Nuts and Volts - November 2010

band scopes). This should handle the normal range of signal levels encountered in high frequency circuitry. It will not handle very low amplitude signals applicable to radio receiver circuits, but that is a job intended for spectrum analyzers only.

As to circuit loading, there really are no high impedance circuits once we enter the VHF spectrum and above. Normally, most circuit source impedances will range from 25-500 ohms at these frequencies. Even reactive LC circuitry will have much lower impedance comparable to low frequency counterparts. So, the 3,400 ohm loading factor of this probe will pose an acceptably small loading on most VHF circuitry. And more importantly, it imposes a very small reactive loading (>>1 pF). In high frequency measurement, the probe’s reactive component is much more critical than its resistive one in terms of circuit loading (as explained earlier). Finally, as to external power, this turned out to be of little inconvenience due to the amplifier’s onboard regulator and power plug. It required only a separate wall plug transformer of adequate voltage. I leave this wall wart plugged into an outlet strip mounted on the back of my scope cart. This makes probe installation quick and easy — just one added cable to connect/disconnect.

Probe Design

In designing this probe, a number of factors had to be considered. It all starts with the MMIC (monolithic microwave integrated circuit) amplifier which has a gain of 31 db (X34), a 1 db compression point (the very onset of saturation) of 12. 5 dbm (approx. 2.5V p-p), and a frequency response roll-off of - 4 db at 500 MHz. It also requires a 50 ohm source and load impedance to maintain these specs, along with good stability. Best performance with these amps is obtained with a bias (B+) current of 36 mA. The B+ supplying this current ideally would be through a load resistance of infinite impedance because this load is effectively placed in parallel with the 50 ohm termination impedance. A constant current source comes to mind but will not work here as per manufacturer’s application notes. If the bandwidth ratio was not quite so wide, a suitable RFC (radio frequency choke) could be used with a bias supply of 5 VDC. However, this circuit requires a very wide bandwidth ratio (1,000:1); no RFC would work due to their inherent self-resonance at some point, rendering them useless at frequencies above that. So, a pure resistance is

ITEM C1 C2 C3 C4 C5 C6 R1 R2 R3 R4 MMIC 18V Regulator Wall plug Power Pack

required here to supply the bias current and I want that value to be as high as possible. Of course, the higher the resistance, the higher the supply voltage has to be to supply that current. A compromise had to be made here in terms of voltage, wattage dissipation, and shunt loading. I chose 18 volts which works out to 392 ohms for the supply resistor with a dissipation of 1/2 watt, without upsetting the output termination to a great degree.

Given that 392 ohms is not the ideal value here, the amp’s maximum output level is degraded, so I decided to spec that figure at + 10 dbm ( 2.0V p-p) for best linearity. Given the amplifier’s gain with a 1:1 attenuation ratio, this would only allow for a max probe input impedance of 850 ohms and its dynamic range limited to 2.0V p-p on the upper end. These limits just would not satisfy a wide range of probing situations, but by increasing probe attenuation to 4:1, I could get the 8V p-p input level I was after at the upper end of range, and increase the input impedance to 3,400 ohms. The actual input impedance works out to 3,400 ohms shunted by <<1 pF. This provides an acceptable loading factor while maintaining the desired dynamic range.

It now becomes a 4x probe and it is easy to calculate the actual display amplitude vs. input signal amplitude — just double the displayed voltage and double it again (X4). Not as easy as a 1X or 10X probe, but still quite simple to compute in one’s head. The amplifier roll off, cabling connections, and board layout combined will add up to a little less than a - 6 db (2x) loss at the upper limit of 500 MHz of the probe’s bandwidth, so the next problem to be solved was how to flatten this response curve. The straightforward answer was to shunt R1 and R2 with a compensation capacitance. The Xc needed for this compensation worked out to be 2,000 ohms at 500 MHz which has a value of 0.18 pF. It worked out in my favor on this one due to the fact that 1/4 watt carbon film resistors have a parasitic capacitance of 0.35 pF. Two in series would give me 0.175 pF — almost exactly what I needed! A call to KOA Speer’s (resistor manufacturer) engineering department confirmed that these resistors are made to exacting mechanical tolerances and that the parasitic is uniform from unit to unit with one minor exception. The laser etching used in the manufacturing process is slightly different from one group of values to the next; a group value being a certain range of resistance in perhaps a 5:1 ratio.

Next I needed to pick out two values of resistance that would give me a combined value of 3,400 ohms and a parasitic capacitance of approximately 0.175 pF. The values chosen were accomplished empirically due to the fact that the exact capacitance could not be predetermined. After considerable experimentation, a set of values was found to be 1,000 ohms in series with 2,400 ohms, with the correct parasitic capacitance to flatten out the response

DESCRIPTION SOURCE O.1 MFD/50V MLC ceramic 0.22 MFD/50V chip size 1812 0.47 MFD/50V chip size 1812 0.33 MFD/50V MLC ceramic 1,000 pF/50V MLC ceramic 4. 7 MFD/50V tantalum electrolytic 1,000 ohm 1/4W carbon film Mouser.com p/n 600 - CF 1/4 102J 2,400 ohm 1/4W carbon film Mouser p/n 600 - CF 1/4 242J 51 ohm 1/4W carbon film 392 ohm 1W 1% metal film MAR-8A mini-circuits 78L18