By Jukka Tolonen
The "magic eye," so characteristic in old time radios, has virtually disappeared, despite the renaissance of other tube equipment. That's a great pity, since the green eye-winks ( jpg image, 57kB) of these lovely devices are certainly something other than customary LED bar displays. Indicator tubes are well-suited for use in fast, accurate level meters for monitoring tape recordings or power amplifier output levels.
An indicator tube is an electron ray device, with a fluorescent, transparent screen instead of the normal anode plate. In addition to the usual grid and cathode electrodes, it includes a small electrode, called a target and usually made of a thin wire, whose potential affects the deflection of the electron beam and thus determines the width of a lighted area on the screen. The grid is usually internally connected to the cathode to safely limit the screen current.
The target potential must change 100-200V for the full deflection, which is also dependent on the screen voltage. Most devices also include a built-in triode amplification stage for driving the target with about 5 V input voltage (Figure 1). The operation is very similar to electrostatically deflected CRTs, except that the deflection occurs in only one dimension, and a single target electrode replaces deflection plates.
Indicator tubes are ideal devices for dynamic display applications, since they are inherently fast. The only moving part is the electron beam, which can easily follow the fastest signals. Unfortunately, the human eye is unable to detect very fast movements, so the early indicator tube circuits contained a simple peak reading circuit, which used a germanium diode as a rectifier (Figure 2).
Originally used for tuning indicators for radios, indicator tubes came in several different forms. The first types used sector-patterned lighted or shaded areas, some types even had dual ranges for course- and fine-tuning. The newest "magic band" versions introduced in the 1960īs were specifically designed for tape recorder use. They had two light bands moving towards the center of the tube. Overlapping bands indicated overmodulation.
To my surprise my local tube dealer stocked in a number of several different European indicator tube types: 6EM80, 6EM81, 6EM84, 6EM85, and 6EM87, mostly made by Philips or Telefunken. Table 1 lists the types I find in my tube manuals, along with possible American equivalents. Types EM81 (6DA5), EM84 (6FG6) and EM87 (6HU6) are probably the most available ones, at least in Europe. These types are bolded in Table 1. The EMM801/803 would be perfect for stereo use, if only I could find any. Unfortunately, I have no information on the availability of these devices in America.
The early Revox 36 series machines, A36...E36, used a single EM71 tube, switchable to either channel. The F36 used an EMM801, a dual indicator tube. The popular G36 used already "modern" VU meters (volume indicators) for recording level indication.
In TAA 1/84 ("A Peak Meter for the ReVox A77," p. 19), Reg Williamson discussed very thoroughly the American-style VU meters and British-style peak reading meters for tape recording. You can be easily modify the circuit in figure 2 to duplicate the performance of the British BBC version, with a rise time about 2.5 ms.
I am sure BBC engineers know their business, and I might swallow, with some difficulty, the theory that the human ear is not able to detected distortion in short transients. However, even if I don't hear something, I'd still like to see what's going on! Audiophiles with limited funds are usually forced to make some compromises in their audio systems. Nevertheless, I don't see any reason to be satisfied with anything less than perfect in such a simple and inexpensive instrument, especially regarding its vital role in quality recordings. With present analog circuits, the cost is certainly not an issue.
A perfect peak level meter must be able to detect a peak of any duration within the audio frequency range. The meter must also have sufficiently long decay time, so peak level display is clearly visible to the human eye. Some CD test records include a test signal consisting of a maximum amplitude pulse of one sample duration; that is, 22.7(s, which is the shortest pulse possible with the standard 44.1kHz sampling frequency. Common VU meters hardly react to these pulses, which clearly demonstrates the inadequacy of these devices.
My overall specifications for a high quality peak level meter are shown in table 2. Achieving this kind of performance with an all-tube design is possible, but not very practical. Even the purists shouldn't object to accepting a hybrid design, in which op amps are employed in realizing the analog functions, and with indicator tubes as display devices. After all, this system is for only monitoring; it does not interact with the actual sound signal, provided that the input does not load the signal source.
Figure 3 shows the meter's functional block diagram. As Mr. Williamson pointed out in his article, the positive- or negative-going peaks may differ as much as 8dB. Performing a full-wave rectification of the signal before the peak detector ensures the unipolar peak detector properly records the transients regardless of their polarity. The schematic of one channel is shown in figure 4, table 3 includes the parts list. The other channel is identical, with component numbering starting from 200. The channels share dual op amp IC1 and quad op amp IC2. Figure 5 is the power supply schematic.
The input buffer amplifier IC1A isolates the rest of the circuit from the signal source. It is a Bessel-type low-pass filter with high input impedance and linear phase characteristics, with a cut-off frequency of about 25kHz. The filter's prime purpose is to reject any remains of the high frequency bias when the unit is used with a tape recorder.
The full wave rectifier and the peak detector are high-speed circuits, operating up to 100kHz frequencies. The upper frequency is limited by the slew rate of the op amp, when the signal commutates from one diode to another.
IC101B and IC101C form a classic full-wave rectifier. With negative signals, diode D101 clamps the output of IC101B slightly positive and D102 isolates it from the summing input of IC101C. Resistors R106 and R109 thus determine the signal at the output of IC101C.
With positive signals, D101 is reverse biased, and D102 closes the feedback loop, with the output taken from the junction of R107/R108. This signal is summed with the original signal in IC101C. With the proper ratio of resistors R105-R108, the summing function accurately reverses the signal polarity in the output of IC101C, and a symmetrical full wave rectification function is achieved.
IC101D and IC101A create a high-accuracy, high-speed peak detector. C106 stores the peak value, and IC101A is a voltage follower acting as buffer and isolating C106 from the output. Whenever the instantaneous input voltage is lower that the output voltage level at the IC101A output, diode D103 is reverse-biased and D104 forces the inverting input of IC101D to follow the input signal.
If the instantaneous input voltage exceeds the output voltage, the IC101D output rapidly rises until D103 is forward-biased and C106 is charged until the output voltage equals the input voltage. In this mode, D104 is reverse-biased, and the inverting input of IC101D closely follows the input signal. Note that all the diodes are inside feedback loops, and the voltage drops of the diodes are eliminated, causing no degradation of accuracy even with small signal amplitudes.
This circuit can detect very narrow peaks, as diode D104 keeps the IC101D output close to the input signal at all times. This output needs to slew only the amount of two diode voltage drops after the input level exceeds that of the output. On rising signal slopes, the output voltage accurately follows the input signal, as C106 charges through D103. The slew rate and settling time characteristics of IC101D determine the speed.
TL074 has a typical slew rate spec of 13V/(s, which enables it detect peak widths less that 1(s. This is more than adequate, considering the input bandwidth is limited to 25kHz. C105 prevents high-frequency oscillations that might otherwise occur in the long feedback loop. R111 along with C106 determine the decay time constant.
Potentiometer P101 sets the input sensitivity, and potentiometer P102 adds a small, adjustable offset voltage, which determines the zero signal display level. IC2A and the triode section of V101 form a high-voltage output amplifier, in which the triode is operated inside the feedback loop of the op amp. This way neither the triode parameter variations nor anode voltage ripple affects the amplified output signal available from the anode, Pin 9.
The triode, operated as a grounded grid amplifier, has wide bandwidth and does not degrade the stability of the op amp. The output voltage range of this hybrid circuit topology is from 0 volts close to the anode voltage of the triode. Grid resistor R116 prevents oscillation in the tube section, and capacitor C107 ensures the stability of the entire stage.
Unfortunately, no way of operating the actual indicator tube in closed loop exists. The screen potential affects both the sensitivity and intensity of the display. To maximize the accuracy, a parallel regulator, R1, IC3 and Q1 (figure 5), stabilizes the screen voltage. R2, R3, and potentiometer P1 determine the output voltage. P1 allows the intensity of the tubes to be adjusted to suitable brightness.
I kept the power supply as simple as possible. The screen voltage is generated with a voltage doubler from 90-110 V transformer secondary voltage. Regulated power supplies for the op amps are not actually necessary, but the regulators (IC2 and IC3) allow some freedom in the secondary voltage, which can be in the range 10-15V. You can take the required secondary voltages from a single transformer with multiple output windings, if available, or from separate units.
You can construct the circuit on a piece of perforated board. The layout of the rectifier and peak detector circuit is not very critical, except for the short leads required for the diodes D101-D104 and capacitors C105 and C107. The high-voltage section is easy to install using a tag board. Transistor Q1 needs a small heat sink. Remember that the collector tab is in 200V potential.
After completing construction, carefully check the wiring, especially the high-voltage section. Use sockets for the op amps and check the power supply voltages before installing them. A signal generator and an oscilloscope greatly help you to check the operation. Start with a 1kHz, 1V sine-wave input. The IC1A output should closely follow the input signal. The output of IC101C should be a perfect full wave rectification of the input.
The gain up to this point is about 3. Finally, the IC101A output should be a DC voltage, equal to about 4.3V. Changing frequency to 10kHz should cause no change in the output voltage level. With the very low input frequencies, there should be a noticeable droop in the output of the peak detector between the signal peaks.
When the unit seems to operate normally, first set the intensity of the indicators with P1. This must be done before any other settings, because changing the screen voltage will affect all other adjustments.
With no input signal, adjust P102 and P202 so the display bands are invisible just outside the edges of the screen. The display seems blank with no signal, but the bands come visible with a signal about 40dB below the full deflection level. The voltage at Pin 9 of the indicator tube should now be around 50V.
The sensitivity is set with P101 and P201. Full deflection, the display bands just touching each other in the middle of the screen, should occur at the absolutely maximum allowed signal level. When the signal exceeds this level, the bands overlap, appearing much brighter. The indicator tube saturates when the overlap width is about 2mm.
The application determines the suitable sensitivity setting. When used as a recording level meter with a tape deck, I set the full display deflection to occur at the maximum recording level, which causes about 2% harmonic distortion in the playback signal. I normally adjust the recording level so the loudest peaks just reach the full deflection, but do not exceed it. This gives the best trade-off between low distortion and maximum signal-to-noise ratio.
When the meter acts as a clipping indicator for a power amplifier, I set the full deflection to occur at the edge of clipping, perhaps using some harmonic distortion level as a limit. The display has 40dB dynamic range. If the amp has the peak power rating of 100W, the peak display stays blank below 1W output levels and indicates even the fastest peaks up to and over 100W. You can make a scale of intermediate power readings beside the tubes. In this application a faster decay time - easily achieved by decreasing the value of R111 and R211 - would be more appropriate.
The display has proven to be a stable and reliable instrument with an unambiguous readout. It impresses even those accustomed to tube amplifiers.
In case you are not able to locate the EM87/6HU6 tubes, you can adapt any of the types listed in Table 1 to the circuit. You can replace the built-in triode with a separate 12AX7 or equivalent dual triode. Types EM84/6FG6, EM85/6DG7/6DU6 and EM840 are pin compatible with the EM87. EM80/6BR5/6M40 and EM81/6DA5 differ in pinout from EM87, see table 4.
Back to Glass Audio Article List
Back to my Home Page
Feedback is highly appreciated: