Hello,
My LTR exploits were on the back burner due to a death in the immediate family. I did listen to the DNC comms and have found some interesting Passport LTR stuff on 217 and 220 MHz. The problem is my LTR-enabled BC780 scanner can not do the 6.25 KHz steps needed for 220 MHz. Last weekend I dug out my protoboard, some components, and my oscilloscope. I was going to build a LTR slicer for my AOR 8600 but I also wanted to get an idea of how well Bessel filters did (2 poles vs 4 poles vs 6 poles, etc).
I found TI has a nice (and free) Windows program for designing filters called FilterPro. To start off I designed a 2 pole 300 Hz Bessel filter using some on-hand components. As expected it did filter but any strong voice buried the LTR data. I got a surprise when I tried a 4 pole 300 Hz Bessel filter. It filtered alot of the voice out. I decided to see how good a 4 pole filter could do and started designing a complete circuit. I discovered another type of filter supported by FilterPro, the linear phase filter. A 4 pole linear phase filter is even better than a 4 pole Bessel filter and still had the group delay response meeded! The reason I used 300 Hz as the corner frequency is I heard that any filtering should pass at least two times the data rate.
Most of the LTR slicer circuits are complex and are designed to be bulletproof. I get the impression that some people are running the LTRDump/LTRTrunk programs with a regular non-filtered slicer because the filtered designs are complex. I stumbled on a simplified design that will work 90 percent plus of the time. I can not say it works 100 percent because it may fail to work on the "Deep Bass Singers" talkgroup where the radios are putting out a very compressed 5 KHz deviation. Using a 4 pole filter allows a circuit design using one Quad Op Amp IC so it can be RS-232 powered.
I remember from my previous life as a 2-way tech that PL tones were set at .75 KHz deviation. If I assume that LTR data is the same amount it means that 5 KHz voice deviation is 6-2/3 times stronger or if I convert to a voltage ratio, 16.5 DB. This means any filter has to reduce the voice frequencies by at least 16.5 DB. You really need to get the voice lower by a couple of more db before you can expect a solid decode. Because the input is assumed to be limited to 5 KHz deviation improving the filter response beyond -20 db (like to -28 db) gains very little. I also suspect that radios tend to hit 5 KNz mainly on voice peaks. There are periods of lower deviation when people take a breath, etc. So a 4 pole filter is good enough most of the time. Also the LTRDump/LTRTrunk program "oversamples" the data bit so a momentary voice peak getting through may not affect decoding.
I could not make this a "Radio Shack" type project because I could not find component values using FilterPro that matched the token values that Radio Shack offers. Another factor is the giant size of the capacitors at Radio Shack. Filters need components with small tolerances and temperature stability so a .1 uf Z5U ceramic capacitor that will vary from .05 to .12 uf will not work very well. Fortunely you can get good resistors and caps easily from Mouser Electronics. AVX makes 63 volt boxed film capacitors that are not very big and have a .2 inch lead spacing. I used the AVX 5% tolerance capacitors and Vishay/Dale 1% tolerance 1/8 watt resistors. I can not fit the circuit into a 9 pin connector shell but it can fit in a small box. Maybe someone skilled with surface mount can make it fit in a 9 pin connector shell.
I do not use a capacitor on the input. I do recommend a resistor in the radio between the discriminator tap and the connector you use. The buffer eliminates the effects of the radio on the filter. Since I also use MTrunk and ETrunk I have a switch to eliminate the filter. The filter is a MFB type to lessen the effect of component tolerances on the filter parameters. By having the 1.0 uf capacitor and the switch at the comparator input the data polarity can easily be inverted.
I hope people find this circuit useful. It is more difficult to build than a Motorola slicer but easier than the other LTR slicers. I am about ready to go from the breadboard to a soldered version in a small box for "Field Testing".
73 Eric
My LTR exploits were on the back burner due to a death in the immediate family. I did listen to the DNC comms and have found some interesting Passport LTR stuff on 217 and 220 MHz. The problem is my LTR-enabled BC780 scanner can not do the 6.25 KHz steps needed for 220 MHz. Last weekend I dug out my protoboard, some components, and my oscilloscope. I was going to build a LTR slicer for my AOR 8600 but I also wanted to get an idea of how well Bessel filters did (2 poles vs 4 poles vs 6 poles, etc).
I found TI has a nice (and free) Windows program for designing filters called FilterPro. To start off I designed a 2 pole 300 Hz Bessel filter using some on-hand components. As expected it did filter but any strong voice buried the LTR data. I got a surprise when I tried a 4 pole 300 Hz Bessel filter. It filtered alot of the voice out. I decided to see how good a 4 pole filter could do and started designing a complete circuit. I discovered another type of filter supported by FilterPro, the linear phase filter. A 4 pole linear phase filter is even better than a 4 pole Bessel filter and still had the group delay response meeded! The reason I used 300 Hz as the corner frequency is I heard that any filtering should pass at least two times the data rate.
Most of the LTR slicer circuits are complex and are designed to be bulletproof. I get the impression that some people are running the LTRDump/LTRTrunk programs with a regular non-filtered slicer because the filtered designs are complex. I stumbled on a simplified design that will work 90 percent plus of the time. I can not say it works 100 percent because it may fail to work on the "Deep Bass Singers" talkgroup where the radios are putting out a very compressed 5 KHz deviation. Using a 4 pole filter allows a circuit design using one Quad Op Amp IC so it can be RS-232 powered.
I remember from my previous life as a 2-way tech that PL tones were set at .75 KHz deviation. If I assume that LTR data is the same amount it means that 5 KHz voice deviation is 6-2/3 times stronger or if I convert to a voltage ratio, 16.5 DB. This means any filter has to reduce the voice frequencies by at least 16.5 DB. You really need to get the voice lower by a couple of more db before you can expect a solid decode. Because the input is assumed to be limited to 5 KHz deviation improving the filter response beyond -20 db (like to -28 db) gains very little. I also suspect that radios tend to hit 5 KNz mainly on voice peaks. There are periods of lower deviation when people take a breath, etc. So a 4 pole filter is good enough most of the time. Also the LTRDump/LTRTrunk program "oversamples" the data bit so a momentary voice peak getting through may not affect decoding.
I could not make this a "Radio Shack" type project because I could not find component values using FilterPro that matched the token values that Radio Shack offers. Another factor is the giant size of the capacitors at Radio Shack. Filters need components with small tolerances and temperature stability so a .1 uf Z5U ceramic capacitor that will vary from .05 to .12 uf will not work very well. Fortunely you can get good resistors and caps easily from Mouser Electronics. AVX makes 63 volt boxed film capacitors that are not very big and have a .2 inch lead spacing. I used the AVX 5% tolerance capacitors and Vishay/Dale 1% tolerance 1/8 watt resistors. I can not fit the circuit into a 9 pin connector shell but it can fit in a small box. Maybe someone skilled with surface mount can make it fit in a 9 pin connector shell.
I do not use a capacitor on the input. I do recommend a resistor in the radio between the discriminator tap and the connector you use. The buffer eliminates the effects of the radio on the filter. Since I also use MTrunk and ETrunk I have a switch to eliminate the filter. The filter is a MFB type to lessen the effect of component tolerances on the filter parameters. By having the 1.0 uf capacitor and the switch at the comparator input the data polarity can easily be inverted.
I hope people find this circuit useful. It is more difficult to build than a Motorola slicer but easier than the other LTR slicers. I am about ready to go from the breadboard to a soldered version in a small box for "Field Testing".
73 Eric