What are the limits of transformer turn ratios?

Question

For applications where very high voltages are needed, like pulse generators, what stops one from simply using a transformer with an absurdly high turns ratio like 1:1000? If this were doable, core saturation would not be an issue when generating high voltage if one could simply use a very large number of secondary turns. This seems too simple to be true, what factors limit generating high voltages using very high turn ratios?

Answer 1

Nothing stops one from using a transformer to generate high voltages, it's done all the time.

A couple of common applications:

• Neon signs
• Strobe lights
• Ignition coils
• Cathode ray tubes in televisions

Some things that would limit high voltage transformers would be insulation breakdown, arcing, and physical construction constraints. You need to have a primary that will handle the output current times the turns ratio, you need to have good coupling between the primary and secondary, you need to have insulation that will withstand the voltages generated.

Answer 2

Specifically for pulse operation, it's winding wire length.

Roughly speaking, a wire is a transmission line (TL), whether it's wrapped in a shield like coax cable, or wound around itself like a big multilayer winding section. The latter is obviously quite different -- any given turn is surrounded by other turns, not a hard ground -- but we can still use a TL argument to analyze it.

Without explaining multilayer TLs further -- that's the jist of it. A TL has electrical length (the propagation delay along the wire), and characteristic impedance (the ratio of voltage to current for a wave traveling along the wire).

A very high ratio necessarily requires a very long winding, if done in a single transformer anyway. Winding length can be minimized somewhat by using a thicker magnetic core, but this also doesn't work once your primary is already a single turn.

This explains why CRT monitor flyback transformers required special designs to run at high scan rates: these achieved bandwidth of several 100 kHz, but also only with a specific design which is made possible by the required DC output.

Arbitrary frequencies are possible, mind -- but bandwidth is still, ultimately, limited. That is, you can have a transformer which passes frequencies of say 1kHz to 101kHz, or 1 to 1.1MHz, which have identical bandwidths, at different center frequencies. We would normally consider the former a wideband transformer, because the ratio of min/max is quite broad (two decades (factor of 100)!), while the latter is quite low (10% of Fc, more-or-less what we would call narrowband).

Indeed, we might find the latter is some variety of "Tesla coil", i.e. a high voltage, resonant transformer, often excited with pulsed energy.

Whether some design is adequate for a pulse generator, you'll have to provide more detail. If you mean more in the sense of, to power some other pulsed mechanism (spark gap, etc.?), then frequency doesn't matter, and even a mains-frequency NST may suffice.

Answer 3

1000:1 turns ratio also implies 1 million to 1 impedance ratio.

To drive a 1 Megohm load you need to drive the primary from something competent to drive 1 ohm.

Then you find the leakage inductance and huge secondary capacitance conspire to eat your signal. But allow for all of these and it can be done.