Do Automatic Shrimp Feeders Really Save Feed? Timers, Feed Trays and Acoustic Sensing Compared
"An automatic feeder pays for itself in saved feed." Equipment brochures repeat that sentence so often that many farmers treat it as a law of nature. The best controlled evidence we found says it is wrong for the simplest kind of automatic feeder. A pond trial published in Aquaculture in 2021 compared automated feeding schedules and found no statistical differences among the timer-feeder treatments, while the one system that listened to the shrimp and fed on demand, an acoustic feedback unit, gave higher yields.
This is a myth-bust, so the claim is stated plainly before it is taken apart: buying an automatic feeder reduces feed use and improves feed conversion. The truth has two halves. A feeder is a delivery device. It delivers the ration somebody programmed, at the times somebody chose, whether or not the shrimp are hungry. Savings come from the decision about how much to deliver, and a timer does not improve that decision; it only repeats it faithfully. What does change the decision is information from the pond: what is left on the feed trays, whether the oxygen can support a meal, and in the best case what the shrimp themselves are doing.
The post sorts feeding control into three levels (timer, feedback from trays and water quality, on-demand acoustic sensing), shows what each level can and cannot save, runs the arithmetic that most proposals skip, and ends with a ration rule engine, a cost model and a checklist for a farm deciding what to buy. It follows our posts on dissolved oxygen and on which water-quality sensors to install, and it uses the same test: what can a farmer do within the hour, and what does it cost to keep right?
A feeder is a delivery device, not a decision. The feed saved by automation comes from better information about how much to deliver, not from the motor that delivers it.
What the Myth Gets Right
Automatic feeders solve real problems, and a fair verdict starts there. Hand feeding in a large farm means a crew walking bunds at fixed hours in heat and rain. Feeders do not get tired, do not skip the 2 a.m. meal, and spread the day's ration over more, smaller meals, which suits an animal that grazes continuously. They free labour for tasks that need a person, such as checking trays, taking samples and watching the water. They produce a log of what was delivered and when, which hand feeding rarely does. And they reduce the variation between crews and shifts.
None of that is a feed saving. Each is a labour, consistency or record-keeping benefit, and all of them justify their price if the farm values them. The myth runs into trouble when those benefits are sold as feed-conversion gains.
Where the Feed Money Is Decided
Feed is the largest operating cost in intensive shrimp culture on most farms, and the feed conversion ratio (FCR), the kilograms of feed given per kilogram of shrimp harvested, is the number that summarises how well it was managed. Two published observations frame the problem. A technical article on feed management in outdoor ponds for the whiteleg shrimp (Litopenaeus vannamei) states that feed management accounts for 65 to 75 percent of the variation in FCRs at harvest, and that the feeding rate giving the lowest FCR and fastest growth varies by 10 to 15 percent. In other words, most of the difference between a good crop and a poor one in feed terms comes from how much and when the farm feeds, and the target is a moving band, not a fixed number.
The arithmetic of a small improvement is large. On a 10-tonne harvest, an FCR of 1.5 uses 15,000 kg of feed and an FCR of 1.4 uses 14,000 kg, a saving of 1,000 kg, or 6.7 percent of the feed. Improving to 1.3 saves 2,000 kg, or 13.3 percent. Multiply by the farm's own feed price to get the rupee value, and compare it with the cost of the equipment. The comparison shows why farms are tempted by feeders and why the comparison has to be honest: a 0.1 improvement in FCR is a real prize, and the question is what produces it.
FCR change on a 10 t harvest | Feed used before | Feed used after | Feed saved | Share of feed saved |
1.5 to 1.4 | 15,000 kg | 14,000 kg | 1,000 kg | 6.7 % |
1.5 to 1.3 | 15,000 kg | 13,000 kg | 2,000 kg | 13.3 % |
1.4 to 1.3 | 14,000 kg | 13,000 kg | 1,000 kg | 7.1 % |
1.3 to 1.2 | 13,000 kg | 12,000 kg | 1,000 kg | 7.7 % |
Three Levels of Feeding Control
Level | How the ration is decided | What it needs | What it cannot see |
1. Timer or programmed feeder | A feeding table and a schedule, set by the farm | A table, an estimate of biomass, a person to update it | Whether shrimp ate, how many survived, whether the water allows feeding |
2. Feedback feeding (trays plus water quality) | The table, corrected by tray leftovers and by oxygen and temperature | Trays, a crew to check them, a DO and temperature sensor | Appetite between checks; depends on tray placement and crew discipline |
3. On-demand acoustic feeding | The shrimp's own feeding activity, heard by a hydrophone | An acoustic unit and controller, plus levels 1 and 2 as a safety net | Anything that makes noise in the pond; needs validation per pond |
Level 1 is what most "automatic feeder" proposals sell. Level 2 is the approach extension services have taught for decades and is where most of the achievable saving sits. Level 3 is the genuinely new capability, and the evidence for it is the strongest, although much of it is vendor-adjacent and should be tested on your own ponds.
Level 1: What a Timer Does and Does Not Do
A timer feeder broadcasts a set amount at set times. If the programme is good, a timer reproduces good feeding without the labour. If the programme is wrong, a timer reproduces the error every meal, all day, for the whole crop, with no human noticing that the trays are full of leftovers because the human is no longer walking the bunds.
The Auburn pond trial speaks directly to this. Its authors compared timer feeding treatments, with different schedules and frequencies, and found no statistically significant differences among them; the acoustic feedback treatment produced higher yields. The lesson is not that timers are useless. It is that tuning a schedule does not substitute for information, and that automating a table does not make the table more accurate.
The first thing that goes wrong with a programmed ration is biomass. A feeding table gives a ration as a percentage of standing biomass at a given shrimp size, and the biomass depends on survival, which the farm does not know. A farm that assumes 90 percent survival and feeds accordingly while the pond holds 75 percent is overfeeding by a fifth, every day, until a sample catches it. An extension article gives a method to back-calculate survival from feed consumption, and the arithmetic shows how much hidden error a table can carry.
# biomass_check.py -- back-calculate biomass and survival from recent feed consumption
def biomass_from_feed(avg_3day_feed_kg: float, feed_rate_pct: float) -> float:
return avg_3day_feed_kg / (feed_rate_pct / 100.0)
def survival(avg_3day_feed_kg, feed_rate_pct, abw_g, stocked):
biomass_kg = biomass_from_feed(avg_3day_feed_kg, feed_rate_pct)
shrimp = biomass_kg * 1000.0 / abw_g
return biomass_kg, shrimp, shrimp / stocked * 100.0
# ABW 14 g, 3-day average feed 65 kg, feeding rate 2.7 %, stocked 200,000
print(survival(65, 2.7, 14, 200000)) # -> (2407.4 kg, 171,958 shrimp, 86.0 % survival)
With an average body weight of 14 grams, a three-day average consumption of 65 kg, a table rate of 2.7 percent and 200,000 shrimp stocked, the calculation gives a standing biomass of about 2,407 kg, a population of about 171,958 shrimp and a survival of 86.0 percent. (The source's own worked example prints 146,214 shrimp and 73.1 percent for the same inputs, which does not follow from its own biomass figure; the numbers above are recomputed, and every farm should check any worked example against a calculator before putting it in a spreadsheet.) The same article notes that experienced technicians can estimate biomass within about 10 percent, which is a reminder that the whole programme rests on a number with a 10 percent error band.
Level 2: Feedback From the Pond
Feedback feeding adds two things to the table: what the shrimp left behind, and whether the water can support a meal.
Feed trays
A feed tray is a small tray or net lowered into the pond with part of each meal on it. After a set time the crew lifts it and scores what is left. The scoring scheme in the technical article is simple. At least three of four trays must be scored for the result to count. If the trays are clean (a score of 0000 or 0001), the next ration rises by 5 percent. If they are heavy with leftovers (1222 or 2222), the next ration drops by 10 percent. The article notes that rations are usually adjusted within a band from plus 5 percent to minus 10 percent. The asymmetry is deliberate: overfeeding costs feed and water quality, underfeeding costs growth, and the rule leans toward caution.
# tray_rule.py -- adjust the next ration from tray scores (0 empty, 1 trace, 2 heavy leftover)
def tray_adjust(scores, min_scored_fraction=0.75):
scored = [s for s in scores if s is not None]
if len(scored) / len(scores) < min_scored_fraction:
return 0.0, "not enough trays scored"
key = "".join(str(s) for s in scored)
if key in ("0000", "0001"): return +5.0, "increase 5%"
if key in ("1222", "2222"): return -10.0, "decrease 10%"
avg = sum(scored) / len(scored)
if 0.4 <= avg <= 1.2: return 0.0, "hold"
return (-5.0, "trim 5%") if avg > 1.2 else (+2.5, "nudge up")
Running the rule on six tray readings gives the outputs below. The two rules in the middle band (hold, trim, nudge) are our additions for the cases the published rule does not name, and they should be tuned with the farm's adviser.
Tray scores (4 trays) | Rule output |
0, 0, 0, 0 | Increase 5 % |
0, 0, 0, 1 | Increase 5 % |
1, 2, 2, 2 | Decrease 10 % |
1, 1, 0, 1 | Hold |
0, none, 1, 1 | Hold (three of four trays scored) |
none, none, 1, 0 | No change: not enough trays scored |
Trays have well-known weaknesses. They depend on placement, because a tray in a dead corner says nothing about the pond. They depend on crew discipline, because a tray lifted late or scored from memory is noise. And they add labour, which is the thing farms buy feeders to remove. A farm that installs feeders and stops checking trays has automated the delivery and abandoned the feedback.
Water quality as a gate
Shrimp eat less, and digest less, when the water is stressful. Oxygen is the strongest signal. A feeding programme that ignores dissolved oxygen will feed into a pre-dawn oxygen sag, which adds oxygen demand from digestion and uneaten feed exactly when the pond can least afford it. A simple gate multiplies the ration by a factor that falls as oxygen falls or as temperature leaves the comfortable band:
# feed_gate.py -- scale the ration by pond conditions (thresholds are the farm adviser's to set)
def feed_gate(do_mgl: float, temp_c: float) -> float:
if do_mgl < 3.5: return 0.0 # do not feed
if do_mgl < 4.5: return 0.5 # half ration
if temp_c < 24 or temp_c > 34: return 0.7
return 1.0
def next_ration(base_kg, tray_change_pct, do_mgl, temp_c):
return base_kg * (1 + tray_change_pct / 100.0) * feed_gate(do_mgl, temp_c)
# next_ration(100, -10, 4.0, 30) -> 100 x 0.90 x 0.5 = 45 kg
The gate costs nothing if the farm already has dissolved-oxygen and temperature probes. Our earlier posts argue that those two sensors are the first ones worth installing, and a feeder that can read them becomes safer and cheaper to run. Where the sensors do not exist, the gate becomes a crew instruction: do not feed when the aerators are struggling.
Feed against the water, not just against the table. A meal fed into a falling oxygen curve is wasted feed and a worse night.
Level 3: Listening to the Shrimp
Shrimp make sounds when they feed. Acoustic feeding systems put a hydrophone in the pond, listen for the sounds associated with feeding activity, and let the feeder deliver feed while the shrimp are feeding and stop when they are not. An industry article describes sensor-based control of multiple feeders within a pond, and a university thesis describes acoustic feedback systems as capable of adjusting the feeding rate to feed only when the shrimp are actively feeding.
The evidence for the approach is the strongest in this post, and it still needs careful reading. The Auburn pond trial found higher yields with the AQ1 acoustic feeding system, and a trade article summarising the research states that on-demand acoustic feeding can significantly increase yield and reduce feed wastage. An industry article reports that productivity gains, reflecting growth increases and FCR reductions, averaged 25 to 30 percent in the cases it describes. That last figure comes from a vendor-linked source and combines growth and conversion, so it should be treated as a claim to test in your own ponds, not a number to put in a business case.
What an acoustic system adds, in terms of the framework, is appetite information at the moment of feeding. Trays tell you what was left after the last meal. Oxygen tells you the pond can digest. Sound tells you whether the shrimp are eating now. That is the missing variable, and it explains why a system with it beat timers whose schedules differed.
Practical issues matter. Aerators and paddlewheels are loud, and a hydrophone near one hears mostly aerator. Pond depth, bottom type, stocking density and the moulting cycle all change what is heard, so a unit needs to be positioned and tuned for each pond, with a trial period in which the acoustic reading is compared with tray scores. The controller should have fall-backs: a maximum ration, a minimum ration, a stop on low oxygen, and an alarm when the hydrophone goes silent for hours because it has been fouled or has failed.
What the Evidence Says in One Table
Claim | Evidence found | Verdict |
"A timer feeder saves feed" | Auburn pond trial: no statistical differences among timer treatments | Not supported as a feed saving; a labour benefit |
"Feed management drives conversion" | Feed management accounts for 65 to 75 percent of FCR variation at harvest | Supported: the decision matters more than the device |
"Feed trays improve the ration" | Standard practice: score trays, adjust plus 5 to minus 10 percent | Supported, with labour and discipline |
"Acoustic on-demand feeding improves yield and waste" | Pond trial (AQ1): higher yields; trade summary: reduced wastage | Supported by research; test on your ponds |
"Acoustic systems deliver 25 to 30 percent productivity gains" | Industry article, growth plus FCR combined | A claim to validate; do not budget on it |
"Biomass estimates from tables are accurate" | Technicians estimate within about 10 percent; survival assumptions can hide large errors | Check with sampling and back-calculation |
The Economics a Proposal Should Show
A proposal that does not state the feed saving per crop, with its assumptions, is a brochure. The calculation is short:
Current feed use and harvest from the last two crops for the target ponds (kg of feed, kg of shrimp, FCR).
Expected FCR improvement with each level of control, from the farm's own history and the adviser's judgement, not the vendor's.
Feed saved per crop: (old FCR minus new FCR) times harvest weight. Convert to rupees at the farm's feed price.
Other benefits valued separately: labour hours saved, survival changes, earlier harvest, reduced water exchange.
Cost of ownership per crop: equipment spread over its life, installation, power, maintenance, replacement of fouled parts, and the labour that feedback still needs (tray checks, calibration, cleaning the hydrophone).
Break-even FCR improvement: the FCR change at which the feed saving equals the cost.
Doing step 6 first is the most useful discipline. If a system costs the equivalent of 600 kg of feed per crop per pond and the pond harvests 3 tonnes, the break-even is a 0.2 improvement in FCR, which is large; if the harvest is 10 tonnes, the break-even is 0.06, which is plausible. The same equipment can be a bad buy on a small pond and a good one on a large pond, and no brochure will say so.
Twelve Ponds, Three Crops
A representative composite, genericised and not a named client, shows how this plays out. A 12-pond brackish-water farm bought programmable feeders for all ponds. In the first crop the farm reported the labour saving it expected, and its FCR was within the normal variation of earlier crops. Feed use per pond was no lower, which matched the Auburn finding. In two ponds, feed trays showed persistent leftovers late in the crop while the timers kept delivering the programmed ration; the farm traced it to survival well below the table's assumption, which the back-calculation in this post confirmed.
For the second crop the farm kept the feeders and added the missing layer: tray scoring four times a day in each pond by a trained crew member, the oxygen and temperature gate on the feeder controller, and a weekly sample to correct the survival estimate. Rations followed the rule engine, with the supervisor able to override. FCR improved by a margin the farm judged worth the extra labour. In the third crop it trialled an acoustic unit in two ponds beside the tray method, and compared three things for each meal: the acoustic reading, the tray score and the oxygen. The units agreed with the trays most of the time, and disagreed in ways that led the crew to look at the pond: once because an aerator had been moved next to the hydrophone, and once because a bloom crash had changed feeding. The figures behind this account are illustrative, not a published result. The pattern is the useful part: delivery automation first changed labour, information changed feed, and the best use of the acoustic unit was as a second opinion that sent a person to look.
Reading Tray Scores Like an Instrument
A feed tray is a measuring instrument, and like any instrument it gives trustworthy readings only when it is placed, used and read the same way every time. Most of the disagreement between farms about whether trays "work" comes from differences in how they are run, not from the idea. A tray lifted at a different interval each day, in a different corner, by a different person who scores by eye in different light, produces a time series dominated by noise.
The scoring scheme in the extension guidance uses at least three of four trays per check, which is a quiet statement about sampling: one tray is an anecdote, four trays give a pattern. Treat the following as the calibration of the instrument.
What to standardise | Why it matters | A workable rule |
Tray position | A tray near the feeder spread sees more feed than one at the pond edge | Fix four positions per pond that cover the feeding area, mark them, and never move them mid-crop |
Tray depth and type | Depth and mesh change what shrimp find and what falls through | Use one tray design across ponds and record any change |
Time of lifting | Leftovers fall with time as shrimp eat and feed dissolves | Lift after the same interval each meal, set by the farm adviser |
Scoring card | "A bit left" means different things to different people | Define 0, 1 and 2 with photographs on a laminated card carried by the crew |
Lighting and rinsing | Murky water and glare hide pellets | Lift gently, tilt the tray to drain, and score in the shade of the body |
Recording | Memory is the enemy of a trend | Enter the scores on a phone at the bund, before walking on |
Calibrate the crew the way you would calibrate a sensor. Once a week, have two people score the same four trays independently and compare. If they disagree on more than one tray in four, the scoring card needs better photographs or the crew needs a refresher. Keep the photographs: a tray image with a time and pond number is cheap evidence when a score is questioned.
A tray also has a built-in bias. It samples the feed that reaches the tray, which is a function of where the feed lands. A feeder with a narrow spread, a pond with a strong wind, or a paddlewheel that herds feed to one side will make the trays unrepresentative. If the scores from two trays are always zero and from two are always two, the feed is not landing where the animals are, and the fix is mechanical, not in the ration.
Four trays, same positions, same interval, same card, same time of day. That is what turns a tray from an anecdote into a measurement.
What Automation Does Change: Frequency, Timing and Night Meals
If a timer feeder does not by itself save feed, what does it change? Four things, and each has value.
Meal size and frequency. Hand feeding in a big farm tends toward a few large meals because each round costs labour. A feeder can deliver the same daily ration as many small meals. Smaller meals leave less uneaten feed in the water at any moment, which helps oxygen and water quality, and they make tray checks more informative because the trays show the result of one modest meal, not a heap. The Auburn trial's finding of no statistical differences among its timer treatments suggests that tuning frequency and schedule is not the main lever. It does not suggest that fewer, larger meals are better.
Night feeding. Shrimp feed around the clock, and a farm that stops feeding at dusk for lack of a night crew is leaving a share of the day's feeding opportunity unused. A feeder makes night meals possible without a person on the bund. The condition is oxygen: night is when oxygen is lowest, so night meals belong behind the oxygen gate, and a feeder that cannot read the oxygen sensor should not feed in the pre-dawn hours.
Distribution. A hand-broadcast meal is spread by arm; a feeder spreads by a mechanism with its own pattern. A narrow spread concentrates feed and shrimp near the feeder, and a wide one covers more of the pond. Mapping the spread area, and placing trays inside it, is part of commissioning. A feeder installed without checking where its feed lands will produce tray scores that describe the feeder, not the pond.
Consistency and records. A feeder does what it was told on the day the supervisor was away, and keeps a log. The log is useful when something goes wrong, because it separates "we fed it" from "we did not."
None of these four is a feed saving by itself, and all four are ways to run the feedback layer more reliably. That is the honest position: the automation makes good feeding easier to do and bad feeding easier to repeat.
Failure Modes: When the Feeder Lies
A feeder reports that it delivered 5 kilograms. Whether it did is a question about hardware, and several things go wrong in a hot, humid, salty environment.
Failure | How it shows up | How to catch it |
Damp feed bridging in the hopper | Dispensed weight falls below the commanded amount; some cycles deliver nothing | Weekly dispense test; hopper covers and desiccant; check after rain |
Calibration drift | A steady shortfall or excess against the setting | Compare commanded and dispensed weight over ten cycles |
Jammed or worn spreader | Feed lands in a clump; trays near the feeder always full | Walk the spread pattern; inspect the spreader and motor |
Fines and dust in the feed | Poor flow, uneven spread, more waste | Check feed quality at delivery; sieve if needed |
Power dips or flat battery | Missed meals; clock errors | Log supply voltage; alarm on missed cycles; see our post on solar and battery sizing |
Lost control link | Feeder runs its last programme, or none | Local fall-back schedule; alarm on lost heartbeat |
Birds and pests | Feed taken from hoppers or the water surface | Covers and inspection |
A short weekly dispense test turns the first two into numbers. Command a fixed amount ten times into a bucket on a scale and compare:
# dispense_check.py -- is the feeder delivering what it says?
def calib_check(commanded_kg, dispensed_kg, tol_pct=5.0):
errs = [(d - c) / c * 100 for c, d in zip(commanded_kg, dispensed_kg)]
mean = sum(errs) / len(errs)
worst = max(errs, key=abs)
status = "ok" if abs(mean) <= tol_pct and abs(worst) <= 2 * tol_pct else "recalibrate"
return round(mean, 2), round(worst, 2), status
# Ten commanded 5 kg cycles:
# good unit -> mean -0.6 %, worst -6.0 %, ok
# drifting unit (4.1 to 4.5 kg delivered) -> mean -13.2 %, worst -18.0 %, recalibrate
A feeder that delivers 13 percent less than the setting is the quiet reason a farm sees "no change" in its feed figures while believing it is feeding 100 kg a day. The ration table and the actual feed differ, and the tray scores will drift to match the real feed while the record shows the nominal one. Weigh the real delivery before judging the feeder, the programme or the shrimp.
Weigh what the feeder delivers before you trust what it reports. A thirteen percent shortfall in the hopper looks exactly like a thirteen percent improvement in conversion on paper.
Twelve Questions for Any Feeder Proposal
Vendors answer the questions they are asked. These are the twelve to ask, with what a good answer sounds like.
Question | A good answer sounds like |
What feed saving do you claim, and on what ponds and crops? | Named trials, with the control and the numbers, ideally independent |
Does the system use tray, oxygen or acoustic feedback, or only a timer? | A clear list, with what is included and what is extra |
How does it read dissolved oxygen and stop feeding when oxygen is low? | A sensor input and a documented rule, with a manual override |
How is delivered weight verified? | A dispense test procedure and a calibration routine |
What happens on power loss, and on loss of the control link? | A defined fall-back and an alarm |
How does it handle damp feed? | A hopper design, covers and a maintenance schedule |
Where does the feed land, and how wide? | A spread map for the pond size |
What does the acoustic unit hear, and what fools it? | Aerator noise, depth and density effects, and a validation period |
What data does it log, and who owns it? | Meal-by-meal logs the farm can export |
What is the cost per pond per crop, including maintenance? | A number with the assumptions behind it |
What is the break-even FCR improvement for my harvest size? | A calculation, not a slogan |
What does the supplier do when it fails at night? | Response time, spares and local service |
Put the answers in a table and compare vendors on the answers, not on the brochure. The vendor who struggles with question 11 has told you most of what you need to know.
Where Each Level Fits
Farm type | Sensible starting point | Why |
Small farm, owner on site, a few ponds | Hand feeding with trays and a DO and temperature probe | The owner is the feedback loop; a feeder adds cost before it adds information |
Mid-size farm with a night crew | Programmable feeders plus trays, DO gate and weekly biomass checks | Feeders remove night labour; feedback keeps the ration honest |
Large farm with technicians | Feeders with tray and oxygen feedback, acoustic units on pilot ponds | The cost can be spread, and technicians can validate the acoustic reading |
Any farm with poor survival data | Fix sampling first | Every ration table depends on biomass |
When Not to Buy Feeders
When the farm cannot sample. Without regular sampling of body weight and a view of survival, a feeder will deliver a confident wrong ration.
When nobody will check trays. A feeder without feedback is the configuration the research suggests gives no feed saving.
When power or maintenance is unreliable. A jammed or unpowered feeder is a missed meal, and a failure alarm is not optional.
When the pond is too small for the economics. Do the break-even calculation before the quote.
When the water is the problem. Feeding more precisely into poor oxygen makes the poor water worse.
Starting on Monday
Compute FCR for each pond from the last two crops, and write down the feed price.
Back-calculate survival in each pond from recent feed consumption, and compare with the table's assumption.
Install DO and temperature probes on the ponds you intend to automate, and build the feed gate into the controller or the crew's instructions.
Start tray scoring with four trays per pond, four checks a day, and record the scores on a phone.
Write the ration rule down, including the plus 5 and minus 10 percent limits and who can override it.
Buy feeders for the labour and consistency they give, and treat any feed saving as something the feedback layer earns.
Pilot acoustic sensing on two ponds for a crop, beside trays, and log every disagreement.
Review each crop: FCR, survival, leftover scores, oxygen events, and what the farm would change.
The Motor and the Mind
A feeder is a good motor. A feeding programme is a mind, and it is built from sampling, trays, oxygen readings and, if the budget allows, the sound of the shrimp themselves. Farms that buy the motor and skip the mind should expect better rested crews and the same feed bill. Farms that build the mind first will find that the motor is the cheap part.
EurthTech is an engineering-first product design and embedded systems company with a development center in Hyderabad. To discuss a project, write to connect@eurthtech.com or see https://www.eurthtech.com/case-studies.
Comments