Which hydroponic system should I build?
The comparisons all give you the same grid — cost, difficulty, yield, "best for beginners" — and none of them tells you the thing that actually separates these systems, which is how they fail. Every one of them holds water in two different places, and the two have nothing to do with each other.
The root-zone buffer is how much water sits around the roots right now. It decides what happens when the power goes out: NFT is a film over bare roots and wilts in half an hour, while a Kratky jar has no pump to lose in the first place.
The reservoir buffer is gallons of solution per plant. It decides how fast the chemistry runs away from you — because plants drink water faster than they take up salt, so a small reservoir concentrates as it empties. That's the buffer nobody quantifies, and it's the one you live with daily.
NFT has a large reservoir and no root buffer. Kratky is almost all root buffer and drifts hard. DWC has a decent amount of both, which is the real reason it's the standard beginner recommendation.
How each one fails
| System | Reservoir per plant | Roots survive a pump failure | What actually goes wrong |
|---|---|---|---|
| Kratky (passive) | 1.0 gal | no pump to fail | No pump, no power, nothing to fail. The solution draws down and concentrates as it goes. |
| Deep water culture | 5 gal | about 12 hours | Roots sit in the reservoir. The pump supplies oxygen, not water, so a failure starves rather than dries. |
| Ebb and flow | 1.5 gal | about 6 hours | The medium holds water between floods, which is the buffer. Cheapest per site at scale. |
| NFT (nutrient film) | 2 gal | 30 minutes | A film of water over bare roots. Nothing is held anywhere; the pump is the system. |
Reservoir volumes are the usual hobby figures for leafy crops — a 5 gal bucket per plant in DWC, a gallon jar in Kratky. Survival times are what growers report and what the mechanism predicts: NFT dries, DWC suffocates (the pump supplies oxygen, not water, so the failure is dissolved oxygen rather than drought), ebb-and-flow coasts on the moisture in the medium.
Read that third column as a shopping list for backup. An NFT system genuinely needs a battery backup or a UPS on the pump — half an hour is not enough time to notice. A DWC bucket needs one much less urgently; a Kratky jar needs nothing at all, which is most of its appeal.
What the reservoir per plant actually decides
Here's the half of the comparison that never gets numbers. Plants take water out faster than they take out salt, so an untouched reservoir doesn't just get emptier — it gets stronger. How fast depends entirely on how many gallons each plant has to work with. Same starting solution at 1.2 mS/cm, same lettuce drinking 0.15 gallons a day, nobody topping anything off:
| System | Start | After 1 day | After 3 days | After 7 days | Half empty at |
|---|---|---|---|---|---|
| Kratky (passive) | 1.2 | 1.41 | 2.18 | run dry | 3.3 days |
| Deep water culture | 1.2 | 1.24 | 1.32 | 1.52 | 16.7 days |
| Ebb and flow | 1.2 | 1.33 | 1.71 | 4.00 | 5.0 days |
| NFT (nutrient film) | 1.2 | 1.30 | 1.55 | 2.53 | 6.7 days |
EC in mS/cm. Computed the same way as the top-off calculator: the salt stays, the volume shrinks, so EC rises by the ratio of the volumes. Late veg wants 1.2–1.8; shaded cells are past 2.4, where most leafy crops are in salt stress. A fruiting plant drinks 0.75 gal a day — five times as fast — so divide every "days" figure by five.
Three days untouched and the Kratky jar has gone from 1.2 to 2.18 mS/cm — through the top of the veg band and into salt stress — while the DWC bucket has crept to 1.32. That is the same plant, drinking the same water. The only difference is that one of them has 5 gallons of buffer and the other has 1.
This is the honest trade nobody prints. Kratky removes the pump and hands you the drift instead. It's a genuinely excellent system for a single fast crop of lettuce — in and out in six weeks, and the concentration climbing as the plant matures is roughly what the plant wants anyway. It is a bad system for anything you'll keep for months, because there is no way to correct the solution without disturbing the air gap the roots depend on. If you're going to top off and adjust, you may as well have the bucket and the airstone, and DWC is that system.
What each costs to build
The cost comparisons quote a price for a system, which hides the thing that matters: some of these scale and some don't. A pump and a reservoir get bought once no matter how many plants they feed; a DWC bucket and airstone get bought again for every single plant.
| System | Fixed | Per site | 1 plant | 2 plants | 4 plants | 8 plants | 16 plants |
|---|---|---|---|---|---|---|---|
| Kratky (passive) | $0 | $6 | $6 | $12 | $24 | $48 | $96 |
| Deep water culture | $15 | $20 | $35 | $55 | $95 | $175 | $335 |
| Ebb and flow | $60 | $5 | $65 | $70 | $80 | $100 | $140 |
| NFT (nutrient film) | $70 | $8 | $78 | $86 | $102 | $134 | $198 |
Typical 2026 hobby prices for the parts, not kits. At 8 sites these land on the per-site figures published elsewhere (DWC $22, NFT $17, Kratky $6 a plant), which is a reasonable check that the split is right.
Read across the DWC row. It's the second-cheapest way to grow one plant ($35) and the most expensive way to grow sixteen ($335, against $198 for NFT) — because nothing about it is shared. NFT overtakes it at 5 plants and ebb-and-flow at 4. Kratky is cheapest at every scale a home grower will ever build; ebb-and-flow's cheaper sites only catch it at around 61 plants, which is a different hobby.
Which crops suit which system
| System | Good for | Bad for | Why |
|---|---|---|---|
| Kratky | Lettuce, herbs, bok choy — anything harvested once | Tomatoes, cucumbers, anything long-season | The solution only goes down. A crop that outlives the reservoir needs a refill you can’t do without disturbing the air gap. |
| Deep water culture | Lettuce, herbs, peppers, one big tomato per bucket | Dozens of small plants | Every site needs its own bucket and airstone, so the cost is linear and the plumbing multiplies. |
| Ebb and flow | Mixed sizes, peppers, strawberries, anything in medium | Very tall top-heavy crops without support | The medium buffers water and holds the plant up, which is why it takes mixed plantings the others can’t. |
| NFT | Lettuce, basil, strawberries — light, fast, uniform | Tomatoes and anything heavy or long-rooted | A shallow channel can’t hold a root mass or a heavy plant, and roots that mat block the film. |
The pattern: light, fast, uniform crops suit the systems with small root buffers (NFT, Kratky); heavy, long-season or mixed plantings need the ones with medium or volume behind them (ebb-and-flow, DWC).
So which one?
Your first ever hydro plant: Kratky. A jar, a net pot and nutrient. It costs $6, there is nothing to fail, and it teaches you what roots and EC do without any equipment in the way. Grow one lettuce. If it bores you, you've lost six dollars.
Your first real system: DWC. Under about five plants it's cheap, it has both buffers, it forgives a missed day, and every problem you'll hit is well documented. The airstone is the only moving part.
A dozen lettuces at a time: NFT or ebb-and-flow. Past the 5-plant crossover the shared pump and reservoir stop being an extravagance and start being the reason it's affordable. Take NFT for uniform leafy crops and ebb-and-flow if you want to grow mixed sizes or anything that needs to be held up — and put the NFT pump on a backup.
Whichever you build, the numbers you'll actually be managing are the same: EC in the range your crop wants, topped off before the drift above gets away from you. The EC ↔ PPM converter handles the meter, the dosing calculator tells you what to mix, and the top-off calculator works out whether today's reservoir needs water or nutrient — which is exactly the question this whole comparison has been about.
Sources: reservoir volumes per plant and system running practice from hobby and extension hydroponics guidance (½–1 gal per plant for leafy crops, 2–5 gal for fruiting, the standard 5 gal DWC bucket); pump-failure survival times as reported for each system type and as the mechanism predicts; component prices are typical 2026 hobby retail and are stated as typical — they reproduce published per-site costs at eight sites. EC drift is computed from conservation of dissolved salts, the same arithmetic as the top-off calculator; stage EC bands as on the EC ↔ PPM page.