Honest Tradeoffs — Where Vendor Wins, Where DIY Wins

No sales pitch. This is where we're honest about what each approach is genuinely better at, because the credibility of the DIY counter-proposal depends on not pretending it wins everything.

Where the vendor's approach is genuinely better

1. Turnkey simplicity

The vendor delivers a box and walks away. The DIY build is a 3-month integration project where the user has to be on-site for install, commissioning, and tuning. For a customer who just wants a facility to appear, the vendor approach is easier. Not cheaper, not safer, not more reliable, but easier in the short term.

If the customer does not have a technical operator, if they don't want to learn Home Assistant, and if they're not paying a dedicated integrator (the user) to build and maintain the facility, the DIY approach is the wrong choice for them. They'll end up with a stack they can't operate.

2. Insurance and commercial-standard compliance

Vendor-installed commercial-grade HVAC units, commercial dosing systems, and named-brand control hardware fit the standard commercial cannabis insurance profile. A DIY facility with Sonoff + Home Assistant + AliExpress-sourced lighting is novel territory for Maltese cannabis insurance.

Action item before committing to DIY: get an insurance quote on both builds. If DIY is uninsurable or materially more expensive to insure, the savings evaporate. This needs to be resolved BEFORE the customer signs anything.

3. Ducted HVAC gives tighter canopy uniformity than splits

Even with 6 wall-split units per flowering room, airflow distribution is less even than a properly-designed ducted system. Expect:

• Ducted: ±1 °C canopy uniformity, ±3 % RH uniformity, even VPD across the entire canopy
• Splits: ±2–3 °C canopy uniformity, ±5–8 % RH uniformity, noticeable variation between wall areas and room centre

For a yield-maximising commercial grow pursuing the absolute peak gram/m², ducted is technically better. The DIY approach trades ~5 % of yield for redundancy and cost. Whether that's the right trade depends on whether "redundancy" or "peak yield" is the customer's constraint.

4. Autogrow IntelliDose is a real commercial dosing controller

We keep the Autogrow in the DIY proposal precisely because it's correctly specced. Replacing it with a DIY Atlas EZO + ESPHome stack saves ~€800 but gives up 10 years of commercial refinement, operator interface polish, and replacement parts supply chain. This is the one subsystem where the vendor got it exactly right.

5. Single point of support (one throat to choke)

Vendor build = one company to call when something breaks. DIY build = "figure it out with the Home Assistant community and the user's cellphone". If the customer values having a commercial support contract more than flexibility, vendor wins.

6. Nanobubble oxygenation at €3 900

Genuinely premium choice. We keep it in the DIY build because there's no point DIYing cheaper — you'd spend €3 000+ building an equivalent and it wouldn't work as well.

7. UCHA60-TDC might be fine if the dehumidifier is ducted outdoors

One optimistic reading of the vendor's HVAC sizing: if the 276 L/day dehumidifier is ducted to reject its sensible heat outside the room (not stated in the offer but possible), then the real cooling demand is ~13.7 kW rather than 20 kW, and the 15 kW nameplate (→ 9.6 kW effective) is only ~30 % under. Still under, but less dramatically. We should ask the vendor directly whether the dehu is ducted out.

Where the DIY approach is genuinely better

1. HVAC is actually sized for the load

The vendor's flowering HVAC is 50 % undersized in the worst case (30 % undersized in the optimistic reading). Our 6-split approach is correctly sized at 20 kW effective. This is the single most important technical win and the easiest to verify with the customer.

2. N+2 HVAC redundancy

Vendor: 1 unit per flowering room. Failure = crop dead in 4–8 hours, 4–6 week replacement. DIY: 6 units per flowering room. 2 can fail simultaneously and the crop survives. 3-hour swap with €500 off-the-shelf parts.

3. Lighting has a real supply chain

Vendor charges €399 per 800 W fixture with no brand, no chip, no efficacy, no warranty. At the same ~€400 price point, Mars Hydro / Spider Farmer EU warehouse delivers named Samsung chips + 2-year warranty + published test data + DLC listing. The money is the same; the product is objectively better.

4. No cloud dependency for critical control loops

Vendor has two separate cloud-dependent control islands (Growlink for climate, Autogrow for dosing). Every cloud outage is a blind spot. Every vendor solvency event is a time bomb. The DIY stack runs entirely local: HA + Tasmota + ESPHome + Sonoff LAN mode + Autogrow standalone (no Intellilink).

5. Facility-wide dashboard + alerting

Vendor gives you 5 separate Growlink per-room dashboards + 1 Autogrow dashboard = 6 places to look. DIY gives one Home Assistant dashboard with everything, plus Grafana for long-term history, plus Pushover/Telegram alerts from a single alert engine.

6. Explicit scope — no hidden gaps

Vendor bid is €82 584 but missing ~€10–12 k of scope (CO₂, carbon, backup, security, drip lines, tank temp, downstream sensors). DIY BOM is explicit about all of these. The customer knows what they're getting.

7. Local sourcing pipeline via Poland

User has AliExpress + Polish distributor access + self-transport to Malta. That logistics edge means the DIY proposal captures prices that a Maltese vendor can't match — e.g. Mars Hydro EU warehouse DDP, Rotenso or TME.eu direct, no Maltese middleman markup on hardware.

8. Upgrade / expansion path

Need to add a new room? DIY: add more Sonoff devices for €50/sensor, configure in HA, done. Vendor: order another €2 500 Growlink kit or negotiate for additional expander modules, bind further into the vendor relationship.

9. UVC safety

Vendor bills 14 UVC lamps with no mention of door interlocks or safety. UVC-254 nm causes corneal burns in 30 seconds of direct exposure. DIY adds a €60 door interlock layer via Sonoff DW2 sensors and HA automation that disables UVC when any door opens.

10. Bugs get fixed

DIY: find a bug, fix it in HA, push to git. Vendor: file a ticket with Growlink, wait for next firmware release, hope it gets prioritised.

Where it's genuinely a coin flip

1. Dosing skid (we keep vendor's Autogrow)

Equal. Both builds use the same IntelliDose + Peripod M4. The only difference: DIY drops the €208 Intellilink module and bridges to HA via Modbus or power-monitoring instead. Minor.

2. Dehumidification

Same sizing in both builds. Same units possible on both. DIY sources directly to skip vendor markup. €40 difference, no meaningful tradeoff.

3. Tables, fans, humidifiers, water tanks, RO

All subsystems where the vendor's spec is either correct or trivially improvable by direct sourcing. Not battlegrounds.

The fundamental tradeoff

Vendor build: pay more (once hidden scope is revealed), get a single-vendor-supported but technically mediocre commercial facility. Simpler to operate. Harder to modify. Likely to miss the first summer heatwave.

DIY build: pay less (once hidden scope is included), get a technically-correct, locally-controlled, fully-explicit facility. Requires integrator involvement during install, 2–4 weeks of operator training post-install, an insurance check, and a commitment to run Home Assistant.

The wrong answer is to pick DIY for the cost savings without being ready for the operational complexity. The right answer is to pick DIY because the customer values control + transparency + redundancy, and to accept the operational ownership that comes with it.

Failure modes for the DIY approach (honest list)

1. Customer can't find a replacement integrator if the user becomes unavailable. Home Assistant is FOSS and there are freelance HA consultants in EU (check Nabu Casa's partner directory), but there's a real dependency.
2. Customer has a technical meltdown trying to read InfluxDB graphs or debug an ESPHome deployment. HA has a learning curve. Design the dashboards for read-only operator use, with all the complexity hidden behind admin-only panels.
3. Insurance or Maltese licensing inspector rejects DIY control panels. This is why the BOM includes a licensed Maltese electrician sign-off on the mains side.
4. A firmware update bricks 4 Sonoff devices in the middle of flower. Disable automatic OTA. Pin versions. Test on a bench unit.
5. A Home Assistant core update breaks an integration the customer depends on. HA releases are frequent and occasionally break things. Run HA behind a 1-month version lag.
6. User's bulk lighting order from Mars Hydro arrives late, delaying build by 3 weeks. Order 4 weeks early.
7. The Polish-transit-to-Malta logistics fail — a damaged LED shipment in transit, a customs question mark on the drivers, an F-gas certificate rejected by the Maltese installer.
8. The customer loses patience 6 weeks into a 12-week build. This is probably the biggest risk. Communicate milestones clearly up front.

These are real. The DIY approach only works when the customer is aligned with the integrator on a realistic build timeline and committed to the operating model. Pitching DIY to a customer who just wants to stop thinking about the project is a setup for failure.