R7100 Discriminator Tap Notes

Just thought I would add my experiences to the confusion with the data slicer circuits. What eventually worked for me may also be appropriate for others having trouble.

I am using a modified R7100 with discrimator tap. When monitoring motorola trunk data channels, this point provides a nominal 500 mV rms. While monitoring POCSAG data, this voltage rises to 1000 mV rms. With no input signal, the noise level was a nominal 430 mV rms.

The data slicer I am using is based on the standard comparator circuit using an LF351 op amp. I say "based on" since I had to modify the circuit slightly to get it to work.

It is important to note, that the standard circuit using 10k resistors to ground from pins 2/3 and a 100K feedback resitor from pin 6 to 3 worked fine for decoding POCSAG signals. However no reliable MDT or trunk signals could be decoded. I was having a better quality of signaldecoding on weaker RF signals than on strong signals. The apparent cause was the amp was driven into saturation and with the feedback, slowed down the response time making higher speed decoding impossible.

I chose the LF351 op amp for three reasons. First it is a J-Fet variety wich means it has a higher input impedance than the garden variety 741. Second reason, the current draw is about 1/10 that of the 741. This should help reduce loading on the PC's serial port and reduce the voltage sag I had noticed with the 741. Third reason, I had several around the shack.

With a 10 k resistor from pin 2 to ground, loading of the discriminator was observed (indicated by a noticeable decrease in audio level) when the discriminator was connected to the slicer. This resistor was increased to 100K which is consistent with the higher input impedance of the LF351.

I concentrated my attention on the feedback resistor in order to speed up the response time. I varied the value from 10 k through 2 M. The results were predicatable. As the voltage at pin 3 approached that of the noise 600mv DC (approx 430 mV rms) the slicer stopped switching with the noise. Data signals would cause switching but still no reliable decoding.

Following normal good op-amp practice, I decided to balance my input resistances to ground. This meant replacing the 10k resistor from pin 3 to ground with a 100k one. I raised the value of the feedback resistor accordingly, still no success.

In a last ditch effort to speed up the response, I removed the feedback resistor. VOILA!!! 100% decoding of trunk data channel info. I varied the resistor from pin 3 to ground between 10 k to 1 M with no difference. When pin 3 was grounded, the amp started to oscillate and decoding stopped.

Decoding of POCSAG was not affected, still reliable decodes from 512 thru 2400.

If others are having trouble with loading and or data slicer problems, this fix may work for you as well. Experiment

Ed Richardson, VE4EAR
Radio/Electronics Engineer
City of Winnipeg Communication Branch
erichard@city.winnipeg.mb.ca