-
There isn't one answer, and anyone telling you there is hasn't worked enough sites
-
Scenario 1: The system is genuinely obsolete — replace it
-
Scenario 2: The metering is fine but the logic is behind — upgrade, don't replace
-
Scenario 3: Data is wrong but the hardware isn't broken — repair and recalibrate before you buy anything
-
Scenario 4: You've outgrown the system — replace, but size it for where you're going
-
How to tell which situation you're in
There isn't one answer, and anyone telling you there is hasn't worked enough sites
I get this call maybe three times a month: "Our demand response system is acting up — do we rip it out and start over?" And I get a version of it from power monitoring and submetering clients too. The honest answer is that it depends on which of three or four situations you're actually in. I've walked into sites where the fix was a $340 part, and I've walked out of sites recommending a full controller replacement within 30 days. The trick is knowing which one you're looking at before you start spending.
Here's the breakdown I use when I'm triaging a call.
Scenario 1: The system is genuinely obsolete — replace it
If your DR controller or metering head is running firmware from before roughly 2018–2019 and the OEM stopped posting security patches, you're on borrowed time. That's not a sales pitch — it's just how long these things stay supported. I've seen sites where the vendor's last firmware release was 2016 and they still expect to pass a utility DR audit with OpenADR 2.0b compliance. That's not going to fly if your utility or ISO has moved to a newer OpenADR profile.
The trigger points I look for:
- No firmware update in 5+ years, and the OEM won't confirm support renewal
- No native support for OpenADR 2.0b, Modbus TCP, BACnet/IP, or MQTT — meaning you can't get it into your BMS or EMS without middleware that itself is aging
- Obsolete CT ratios or voltage sensing ranges that don't match your current service
- The OEM can't sell you a replacement board, only a full unit — and they're charging 2016-era prices for it
In March 2024, I got a call from a light industrial client 36 hours before a utility DR audit. Their controller was on 2015 firmware, the OEM had stopped supporting the model, and the auditor was asking for OpenADR 2.0b certification. We found a compatible gateway from stock, paid roughly $1,400 in rush freight, and had it installed and commissioned in 29 hours. The client's alternative was failing the audit and losing about $18,000 in annual DR revenue. That one was a straightforward replace. The system wasn't going to get compliant with a patch.
Scenario 2: The metering is fine but the logic is behind — upgrade, don't replace
This is the one most people get wrong. The CTs are good. The meter heads are reading accurately. The panel is clean. But the DR logic is stuck on fixed setpoints that don't respond to the utility's new time-of-use windows, or it can't accept a simple load-shed command from your building automation system.
Here's the counterintuitive part: replacing the whole system won't fix this. You'll spend $15,000–$40,000 on new hardware and still have mediocre DR logic if the new gear is configured by the same person using the same old approach. What you actually need is a new controller or gateway that sits in front of the existing meters and does the thinking. On a mid-size commercial site, that's usually $2,000–$5,000 installed, not a full rip-and-replace.
The way to check: can your existing meter heads talk Modbus or BACnet to a new controller? If yes, add the controller. If no, you're probably back in Scenario 1. My rule of thumb — if the meters and CTs are under 7 years old and calibrating within ±0.5% on a recent check, keep them.
Scenario 3: Data is wrong but the hardware isn't broken — repair and recalibrate before you buy anything
I used to assume that drifting readings meant a dying system. Half of my early service calls were just loose CT wiring, a failed gateway, or a comms setting that got changed during a network upgrade. I've since learned to check those three things before I even quote replacement hardware.
The failure modes I see most often:
- Loose or reversed CT leads — reads as a phase imbalance or a negative reading on one channel
- Gateway failure on the submetering side — the meters are fine but data stops flowing to the dashboard
- Comms misconfiguration — Modbus ID conflicts, changed IP ranges, an updated BMS that dropped the old polling address
- Out-of-range CTs after a service upgrade — the CT ratio wasn't updated when the panel was re-fed
A power quality analyzer GridSense or any comparable unit gets blamed for a lot of things that turn out to be wiring. If your readings drift on one channel but three others are clean, that's not a dying analyzer — that's a wiring or CT problem. Bring in a handheld analyzer and do a spot check against a known reference before you decide the entire system needs to go.
Scenario 4: You've outgrown the system — replace, but size it for where you're going
Adding a building, a tenant, or a major load (EV chargers, a new chiller, a second service) can quietly push a small submetering setup past its channel count or its panel capacity. This one does usually end in a replacement — but the mistake is sizing for today's count instead of next year's.
I'll be blunt here about a pattern that bugs me: small facilities — a 10-channel submetering job for a 40,000 sq ft tenant space, say — routinely get told "we don't do projects that small" or get quoted list price plus markup because the order isn't worth the paperwork. That's a bad way to do business, and I've watched it cost bigger vendors accounts. Smaller doesn't mean unimportant — it means the client is likely testing you before they trust you with the next building. I've had $340 diagnostic calls turn into $20,000 retrofits two years later, and I've had it go the other way with vendors who brushed off the small job.
If you're a small facility being told you don't qualify for the product line, look for a distributor or OEM that sells the same tier of gear into small projects. The hardware exists. The willingness is often what's missing.
How to tell which situation you're in
Three questions, in this order. Answer them honestly and you'll usually land in the right scenario within ten minutes.
- Is your OEM still supporting the hardware? If the answer is no — no firmware updates, no replacement parts, no compliance statements — you're in Scenario 1. Replace it.
- Has your utility or ISO changed its DR protocol, reporting format, or dispatch window recently? If yes, check whether your controller or gateway can adapt. If it can, that's Scenario 2 — upgrade the controller, keep the meters. If it can't, back to Scenario 1.
- Are the problems in the data or in the logic? Data drifting on specific channels is Scenario 3 — check CTs, wiring, comms, calibration. Logic that doesn't respond to new signals is Scenario 2. Whole-system capacity shortfalls are Scenario 4.
One caveat on my own experience: most of what I work with is mid-size commercial and light industrial — roughly 200 emergency dispatch calls over the last three years, maybe 60% of them in the 50–400 kVA range. If you're running utility-scale substation DR or a multi-megawatt industrial site, the protocol and hardware considerations shift enough that some of this won't map cleanly. For the typical facility operator or electrical contractor trying to make a Tuesday afternoon decision — this is the framework I'd use.
There's a certain satisfaction in getting it right without overbuying. After three years of rush calls, the clients I respect most are the ones who push back on replacement quotes and ask which scenario they're actually in. They usually end up spending less and getting a system that fits.

