As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
Prove the repair with a measurement that does not depend on the thing you are about to restore. A sealed circuit is proven by a timed pressure hold while it is empty, a repaired direct current string by continuity, polarity and insulation resistance while it is isolated, a battery by internal resistance before it goes back on charge, and a worked on space by a particle count before it is handed back. Each of those answers the question while a failure is still cheap to find, which is the whole argument against refilling the system and watching to see what happens.
A repair is proven by a measurement that does not need the charge
The failure mode in every one of these trades is the same. Work is completed, the system is restored, and the restored system is used as the test. It is a poor test for three reasons: it is slow, because a marginal fault may take days to express itself; it is destructive, because whatever the fault was is now doing harm again; and it is ambiguous, because by the time the symptom returns there are several candidate explanations and only one of them is the repair you just made.
The alternative is to pick a quantity that can be measured in the absent state. Pressure can be measured with no refrigerant present. Resistance and continuity can be measured with no current flowing. Insulation can be measured with the array isolated. Airborne particle concentration can be measured with nobody in the room. In each case the measurement is made against a threshold agreed before the work started, and the result is a pass or a fail rather than an impression.
Pressure decay tells you a rate, not a location
A sealed circuit that has been opened and closed again gets a hold before it gets a charge. The circuit is evacuated, brought to a test pressure within the limit published for its weakest component, isolated, and watched. What the gauge reports is a rate of loss, which answers the question of whether the repair sealed. It does not answer where a surviving leak is, and expecting it to is the commonest misreading of the method.
Two corrections make the figure trustworthy. The first is temperature: a fixed volume of gas changes pressure as it warms and cools, so a drop that matches the ambient swing is not a leak. Record the temperature at both ends of the hold. The second is the test equipment itself, since a weeping hose or a loose adapter produces a decay curve identical to the one a bad joint produces. Capping the manifold and holding the test rig alone for a short period separates the two.
The same rule on a direct current string
Replace a connector, splice a cable or remount a module and the string is in exactly the analogous state: repaired, not yet proven, and about to be reconnected to something that will immediately impose full voltage on it. The measurements that belong before reconnection are continuity through the repaired section, polarity at the terminations, open circuit voltage against the expected value for the number of modules, and insulation resistance between each conductor and earth.
Reaching those terminations without cutting them is a connector problem rather than a meter problem. A lead set such as the BOCEUC MC4 to banana plug test lead exists for that, and its listing states a 1500 V and 20 A rating, an IP68 weatherproof classification and a housing in PC and ABS chosen for insulation, impact resistance and ultraviolet exposure. What it does not state is any accuracy figure, which is correct, because a lead has no accuracy of its own and contributes only the small series resistance of its conductors. Assembling the rest of that kit is covered in testing kits for system commissioning and building a basic solar testing toolkit.
On the alternating current side of the same installation the equivalent is a proving instrument rather than a measuring one. The Testo 750-3 voltage tester states a three digit display of the actual voltage, an all around viewable light bar, polarity detection on direct current and the ability to perform a ground fault circuit interrupter test on an outlet. That last function is itself a before and after measurement: it proves a protective device operates rather than proving it is present. Earth fault testing at array scale needs different instruments, described in ground fault testing tools, and the very low resistances found in bonding conductors need the dedicated instruments covered in milliohm meters for low resistance testing.
Proving a battery before it goes back on charge
A repaired or rebuilt battery pack is the clearest case of all, because the quantity that predicts failure is not the quantity that is easy to read. Terminal voltage reports state of charge and almost nothing else. A cell whose internal resistance has risen will read a normal open circuit voltage and then sag badly under the first serious load, and putting a pack like that back on charge hides the evidence rather than revealing it.
Internal resistance is the diagnostic, and it has to be measured with four terminals rather than two. The YR1035+ internal resistance tester states a true four wire system that measures internal resistance and voltage at the same time, with probes and clips supplied and a display that shows both values together. Measuring both simultaneously is the point: a resistance figure without the voltage it was taken at is not comparable to the next cell, and a voltage without the resistance is the reading that let the bad cell through in the first place.
Writing down the pass criteria before you start
Every measurement above is only as good as the threshold it is compared against, and the threshold has to be chosen while nobody has an interest in the answer. Decide the hold duration and the acceptable pressure drop before the hold begins. Decide the insulation resistance figure before the megohmmeter comes out. Decide the internal resistance ceiling before the first cell is clipped.
The same applies to the space the work happened in. Cutting and brazing shed particles that settle and then recirculate, and in an area with a cleanliness requirement the handover evidence is a count rather than an inspection. An instrument such as the six channel particle counter states channels at 0.3, 0.5, 1.0, 3.0, 5.0 and 10 micrometres, a pump drawing 2.83 litres per minute, independent alarm thresholds for each of the six channels and stored records for later comparison. Independent thresholds matter here because the pass criterion is usually written for one or two sizes rather than for a total, and a counter that only alarms on a combined figure cannot express that criterion.
| System just repaired | Measurement before restoring it | What a pass proves | What it does not prove |
|---|---|---|---|
| Sealed refrigeration circuit | Timed pressure hold with temperature recorded | The circuit retains pressure at the test figure | That it will hold at operating pressure over a season |
| Repaired direct current string | Continuity, polarity, open circuit voltage, insulation | The conductors are intact and isolated from earth | That the modules produce their rated power |
| Rebuilt battery pack | Four wire internal resistance with voltage | Every cell can deliver current, not just hold charge | Capacity, which needs a discharge test |
| Worked on clean area | Particle count per channel after cleaning | The airborne load is back under the agreed limit | That surfaces were cleaned to the same standard |
Record the number, not the verdict. Someone will eventually need to know whether the hold lost two units of pressure or twenty, and a note that says the system passed is worth nothing to them. Technicians who maintain systems rather than install them need the history more than anyone, which is the argument made in testing tools for operations and maintenance technicians.
Frequently Asked Questions
Why not just recharge and watch for the fault to come back?
Because the system becomes the instrument, and it is a slow one with a poor readout. A slow leak or a marginal joint can take weeks to show, during which the fault is doing damage and the evidence of what caused it is being erased. A measurement taken on an empty or dead system gives an answer in minutes and gives it before anything is at risk.
How long should a pressure hold run for?
Long enough that the resolution of the gauge can separate a real decay from the noise, which usually means at least an hour on a small circuit and considerably longer on a large one. The useful discipline is to decide the duration and the acceptance figure before the hold starts, rather than deciding afterwards that whatever happened was close enough.
Does a terminal voltage prove a battery is healthy?
No. Terminal voltage reports state of charge under no load and almost nothing about the ability to deliver current. A cell with high internal resistance can sit at a perfectly ordinary open circuit voltage and collapse the moment it is asked for current. Internal resistance measured with a four wire method is the figure that separates the two cases.
Why does a four wire measurement matter for small resistances?
Because at milliohm level the resistance of the test leads and the contact resistance at the clips are comparable to the thing being measured. A four wire arrangement forces current through one pair and senses voltage through another, so the lead drop falls outside the measured loop. Two wire readings at that scale measure the leads as much as the sample.
Should a repaired string be tested before or after the inverter is reconnected?
Before. An isolated string can be checked for polarity, open circuit voltage and insulation resistance with the inverter out of the picture, and a fault found then is attributable to the repair. Once the inverter is connected, the same symptom has two possible homes and the diagnosis takes twice as long.
What is the point of measuring the air after mechanical work?
Cutting, grinding and brazing release particles that settle on surfaces and get drawn into equipment long after the work is finished. In a space with a cleanliness requirement, a particle count taken after cleaning is the evidence that the area was returned to the condition it was signed off at. It is the same logic as a pressure hold, applied to the room instead of the circuit.