A ferrite ring with a few turns of wire on it, threaded onto one of the thick conductors going to the primary. It is how the driver finds out what the tank is doing, and on a machine that takes its feedback from current it is the only sense there is.
What it is and why
The thick conductor passing through the ring is the primary winding, and it has exactly one turn: itself. The wire wound round the ring is the secondary, with N turns.
I_secondary = I_primary / NSo the ring divides the tank's current by its turns count, and what comes out is small enough to work with. That current is then run through a resistor, called the burden, and what the driver actually reads is the voltage across it:
U = I_secondary × R_burdenHundreds of amps in the tank become a few volts on a comparator input. That is the whole device.
And the current through the hole is not what saturates it. A ring with seven hundred amps passing through it looks like it should be deep in saturation, and the intuition is right about inductors, where the flux really does follow the current. A transformer is the other case, and Steve Ward puts the difference in one paragraph:
Saturation of a transformer happens because there is too much magnetic flux in the core. Flux is proportional voltstime/number_of_turns. For a simple inductor, the current is proportional to voltstime/number_of_turns... which means flux is proportional to current. Now for the "confusing" part... when looking at a transformer, there is another coil coupled in there (the secondary) and it's current interacts with the primary coil current to give a resulting flux in the core. Any secondary current will generally cancel out the additional primary current, and so despite operating at a high level of current, the magnetic flux in the core is the same as if there were 0 secondary current and only the tiny "magnetizing current" that would be in the primary coil as its energized by the source voltage.
He was answering it about the current sharing transformers in a paralleled QCW inverter (HVF 294), and it holds here for the same reason. So the core is sized by volt seconds per turn and not by amps, which is why a ring you can hold between two fingers sits on a conductor carrying hundreds of them.
Read the two warnings below against that, because both are the same fact turned over: the way to saturate this ring is to take the secondary current away. Open the burden and there is nothing left to cancel the primary, so the whole of it becomes magnetizing current. That connection is ours rather than Ward's, who was answering a different question about a different transformer.
Why the driver needs it at all. A Tesla coil's working frequency is not fixed. The arc adds capacitance as it grows and pulls the frequency down, several tens of kilohertz across a single bang. A driver that switched at a frequency you set in advance would be right for the first few cycles and wrong for the rest. So it does not set a frequency: it watches the current and switches in step with it, which keeps it on the coil's frequency wherever that has got to.
What one actually looks like
A ferrite ring you can hold between two fingers, with a few tens of turns of light wire on it and two leads coming off. The thick conductor of the tank goes through the hole and touches nothing: the whole coupling is magnetic, which is why the part can sit on a conductor swinging at hundreds of volts and hand a few volts to logic.
There are two of them on a working coil, and they are usually the same part. One feeds the driver's feedback input, the other its overcurrent input. They see the same current and answer different questions, which is why the next section says not to economise by fitting one.
Where they sit is fixed by the voltage, not by convenience. On the return lead, after the tank capacitor and before the bridge, because that is the low potential end of the loop. The other end of the same loop is at the tank's full voltage, and the warning below is about what happens if the ring goes there instead.
And it has a direction. A transformer wound one way round gives the signal the driver expects and the other way round gives its inverse, which on a self-oscillating driver means the coil does not start at all. Turning the ring over is the whole fix, ten seconds, and some boards will do it for you with a switch instead. That it is a sign reversal rather than a phase error, and that no amount of lead inductance corrects it, is with the two rings.
You can wind your own, and on a bench you often have to, because a measurement transformer is not the same part as the two the driver runs on. That is with the scope.
The one decision this page is enough to make
Two rings, not one. The second thing the current tells you is whether it is too high, and it is tempting to take both readings off the same ring.
Do not. The feedback path has a phase lead network hanging off it, which loads the transformer's first stage; the core saturates, and the protection gets weaker exactly when it is needed.
And the burden is not a free number, which is easiest to see on somebody else's arithmetic. Gao Guangyan works one through on his UD2.7 page, designing for a bridge that must not see more than 700 A of primary current:
650 A 700 A, the design ceiling
through 625:1 1.04 A 1.12 A
across R17 = 5.1 Ω 5.30 V 5.71 V [derived, both columns]
against a 7 V ceiling 1.32x 1.23xRead it as his worked example rather than as a measurement: he writes "Lets say we have a 625:1 current transformer" and "Lets say we are building a DRSSTC", and the one figure that is a board rather than an illustration is R17, "a single 5.1R 2W by default". What survives whichever numbers you put in is the shape: twenty three per cent of headroom at his own ceiling, and the burden is what spends it. Note also that 625:1 is not 625 turns on a core. It is two 25:1 rings in cascade, and the ratios multiply.
How many turns, what burden, and the number that actually compares one installation with another are all in two rings, not one.
What will get you
And only one of the two burdens can be probed with an ordinary probe. The feedback burden has one side on ground and an ordinary probe is fine. The overcurrent burden has neither side on ground and needs a differential probe.
What goes wrong
- The coil will not start at all. Wound the wrong way round, which is a sign and not a setting. Turning the ring over is the test and it takes ten seconds; a board with a phase select switch does the same thing without the screwdriver.
- The insulation punctured. It was in the high potential position, between the primary and the tank capacitor.
- The protection is weak exactly when the current is highest. One ring doing both jobs.
- New faults appeared after probing. A probe ground on the overcurrent burden, where neither side is ground.
- The protection reads a current the coil is not drawing. The burden, or the ratio, or a threshold set against neither. It is not a knob position.
- The first cycles of every bang look wrong and the rest look right. Expected: there is no current for the ring to report until there is current, so every bang starts open loop.
- The signal clips flat before the coil is anywhere near its limit. The feedback is too strong for this machine, which is a number rather than a fault and is the one to compare.
Where next
- Two rings, not one, the same part with the numbers, and the feedback strength figure that matters more than the turns count.
- What a phase lead actually buys, what the driver does with the signal.
- Read the envelope, because the picture this ring gives you is the cheapest tuning tool on the bench.
- Scoping a coil without killing it, for probing these two safely and for winding a third to measure with.
The two rings on the department diagram sit on the return lead, after the tank capacitor and before the bridge, which is the low potential point named in the warning above.
- Current transformer does the cascade ratio and the burden voltage worked above, and puts a figure on what stray series inductance costs you.