A lab I work with lost a 96-well sequence at 3 a.m. because the autosampler reported error 0x76 “gripper failed to grasp plate” on well 61, after 60 clean injections. The plates were the same lot that had run fine the week before. The difference was the tray: a technician had swapped a 12 mm plate nest for a 9 mm nest during method setup and left the stack height 3 mm short. The instrument kept closing on air. That is an autosampler pickup error and misalignment problem, not a filter problem, and it took two hours to find because everyone on the shift blamed the filter frits. This article is about the mechanical stack above the filter: tray height, gripper tolerance, cap seating, and the exact error codes each misalignment throws on Agilent, Thermo, and Waters instruments.
The autosampler does not “know” where the plate is. It knows where the deck says the plate should be, and it trusts the gripper’s closed position to confirm it got there. When those two numbers disagree by more than the gripper’s compliance, you get a pickup error. The stack is: deck surface + nest or adapter + plate + cap or mat + septum. Every layer has a tolerance, and they add.
A typical 96-well deep-well plate is 44.0 mm ± 0.3 mm tall at the flange. A standard 2 mL glass vial in a 6×8 tray is 32 mm ± 0.5 mm to the crimp top. Add a 0.5 mm septum and you are at 32.5 mm. Now add a vial insert that sits proud of the vial mouth by 0.4 mm because it was pushed in by hand instead of seated with a tool, and your effective height is 32.9 mm against a nominal 32.0 mm. That 0.9 mm is inside most grippers’ grasp window, but stack three such errors across a tray and the outer rows fail first.
I keep a stack-up worksheet for every method. It lists each layer with its nominal and tolerance, sums the worst case, and compares that sum to the instrument’s specified grasp window. For a Thermo Vanquish with the standard plate handler the window is roughly ±1.5 mm around the taught height. For an Agilent 1290 Infinity II multisampler with the plate handler it is tighter on the Z axis, about ±1.0 mm. If your worst-case sum exceeds half the window, you have a latent misalignment that will surface as an intermittent pickup error, usually on the row furthest from the teach point.
The teach point drifts when you change consumables
The teach point is set once, often by the installing engineer, using whatever plate was in the box that day. Change to a different plate brand with a 0.6 mm taller flange and the teach point is now wrong by more than half the window. I have seen this exact failure on a Waters Arc HPLC with the 96-well sample handler: the lab switched from one polypropylene plate to another and the gripper started dropping plates at row H. The fix was a 15-minute re-teach, not a hardware repair.
Re-teach whenever you change plate manufacturer, cap type, or nest. Write the teach date on the tray with a paint pen. It sounds crude. It has saved more runs than any SOP.
Gripper tolerance: what the jaws can and cannot forgive
The gripper is a two-finger parallel jaw with a spring or servo close. Its grasp tolerance is the range of plate widths it can grip without crushing or dropping. A standard 96-well plate is 127.76 mm ± 0.25 mm wide per the ANSI/SLAS footprint (formerly SBS). The gripper’s nominal open width is 128.5 mm, closing to a commanded position. If the plate is 127.5 mm, the jaws close 1 mm further than expected and the grasp sensor reads “plate present” correctly. If the plate is 128.0 mm, the jaws may not close fully and the sensor reads “no plate” even though the plate is right there.
This is where filter frits enter the story. A HPLC filter that is slightly oversized in its housing, or a filter plate whose wells are not flush with the frame, changes the plate’s effective width and flatness. I have measured filter plates where the membrane sat 0.3 mm proud of the well rim because the frit was not fully seated during assembly. The plate would not sit flat in the nest. The gripper closed on a tilted plate, the grasp sensor read marginal, and the instrument threw a pickup error on every other plate. The measurement that found it was a simple feeler gauge across the plate face: 0.3 mm of rock at one corner.
Cap seating is the most common single cause
Crimp caps and screw caps both fail the same way: they are not seated to the same depth as the teach point assumed. A crimp cap that is 0.2 mm high because the crimper jaw is worn will lift the vial’s effective top by 0.2 mm. Across a tray of 48 vials, the variance between the highest and lowest cap can be 0.5 mm. The gripper does not care about the average. It cares about the worst one.
For screw caps, the torque matters. A 9 mm screw cap on a 2 mL vial should be torqued to 0.6 to 0.8 N·m with a calibrated capper. Below 0.5 N·m the cap sits high and the septum is not compressed. Above 1.0 N·m you deform the cap skirt and the cap sits low, which changes the septum height and the needle penetration depth. I have a torque wrench on the bench for exactly this reason. It cost less than one lost sequence.
For crimp caps, the crimper jaw height is set with a gauge. The standard is a 11 mm crimp with the cap top sitting 0.0 to 0.1 mm below the vial mouth reference. Check it with a crimp gauge every 200 caps. If you do not have a gauge, crimp ten caps and measure the cap-to-vial height with a digital caliper. The spread should be under 0.1 mm. If it is 0.3 mm, your crimper needs service.
The error codes each misalignment produces, by instrument
This is the part most troubleshooting guides skip. The instrument tells you what it thinks went wrong, and the code points at the layer. Here is what I have logged across three platforms.
| Instrument / handler | Error code or message | Most likely misalignment layer | First check |
|---|---|---|---|
| Agilent 1290 Infinity II multisampler | “Gripper failed to grasp” / 0x76 class | Plate height (stack-up) | Re-teach Z; check nest height |
| Agilent 1260 autosampler | “Vial not found” / tray position error | Vial height or cap seating | Crimp height gauge; cap torque |
| Thermo Vanquish / TriPlus | “Plate not detected” / “gripper error” | Plate width or flatness | Feeler gauge across plate face |
| Waters Arc / Alliance | “Sample not found” / “carrier error” | Teach point drift after consumable change | Re-teach; check plate brand |
| Generic 96-well handler | “Z-axis limit” / “crash detected” | Insert proud of vial mouth | Insert seating tool; measure insert height |
The codes are not universal. Agilent publishes its autosampler error lists in the instrument firmware help and in the Agilent consumables and accessories documentation. Thermo publishes its sample handling error tables in the chromatography and sample handling support library. Read your instrument’s list once and keep it in the method binder. Half the pickup errors I have seen were resolved by matching the code to the layer before touching hardware.
Why the same code can mean two different things
“Gripper failed to grasp” means the grasp sensor did not confirm closure. That can be a plate that is too low (jaws close on air), too high (jaws hit the plate before the close command completes), or too wide (jaws cannot close). The code is the same. The layer is different. The way to tell them apart is to watch the gripper during a teach cycle with a plate in the nest and a plate on the deck. If it closes fully on the deck plate and not on the nest plate, the nest height is wrong. If it fails on both, the plate width or the gripper itself is the problem.
Two field cases where the filter was blamed and the tray was guilty
Case 1: the intermittent pickup error that was a proud frit
Symptom: A contract lab running EPA method 8270 on a Thermo Vanquish had intermittent “plate not detected” errors on a 96-well filter plate, roughly one plate in eight, always at the row nearest the operator. They had replaced the filter plate lot twice and the gripper once.
Measurement: I put a dial indicator on the plate face in the nest and read 0.4 mm of rock across the diagonal. A feeler gauge confirmed the high corner. The cause was a filter plate where the membrane frit was not fully seated in one well, lifting that corner of the plate above the nest datum. The PTFE membrane filter itself was fine. The assembly was not.
Fix: Incoming inspection now checks plate flatness on a granite surface plate with a 0.05 mm feeler gauge on a 5-plate sample per lot. Any plate with more than 0.15 mm of rock is rejected. The error rate went to zero. The lesson: the filter plate is part of the mechanical stack, and its flatness is a tolerance you have to specify.
Case 2: the crimp cap that lifted the vial 0.3 mm
Symptom: A pharmaceutical QC lab on an Agilent 1260 autosampler had “vial not found” errors on the last two rows of a 100-vial tray, always after a capper service. The capper had been returned from calibration with a new jaw set.
Measurement: Ten crimped vials measured with a digital caliper showed a cap-to-vial height spread of 0.35 mm, against a spec of 0.10 mm. The new jaws were 0.2 mm taller than the old set, so every cap sat high. The autosampler’s needle guide was catching the cap shoulder on the outer rows where the tray had a slight tilt.
Fix: Re-shim the crimper jaw height to the gauge, then crimp a 10-vial check every shift. The error stopped. The lab also added a COD vial height check to the same shift routine, since those vials use the same 11 mm crimp and the same failure mode.
Cap seating and septum compression: the numbers that matter
Septum compression is the variable nobody measures and everybody assumes. A properly crimped 11 mm cap compresses the septum by about 0.3 mm. That compression is what seals the vial and what sets the needle penetration depth. If the compression is 0.1 mm, the seal leaks and you get evaporative loss on long sequences. If it is 0.5 mm, the septum is over-compressed, the needle cores it, and you get a filter full of septum fragments.
The needle penetration depth on most autosamplers is set to reach 1 to 2 mm above the vial bottom. If the cap sits 0.3 mm high, the needle reaches 0.3 mm less deep. On a 2 mL vial with a 300 µL sample, that is the difference between sampling the bulk and sampling the surface. I have seen RSD% climb from 0.8% to 2.4% on the same method because the capper drifted 0.2 mm. The chromatograms looked fine. The numbers did not.
Check septum compression with a depth gauge: measure the cap top to the vial mouth on a crimped vial and compare to the uncrimped cap height. The difference is your compression. Keep it between 0.25 and 0.35 mm. Write it in the method.
A troubleshooting checklist for autosampler pickup errors
Work top-down. Do not start with the filter. The filter is downstream of every mechanical error in this list.
- Confirm the error code against the instrument’s published list. Match the code to the layer before touching hardware.
- Measure the stack-up. Deck + nest + plate + cap + septum, worst case, against the grasp window (±1.0 mm Agilent, ±1.5 mm Thermo as starting points).
- Check plate flatness. Granite surface plate, 0.05 mm feeler gauge, 5 plates per lot. Reject above 0.15 mm rock.
- Check plate width. 127.76 mm ± 0.25 mm per ANSI/SLAS. Measure the widest point, not the average.
- Check cap height. Crimp gauge or caliper, 10-vial sample, spread under 0.10 mm.
- Check cap torque. 0.6 to 0.8 N·m for 9 mm screw caps, calibrated capper.
- Check insert seating. Insert flush to 0.1 mm proud, seated with a tool, not by hand.
- Re-teach after any consumable change. Plate brand, cap type, nest, insert. Write the date on the tray.
- Watch a teach cycle. Plate on deck vs plate in nest. If it closes on one and not the other, the nest is the problem.
- Only then look at the filter. If the plate is flat, the caps are seated, and the teach is current, the filter housing is the next suspect.
For method-level guidance on the filter itself, the U.S. EPA SW-846 test methods and the Restek technical resources are the references I hand to new analysts. Neither will fix a tray height problem, but both will tell you what the filter is supposed to do once the plate is seated correctly.
The decision rule: fix the stack before you replace the filter
When an autosampler throws a pickup error or a misalignment fault, the probability that the filter is the cause is low. The probability that the mechanical stack above the filter is out of tolerance is high. The decision rule is simple: if the error code is a grasp, detection, or Z-axis fault, measure the stack-up and the cap seating before you open a filter box. Specify plate flatness at 0.15 mm, cap height spread at 0.10 mm, and cap torque at 0.6 to 0.8 N·m, and re-teach after every consumable change. Do that and the filter stops being the suspect, which is exactly where it belongs.
References & further reading
Related resources from our lab consumables network:
- Hplc Filter at hplcfilters.com
- Ptfe Membrane Filter at ptfemembranefilter.com
- Cod Vial at codvial.com
- Vial Insert at hplcinsert.com
External references and standards: