We sell SPC software, so read this with that in mind. It is still worth writing honestly, because the alternative — pretending free options are bad — is both untrue and easy to check.
There are four realistic free routes. Each is genuinely viable for somebody.
1. A spreadsheet
What you give up: limits that stay put, rules 2 to 8, anything watching overnight, and continuity when the author leaves. What you keep is total flexibility and zero cost.
Right for: one characteristic, one engineer, a weekly review.
The failure mode is specific and silent: limits computed over the same range the
chart plots drift along with the process, so the chart never signals. If your
spreadsheet's limits are =AVERAGE(B:B) rather than a frozen number, that is
already happening. More on that here.
2. R or Python (qcc, spc, statsmodels)
What you give up: nothing statistically — qcc is excellent and better
validated than most commercial code. What you give up is everything around it:
a scheduler, a database, alerting, access control, and a screen an operator can
read. You are not choosing a tool, you are choosing to build one.
Right for: an engineer who codes, doing analysis rather than monitoring.
The honest failure mode is a notebook that runs on one laptop and stops when that person is on holiday.
3. A free tier of a commercial product
What you give up: usually characteristic count, history retention, or user count — and you find out which at the point where the thing has become load-bearing. Worth reading the limits before building on it, not after.
Right for: evaluating, and for genuinely tiny use.
4. Open-source SPC applications
There are a few. What you give up is hosting, backups, upgrades, TLS and somebody to ask, and the smaller projects can go quiet without warning. If you have an IT function that already runs internal apps, that cost is low. If you do not, it is not free at all — it is unpaid, which is different.
Right for: organisations with IT capacity and a preference for owning the stack.
What is actually free here
No account required, nothing stored:
- Cp/Cpk calculator — reports within-subgroup and overall side by side, because quoting one without the other is how capability misleads.
- Control limit calculator — shows the constants used, so you can check it by hand.
- Control chart generator — paste a column, get a chart with the rules evaluated.
- Constants table.
- Five control charts running on live public data.
Those are complete tools, not teasers, and there is no email wall on any of them.
So when is paying worth it?
One question decides it: is anything watching the process when you are not?
If a rule tripping at 02:00 on Sunday would cost you a shift of scrap, the value of continuous evaluation and an email is easy to work out, and it is usually larger than the subscription. If it would cost you a shrug, the free routes are genuinely fine and you should take one.
That is the whole case. From €169 a month per line, with a trial that needs no card and no call.