Frequently Asked Questions
Table of Contents
- What kind of circuits need matched tubes?
- How does push-pull work?
- What happens when output tubes are mis-matched?
- How do I measure tube plate current in an amp?
- What is the difference between a tube tester and tube matcher?
- What kinds of tube testers are there?
- How does an emission tester work?
- How does a mutual conductance tube tester work?
- What other tests are useful?
- How does a laboratory tester work?
- How does a curve tracer work?
- How does the MaxiMatcher work?
- How does the Apex system work?
What kind of circuits need matched tubes?
The variability of vacuum tube characteristics of a given type can be from ± 5% to ± 20%. This contrasts with up to ± 200% for bipolar transistors and MOSFETs. The need for using tubes with tighter characteristics than average depends on the circuit used and the performance expected from the circuit.
Single-triode or single-pentode amplifier stages generally do not require matching. The only possible exception is if you want the gain on two channels of a stereo amplifier or preamp to have exactly the same gain. This is exceptional, since using unmatched tubes does not give that much variation.
Paralleled Tubes
When two or more tubes are paralleled for either increasing the transconductance (gain) or power output, it is desirable to have the tube have matched characteristics.
One reason for matching is to even out the wear on the tubes, particularly with power tubes. If one tube has higher gain or higher idle current than the other(s), then it will run hotter and wear out sooner.
Another reason for matching paralleled tubes is to reduce distortion. The composite transfer characteristic of paralleled tubes is the sum of each one. If the linear portion of paralleled tubes is identical, the composite characteristic remains linear. If the characteristics are different, though, a non-linear characteristic results. The drawing below shows how this happens:
The green and blue curves are for the different tubes. The red curve is the composite - note the “kink” in the curve. The amount of difference is exaggerated here, but the concept still holds. Matching for linearity is generally most important in hi-fi amplifiers and in SSB linear amplifiers for radio transmitters.
Push-Pull Driver Circuits
Driver circuits here mean low-level input or driver amplifiers, not power amplifiers.
When a differential amplifier or any other kind of amplifier using a push-pull arrangement is used matched tubes are helpful in reducing distortion and drift.
Below is an amplifier schematic from the 1955 Acrosound catalog:
The first two driver stages are balanced and the third is a form of differential amplifier. The triode sections in each of the tubes should be matched (i.e. The two triodes within the same envelope). This is especially important here, since the driver stages are direct-coupled. Any imbalance in the first stage gets amplified, which could severely unbalance the current in the last driver stage.
A more demanding requirement for matched low-level tubes is in vacuum-tube oscilloscopes. Here is the vertical amplifier of the Tektronix 360 oscilloscope. (This rather obscure scope was picked because the circuit is simple - most Tektronix scopes have complex front-end circuits).
This scope has a vertical sensitivity of 50mV/division and a bandwidth of 100KHz. Since this is direct-coupled all the way, balance is important. Imbalance that drifts will cause the trace to move around on the screen. In this circuit, tubes V30 and V32 should be matched and tubes V50 and V52 should be matched.
Push-Pull Output Circuits
Push-pull audio output stages are the prime reason that matched output exist. In general, all push-pull output stages should use matched tubes. However, the degree of matching needed depends on the output circuit and any means of adjusting bias to the tubes.
Cathode Bias Push-Pull
Cathode bias has the beneficial effect of auto-correcting, to an extent, the bias point of the tube. If the tube draws more current, the voltage drop across the cathode resistor increases, which increases the negative grid bias that the tube sees, thus reducing the plate current. This compensation is not perfect, but does help the stabilize the tube’s operating point.
Why isn’t cathode bias universally used for output stages? Two reasons. The first is that significant power can be dissipated in the cathode resistor, especially in high power amplifiers. This is wasted power. The second and more important reason is that cathode bias is only really usable when the cathode current doesn’t change much with different signal levels. This means class A operation or a mild class AB1 mode. Big changes in current will move the operating point around, with the slow time constant of the cathode resistor/cathode bypass combination. For example, if the signal increased, the cathode bias voltage would increase. However, if the signal was suddenly reduced, the bias is still high for a fraction of a second or so, and during this time the output stage gets cutoff. This is an ugly type of distortion. For these reasons, high power amps generally use fixed bias.
The most resilient type of cathode bias for a push-pull amp is when each tube has its own cathode resistor and bypass capacitor. Here is an example of this from the schematic of a 10-Watt Mullard amplifier:
In this circuit arrangement each tube self-biases, and matching is less critical. By the way, if the two cathode resistors are accurate, the voltage drop across them can be used to measure cathode current.
In amplifiers with a common cathode resistor, the self-bias compensation works on the average of the two tubes current, so differences between the tube’s operating point is not compensated. In this case well-matched tubes should be used, unless there is provision for balancing the bias between tubes. Here is an example of the well-know Williamson amplifier implemented as the Heathkit W-3M:
The 100 Ohm potentiometer allows the bias to be balanced. Current is measured by inserting a phone plug into the jacks in each cathode connected to an ammeter. If only one meter is available, this balance adjustment can be tedious, since the meter will have to be repeatedly switched between jacks. A better scheme was used in Heathkit’s W-5M Williamson amplifier:
30 Ohm resistors are inserted in series with each cathode, and a voltmeter connected from cathode to cathode is adjusted for zero volts for a perfect balance. (Of course the accuracy of the balance depends on the accuracy of the resistors). The 100uF bypass capacitors eliminate any cathode degeneration caused by the 30 Ohm resistors.
Fixed Bias with bias balance adjustment
In a fixed bias circuit, the idle current is dependent only on the tube’s characteristics and the negative grid bias voltage. If there is a difference between the pair of tubes in a push-pull circuit, some, but not all of the difference can be corrected if the bias for each tube is adjustable. Thus strict tube matching is not required, although is helpful in keeping distortion down.
There are two main ways of adjusting the bias in a push-pull amp: separate bias adjustments for each tube or a single “balance” adjustment that increases the bias voltage on one tube while decreasing it on the other as the control is rotated. Both methods work well, but involve different adjustment techniques.
For amps with a bias balance control, the control is rotated until the plate or cathode current is the same on both tubes. This can be measured at the plate or across separate cathode sense resistors, as described {here}. Note that the balance control only changes the tube currents relative to each other - it doesn’t affect the absolute current level of the tubes. For this, a separate “bias” level control is needed.
Bias Balance adjustment example:
Before adjustment:
Tube A current = 54 mA, Tube B current = 48 mA.
Adjusting the bias balance control for equal current gives:
Tube A current = 51 mA, Tube B current = 51 mA.
However the correct current is 60 mA. The bias adjust control is adjusted to give:
Tube A current = 60 mA, Tube B current = 60 mA.
Here is an example of an amp with a bias balance control, the Pilot SA-260:
Not shown on the schematic is a bias adjust control, which changes the level of the “T −45V” line.
For amplifiers with a separate bias adjustment per tube, the adjustment procedure is as follows:
Separate Bias Adjustment Example:
Before adjustment:
Tube A current = 54 mA, Tube B current = 48 mA.
Adjust Tube A current to 60 mA:
Tube A current = 60 mA, Tube B current = 46 mA.
Adjust Tube B current to 60 mA:
Tube A current = 58 mA, Tube B current = 60 mA.
Note that the two bias adjustments interact. This requires repeating the adjustments.
Adjust Tube A current to 60 mA:
Tube A current = 60 mA, Tube B current = 59 mA.
Adjust Tube B current to 60 mA:
Tube A current = 60 mA, Tube B current = 60 mA.
The interacting effects of the bias adjustments, as shown above, is typical in most amps. This is caused by the high voltage B+ shifting as the tube currents change.
Fixed Bias with no balance adjustment
Many amplifiers have no bias balance adjustment. This requires output tubes with good matching.
If the amplifier has a single bias adjust control, then only tube-to-tube matching is needed. The absolute current in the tube is not critical.
Here is an example of an amp with a single bias adjust control, a Fender Showman amp:
The bias adjustment is the 10K pot near the bottom of the schematic. Since both push-pull and parallel tubes are used here, the output tubes must be replaced by a matched quartet.
Some amplifiers have no bias adjustments at all. In this case matched tubes are required, but also tubes selected for the correct current would be needed.
Here is an amp with no adjustments, the Fender 6G6 Bassman:
Another example is the Fisher 500C stereo receiver:
In amplifiers with no bias adjustments it is crucial that the bias power supply be working correctly. In many older amps the bias rectifier is a selenium rectifier. These either have metal fins or appear as a small block with terminals coming out of it. Selenium rectifiers degrade with age, reducing the bias voltage. This causes the output tubes to run hot, significantly shortening their lives. It is recommended that all selenium rectifiers be replaced with silicon rectifiers. However, silicon rectifiers have lower voltage drop than seleniums, so some extra resistance may need to be added in series with the silicon rectifier to achieve the proper bias voltage. This usually requires a “cut and try” approach to get the right value.
How does push-pull work?
Most vacuum tube power amplifiers, both high-fidelity and music amps, use a pair or multiple pairs of output tubes in a “push-pull” arrangement. To work the best, the tubes in a push-pull output stage should have identical characteristics. The most important characteristic is the relation of plate current to plate voltage and grid voltage. A secondary characteristic is the effect of screen voltage on plate current. These characteristics are given graphically in a set of curves:
There are separate plate current vs voltage curves for each grid voltage. To find a current for an arbitrary grid voltage, interpolate between the two nearest grid curves. This set of curves is for a screen voltage of 250 Volts. A higher screen voltage will move all the curves up (higher plate current) and a lower screen voltage will move the curves down (lower plate current).
The hyperbola going from the upper left to the lower right is the maximum average power dissipation allowed. The tube current can go above this curve momentarily, but the long-term average must be below the curve.
Now we can look at how the actual operation of the amplifier uses the plate characteristic curves. Below are the curves for a 6L6GC:
The red dot is the idle (no signal) current. This is the operating point set by the grid bias voltage. The red line is the “Load Line” - the path taken as the amplifier is driven to full output. At the upper left end, the curve stops when it hits the Ec1 = 0 Volts (zero grid voltage). If driven beyond this, the amplifier will clip since grid current will load down the driver stage. [This is for a class AB1 amplifier, the most common type. A class AB2 amplifier has a special beefy driver stage that allows the grid to be driven positive. This is not common in most hi-fi and musical amps.]
At the lower right end, the plate current goes to zero. If this was just a single-tube amplifier (“single-ended amp”), then the amp would clip here. But, in a push-pull amplifier, the other tube picks up the current, and the output is undistorted. The load line is a straight line only for purely resistive loads. With a real-world speaker as the load, with its inductances and capacitances that vary with frequency, the line opens up into an oval.
Each side of a push-pull amplifier is driven 180 degrees out of phase. The phases are lined-up again in the output transformer. The idea is that during one half of a cycle, one tube is “pushing” and the other “pulling”, then in the next half cycle the operation is reversed. Push-pull amplifiers have several advantages over single-ended amplifiers:
- Class B operation is permitted without distortion, increasing the maximum power output.
- All even order (2nd, 4th, etc.) harmonics generated within the output tubes are cancelled.
- The DC magnetizing current is cancelled in the output transformer, permitting a smaller and broader-bandwidth transformer.
- The output stage is much more immune to hum and voltage changes on the B+ supply line.
The operation of a push-pull output stage is predicated on both tubes being identical. The next question will explore what happens when they are not.
What happens when output tubes are mis-matched?
There are several ways two tubes can be mis-matched. One way is if the idle current at a given grid bias voltage is different. This means that the current through the output transformer is unbalanced and one tube is running hotter than the other, even if there is no signal. If each tube has its own grid bias adjustment or there is a “bias balance” adjustment, then this type of mis-match can be compensated.
Another form of mis-match is if the tubes have different gain or transconductance. Even if the idle current is the same, the current drawn by the tubes during large signals will be different, leading to unbalance in the transformer. This is why amps with separate bias controls or bias balance controls should still use matched tubes.
There are three main effects of current unbalance in a push-pull amp:
- Increased even-order (2nd, 4th, etc.) harmonic distortion. Even order distortion is, in general euphonic, i.e. sounds good, so this is not usually a major problem.
- Transformer core saturation. Unbalanced DC current in a transformer with no air gap will cause the iron to saturate, thus lowering the transformer’s inductance. This can cause gross distortion at low frequencies and high power levels. Most output transformers using the E and I laminations have an inherent small air gap, so can tolerate a small amount of DC current unbalance. Toroidal output transformers (uncommon) need absolute current balance.
- Uneven tube wear-out. If one tube runs hotter than the other, it will wear out faster. Since output tubes should be replaced as pairs, this is not too bad of a problem, but this can shorten the overall life of the pair.
The need for matching depends a lot on the design of the power amplifier.
How do I measure tube plate current in an amp?
There are two methods of measuring plate current in an amplifier: inserting a milliammeter directly into the plate circuit of each tube, or put a small resistance in the plate circuit for each tube and measure the voltage across this sense resistor. Using Ohm’s Law, the current = voltage / resistance. However, there can be complications doing either of these methods that depend on the design of the amplifier and how many modifications you want to do to it, if any.
A problem of putting a meter or sense resistor in the plate circuit is that this is at a dangerously high voltage. The utmost care must be taken to avoid getting an electrical shock. A safer alternative is to measure the current in the cathode circuit, which is much closer to ground potential. The complication here is that both the screen current and plate current flows through the cathode. The screen current is a small percentage of the plate current (about 5 to 15%, depending on the tube), but varies inversely with the plate voltage, so can throw off readings at low plate voltage. In many cases, this is an acceptable trade-off. In an “ultra-linear” circuit, where the screen grids are connected to taps on the output transformer, measuring current in the cathode is closer to optimum.
Using cathode sense resistors
Sometimes the cathode sense resistors are already built into the amplifier. Here is an example from the Pilot SA-260 hi-fi amplifier:
The sense resistors are R39 and R43, 20 ohms each. The voltage is measured from each test point to ground (the chassis here). Since the cathode voltage here is shown as 1 Volt, from Ohm’s Law, the idle current through each tube is I = E / R or I = 1.0V / 20 Ohm = 0.02A = 20mA.
Note 1: The sense resistors shown here are spec’ed at 5% tolerance. For a vintage amplifier like this, this may have been acceptable, but could cause a misreading of current between the tubes of up to 10%. Tighter tolerance resistors, say 1%, are easily available today, and are recommended.
Note 2: Always verify the value of sense resistors in vintage equipment, especially if they are of the carbon composition type. These can drift up in value with age, sometimes by as much as a factor of two! This will make tube current readings meaningless. Check those resistors!
If you want to measure the cathode current in output tubes in an amp without sense resistors, they can be added in at the points shown in the schematic above. The 20 Ohm values shown above are near the high end of the range that can be used. Much higher and the cathode degeneration caused will start to affect the performance of the amplifier. More commonly resistors from 2 to 10 Ohms are used. The resistors can be any value (but should be the same for both sides), but by picking an even number, the math is calculating the current is easier.
Normally the power dissipated in the sense resistor is small - 50 mW at idle in the example above. However, the current will increase by a factor of two or more with a strong signal. Even worse is if the tube shorts out or goes into “melt-down”. Then many watts could be dissipated. In this case the sense resistor would become a fuse! Metal-oxide or wire-wound resistors of 1 or 2 Watts are recommended. Stay away from carbon composition resistors.
If the sense resistors in the circuit above are well matched, then at perfect current balance, the voltage between the two test points will be zero. This is a handy way of adjusting a balance control, if available.
Using output transformer resistance
The windings on any transformer have a finite DC resistance due to the resistance of the copper wire in the transformer. Using Ohm’s Law, this resistance can be used as a built-in sense resistor to make plate current measurements. The downsides with method include: odd-ball resistance values, possibly different resistances for the two halves of the push-pull transformer, and the high voltages on the voltmeter leads. However, since it is easy to do and using a calculator (available on any smart phone) make the Ohm’s Law calculations easy, this method is frequently used.
Here is an example with a T-808-127 output transformer removed from an old Fisher receiver. One side of the primary is 100.9 Ohms and the other is 88.7 Ohms:
The plate current in the upper tube is voltage / 100.9 and the lower tube’s current is voltage / 88.7. As long as you keep the different resistance straight, the current measurement can be easily done. On the other hand, the different resistances keep you from using the technique of measuring plate-to-plate and adjusting for zero voltage, hence balanced current. The solution is to “build-out” the resistance on the low resistance side:
The 12.2 Ohm resistor now makes the resistance to each plate from the center-tap be the same - 100.9 Ohms. This added resistor doesn’t affect the circuit much, and in fact makes the circuit more balanced. As with a cathode sense resistor, a 1 to 2 watt wire-wound or metal-oxide resistor is recommended. If you cannot find the exact value you need, you can parallel or series resistors to get the right value.
Why is the transformer resistance different on each side of the center-tap?
In transformers, the quantity that determines the basic operation of the transformer is the number of turns around the core. The length of the wire in the transformer isn’t that important. In a push-pull output transformer, the two halves of the primary must have the same number of turns.
In audio transformers the windings are broken up or "interleaved" to reduce the leakage inductance, which improves the high frequency response. A simple version of this is shown in this cross-section of a push-pull output transformer below:
The secondary is sandwiched between the two halves of the primary. The number of turns in each primary section is the same, but clearly, Primary 2 has a longer path than Primary 1, since it is further from the center of the core. Thus it has more wire than Primary 1, and hence higher resistance.
Hi-Fi transformers often have more complicated interleaving in order to achieve better frequency response, but most instrument amps use the simple interleaving shown above.
Using a test socket/plug
If there are no convenient means of monitoring the plate current, or you don’t want to make any modifications to the amplifier, a test socket can be used to monitor current. This is simply a plug and socket that carries each wire from pin-to-pin with a sense resistor inserted in either the plate or cathode lead. Here is a schematic for a test socket for tubes with a 7AC pin-out (6V6, 6L6, EL34, KT88, etc,) with a sense resistor in the cathode lead:
The 10 ohm resistor is low enough to not affect the circuit much - the voltage drop is less than one volt, but leaves enough voltage to be accurately measured by a modern DVM. One volt = 100mA.
The optional grid stopper resistor is recommended to help suppress high-frequency oscillations. These can happen on high gain tubes such as the EL34 and 8417, and can cause bizarre effects that are hard to track down.
The sense resistor can also be inserted in the plate lead (pin 3), but is at a dangerously high voltage, so should be used only with the utmost care.
What is the difference between a tube tester and tube matcher?
Vacuum tubes suffer from various degradations over time: loss of cathode emission, gas, electrical leakage, mechanical failure, etc. A tube tester is designed to verify that a tube basically works and gives an estimate of its strength. The typical read-out of a traditional tube tester is a meter that either gives a relative reading (e.g. 0 −150) or an actual gain reading that can be compared against a minimum value. Operation at the actual operating voltages and currents is not attempted.
When matching tubes, more than one parameter must be measured, and operation at the actual operating condition must be attempted. This requires a more powerful power supply, the ability to measure at multiple operating points and accurate voltage settings and current readings.
Most tube testers are vintage units, since many thousands of testers were made during the heyday of vacuum tubes. Ones made before World War II are of limited value, since they cannot test most modern tubes. Tube testers made from the late 1940s until the end of general tube usage in the 1970s are the most useful. Most common are the ones made for TV and radio servicemen or the military testers, which were often just ruggedized versions of commercial units. There were some more exotic testers made for laboratory or specialized testing. These would fall into the category of tube matchers, due to their flexibility and accuracy.
What kinds of tube testers are there?
There are two main varieties of tube testers: emission testers and mutual conductance testers. Both types were made in great quantities during the tube era. Both can determine the state of a tube, but the mutual conductance type are preferred for general testing.
How does an emission tester work?
An emission tube tester basically just measures how much current can be pulled from the cathode and compares this against the average current value of a new tube. The tester inserts different voltages and series resistances appropriate for the tube being tested - these would be much different for a 12AX7 than a KT88. The readout is a meter that is either marked in a red/green scale of BAD-?-GOOD, or on a relative scale of 100%.
Serviceman-type American emission testers are not recommended for general use for two reasons: they don’t test the tube in a realistic operating condition, and they can be hard on tubes, especially small, delicate tubes. They will help find weak or dead tubes, but not much else.
The classic "drug-store" tube tester was almost always an emission tester. It had many sockets, each wired for a particular tube type, so the users wouldn’t have to fiddle with rotary switches.
In Europe, most tube testers are the emission type. However, they usually have more control over the operating point than the American ones and some even verge into the sophistication of laboratory testers.
Notable European tube testers include the British AVO testers, the French Metrix testers, many German types and the Soviet L3-3 tester. The latter tester, as well as some of the German testers use a plug-board to configure the socket connections and have cards for each tube type with holes for the appropriate plugs. The European testers have more socket types than the American testers, since many different sockets were used that never were seen in America.
How does a mutual conductance tube tester work?
A vacuum tube amplifies by changing its plate current in response to its negative grid voltage. A less negative grid gives more plate current. This relationship is called, traditionally, mutual conductance, or more modernly, transconductance. A mutual conductance tube tester measures this amplification and base the merit of the tube on this reading. This is a more comprehensive test than the emission tester gives, since it is closer to how a tube is actually used. As the cathode emission of a tube falls with age, the transconductance also falls, so this is a useful test of tube performance.
Lab-grade tube testers measure transconductance by biasing the tube to a class-A operating point, then applying a small AC voltage to the grid and measuring the AC voltage on the plate. This is a direct measurement of the tube’s gain, but requires an AC signal source, an AC meter, and ability to adjust the operating point for vastly different tubes. This type of test wasn’t feasible for a moderately-priced portable tube tester.
This problem was solved by a clever invention by J.R. Barnhart in U.S. Patent number 1,999,858, issued in 1935. By applying a full-wave rectified but unfiltered waveform to the tube under test, applying a small AC voltage at the grid (at the power line frequency), the transconductance could be measured by putting a DC ammeter at what would normally be the center-tap of the full-wave rectifier. Here it is from the original patent:
The triode is the tube under test. The coils are windings on a transformer - “P” is the primary and “S2”, etc. are the secondaries. “V” is a full-wave rectifier, typically a type 83 mercury-vapor tube, and “M” is the DC meter.
This mutual conductance circuit was licensed to the Hickok Electrical Instrument Co. in Cleveland, Ohio, who became famous for their implementation of this circuit. They extended their lock on the tube tester market with U.S. patent number 2,440,607, issued to Robert D. Hickok in 1948. They licensed this circuit to a few companies that weren’t direct competitors, such as Sylvania and Philco for their service equipment but likely charged a lot for the license. They also made military versions: the World War II era I-177, and the post-war TV-3/U, TV-7/U and TV-10/U testers. Popular testers intended for the radio/TV service market include the 539A, 600A, 800A, and 6000. After the Hickok patents expired, other companies began using the Hickok circuit, most notably B&K in their 707 and 747 testers.
Hickok-type tube testers go for serious money if they are in good working condition - $300 and up. Why aren’t there cheap Chinese knock-offs flooding the market? Well, a look at the schematic shows at least nine rotary switches with incredibly complex switching functions. Making these custom rotary switches is a lost art and would be expensive today, if revived. Also, there still are a lot of military and serviceman’s testers out there in various states of repair.
Limitations of the Hickok Circuit
While able to measure the transconductance of a tube, the Hickok circuit does not run the tube near its usual operating condition, especially for power tubes. This does not allow the tube to be tested under stress, either in voltage, current, or power dissipation. It also only exercises a small area of the tube’s characteristic curves, as shown in the follow set of curves for the 6L6GC:
The diagonal straight line is the normal class AB1 load line in a push-pull power amplifier, with the red dot being the idle operating point. The red curves in the lower left corner are load lines as presented by a TV-10/U Hickok-style tester. The two curves are the two halves of the rectified sine wave - the transconductance is the vertical difference between the two curves. This certainly isn’t testing the tube under realistic conditions! These curves were actually measured on a calibrated TV-10/U (quite similar to the TV-7/U). The peak plate voltage was about 165 volts and the peak current about 70mA.
Another limitation of the Hickok circuit is that pentodes and beam power tubes (tetrodes) are tested in a triode-connected mode, that is, the screen grid is connected to the plate. This still gives valid transconductance readings, but the effect of the screen grid is lost.
What other tests are useful?
The emission and mutual conductance (transconductance) tests are generally the most useful in evaluating a tube, but there are some other parameters that are worth testing.
Nearly all tube testers have a short-circuit test. This is generally implemented by rotating a switch through several position and looking for a neon bulb to extinguish. If you are testing a doubtful tube, this test is worth running, since a short could blow a fuse on the tube tester in later tests. It will also reveal gross leakage between the heater and cathode.
A "gas" test is frequently offered in a tube tester. Trace amounts of gas in a tube will increase the grid current in a tube, which, depending on the amplifier circuit, could cause the plate current to run away and cause a "melt-down". The problem with this test is that most tube testers don’t run the tube at high voltage or high power, which is when the gas in a tube becomes most evident. In an amplifier a gassy tube will usually show a diffuse blue or purple glow between the elements of the tube and often make the plate current unstable.
How does a laboratory tester work?
A laboratory tester allows detailed and accurate measurements of tube characteristics to be made. It can apply precise voltages to the elements of the tubes and then measure the current on each of the tube’s elements. Since this requires accuracy and adjustability, lab testers are expensive, not often used by servicemen or hobbyists.
The grand-daddy of laboratory testers is the Weston model 686. It has separate panels for the tube sockets and the power supply/measurement sections. There are three meters for setting up the voltages and three meters for measuring current and transconductance. The power supplies are not regulated, so don’t use this where the power line voltage changes!
Another old classic is the General Radio 561 Vacuum-Tube Bridge. It uses a balanced bridge circuit to measure amplification factor, plate resistance, and transconductance. As with most old-school bridges, it needs to be surrounded by lots of other equipment: a 1000Hz source, a null detector, and all required power supplies.
A more modern lab tester is the RCA WT-100A. It has built-in regulated supplies and AC sources and is easier to use than the earlier lab testers. However, they all are relatively tedious to use, since each operating point must be set up by hand and an the current read out. To collect all the data needed to make a complete set of characteristic curves could take all day!
How does a curve tracer work?
A curve tracer displays the entire set of characteristic curve on a screen in an instant. While each data point may not be able to be measured with the same accuracy as with a laboratory tester, any problems with the tube’s characteristics can bee seen at a glance. If there is a way of quickly switching between two tube, then matching can be achieved by simply selecting tubes where the curves line up.
Various custom-built tube curve tracers were made from the 1920s and on. They worked by applying a rapidly varying voltage to the plate, typically rectified but unfiltered AC from the power line, then feeding the plate voltage and current readings to an oscilloscope. To get a family of curves, the negative grid voltage was increased step-wise with every pulsation of plate voltage. If each step lasted 1/120th of a second (for 60Hz power) then a family of 10 curves could be viewed on the oscilloscope screen without much flicker.
The ultimate analog tube curve tracer is the Tektronix model 570, which was introduced in 1955. It had all the required power supplies built-in and regulated and had provision for doing A-B testing on two tubes. It could test tubes at full voltage and current - 500 volts maximum and 500 mA maximum. This was accomplished with a design that used 45 vacuum tubes! The only semiconductor was a 1N34 diode. Unfortunately, the Tektronix 570 is quite rare. It came out just as electronic designs were shifting away from tubes to transistors, and it was expensive, so not many labs bought it
However, the semiconductor curve tracers, such as the Tektronix 575 and 576 and units made by many other test equipment companies are fairly common. They can be used to test tubes in a limited way, but a filament supply is needed, and the maximum voltage is often limited. Some people have modified semiconductor curve tracers to better handle tubes, but this requires good engineering skill.
With modern solid state analog design, A/D and D/A converters and cheap computing, a decent vacuum tube curve tracer that can display the curves on a computer screen is quite feasible. The most difficult aspects of this are the high-voltage power supplies. Several commercial tube curve tracers have come out in the last twenty years or so, but typically didn’t last long on the market, because of limited demand.
The Apex tube matching system is essentially a version of a modernized tube curve tracing system, optimized for testing many tubes at a time. See [section] for more details.
How does the MaxiMatcher work?
The Maxi-Matcher, made by Maxi-Test of Seattle, Washington (www.maximatcher.com) tests up to four tubes for both idle current and transconductance. The built-in sockets are for the octal 6L6/EL34/6550 type tubes, but adapters are available to test other types, such as the EL84, 7591, 2A3, etc. It has a built-in power supply and has two plate voltages (325 and 400 Volts) and five fixed bias voltages (-14, −24, −36, −48, and −60 Volts). A short-circuit test socket is provided. A built in overcurrent circuit is provided which protects both the tube or tester if something shorts-out during test. A 3 1/2 digit LED meter can measure either plate current and transconductance. By measuring both, a simple multi-point test, as used in Apex testing, is achieved, although crudely. By matching both plate current and transconductance, a reasonable match can be made.
One problem with the Maxi-Matcher is that the voltages are not regulated. The result is that changes in the line voltage change the readings. If a pair or quad is to be matched in a single session and the line voltage doesn’t change, then a reasonable match can be made. However, if the plate current and transconductance are marked on the tube in order to match with tubes tested in other sessions, incorrect matches could result unless the line voltage is rock-steady or regulated.
Another side-effect of the lack of regulation is that the readings change slightly if four tubes are inserted versus two tubes inserted. This is caused by the heater voltage being dragged-down due to the current load of four tubes.
To evaluate the regulation issue, tests were made with a stock Maxi-Matcher (original version). Two or four EL34s were used. The tube in socket was instrumented to measure heater voltage and plate current. The Maxi-Matcher’s plate current readings were virtually identical to the measured reading, so the Maxi-Matcher’s plate current and transconductance reading were used. Here is the results (Ef = heater voltage, Ip = plate current, and gm = transconductance):
4 tubes | 2 tubes | |||||
$$Line$$ | $$E_f$$ | $$I_p$$ | $$g_m$$ | $$E_f$$ | $$I_p$$ | $$g_m$$ |
$$v_{rms}$$ | $$v_{rms}$$ | $$mA$$ | $$mS$$ | $$v_{rms}$$ | $$mA$$ | $$mS$$ |
110 | 5.91 | 27.3 | 4.87 | 5.62 | 27.1 | 4.79 |
115 | 6.19 | 29.3 | 4.97 | 5.88 | 29 | 4.92 |
120 | 6.45 | 31.5 | 5.07 | 6.17 | 31.4 | 5.02 |
125 | 6.81 | 34.1 | 5.21 | 6.47 | 33.8 | 5.13 |
130 | 7.08 | 36.5 | 5.37 | 6.75 | 35.9 | 5.24 |
For the usual heater voltage tolerance of ± 5%, it shouldn’t be below 6.0V or above 6.6V. The line voltage would have to be kept close to 120V for this to hold true. Since the plate and screen voltages are not regulated, either, the plate current changes a lot with line voltage. The change in transconductance is less. The differences due to 4 tubes vs 2 tubes is much less, since changes in heater voltage don’t affect the tube parameters as much.
So, in summary, the Maxi-Matcher can do matches moderately well in a single session, but special care needs to be given for regulating the line voltage.
How does the Apex system work?
For a full explanation of how the Apex tube matching system works, see our How Apex Tube Matching Works page.