Sondermaschinenbau (custom single-purpose machines) is the most expensive way to produce a piece of motion, gluing, or screwing. It is also the cheapest, if you use it in the right place. The difference between these two sentences is seven questions nobody usually asks — and six invoices later the client finds out why the CFO is cutting further investment for a year.
The first real decision point isn't "custom vs. buy"
It is: **how many pieces per year, and what tolerance.**
- < 50,000 pieces/year at ±0.5 mm tolerance and above → manual line or cobot. ROI of a custom machine has nothing to come from.
- 50,000 – 500,000 pieces/year at ±0.1 mm tolerance → standard machine with custom fixture. Usually a Bosch Rexroth / Festo modular platform, $40k–120k fully equipped.
- > 500,000 pieces/year at ±0.02 mm tolerance or geometrically non-trivial motion → that's when custom machine building makes sense. Custom frame, custom drive, custom HMI.
- Any combination with **regulatory validation** (medical device, food contact, ATEX zones) → custom is often the only route, because a standard machine can't be validated for an undefined process.
This simple filter eliminates 60% of inquiries that reach us under the heading "we want our own machine". After two hours of conversation, most switch to "we want a standard line with a custom fixture".
Three hidden cost blocks that don't show up in the first quote
A custom machine quote from us is a technical exercise. The hour-long conversation before should cover these three blocks — otherwise the number is meaningless.
Validation and CE marking
A custom machine doesn't have CE from production. It requires a **risk assessment per EN ISO 12100**, technical documentation, declaration of conformity, often also a notified body (at category M). For a machine with 3D kinematics and a robotic cell, we budget 6–12% of the machine's value just for this block. Real number clients forget to plan: **€8,000–15,000 for audit + documentation**, sometimes with a 4–8 week schedule extension for the first prototype.
Process validation (not machine validation)
In food, pharma, medical, the process has its own validation — IQ/OQ/PQ. The machine can work perfectly and the process still doesn't validate. This is a process engineer's problem, not the mechanical engineer's, but the same project pays for it. Counting: **20–60% of the machine's value in sectors with GMP.**
Spare parts and service lifecycle
Sondermaschinenbau means components are picked for a specific purpose — they are often end-of-life before the end of the machine's first lifecycle. A Beckhoff AM8131 servo motor may not be replaceable with the same type after 7 years; TwinCAT 3.1 controller firmware may require an upgrade of the whole package. We plan **10–15% CAPEX reserve for service over 10 years**, plus an agreement with the manufacturer for 7+ year parts availability (Festo, Bosch Rexroth, KEB offer this contractually, you need to ask at order time).
A client who counts "life-cost = CAPEX × 1.1" gets a 35% CAPEX invoice for line rework after 5 years.
Four questions that decide on price more than technology
1. What is the cycle time — and what is its variability?
An 8-second cycle with ±0.2 s tolerance is a different machine than an 8-second cycle with ±2 s. Tight cycle means servo instead of pneumatics, servo means regeneration, regeneration means three-phase + EMC filter, that means a larger switchgear, the larger switchgear means a different shop layout. A chain of four decisions that can't be solved retroactively.
2. Will the machine be in a production line or stand alone?
A standalone machine is 30% cheaper. A line means synchronization with upstream/downstream (PLC handshake, buffer queue, fail-modes when a neighbor stops), which multiplies PLC code complexity from 2,000 lines to 8,000+. After adding one line-level interlock, the quote changes by 15–30%.
3. What materials and what is their tolerance?
"2 mm sheet metal" is a sentence that covers 6 ČSN standards. If thickness in the spec is ±0.1 mm and actual deliveries have ±0.3 mm, the machine jams on 5–10% of pieces. The fixed construction jams, the client shouts, we deliver adaptive compensation — which should have been in the first iteration, if we had received a realistic sample set from real production. **We require 100+ pieces of real material before construction.** Not 5 idealized samples from the lab.
4. What is the expected product change over 5 years?
A custom machine built for one product is a good CAPEX argument until the product changes. After 18 months the client announces "we want to make variant B too" and the machine can't be modified without new validation. Solution: **modular construction with interchangeable fixtures and a parameterized PLC program** — adds 12–18% CAPEX, but saves 60–80% of costs at the first product change. For a stable product it makes no sense; for an innovative one it's mandatory.
When custom, never
- The ROI of securities doesn't rest on our machine as a risk — it rests on a different number: reliability of component suppliers. If your vision is "we'll build the machine and not think about it for 10 years", custom isn't the answer.
- If the process is in a standardized segment (CNC machining, packaging, palletizing) — DMG Mori, Mazak, KraussMaffei, Engel, Bosch have standard machines at 70–90% of custom solution performance, for 40–60% of the price.
- If your know-how is more in software (vision system, MES integration, AI quality control) than in mechanics — buy a standard machine, put an eye on top of it. Sondermaschinenbau would occupy capacity you have nowhere to spend.
When custom, unambiguously
- Motion geometry that doesn't exist in any manufacturer's catalog. (Real example: 5-axis pneumatic head with rotational compensation for gluing polyurethane sealant onto an unusual 3D surface — no catalog solution exists.)
- A combination of speed + tolerance above the limit of standard platforms. (Example: 30 cycles/s at ±0.03 mm position. Festo's catalog ends at ~20 c/s, or at 0.1 mm tolerance.)
- Regulatory validation that requires full control over every decision in the signal path (typical medical device, ATEX zone classification).
- Existing production lines, where the new machine is the last link and must precisely fit into the available space + handshake with existing PLCs. Standard machines rarely adapt to the dimensions of an existing line.
A calculator that helps you decide in 30 minutes
Take these numbers: - Hourly rate of a line operator in manual production (€/h) - Number of operators needed for manual production of the process - Expected number of pieces per year - Expected product lifespan (years until change) - Expected machine amortization (5, 7, or 10 years per tax depreciation)
Calculation: - **Manual cost over lifespan** = rate × number of operators × 1,600 h/year × years of lifespan - **Custom machine cost** = CAPEX + 30% CAPEX (validation + service + spares) + 5% annual energy
Custom makes sense if the difference is **> €200,000** in favor of the machine. At smaller differences, the client comes out better with a manual line (can change the process) or with a standard line (can change the product). The €200k reserve covers risks that show up only after 18 months of operation.
Our workflow when the client accepts the custom path
1. **Discovery (2 weeks, fixed price)** — process FMEA, material samples, kinematics simulation in SolidWorks/EPLAN, initial CE categorization. 2. **Concept design (3–4 weeks)** — 3D model, motor/sensor selection, capital quote within ±15%. 3. **Detailed engineering (8–12 weeks)** — production drawings, BOM, EPLAN schematic, validation plan. 4. **Manufacturing and assembly (12–20 weeks)** — frame, mechanics, electrical installation, PLC programming in parallel. 5. **FAT (Factory Acceptance Test) at our shop (1–2 weeks)** — the client validates at our shop before transport. 6. **SAT (Site Acceptance Test) at the client (2–3 weeks)** — final tuning under real production. 7. **Hyper-care 90 days** — 24/7 standby, 4-hour response, fully covered by CAPEX.
**Whole cycle: 9–15 months. Not 3–6.** A client who is promised 6 months gets either a bad machine or a broken supplier — or both.
---
*We write this as a technical partner who in 15 years has lost more money on poorly specified custom machines than earned. If you're considering a sondermaschinenbau project, the first consultation (90 minutes, no commitment) walks the calculator above on your numbers.*