Cutting glass vacuum tubes using hot wire

02 Mar 2021 - tsp
Last update 04 Mar 2021

This blog post is about something really simple. How to cut glass tubes with an electrically heated hot wire. In my case this has been used to cut open old vacuum tubes - maybe there will be a blog post about the project that this took place in at some point later in time.

Wire resistance

There are multiple possible choices for cutting wire:

• Nichrome (which is cheap and readily available)
• Tungsten (More expensive, higher temperatures)

The advantage of Nichrome is that it’s cheap, readily available (nearly every heater uses it) and has - compared to tungsten wire - a high resistance..

Wire Resistance $\frac{\Omega m}{m^2}$ Melting point (Celsius)
Tungsten (pure) $5.65 * 10^{-8}$ $3422$
Nichrome   $1400$
Konstantan $4.9 * 10^{-7}$ $1280$

The process

Basically all heating processes using these wires are Joule heating processes - i.e. the resistance of the wire converts the power loss (nearly perfectly) into heat. The heating power can be estimated as

[ P = U * I \\ P = R * I^2 = \frac{U^2}{R} ]

When doing calculations the main problem is the change of resistance with increasing temperature. The other factor is that heat dissipation into the environment by enclosing gas as well as any contact points would have to be taken into account. Since the only target is a fast rising temperature on top of the glass surface and glass being a bad thermal conductor no calculations have been performed when doing this on a laboratory or hobbyist scale. Before doing the cut itself heating the wire with up to $5A$ has been tried which works out pretty well leading to a nearly white glowing wire which allows one to estimate the temperature being somewhere above 1000 Celsius which should be sufficient for this application.

The setup

The setup consists - a little bit over engineered - of two posts that are insulated against the base aluminum plate. One post is fixed, the other is attached in a movable fashion - in this case with a “spare” Thorlabs linear translation stage - but the be fair pulling on the wire by hand or using a cheap screw from the hardware store would also have been sufficient, but you built such tooling with what’s at your hands. If one’s using higher voltages one has to be cautious about electrocution, when using voltages such as 5V this doesn’t matter anyways (but keep in mind that the wire gets really hot so don’t try to grab the wire directly by hand).

Wire diameter ($mm$) AWG cross section ($m^2$) Length ($m$) resistance ($\Omega$)
Tungsten $0.22 mm$ $31$ $3.80e-8 m^2$ $0.46 m$ $0.55 \Omega$
Constantan $0.125 mm$   $1.23e-8 m^2$ $0.46 m$ $40 \Omega$

The choice fell onto two strands of $0.46 m$ Constantan wire (with a total resistance of about $20 \Omega$).

The used power supply was a common lab power supply in current limiting mode that could provide up to $30V$ at up to $5A$. This type of power supply is readily available (Note: Amazon affiliate link, this pages author profits from qualified purchases)

Procedure

The procedure is rather simple:

• Use a tool such as a diamond file or a glass cutting tool to scratch a path into the glass
• Optionally make the glass wet using water
• Wrap the two strands of wire around the glass cylinder that should be cut - avoid the wires to be toughing themselves at the intersection though since even when they’re coated with some kind of insulator that insulator will burn away at pretty low temperatures
• Apply current. This is done by putting the current limiting value as low as possible, setting the voltage at maximum and then enabling the output. Then one can slowly increase the current limit. The supply will ramp up the voltage till it reaches the set current limit - in case of the mentioned two strands of $0.125mm$ Constantan wire the target has been up to $4.48A$.
• Apply slight tension if necessary. At some point the glass will crack due to internal stress induced by the scratches and the high heat gradient. If you didn’t scratch the path before the glass may splinter randomly or not crack at all.