How to Work a Job With No Manual at All

Why this matters

Sooner or later you are standing in front of equipment nobody publishes for. The manufacturer was absorbed twice, the model was built for four years two decades ago, the nameplate has been painted over, and the building's binder is a stack of warranty cards. The tech who treats that as a dead end quotes a replacement. The tech who knows the substitution ladder finishes the call, because almost every number in a manual can be recovered from somewhere else if you know which somewhere and how much to trust it.

Before you open a drive guard on a roof

Two hazards stack on this kind of work and they need separate handling.

Height comes first because it kills faster. On a roof, stay back from any unprotected edge and any skylight, and if the work puts you inside the fall-hazard zone, use the fall protection the site provides and tie off before you start, not after you decide you need it.

Then the drive. Measuring a sheave means the guard comes off, and the guard comes off after you open the disconnect, apply your own lock and tag, and confirm the shaft is stationary. That is 29 CFR 1910.147, and a rooftop unit with a remote start signal is exactly the case the standard was written for, because the thing that starts it is not in the room with you.

The call

A small commercial tenant complained the space had gone lukewarm. The rooftop unit ran, the compressor ran, nothing was locked out, no code was displayed. Nameplate was legible for voltage and amps and nothing else. The manufacturer name on the cabinet no longer exists as a company. There was no service literature to be found for the model, and the customer's binder held the roofing warranty and nothing mechanical.

The symptom the tech could actually measure was airflow: the supply registers were moving noticeably less air than the returns and the coil temperature split was wider than it should be. Low airflow. In a normal call that is filters, coil, or duct. All three checked out.

What we used instead of a manual, in the order we used it

This ladder works across trades. Each rung gives you information the rung above it does not, and you go down it only as far as you need to.

1. The machine's own surfaces. Clean the nameplate rather than assuming it is gone; a rag and a solvent recover painted-over stamping more often than not. Then look for the second data source almost every machine carries: a wiring diagram glued inside the control panel door, a sequence card, factory harness tags, a balance or startup sticker.
2. The components, not the assembly. This is the rung most techs skip and it is the richest. The assembly manufacturer may be gone, but the motor, the contactor, the transformer, the gas valve, the pump seal and the control board were each made by a component manufacturer, most of whom are still trading and still publish. A motor nameplate alone gives you horsepower, speed, frame, voltage, full-load amps, service factor and rated ambient.
3. The listing mark. A certification label carries a listee name and a file identifier that survive brand changes and acquisitions, and it names the standard the machine was built to. That standard tells you which protective devices the machine is required to have, which is often exactly the question you are trying to answer.
4. The building's own paperwork. Permit records, the original mechanical schedule on the as-builts, a commissioning or air balance report, the operations and maintenance binder from the general contractor. These are written about your specific installed unit, which makes them better than a generic manual for anything about setpoints and design conditions.
5. The sister unit. An identical or near-identical machine in the same building, still working, is a live reference standard. Every comparative measurement you take across the two is worth more than a published tolerance, because it holds the installation constant.
6. The parts supplier cross-reference. Counter systems cross obsolete part numbers to current equivalents, and the cross itself often reveals the original spec.
7. Reconstruct it. When none of the above holds the answer, you draw it yourself from the hardware.

We got what we needed on rungs two and four.

The number the balance report gave back

The building binder did have one useful document: a commissioning air balance report from the original install. It recorded design airflow for that unit and, critically, the fan speed measured at commissioning: 875 rpm.

That is the number no generic manual would have given us, because it is about this unit in this duct system.

Then rung two. The motor nameplate read 1,750 rpm. With the drive locked out and the guard off, we measured pitch diameters on both sheaves: 3.5 inches on the motor, 8.0 inches on the fan.

The correction, worked

Fan speed on a belt drive is motor speed times the ratio of the two pitch diameters.

Fan speed equals 1,750 times 3.5 divided by 8.0, which is 765.6 rpm.

Against the commissioned 875 rpm, that is 0.875, or 87.5 percent of design speed. The fan laws carry the rest, and they are why this matters more than it first looks:

• Airflow moves with speed. About 87.5 percent of design airflow.
• Static pressure moves with the square of speed. 0.875 squared is 0.766, so about 77 percent of design static pressure.
• Power moves with the cube of speed. 0.875 cubed is 0.670, so about 67 percent of design shaft power.

That is a fan doing noticeably less work than it was commissioned to do, and it explains a lukewarm space exactly. Someone had replaced the motor at some point and fitted whatever sheave came with the replacement, and nobody checked the ratio because there was no manual to check it against.

The correction comes out of the same equation, solved the other way. To get 875 rpm at the fan with the existing 8.0 inch fan sheave and the 1,750 rpm motor, the motor sheave needs to be 875 times 8.0 divided by 1,750, which is 4.0 inches.

We ordered the 4.0 inch sheave, and before it went on we redid the arithmetic in the other direction as a check: 1,750 times 4.0 divided by 8.0 is 875 rpm. It closes. That check takes fifteen seconds and it is the difference between a correct order and a second trip, because a sheave ratio error of half an inch is invisible until the belt is on and the tachometer says so.

The failure mode of not doing this is not subtle and it is not rare. Without the balance report, the tech in front of that unit has no design speed, sees a drive that looks factory, and concludes the fan is fine. He then spends the call and probably a second call on the coil, and eventually recommends replacing a unit whose only fault was a mismatched sheave.

The specs you cannot derive and must not guess

Being good at this ladder creates its own hazard, which is the confidence to derive something that is not derivable. Draw the line here.

• Anything in a fuel train. Manifold pressure, orifice size, and combustion air requirements are appliance-specific and not recoverable from the hardware. Guessing here produces incomplete combustion and carbon monoxide, and the customer is the one who breathes it. If you cannot get the appliance-specific values, the appliance does not run.
• Refrigerant charge by weight. A weighed-in charge is a factory number. You can charge by superheat or subcooling against the metering device type, which is a method rather than a spec, but you cannot invent a charge weight.
• Torque values on anything that seals or carries structure. A guessed torque on a head bolt, a flange, or a structural fastener fails at the worst moment.
• Protective device settings. Overload settings, relief valve settings, limit setpoints and pressure switch settings are safety parameters. Where a device is adjustable and the setting is unknown, the device gets replaced with one of a known rating rather than adjusted to whatever makes the machine run.
• Clearances to combustibles and venting configuration. These come from the listing, not from what fits.

The rule that separates the two lists: derive a value only when it follows from measurements and physical law, and never when it is a design choice the manufacturer made. A sheave ratio follows from geometry. An orifice size does not follow from anything you can see.

How to verify a derived spec before you act on it

1. Solve it backwards. Take your derived answer, put it back into the relationship, and confirm you land on the input you started from. The sheave check above is the pattern, and it catches transposed numbers.
2. Sanity-check against the component's own limits. A derived value that pushes a component past its nameplate rating is wrong, or the component is wrong. Restoring design fan speed raises shaft power, so confirm the motor's nameplate full-load amps still cover it before you fit the new sheave, and measure amps after.
3. Compare against the sister unit. If a near-identical machine sits beside it, your derived value should be close to what that machine measures. A large gap means one of the two has been modified and you need to know which before you act.
4. Write down the derivation, not just the answer. The next tech needs to know that 875 came from a commissioning report and 4.0 inches came from arithmetic, because a number with a source can be checked and a number without one can only be believed.

References

• 29 CFR 1910.147, control of hazardous energy, for locking and tagging a drive before a guard is removed
• 29 CFR 1910 Subpart D, walking-working surfaces and fall protection, for work on a roof near an unprotected edge or skylight
• Component manufacturer documentation for motors, controls and valves carried on the assembly's own parts
• Certification listing label for the standard of construction and the listee identifier
• See related: When the Manual Is Wrong or Doesn't Exist for This Unit; How to Compare a Suspect Unit Against a Known Good One