One common opinion is to test for CO in the ambient air. While this is a valuable test, it only shows if CO is in the air you’re breathing. Another shared opinion is testing around the cabinets of fuel-fired equipment. Techs often assume this is where they will pick up flue gas spillage. Unfortunately, neither of these tests identifies a CO source. They only tell you CO is present.

I once encountered a situation where a company, responding to a CO alarm, red-tagged the hall closet furnace because they measured 5 ppm near the cabinet. Upon further investigation, we found the gas oven dumping more than 1,000 ppm into the living space. There was nothing wrong with the furnace. This initial test at the furnace cabinet could have proven fatal without a second opinion. It’s essential to test in the correct locations to avoid dangerous inaccuracies like these.

Testing for CO with a combustion analyzer above the drafthood of natural draft equipment is also incorrect. Air entering the drafthood dilutes how much CO the combustion analyzer measures. I’ve measured CO readings of 15 ppm above a drafthood and over 1,500 ppm under the drafthood, in the heat exchanger outlet.

You should measure CO in the flue gas before additional air dilutes it. You can test fan-assisted (80%+) and condensing (90%+) equipment in the flue since they don’t have drafthoods.

Wrong Test Equipment

Besides testing in the right location, you also need to use proper test equipment. Using the wrong test equipment usually leads to misdiagnosed systems.

Heat exchanger inspections with a flashlight and mirror are a perfect example. If a technician doesn’t discover cracks in the heat exchanger with these tools, they assume the furnace is safe. Unfortunately, that isn’t enough information to make the right diagnosis. A technician won’t know how much CO is in the flue gas or if it is spilling into the living space. We’ll dig deeper into heat exchanger cracks later.

Another commonly misapplied test is to place a store-bought CO alarm on top of a supply register. If the alarm remains silent, technicians assume the heat exchanger is intact, and no CO is “leaking” from it. Remember, most store-bought CO alarms don’t read low levels of CO. The first alarm level for these devices is typically 70 ppm for one to four hours of exposure.

Testing for CO the right way means using the correct test instruments. At a minimum, you’ll need a combustion analyzer and draft gauge to test correctly. Unless you measure the right way with these instruments, you’re misdiagnosing CO problems.

Smoking the Drafthood

A commonly taught test is smoking the drafthood of natural draft equipment. This test assumes if smoke draws (drafts) up the hood, the equipment is venting correctly. However, if smoke spills out from the hood, the equipment is spilling and dangerous.

This test assumes two conditions. One is that spillage only occurs at one location — the drafthood. Unfortunately, this assumption is wrong — flue gases can also spill from the burner compartment. The second assumption is that flue gas and dilution air flow evenly into the drafthood.

Air takes the path of least resistance. If too much dilution air enters the drafthood, there isn’t any room left for flue gases to draw from the equipment. When this occurs, they spill from the burner compartment. Ever seen discoloration or rust around a burner compartment? If so, you’ve seen the results of spillage.

You’ll need additional tests to discover this phenomenon. It requires a minimum of three flue gas CO readings during the run cycle to diagnose burner compartment spillage. If CO readings rise during the run cycle, carbon dioxide (CO₂) is backing up in the burner compartment and displacing oxygen. But if CO readings are stable, flue gases are venting properly with no spillage.

Eyeballing the Flame

A common myth found in textbooks teaches technicians to “eyeball” the flame. If the flame is blue, we assume it is safe and clean burning. If the flame has yellow tips or burns yellow, it is an incomplete burn that produces CO. While this sounds good, it’s not a measurement, it’s a guess.

Flame color doesn’t determine a safe burn — CO readings do. Flue gas CO levels under 100 ppm and stable during equipment operation indicate a safe burning flame. CO levels that are excessive and/or rising indicate an unsafe burning flame.

While you can’t trust flame color to determine safe operation, you can trust combustion analyzer readings. They tell you if a flame is the right color or not. For example, if flue gas CO readings are less than 100 ppm and stable during the run cycle, the flame color is right, even if it’s purple with neon green sparkles at the flame tips.

No Crack, No Problem

The last testing error is to assume that no heat exchanger cracks equal a safe operating furnace. The belief here is that heat exchanger cracks are primary sources of CO. Such an assumption usually stops technicians in their tracks from looking for additional CO sources.

It’s possible to have a furnace with a perfectly sound heat exchanger that still produces dangerous CO levels. This happens because of other problems with fuel input, cleanliness, and other mechanical issues that may be present. It’s also easy to overlook other CO sources with this line of thinking, such as water heaters, gas ovens, generators, unvented heaters, and attached garages.

Don’t let this common error keep you from looking at the entire picture and prevent you from correctly diagnosing a problem.

Fixing the Top Testing Errors

The top CO testing errors described in this article have been around our industry for decades. Yet, field testing has proven all of them are unreliable. Why do we still depend on them?

We’re one of the few industries crazy enough to start a fire in someone’s home and assume it’s safe. If something goes wrong, we toss around blame and opinions instead of facts and accurate test results.

Just because you install a system according to current codes and standards, it doesn’t always mean it’s safe. Start testing the right way and prove it for yourself. Don’t take my word for it. Your customers deserve your very best. Let’s work to fix these errors and replace them with live measurements and proper diagnosis.