Sooty Solutions - Burnaby BC Consulting Company - Advising Business Managers on Security, Information Technology, Business Process Performance, & Best Practices
Sooty HomeContractors are BetterThe Sooty ApproachLong Range PlanningThe Sooty PartnersContact Us
   

Vacuum Gauges

Quick Links
 
  Background
  Microcontrollers
  Vacuum Pumps
  Vacuum Chambers
  Noble Gases
  Power Supplies
  Vacuum Gauges
  Plasma Tests
  Theory
  Simulators
 
 

Atmospheric Pressure (760 Torr) to 100 Torr

You need a vacuum gauge that works at one atmosphere an shows whether your vacuum pump is working. These are usually based on a bourdon tube with a flexible metal chamber or finger that changes dimension when there is a pressure change. The chamber is mechanically connected to a mechanical or electronic meter face to show the pressure. All you need is a crude instrument with limited accuracy because gauges that work in the interesting range for plasmas often do not work well at one atmosphere.

If you are using an automotive A/C repair manifold you will get two gauges a red gauge with a -30 to 120 psi range and a blue gauge that measures pressure above one atmosphere 0 to 500 psi (blue). These are pretty much useless because you will be working in the -29.9 to -30 psi range and 0 to -30 is only 20o movement on the red gauge. If you want to watch the pumpdown you will need is a gauge that goes 270o from one atmosphere to -30 psi.

I bought an automotive vacuum gauge ($35) for chamber testing that covers the pump down range. For most experiments the red gauge is good enough to tell me whether the pump is working and the valves are set right and I do not bother the drill a port into my vacuum chamber for a gauge to just measure the pump down.

Plasma Range 10 Torr to 10 milliTorr

Most of the gauges in this range are based on the ability of the gas molecules to remove energy from an electrically heated wire. In a hard vacuum the wire can only lose heat by radiation and with a small surface area the wire has to be very hot to lose much heat at all. The wire is chosen with a resistance that is has a high temperature coefficient.

If a constant voltage is placed across the wire it will remain at a low temperature and low resistance at one atmosphere but its temperature and resistance will rise quickly as the pressure approaches a hard vacuum. The effect is even more dramatic if a constant current is used but the electronics are more complicated.

The best known gauge using this effect was invented in 1906 by Marcello Pirani and most heated wire gauges are called "Pirani gauges". The basic technology has changed little in a hundred years and most of the advances have been in making a gauge that can be mass manufactured (less variation from unit to unit) and holds its calibration (is not affected by age or physical abuse).

You can buy a quality Pirani gauge for about $300 but you really need the $2000 meter to make sure you get the accuracy and more important so you do not stress the fine heated wire and burn it out. For the poor starving inventor this is a little pricey so like most of this site I am going to show you how to build your own for a couple of bucks.

Light bulbs have tungsten wires that have a resistance with a high temperature coefficient. If you punch a hole in an ordinary light bulb you have a Priani gauge. The problem becomes one of choosing a light bulb with the right properties, building the electronics to apply a constant current, and calibrating the gauge.

Finding the Right Light Bulb for Your Pirani Gauge

An ordinary 100 watt bulb could be used as a Pirani gauge but it is too big, too fragile and would consume too much power. You want a small, low wattage bulb with a glass shell that will hold up after you drill a hole in it. Flashlight bulbs are small and durable but it turns out their resistance is too low. What you need is something the size of a flashlight bulb but uses much less current. This means you are looking at decorative lights, Christmas tree lights and industrial panel lights.

Your choice is limited by the fact that most small operational amplifier chips have a 50 ma maximum current and 20 volts maximum voltage. A light bulb has its highest resistance in a vacuum and lowest resistance when cold. Most light bulbs are sealed with 10 milliTorr argon gas so a measure of the cold resistance and the voltage/current rating of the bulb is a good place to start.

A 12 volt 0.78 watt (65 ma) bulb like the IKEA E512V has a hot vacuum resistance of 12/.065 = 180 ohms. The measured cold resistance is 20 ohms which makes it an good candidate with a 9:1 resistance ratio. A 48 volt 38 ma telephone office panel bulb has a 48/.038 = 1200 ohm hot resistance and a measured cold resistance of 96 ohms giving it a 13:1 ratio. In a Pirani circuit you get about half the hot/cold ratios (about 5:1) so both of these are good choices.

The operating range of our opamps is a critical factor. An 8D 12 volt lantern bulb that needs 500 ma for a hot resistance of 24 ohms and has a measured cold resistance of 1.4 ohms which is a very nice 17:1 ratio but it would a lot of current to drive the circuit. A 7 watt (7/120 = 58 ma) night light has a hot resistance of 120/.058 = 2070 ohms and a measured cold resistance of 192 ohms giving it a 11:1 ratio but it would need a lot of voltage to drive the circuit. In fact over a 0 to 12 volt range the 7 watt nightlight only has 1.4:1 ratio because you cannot get the filament hot enough to get the Pirani effect.

A simple rule of thumb is you will be operating your gauge at roughly one third the rated voltage and current for the bulb. For the IKEA bulb this would be 4 volts and 20 ma and the telco bulb would be 16 volts and 12 ma. Both are in the opamp range. Compare this to the lantern bulb (4 volts and 170 ma) and night light (40 volts and 19 ma) both which fall outside the range of our opamps. Christmas tree bulbs come 50 on a 120 volt string so operating them at one third of their 120/50 = 2.4 volt rating would mean 0.8 volts which is probably too low.

Turning your Light Bulb into a Pirani Gauge Sensor

To turn a light bulb into a Pirani gauge you have to drill a hole in the glass shell. Grinding or drilling the bulb is often causes the glass to shatter. If you just heat the glass it will form a dimple that will collapse into the bulb.

The simplest way to heat a wire white hot and push it into the glass to melt a hole. I use a stainless steel turkey lacer ($1 for 6) and heat it with a propane torch. After destroying a few bulbs I was able to make a fairly neat 1 mm hole in the glass shell. The result is an extremely durable sensor that can be soldered or screwed into a mount.

You can just make out the holes in the shells of the IKEA bulb (on the top) and the telco bulb (on the bottom) in the following picture:

 

Building a Pirani Sensor Circuit

You can use a constant current or a constant voltage but the most sensitive circuit uses a Wheatstone Bridge to maintain a constant resistance and the measures the current on the bridge to provide a sensor value. The circuit I designed has the advantage that one end of the Pirani bulb is grounded making it easier to install in the vacuum chamber.



Pirani Gauge Calibration Curve

The sensor and circuit give the most sensitive measure of pressure but he Pirani curve is highly non-linear. In fact even on a log plot the curve is still highly non-linear. To appreciate the technology it is a good time to look at some commercial documentation and actual commercial product graphs. The first is a standard plot of voltage against log of pressure:

The non-linear nature of the sensor is even more apparent in when pressure is plotted on a linear scale:


Expanding our experimental region from 1 to 20 torr we still have a lot of non-linearity but there is enough voltage change to create a reasonable calibration chart:

An actual measurement of a IKEA sensor with my bridge circuit using the automotive pressure gauge as a standard from one atmosphere to 50 torr:

Using the relaxation oscillator circuit described at the bottom of the gases page the curve below 50 torr is:

With these curves I was able to spline fit and create a table I use for measuring the gas pressure in all subsequent experiments.