Decent Espresso

MTBF (mean time between failure) testing at Decent Espresso


Given enough time and stress, everything eventually breaks.

At Decent Espresso, we have a room where we torture test our machines between extremes of temperature and pressure. We've run over 3 million stress test espressos on our 4 test machines.

Whenever we are planning on putting something new into our machines, it goes through these tests. The new part needs to last as long as everything else. In practice, this literally means that we're no longer able to break it.

After 4 years of doing this, we're now finding that the “no hole” (blind) portafilter baskets that we use to create this test, are the first thing to fail, between 30,000 and 154,000 espressos.

It's really quite something that a 1.2mm thick piece of stainless steel, breaks first.

We tried switching to plastic nozzles on a basket (to relieve pressure) but that test gear couldn't withstand the test either. Last week, we ordered custom-designed blank baskets to be made for us, which we hope can survive our stress testing.

This is why only 0.3% per year, of all the espresso machines we've ever sold, need us to repair them.

We're now at a point where nothing from our standard machines breaks in our test machines. Only new parts that are still in R&D, break when they go into the test lab. We then iterate on the design until we can no longer make it fail. In the photo at the top, you can see that there is steam under the plastic cover of the center machine: this is from a new group head heater design that we are still working on.

The key takeaways:

  • OLAB Italy redesigned their valve bodies for us based on our observations of how they were failing.
  • To make steaming as fast as we could, we tested heaters at 2800W, 3000W and 3200W, right at the limit of what our supplier said was possible. We learned that 3200W heaters would fail quickly, but our 3 machines with 3000W heaters have been going constantly now for over a year. Those are the heaters in our new Bengle model. We also worked with select cafés, making custom DE1XXXL machines with these heaters, so they could give them real world tests for several years, before they made their way into our production machines.
  • our CNC manufactured “manifold” (it directs water to many different destinations) had initially (when we launched) struggled with temperature cycling causing cracks over time. We'd fixed that with redesigns, but now we wanted to move to a “designed for manufacturing” compact “All in one” design that was moulded and easily repaired. It took us 4 years of testing, and 3 major revisions, to get to a point where it's now doesn't fail.
  • we had one valve eventually fail from overuse in 2023. We tweaked the firmware, in how the valve timings were sent and that brought the previously-failing valve back into function.
  • the rubber seals in our plumbing kit will leak after several years of constant use, and will need to be replaced.
  • that's where we currently stand, in terms of parts that have a known MTBF that go into our espresso machines.

Key Points Summary:

  1. Cycling Test Process:
    • Temperature Cycles: 5 cold (~50°C) and 5 hot (~85°C) cycles, repeated continuously.
    • Pressure Profile:
      • Ramp up to ~13 bars, hold for ~6s, then release in ~2s.
      • Each cycle lasts ~8s.
      • Transitioning between hot/cold cycles extends pressure release to 12s for faster cooling.
  2. Design & Testing Timeline:
    • Initial Design (Dec 2021): we used OLAB valve with tapered PTFE seal.
    • Redesign (June 2023): OLAB optimized for plastic body valve with o-ring seal; AIO manifold adapted to this change.
    • Testing Phases (Sept 2023–July 2024):
      • Conducted on DE1XXXL with AIO manifold batches v2–v4.
      • Cycle counts:
        • Manifold v2: 337k test cycles
        • Manifold v3: 1151k test cycles
        • Manifold v4: 1463k test cycles
    • Final Design (Aug 2024): Manifold v5 AIO manifold with creamy white PPS material.
  3. Post-v5 Testing (Aug 2024–May 2025):
    • Machines and cycles:
      • DEXXXL 220V: 1,084k test cycles
      • DE1PRO 220V: 1136k test cycles
      • DE1PRO 120V: 540k test cycles
      • DE1XL 220V 737K test cycles
  4. Failure Observations:
    • DE1XXXL Error (Feb 2025): “Zero flow, valve open, pressure > 1.0 bar” due to valves not opening fully before pump activation.
      • Cause: Long-term wear prevented pressure release, leading to abnormal operation.
      • Solution: Fixed via software programming, change of valve power timings.
  5. Component Durability Issues:
    • Blind Basket for testing: cracks consistently between 30K–154K cycles (high-pressure stress).
    • Portafilter with Plastic Nozzle for testing: consistently leaked/cracked after months.
    • Plumbing Kits: after years of use, seals degraded, causing slow leaks (water dripped into the tray). Seals need to be replaced when this happens.
  6. Visual Evidence:
    • Timing diagrams and photos documented failures (e.g., cracked baskets, leaking plumbing).

Key Takeaways:

  • Design Iterations: Improved valve/manifold durability (v2 → v5).
  • Stress Points: Baskets and plumbing kits are long-term wear concerns.
  • Software Fix: Valve timing adjustment can prevent pressure-related failures.
  • Material Choice: PPS (v5) shows promise for extended cycle life.

  • Chinese (simplified): Decent Espresso 的 MTBF（平均故障间隔时间）测试

    Updated 2025/07/24