mm tinned copper wire, and pass the U through the drilled holes.
Carefully twist the free ends together with pliers, (don't break anything!) cut the ends back, and bend the twisted wire down flat. Ream out the lock pin’s hole with the 0.95 mm drill, and clean out the plunger hole with a 3.5 mm drill. Insert the coil spring into the hole in the toggle, followed by the plunger and over-centre link. Start the pin squarely in the hole, and using a vice, carefully press the pin back into place. Reassemble the toggle onto the load.
Appendix 2: How to Disassemble a Sliding Load
Warning: this is not for the faint-hearted. If your sliding load meets spec, do not pull it apart. You will only need to do this if your cal kit's foam has fallen to bits, and you have black goo all inside the sliding load which you can't clean out. Or maybe you have a damaged sliding load, and you're curious to see what’s inside. In any case, you only have yourself to blame if you damage it further or break parts.
HP: Remove the toggle from the sliding load.
Slide the ring as far as it will go toward the connector end.
Extracting the toggle clamp's pin from the centre conductor is tricky. It's best to draw it out. Set up a milling machine with a small drill chuck in the spindle and a machine vice on the table. Edge-find the fixed jaw of the machine vice, then translate the table 4.55 mm inwards. Loosely grip the pin in the drill chuck, then bring the spindle down enough to hold the body of the sliding load squarely in the machine vice. Tighten the drill chuck and lift the spindle to extract the pin.
While the sliding load is still held in the machine vice, remove the backstop screw and plug from the rear of the body.
Working from within the slot, ease the centre conductor out of the rear of the sliding load. Remove the locating screw from the slide ring, and slide the ring off the back of the body.
Turn the centre conductor around backwards, and introduce it back into the body from the rear. Push the reversed centre conductor in to slide the absorptive element out from the connector end.
Reassembly is the reverse of disassembly, but take care to line the parts up correctly. When replacing the toggle clamp's pin into the centre conductor, a vice and V-blocks are recommended.
Maury 8034: Remove the two grub screws from inside the blue outer housing (0.050" hex key).
Remove the locating screw from the blue slide ring, and slide the ring off the rear of the body.
Unscrew the large stainless knurled nut from the body. The centre conductor can now be withdrawn from the rear of the body. The absorptive element should also come out with the centre conductor. Slide the absorptive element off the centre conductor.
When reassembling a Maury 8034, take note that the two grub screws which are inserted through the blue outer housing control the pre-load of the large stainless knurled nut. The tighter the grub screws, the harder to turn the large knurled nut. Set the inner conductor protrusion with a connector gauge, then tighten the two grub screws to fix the knurled nut in position.