Precision Detector pcb

A careful study of the precision detector pcb has revealed that the power supplies necessary are +8 volts and -5volts. The circuit will work with +9volts but very much more latitude in supplies and the zero adjust will not cover sufficient range. Some data I have suggests that the two diode biassing resistors can be changed for different power supply voltages (see the circuit diagram)

Thanks to some assistance from Sam G4DDK we were able to investigate the performance of the pcb between 1 and 4GHz. The circuit will give relative power indications down to about 500MHz. It seems as though the sensitivity is waning at above 3GHz (possibly due to the spec of the diodes but more likely the effect of the meander-line inductor). We believe that the input circuit may be configured to compensate, to some degree, for the change in sensitivity with frequency. I think the board can be optimised for 3.5GHz by shorting the second loop of the meander to ground. The meander is a quarter-wave line track providing the DC path for the diode. The package has two diodes which are closely thermally coupled and matched. the second diode is used to provide thermal compensation and minimise thermal drift. The board can be used with between 5 and 12v supplies, though this build is for +8volt supply. The current source resistors biassing the diodes are changed for the different supply voltages, and though I have no information it may also be necessary to change the trim-pots. These details were obtained from an earlier version of the circuit, but with the same diode arrangement are given on the circuit diagram. The circuit shown below is from my examination of the pcb and the the circuit of an earlier version. (there may be errors!)

On the input after the 50 ohm pad there are a couple of pairs of pads that are marked L1 and L2 in the silk screening (these are not shown on the circuit diagram). With the top capacitor these would make a high-pass pi filter which help to roll-off the high sensitivity at UHF. The meander-line inductor has has two ^ marks one labelled 1.6 and the other 3.4. I believe this indicates that the line should be shorted to ground at those points to improve the sensitivity at or around those frequencies. The earlier circuit diagram confirms that the meander line is a quarter-wave track. Note this line will be a short circuit at frequencies below 1GHz and will need to be cut and replaced with an inductor ( at L2 ? ) for low frequency applications. I think a suitable inductor or even resistor would allow the pcb to be used at lower frequencies. I believe one has been used at VHF. Some experimentation may be required to use the circuit below 500MHz. The circuit of the pcb is a standard high input-impedance instrumentation amplifier following the application note in the Nat Semi databook, which is reproduced below. The 4th op-amp is used to allow for offset or zero adjust of the meter on the output..

As it stands the board is quite good at 23cms and 13cms, but a bit marginal at 9cms, this reflects the fact that the meanderline is about a quarter-wave at 1500 MHz. It should be shorted to ground at the marked places to peak at higher frequencies. It should make a useful power indicator as it gives an output voltage in the order of 0.5 to 1 volt for 0dBm at those frequencies. The multiturn pot can be adjusted to give a condition were one volt equals 0dBm. Some further experimentation is required around the input circuit to get best performance. The 6dB pad on the input suggests that this should show a good return loss at these frequencies. I am not sure on this point but it is possible that the bias has been selected to ensure the diodes are in their square-law operating region. If so the response will be approximately linear with power input. I hope to check this later.

Note there are also mirror images of this pcb, reflected about the long side, and for use with right-handed and left-handed Power Amplifier boards. The circuit below does not show the capacitor between the pad and meander-line L1and the unpopulated inductor positions.

If this pcb is used in conjunction with the 30dB coupler, which can be "cracked off" (after scribing) one of the "sweepings" PA pcbs A useful 3Ghz through-line directional power indicator can be constructed. These will fit nicely in the tin-plate boxes !!

I have now attempted to measure the frequency response of the board. I do not have levelled signal generators out into this region so the method I used was as follows. I connected my Marconi 605x source to a 2-way splitter, and connected one output to the detector pcb, the other outpur was connected to an HP 432A power-meter with a HP 478A thermistor head. Because I am not sure of the head calibration I adjusted the signal generator level to give the same reading on the power meter for each frequency. My thought was that the splitters have good amplitude balance, often even outside their specified range. Two different Marconi signal sources were used and these needed to be set for different levels on the power meter. The output was measured on a 10 megohm input impedance DMM on the 2volt range (resolution 1mV). The gain setting was wherever the pot happened to be, so this is not intended as a final measure of sensitivity. On the sample I was measuring at 3.5GHz in the same condition as plot D below I measure 66mv at maximum gain. This is for a drive level approximately -3dBm (give or take half a dB max for the splitter) At -10dBm the output was 14mV. The performance even with the shortened line is much better at 2.4GHz, +2dBm gives 1839mV, 0dBm gives 1107mV, -6dBm 237mV and -10dBm 71mV (assuming the amplitude balance on the CHC3500 is still good)

Plot A and B is with a 20dB Tektronix 20dB attenuator in front of the power meter, and using a Minicircuits ZAPD 21-2 splitter specified from 500 to 2000MHz.. Plot C uses the DML CHC3500 power splitter I have been selling, the attenuator was removed, because I was not sure of its frequency range. For Plot D I soldered a short across the meander line, near the "3.5" legend in the silk-screening. Plots C and D are done with the same constant power level. You will see that the response at the top frequency is improved but not by much at this level.