This is a copy of Hubert Houtman's artical in "Electronic Design" magazine, September 18, 2000.
This sequential-sampling oscilloscope plug-in module can significantly increase the display bandwidth of an ordinary 10-MHz oscilloscope for repetitive signals (Fig. 1).
Using the 5KΩ potentiometer and oscilloscope adjustments, the time base is adjustable from 1 to 50nS/div. For longer sweeps, the circuit should simply be bypassed, and the oscilloscope must be used directly. By adding a DSO board and computer, it can be converted into a digital sampling oscilloscope as well. Future modifications, such as the use of the strobe as a variable delay generator, are simple to put into effect. Also easily implemented is the “synchroscope” mode, with triggerable components installed within the delay line.
The input signal enters via the compensator box, which is a 4.6-dB pole zero cancellation attenuator.(ref 1) In essence, the 40pF capacitor diverts fast changes into the 40nS RG58A/U 50Ω delay line, (about 26' or 27' for teflon), and away from the inductor arm. As was verified by using this sampler as a time-domain reflectometer (TDR), the two main ports are 50Ω up to 1 GHz. Frequency-dependent losses in the delay cable are quite accurately compensated, as shown by the reed-relay (Radio Shack part number 275-232) step response (Fig. 2, trace A). When an ordinary 4.6dB attenuator is implemented, a rounded step response results (Fig. 2, trace B).
Although the inductor arm derives the trigger from the input signal, other triggers can be plugged in as needed. These triggers also can be displayed by routing them to the sampler via the delay line. For each input pulse, the MAX961 trigger-and-hold comparator triggers an adjustable-rate, fast-ramp circuit by switching the two Schottky clamped MPSH10 transistors. At the same time, its 4V complementary-output step quickly enables its own latch (LE), locking out further changes. This locked comparator allows the switched MPSH81 current source to ramp up the 82pF capacitor from -2V to 2V, without interruption. After 500nS, the resistor discharges the 1nF capacitor to 0V, thereby reinstating the comparator and re-arming the ramp.
The plus input of the MAX961 strobe comparator is scanned by either a slow ramp from the oscilloscope time-base output (T), or the 250-WO manual control (X) for the X-Y plotter. It’s scanned over the same voltage range of -2V to 2V, which also is used as the display horizontal signal. Each time the fast ramp crosses this voltage, the comparator output switches from 2V to -2V. Meanwhile, its complementary output switches from -2V to 2V. Both outputs maintain a specified transition duration of 2.3nS. They also symmetrically drive the master/slave track-and-hold bridges that are connected to sequentially sample VIN in the delay cable. The actual sampling event happens in the track-to-hold transition. It takes place during the few hundred picoseconds in which the master MBD301 Schottky diode bridge resistance switches from low to high. Such switching occurs within a small, central part of the full 8V applied step.
Consequently, its switching duration must be only a small fraction of the strobe comparator transition duration. The RC delay network controls the strobe comparator's latch. While the slave is in the track mode, this latch locks the master T/H, (Track and Hold), in the hold state for 650nS.
During this 650nS period, the master BIFET charge amplifier (½TL082 4MHz unity gain) can buffer the sample, pulsed onto its input capacitance of about 15pF. This voltage is a scaled-down replica of the sampled voltage entering the 1pF/15pF capacitive divider. A 15MΩ/1MΩ resistive divider is connected across it, which completes this wideband compensated attenuator. Therefore, the master bridge is loaded with only a low-input capacitance of about 1pF. Because the bridge's on-resistance is about 100Ω, the RC time constant is around 100pS.
Continuously buffering the 1nF memory, the slave T/H BIFET amplifier (½TL082) produces a step-like, down converted representation of the input waveform (VIN). It does so at a low intermediate frequency (IF) for the oscilloscope, or a zero IF for the X-Y plotter. Switching transients are eliminated via the adjustable low-pass filter (LPF). By averaging the output over as many input pulses as desired, this device strongly reduces noise.
The bandwidth of this sampler is over 1GHz. Its transition duration is about 300pS which, together with the approximately 250pS intrinsic risetime of the reed relay, yields the measured display transition duration of 400pS (Fig. 2, trace A, again). Trace C illustrates the time-scale calibration and vertical-scale linearity by showing a sampled 100MHz sine wave with a sampled zero line. For this RF measurement, a countdown trigger was used, prior to the trigger-and-hold.(ref 2) To display higher frequencies with this sampler, the countdown trigger should be preceded by a prescaler, such as the emitter-coupled logic (ECL) circuitry.(ref 3) Additional experiments have verified that its fairly easy to add more synchronized channels, using the strobe comparator to fan out and drive two or more similar comparators. Via separate delay lines, each of these comparators drives individual T/H circuits to sample multiple inputs.
Comments by me, redrok:
I believe the "Resistive Splitter" started out as an "Adams' type Un-Equal resistive splitter". It probably had a 4.6db main loss with a 10db pick off for the compensator. This would have had a pair of 13Ω and a 54Ω resister in the "Y" connection. See .
However, as with all microwave circuits, there are numerous "Magic" components we can't see. So Houtman tweaked the design with:
1. Adding a 40pF speedup cap.
2. The 54Ω was changed to 40Ω and a matching inductor to the 50Ω coax going to the trigger comparator.
This probably results in less main loss and less going to the pick off.
The MAX961 trigger comparator doesn't need very much amplitude to work well. Also the MAX961 has about 3.5mV of hysteresis.
The MAX961 trigger comparator triggers on a rising edge. If you want to trigger on a falling edge just reverse the inputs. This could be done with a small DPDT reed relay or tiny switch.
The AVAGO HSMS-282x series of Schottky diodes are superior to the MBD301 especially for the sampler bridges. Use the HSMS-282P or HSMS-2828.
I intend to use a MicroMite MK2 controller to drive the sampler. It is programmed in a very capable version of BASIC. The uMite has a 10bit AtoD converter and an OLED graphic display. I will be using several I2C 10bit DtoA to control the trigger level, Slope, and Sample level. I like the MCP4716 although 8bit or 12bit versions could also be used.
 G. Amsel, R. Bosshard, R. Rausch, and C. Zajde, “Time Domain Compensation Of Cable Induced Distortions Using Passive Filters For The Transmission Of Fast Pulses,” Review of Scientific Instruments, Vol. 42, No. 8, p. 1237-1246, August 1971.
 H. Houtman, “Counter Circuit Improves Oscilloscope Triggering,” Electronic Design, Vol. 48, No. 15, p. 126, July 2000.
 F.J. Hufft, “Build This 1.6-GHz Counter Presale,” Radio-Electronics, Vol. 61, 10, p. 47-54, October 1990.
More references added by me:
 H. Howe, “Stripline circuit design”, 1974, Artech House.
(Invaluable "Must Have" book!!!!)
 "Adams' Un-Equal resistive splitter"
"Adams Calculator for un-equal splitters"
 "DIY Ghz sampling head for greater than 100Mhz scopes"
 "Generation of short electrical pulses based on bipolar transistors"
 "SAMPLING NOTES"