Tuesday, March 14, 2023
I've been handling solar equipment orders for installers and system integrators for nine years. In that stretch I've personally made — and documented — eleven significant mistakes, totaling roughly $47,000 in wasted budget. The list lives in a shared doc that new hires read in their first week. It's the most useful thing I've ever built, and every line on it cost somebody real money.
The one I keep coming back to started on a Tuesday afternoon in March 2023. Our Q1 numbers were due the next morning, everyone else was in a conference room, and I was answering the phone because there was nobody else to do it.
The caller was a systems integrator I'd worked with since 2019. Careful guy. Asks good questions. He was quoting backup power for a rural machine shop — about 6,000 square feet, three CNC machines, and a well pump. The shop owner had been doing his own research, which is usually a good sign and occasionally a disaster. The first thing out of my integrator's mouth:
"What's the difference between an inverter and a generator? Because my customer keeps saying he wants the most powerful solar generator he can buy, and I'm not sure he means what I think he means."
I gave him the thirty-second version. A generator burns fuel, spins an alternator, makes AC. An inverter takes DC — from panels or from a battery — and converts it to AC. A "solar generator" is a battery and an inverter in a plastic case with a handle. Marketing term, not a product category.
He said, "Okay, so we need an inverter." We moved on to sizing, which is the part of the job I actually like.
That was mistake number one, and it took about fifteen seconds. I answered the question he asked instead of the question underneath it.
Sizing the system — and the assumption I didn't check
The shop's load list was straightforward. Three CNC machines, air compressor, well pump, lights, office equipment. Peak demand maybe 22 kW if everything ran at once, which it never does. Steady-state, closer to 9 kW.
I asked for nameplate data on everything. I got it for everything except the pump. The integrator said, "It's a standard 1.5 hp submersible. Nothing weird."
I knew I should ask for a photo of that nameplate. I've been burned by that exact sentence before. But it was 4:40 in the afternoon, I had a report to finish, and I thought — what are the odds? It's a residential well pump. They're all basically the same.
The odds caught up with me eight weeks later.
On the equipment side, we'd standardized on SMA inverters for commercial work years ago. Part of that is reliability, part of it is that I can pull a replacement in a day instead of a month. When I'm comparing brands I do look at things like the SMA Energy AG inverter shipments 2023 figures — don't hold me to the exact number, but I believe it was somewhere in the 20 GW range — not because shipment volume makes a product good, but because it tells me there's an installed base. An installed base means spare parts in five years, and spare parts in five years is a line item almost nobody puts in their quote comparison.
We spec'd a hybrid SMA inverter with a battery bank. The SMA Flex inverter line was the natural fit for a job this size, since it handles the grid-tied side and the backup side without a separate box. On paper, it was correct. Continuous rating covered the load, surge rating covered the compressor startup, all the numbers matched.
On paper.
The pump kicked on at 3:47 p.m.
Commissioning day, mid-May. Beautiful weather, which I mention only because it meant the shop was running at full tilt with the doors open. Everything came up clean. Grid-tied, batteries charged, transfer tested, all fine.
Then the well pump's pressure switch closed.
The inverter display threw a fault, went dark for about two seconds, and came back. Every other load in the building stayed up. Just the pump.
We cycled it four times. Same result every time.
The pump wasn't 1.5 hp. It was 2 hp, and — more importantly — it had a locked rotor amp draw peaking around six times its running current for roughly a second on startup. Our inverter's surge rating was 1.5x its continuous rating for three seconds. That's a perfectly legitimate spec. It's also completely the wrong spec for starting a motor.
I'd verified every assumption except the one I was least sure about. That's not an oversight. That's a decision to skip the work.
The fix at the pump was a soft starter — $1,480 installed, which brings startup current down to something the inverter can actually handle. But when we re-ran the load calc with real numbers, we found the CNC spindle motors had similar startup characteristics that I'd modeled too optimistically. The whole system was one size under.
So we didn't just fix the pump. We swapped the inverter for the next size up, re-filed the permit, re-scheduled the inspection, and pushed the owner's go-live back two weeks.
What that actually cost
The original quote was about $1,150 cheaper than the correctly sized one. Here's what the "cheaper" version ended up costing:
- Inverter size upgrade: $1,150
- Return freight on the original unit: $340
- Soft starter plus installation: $1,480
- Re-permit and inspection re-scheduling: $400
- Roughly 30 hours of my time and our field tech's time: not billed, but not free either
Call it $4,200 in avoidable cost, plus two weeks of delay the shop owner never got back. The cheaper quote was not the cheaper system. It wasn't even close.
The thing that actually changed how I work
Two months later I was back on site for the follow-up inspection, and the owner showed me his phone. He had a Filtrete smart air filter in the shop's HVAC unit — the kind that pings you when the filter needs changing. He was genuinely pleased with it. Thirty dollars of filter, and he could tell you its status at any moment of any day.
Behind him, through the door, sat about $28,000 of power equipment with no monitoring on it at all. No alerts. No remote visibility. He'd find out something was wrong the same way he found out in May — by walking back there and hearing a fault beep.
I don't think that's a discipline problem. I think it's a sales problem. Nobody on my side of the table had ever put monitoring in the quote as a line item, so it never entered the comparison. It was invisible, and invisible costs don't get evaluated.
Monitoring is part of total cost of ownership. So is spare parts availability. So is the lead time on a replacement unit. So is the labor to redo something. The invoice price is just the first number, and it's usually the least interesting one.
What goes on the checklist now
- Ask what the equipment is for before explaining what it is. "What's the difference between an inverter and a generator" is basically never a definitions question. It's a question about a job that needs doing.
- Get the nameplate. Every time. Not the spec the customer remembers. Not the one in the listing. The plate on the motor.
- Size for startup, not for steady state. Steady-state numbers are what make quotes look attractive. Startup numbers are what make systems work.
- Quote the monitoring. Put it in writing, even if the customer declines it. Then it's a decision instead of an omission.
- Compare total cost, not unit price. I keep a short list of costs people leave out: freight both directions, labor, permitting, downtime, rework, and the value of knowing your system is fine without driving out to look at it.
One note on spec sheets, since this is exactly where I got fooled. The FTC requires advertisers to substantiate performance claims, and that's a real requirement with real teeth. But "substantiated" is a floor, not a ceiling. A surge rating can be perfectly honest, properly tested, and still tell you almost nothing about starting a 2 hp submersible at 40°C in a metal building in May. Read the test conditions, not just the number.
Where this applies and where it doesn't
I'm not an electrical engineer, so I can't walk you through motor starting theory or tell you how to size a soft starter for a specific pump. That's a conversation for whoever stamps your drawings. What I can speak to, from nine years of quoting and ordering this equipment, is how load lists get built wrong and how quotes hide cost.
And this is all from my corner of the market — commercial and light industrial projects in the 5 kW to 150 kW range, mostly in North America. If you're working utility-scale, or off-grid somewhere with no service infrastructure, the priorities shift in ways I'd just be guessing at.
What I can tell you is that the checklist works. We've caught 41 potential sizing and spec errors with it since I wrote it in late 2023. Every one of those is a phone call somebody didn't have to make, standing in a shop, listening to a fault beep.
There's something satisfying about a commissioning day where nothing goes wrong. It took me a $4,200 mistake and a customer's two lost weeks to earn that feeling, but I'll take it.
