Author: admin

I have posted a simulation for investigating energy dissipation when a high power transmitting vacuum tube arcs over from anode to grid (A-G arcing). I have simulated energy dissipation in the glitch resistor, in a simulated SCR based crowbar circuit and finally in the vacuum tube itself. In this case an Eimac YC-179 high power triode with a 6KV DC supply is investigated. The PSU storage capacitor is 20 uF.

Schematic:

Curves:

According to an Eimac engineering note, the maximum energy the grid circuit of a typical Eimac YC-179 or 8877 can take is 4 Joules. With only a fuse and a glitch resistor it looks like it will be hard to prevent the grid taking less than 4 Joules of energy. A typical Elschukom HV fuse opens after 10ms at 10x rated current according Elschukom datasheets. This is too long time to prevent the tube becoming damaged, if Eimacs data is not too conservative. I suspect the opening time will be less if loaded by 100-200x the rated current. I have contacted Elschukom engineering department to get more data.

In this simulation I have taken the time integral of the current x voltage product to investigate energy dissipation in Joules. (A not very well documented trick in LTspice is to CTRL+left click on the trace you would like to investigate to get the numerical time itegral of the trace data. If the trace data is a compound statement, the integral engine will give you correct units. For example P in Watts x t in seconds will give you Joules).

The findings in this simulation may be incorrect. Please contact me if you have comments or would like to share experiences from practical field tests.

Instructions:
===========

If you want to disable the crowbar to see dissipation with just a glitch resistor, just change R1 to 10 MEG.

If you want to change the PSU DC voltage, right click V1 and change to desired value.

In case you would like to increase the capacitance of the capacitor bank, right click C1 and enter your new value.

If you want to change timings, please change pulse sources (they control the voltage controlled switches to simulate the sequencing in a real HV PSU).

You can download the simulation files here:
hv_crowbar_switch_test_with_tube.asc

Here is the LTspice plot settings control file (Rename the file so it has a .plt extension and load it via Plot Settings / Open Plot Settings file) :
energy_in_glitch_res_and_tube.txt

These files are licensed to you free of charge for hobby use. If you change them you have to share the results back. If you would like to use these files for commercial design / R&D purposes, please contact the author for more information.

I have posted the equivalent schematic for my hipot tester.

The design is based loosely on some ideas from K8CU and AG6K.  However, I have simplified the circuit somewhat and used different components. The mechanical encapsulation is also different. I used LTspice to verify the tester before building it. The schematic is from the LTspice simulation (see below). I have also posted the simulation files so you can load them into your own LTspice simulator.

The unit consists of a 700VA 12KV AC RMS output (nominal) center tapped neon sign transformer that I purchased on Ebay. The trafo was designed for less than 230V AC RMS nominal primary voltage. Therefore the transformer gives out a bit more than 12KV AC RMS when supplied from the 230V AC RMS mains in Norway. After the rectifier I get approx 17-18 KV DC. The transformer is simulated by a sine voltage source in LTSpice. The transformer is insulated from ground in the tester for safety reasons. The schematic shows a gnd connection but that is only to get the Spice 3F4 solver to accept the netlist. It is recommended to insulate your tester from ground if you build one. There is a variac (not drawn on the schematic) in front of the transformer and the input is current limited by two 100W light bulbs in parallel (bulbs not drawn on the schematic). The voltage metering circuit is a uA meter with a shunt resistor and the current measurement instrument is a std. 100uA instrument protected by back to back diodes (not drawn on the schematic). The instruments are at high potential so adequate protection must be implemented to prevent arcs to the operator.  The device under test (DUT) is drawn. Note that a working tube has much higher resistance than 100 megs that is the load in the simulation. However this is placed as a load test for the simulation to check smoothing of the output voltage. The chassis is built in plastic and lexan and I have ensured long creepage distances by utilisation of vertical lexan insulators. There are no detected creepage currents at all even at full DC HV potential with my current design.

WARNING: do not attempt to build this high voltage device unless you are experienced in electronics and high voltage circuits. Touching the HV leads may instantly kill you. I do not accept any responsibility for your use of this device. The schematic is not a complete circuit description.

Download the LTspice simulation file here:  LTspice schematic of hipot tester

Video of the hipot tester in operation (test of an air variable capacitor):

I have posted some pictures of the hi pot test setup of my new Eimac YC-179 vacuum tube (MRI pull). This is for a future legal limit (LA amateur regulations) amplifier for digimode and contest CCS amateur radio use. It is better to run a tube cool than to stress a smaller tube for high dutycycle modes. Let’s hope the tube behaves OK. It measured less than 15uA at 16KV DC from A to G so I might have a good pull. (The reason for measuring at such a high voltage is to ensure that the swing that the PI circuit contributes when driven by the tube can be tolerated w/o breakdown).

HV setup to test to 16 KV (below)

Tube in the test rig (below)

Click on the pictures to see the gallery (below)

The HV meter to the left is on the hot side of the rectifier / smoothing cap and has a multiplier shunt that gives 200 x. That means that 80 on the meter on the left indicates 16KV (DC) output. The uA meter on the right is in line and measures leakage current. (uA)  The o/p is current limited on this output. The meter panels are insulated with plexiglass and air gaps to prevent arc overs to the operator. Not that this would help since the HV leads are exposed. However in a future version… (below)