067-0681-01
The Tektronix 067-0681-01 is a pulse generator for adjusting the transient response of high frequency oscilloscopes such as the 475, 485, 7A11, 7A24, 7A19/7904, and 7A29/7104. It was designed by John Addis.
Key Specifications
Stimulus | 60-100 V square wave providing at least 11 mA, 50 Hz to 100 kHz |
---|---|
Pulse amplitude | ≈250 mV into 50 Ω |
Rise time | ≤150 ps |
Abberations | < 4% including < 8% peak-to peak ring at (typically) 5 GHz measured with 11.5 GHz bandwidth |
The original TD pulser was the TU-5, introduced in 1962. It had a TD, two resistors and a "pot" (variable resistor or potentiometer) and no transistor.
It takes in a 60 to 100 Vpp rectangular waveform (e.g. from the calibrator output of an oscilloscope) and produces a 250 mVpp rectangular waveform with rise time less than or equal to 125 picoseconds.
The output pulse is produced by a 10 mA, 2 pF tunnel diode, part number 152-0177-02. The tunnel diode may be replaced with a TD-263B from American Microsemiconductor, if available.
https://www.americanmicrosemi.com/197602-tutorial-on-tunnel-diode-and-back-diode/
The input and output connectors are BNC. The male BNC of the pulse output connector is appropriate for connecting directly to the female BNC input of an oscilloscope (i.e., with no cable between the 067-0681-01 and the oscilloscope to spoil the risetime).
A regular BNC cable is typically used for connecting the 067-0681-01 input from the high voltage pulser that drives the 067-0681-01. The risetime of the input to to the 067-0681-01 is not important if driven at less than 100kHz.
Links
Internals
The input circuit is a rectifier. This allows the 067-0681-01 pulser to operate from a negative input pulse or a positive input pulse. The output of the rectifier turns the current source on and off.
The current produced by the current source (at the collector of PNP Q1) is set by controlling the emitter current using 6.3 V Zener diode CR3, R1, R2, and VBE of Q3.
R1 is a 200 Ω potentiometer (assume it is set to 105 Ω). R2 is 390 Ω.
- IE (input waveform high) = (6.2 V - 0.7 V) / (120 Ω + 390 Ω) = 10.8 mA
Q1 is Tektronix part number 151-0410-00, a 2N5087, which has a minimum beta of 250, so we can assume IC = IE.
Starting with the tunnel diode in its low state, 90 mV and 10.0 mA. Of the 10.8 mA, from the current source, 10 mA is flowing in the tunnel diode and 0.8mA is in the load. 0.8 mA X 50 Ω = 40 mV at the 067-0681-01's output. The TD's current is just barely below that which is required to switch it to it high state.
Just a touch on the potentiometer switches the tunnel diode to its high voltage state, more current is diverted into the load instead of the tunnel diode. With the output is terminated in 50 Ω ohms, the TD's current drops from 10.0 mA to about 3.25 mA and the output current increases to about 6.75 mA, resulting in an output voltage of about 0.34 V. The output has stepped from 90 mV to 340 mV.
Between the collector of the current source transistor Q1 and the output tunnel diode, there is a chain of three 1 kΩ composition resistors in series. The purpose of this is to remove Q1's parasitic capacitance and eliminate the inductance of the lead to Q1. This makes the current source (Q1) nearly perfect at high frequencies.
Output
The dynamic resistance of the tunnel diode in the high voltage state is low, about 7 ohms. This, plus the 43 Ω output resistor, makes for a 50 Ω output resistance when the tunnel diode is in the high-voltage state. To prevent the 43 Ω output resistor from appearing inductive and spoiling the flat top of the output waveform, the resistor is actually mounted inside the BNC output connector. The BNC center conductor has a small hole to accept the 43 Ω resistor’s lead.
Because of this, the resonant frequency of the TD and its parasitics is around 5 GHz and well damped, though the exact resonant frequency depends on exact lead length of components which varies somewhat during production.
For any scope with less than about 1.5 GHz BW, the ring would likely be invisible.
Transient Response
John Addis writes: I measured the 067-0681-01 transient response printed in the datasheet. It was not an easy measurement to make to 0.2% at 50 ns or less. I measured using at least two scopes, one was the 10A1 which I held in high regard. The result is in the datasheet. There is an entirely plausible explanation for the 0.25% peak in the response at about 28 ns. Two competing phenomena are responsible.
At t = 0 when the tunnel diode switches to its high state, there is a load to ground on the TD due to the three 1 kohm resistors in series with the transistor current source's collector-base capacitance. The load is 3 kohms at t = 0 but goes away with a time constant of 3 kΩ X 3 pF = 9 ns. After three time constants (27 ns) that load is almost entirely gone. So from t = 0 to about 27 ns, the pulser output is rising, as shown at the extreme left end of the graph. The 3 kohm load on the TD at t = 0 reduces the pulser output by an easily calculated 0.20%.
The second phenomenon makes the pulser output decline starting at t = 0. This is a thermal effect. When the TD first enters the high state, it is cold, room temperature. Over time the TD junction heats up due to the power it dissipates. Germanium TD junctions have a negative temperature coefficient (as do silicon transistor junctions), so the voltage across the TD decreases as it heats. That it takes 100 μs to reach thermal equilibrium should not be a surprise. It is a very small junction that heats up as the case and leads warm up at different rates.
That the thermal and loading effects are both about 0.25% is entirely coincidental.
The tunnel diode itself has a finite rise time. The TD has a unique VI characteristic. In its low state the TD's voltage is about 90 mV maximum. When the voltage rises above that, the TD's current decreases. The constant current driving the pulser's TD is instead diverted into the junction capacitance. As the junction capacitance is charged the diode current ultimately reaches its high state and the supplied current again all flows into the junction and any load. When observing this change of state, one can see that the TD voltage is changing most rapidly when the junction current is at its minimum and the supplied current is mostly flowing into the junction capacitance.
The 067-0681-01's risetime is dominated by the lead inductance. When the TD body is soldered to the BNC case and the TD's remaining lead is as short as possible, the risetime is about 80 ps. When manufacturing later soldered the lead to the BNC (or worse, left a loop of the TD's lead connected to the 43 ohm resistor that is inside the BNC) and replaced the original ⅛ W 43 ohm resistor with a larger ¼ W resistor, the risetime became 125 ps with a low amplitude ring at 4-5 GHz.