Seeking Transient Voltage Surge Suppressor testing circuits

How does one measure the clamping voltage of a transient voltage surge suppressor at a specific current limit?

Application should clamp at 1100 volts @ 3Amps.

What would this measuring setup look like?

Jerry,

the TVS devices I have tested are specified for a 10/1000 us waveform. This is a waveform having a 10 us rise time and a 1000 us exponential decay to the 1/2 current point. Power-wise it is equivalent to a 1.4 ms square wave.

You need a constant current source of 3 amps with a compliance of over 2 kV. A pseudo constant current source will do the job. A capacitor charged to a high voltage and a series resistor will work. To connect the circuit to the TVS a wire can be touched to the TVS. The wire can be taped to an insulating stick a couple of feet long. Be careful not to get shocked with this lash-up. Wear googles as sometimes a TVS will explode when it fails.

A 2 uF capacitor charged to 2500 volts and a 500 ohm series reistor is needed. The charging resistor between the HV power supply and the cap can be 100 k ohms or greater.

A 100:1 Tektronix oscilloscope probe is connected across the TVS to monitor the voltage. To determine the current another 100:1 voltage probe can be connected across the capacitor and the current calculated.

Hello Dave,

It's great to hear from you!

I recognize your 10/1000 us waveform as typical for testing Silicon TVSS. I'm working with Selenium types with ms turnon capability. In fact, I'm looking at a test voltage pulse of about 2.25ms measured at 3A peakpulse mid points. I have reason to believe that a 1/2 60Hz cycle or 8.3 ms pulse should work.

Because the TVSS should clamp at 1100v, I think I need to charge your 2uF capacitor to something higher than 2500v.

Dave,

Let me try this again... 2500 is too large.

I think you meant ... charge cap to 1500V and discharge through a 500 series resistor. How else will the current be limited to 3A?

The TVS will just be starting to conduct at 1500 volts. To obtain a psuedo constant current source the HV needs to be much higher than the TVS clamp voltage. 2500 volts is a mininum.

A half cycle of 60 Hz might work as you suggest. The trick here is to switch ON the source and switch it off after one half cycle. An SCR on the AC primary will do that. You need over 5 kW peak and this will require a 240 VAC, 30 amp circuit to power a step-up transformer.

If you do want to use a capacitor as a one-shot circuit 5 uF is a good value. Six 33 uF, 450 VDC electrolytic caps in series (with equalizing resistors) will work. Or use HV ceramic or plastic dialectric caps.

Where to get the 2500 (or more) volts? Ultravolt has suitable HV bricks. A 20 watt, 5 kV brick (dialed to 2.5 kV) with a 750 k charging resistor will charge the cap in 15 seconds.

Thanks again Dave.

I assume your one-shot capacitor circuit is for firing the SCR?

Why equalizing resistors?

I suppose this is where a picture (schematic) is worth a thousands words.

The equalizing resistors help ensure that the 450 volt capacitors, which have unequal leakage current, do not exceed 450 volts.

The 5 uF cap is the energy storage for the pulser.

Another way to do this is to use an inductor. This can be exciting. You need a 2H inductor. Two paralleled Hammond Manufacturing 193S inductors will work. The risetime might be in the tens of microseconds range. A power supply is used to charge the inductors to 3 amps. The TVS is placed across the inductors. The trick here is to disconnect the power supply from the inductors. As it is disconnected the voltage will rise until the TVS conducts at over 1500 volts.

HV DC disconnect relays are available but are not an everyday item. Pulling a wire away fast enough as it's arcing is a problem. So, I think the capacitive discharge method is easier.

A few years ago I designed a linear current amplifier for testing TVS devices. It can test a 30 Joule TVS from 20 to 600 volts at up to 1700 amps or 100 kW, whichever is less.

If you're looking for a production test method something like this is what you need and not a piece of wire as I described. At the 1500 volt, 3 amp level this is not a difficult design.

The nice thing about the linear current amplifier is that you can test with any waveform you want. An arbitrary waveform generator drives it.

Your 1500 volt requirement (probably closer to 2 kV) is an awkward voltage. It's too high for 1500 volt MOSFETs leaving IXYS 2.5 kV and 4 kV MOSFETs as possible devices. These are not hardy MOSFETs.

Another topology is an autotransformer driven with a 500 volt supply and 600 volt MOSFETs. This requires a customer transformer but that's not such a hurdle.

And another topology is vacuum tube. Four 4CX250B vacuum tubes with 350 VDC on the screens will run 4 amps peak. These would be driven by a single 200 volt MOSFET in a cascode configuration. A 3 kV supply charges a 10 uF coupling cap to the TVS. This is a bullet proof design capable of surving a 24/7 production test environment with constant TVS failures.