BER Performance of TAPR TNC Modem

(continued from previous page)

and so on. Thus, the probability of receiving N consecutive bits is:

_3 1240

All flags, control bits and data bits must be obtained correctly before receiving a packet. For a maximum 128 length packet through a digipeater, this is 1240 bits. Therefore, the Packet Probability of Reception (PPR) is:

The probability of receiving a packet for a given signal level and packet length can now be calculated. Let's take a signal at 20 dBQ, on the 1200 bps curve. It has a BER of:

For a 1240 length packet, this gives a PPR of:

or 2 packets received in every billion sent. Not very good odds. We can convert from a given PPR to a BER by solving equation 1 for BER. The BER required for a PPR of 10% is:

It corresponds to about -119 dBm, and to 23 dBQ in the Syntor. We can also calculate the level for a PPR of 98% to be:

or about -116 dBm. This corresponds to 25 dBQ in the Syntor. This shows that the fast quieting effect of fm takes the PPR from 10% to 98% for a change of only 3 dB in input signal strength.

"What are my odds of maintaining a packet QS0 for 15 packets?" We must first know the odds of not retrying out for the current packet. [Ed. Note: "Retrying out" is assumed to mean "exceeding the maximum number of retries allowed in the protocol."] To maintain a packet QSO, one packet should be received correctly in every set retry count or it will retry out. Let's call this the Remain Connected Probability (RCP). This is the probability that the packet will be correctly received at least once for the set retry count (L). Since the probability of not receiving a packet is 100% - PPR, the RCP is:

Assuming the other station receives your acknowledgements, calculate the probability of maintaining a QSO for M number of packets. . This is called the QSO probability for M packets (QPM) and is obtained by raising the RCP to M:

If we assume a maximum retry of 10 at -119 dBm, the QP15 (QSO probability for 15 packets), would be:

15 15

or about 2 chances in every 100,000. Again, not very good odds. At -118 dBm, the BER is:

for a PPR of 52.5%, an RCP of 99.94% and a QP15 of 99%. So, in a change of one dB, we go from an extremely small chance of not retrying out to a high probability of sending 15 packets without error. With these formulas, we should be able to calculate the probability of staying connected for any number of packets, at any set retry count and any received signal level.

Since it is easy to send unconnected packets and the TNC can receive and transmit simultaneously, the calculations shown were checked by keying the HP8640B pulse input with the transmitter line from the TNC. This allowed the TNC to receive its own unconnected packets in the C0NVERS monitor all mode. The generator level was set at -119 dBrn. Only one data character was sent per packet so the packet length was 168 bits (8 Flag + 112 Address + 8 Control +'8 PID + 8 Data + 16 FCS + 8 Flags). This gives a PPR of:

One hundred packets were sent with 66 being received for a measured PPR of 66%.

### Conclusions

We now know that at the 20 dB quieting level of an fm receiver, the TAPR TNC has an extremely small chance of receiving a packet. At 23 dBQ the packet probability of reception for a 128 byte packet is 10% and at the 25 dBQ level the PPR is 98%. Therefore, useful sensitivity of the TAPR TNC is about 23 dBQ. Above 25 dBQ a packet QSO has a high reliability. As a general rule of thumb, the TAPR TNC will require a stronger signal than a voice contact since 20 dBQ is normally defined as usable voice sensitivity. The TAPR board requires about 23 dBQ to print. The dBQ of a station being received is easily seen by measuring the voltage out of your receiver audio with no signal on channel. The audio level is then measured with the station not modulating, taking that ratio and expressing it in dB volts. For example, if the output of your receiver is one volt on noise and 0.05 volts with the unmodulated station:

or 26 dBQ. By measuring the dBQ of the packet station you are trying to contact, you should get a good idea if you will be able to carry on a packet QSO. (continued on page 7)

My last column asked for information about silver-bearing solder sources. One such solder is Ersin "Multicore" SN62 (2% silver), which K2LGJ informs me is available from Newark Electronics at a price around \$30 for a 1-lb spool. W0GNV writes that this Ersin product and Kester 3% silver solder, both with flux core, are also available from Jensen Tools, Inc., 7815 S. 46th Street, Phoenix, AZ 85040. If a full lb spool is not required, KA8QKY mentions that small rolls of 2% silver solder, with separate flux, are normally sold in hardware stores. They are also available as kit 53013 from the manufacturer,Oatey Co., 4700 W. 160th Street, Cleveland, OH 44135. Everyone will now be able to use the proper silver-bearing solder for chip components at VHF+ frequencies.

### New Low-Noise GaAs FETs

A pair of new low-noise GaAs FETs to consider for the 144-1296 MHz bands are the Mitsubishi MGF-1202, a single-gate device utilizing the MGF-1402 chip in a lower cost micro-X package, and the MGF-1100, a dual-gate device. Both are in the under-\$10 class and available from Applied Invention, R.D. 2, Rte. 21, Hillsdale, NY 12529. Some preliminary measurements, at 432 MHz, indicate that the MGF-1100 should have an even lower noise figure than the 3SK97, another well known dual-gate GaAs FET in the under-\$10 class, and does not appear to have the self-oscillation problems which often destroy a '97 device (my experience being the loss of five of six devices of that type!). Thus, we now have more candidate parts for use in the 902MHz range when that band is authorized.

### 902-MHz Relay

Other potential 902-MHz parts recently tested by this writer include the Omron G4Y UHF relay. This relatively small (1 in. L X 0.83 in. W X 0.41 in. H) electro-mechanical unit is designed for microstrip mounting and is useful to beyond 900 MHs. The particular relay tested, model G4Y-152P-DCI2, has a list price of about \$5 and is available from Key Components, 365 Karlough Rd., Bohemia, NY 11716. While the 15-W maximum rf power rating of this relay is not comparable to that of a much larger coaxial unit, neither is the price nor the coil power (less than 1/2 W) required for switching. The relay has very impressive shock and vibration ratings, which, along with the power rating and size, point toward use in mobile VHF+ applications. The relay has its pair of dc coil terminals placed along the opposite edge of the same package surface from which the three rf contact pins protrude. At least one grounding pin is provided adjacent to each of the rf contact pins and each of these ground pins should be used, as should a continuation of the ground plane on the microstrip side of the board, with many relatively—thick ground jumpers between

*16 Riviera Drive, Latham, NY 12110 QEX August 1983 Page 5, QEX, August 1983, published by The American Radio Relay League, Inc.

the ground planes on both sides and adjacent to the relay case. I tested a total of four units on 50-ohm microstrip at 220 and 910 MHz using 1/16 inch G-10 pc board and found the following average insertion loss, isolation, and VSWR:

Frequency Insertion Loss Isolation VSWR

A separate metallic plate (not tested) is available for increasing the isolation when this relay is mounted on a single-sided pc board. One final note, since my test microstrip boards were quickly "hand cut" and probably had unaccounted-for discontinuities and the like, I would expect that a properly laid-out microstrip transceiver would obtain even lower insertion loss and higher isolation from these relays.

With the addition of a stable L0, a pair of double-balanced mixers and a linear transmit amplifier, the above relay and FETs provide the basis for a 902-MHz ssb transverter. All we need now is authorization for use of this band. I have heard a rumor that someone has petitioned the FCC for the addition of this band to the authorized frequency table, and would appreciate receiving information from any reader concerning a 902-MHz Petition for Rulemaking (actual or contemplated). In the meantime, anyone desiring to try 902-MHz can still apply to the FCC for a license in the Experimental Service.

### SWR Meters

There is one piece of test equipment just about every VHF+er owns: an output-power/SWR meter. Having recently built a 220-MHz 40-W ssb transverter, I was looking for a power/SWR meter that could be permanently installed between the transverter output and the input of a subsequent power amplifier. For tune-up purposes, I wanted to monitor the forward and reflected power (and therefore the SWR of the amplifier input), but was hesitant to purchase the best-known meter that does not simultaneously read forward and reverse power. After much testing, I acquired the Daiwa CN540 and CN550 meters, respectively covering 50150 MHz and 144-250 MHz. Each of these power/SWR meters is a small (3 in. square X 3-3/4 in. deep) unit, having a single push-button located on the top of the case. The button selects between the higher-power 200-W forward power and 40-W reflected power indicating ranges, and the lower-power 20-W forward power and 4 W reflected power indicating ranges. Each unit is one of the "crossed-needle" type, with forward and reflected power magnitude being simultaneously indicated on the unit face, and with the point at which the needles cross indicating the SWR. Price range for the CN540 and CN550 is \$80 and \$90, respectively. The units are distributed by MCM Communications, 858 E. Congress Park Dr., Centerville, OH 45459.

(continued on page 6)

Correspondence

(continued from page 2)

One other small item I would like to note is that logic signals tied active high should never be tied directly to +5 V, but rather through a series resistor of 1000-3300 ohms. This resistor acts as a current limiter to prevent noise or voltage spikes from damaging sensitive LSTTL inputs. - Steven E. Margison, WA9DRE, 704 Franklin St., Downers Grove, IL 60515.

### Inductor Stack Revisited

An inquiry recently received questioned the wiring of the inductor stack shown in Fig. 2, p. 11 of the June 1983 issue of QKX. The jumper wiring of the inductor stack for the 88-rnH and 22-mH connections is correct as shown. The internal connections of the two 22-mH windings on each inductor core were not shown because this detail was unnecessary for the external wiring of the filter. However, to anticipate questions and eliminate any possible confusion, it appears advisable that the internal wiring of the inductor stack be shown.

Shown below is a diagram taken from my arti cle published in Radio Communication, April 1983. In Fig. 1(b), you will find the schematic diagram with polarity markings shows the winding connections of a single inductor in a typical stack. You can now verify that the jumper connections shown in Fig. 2 of my (£EX article do in fact produce the expected 88- and 22-mH values as shown. - Ed Wetherhold, W3NQN, 102 Archwood Ave., Annapolis, MD 21401.

Leads are shown In the same order as they leave the potted surface

¡_-Case of potted inductor

Leads are shown In the same order as they leave the potted surface

¡_-Case of potted inductor

. V<i inch dia stud for mounting inductor

Solder terminals a/ \b

Solder terminals a/ \b

D I C Terminal strip

D I C Terminal strip

Fig 1. The internal wiring of the two coils found in the potted (a) and stack (b! inductors, and the external connections required for either the 22 or 88mH inductance vaiues. For series-aiding connection (88mH), connect A to B to give 88mH between C and D. or connect C to D to give B3mH between A and B. For parallel-aiding connection (22mH) connect A to D and B to C to give 22ftiH between A and B or C and D

VHF+ Technology (continued from page 5)

The CN540 and CN550 were tested against a Bird 43 wattmeter with the appropriate slugs and a labtype rf calorimeter with the results shown in the Table.

Unfortunately, both units use UHF-type connectors. Even this manufacturer's type CN720B (a 1.8-150-MHz unit rated to 2 kW) utilizes the same S0-239s. The same manufacturer also makes the CN-630N (140-450 MHz, 20/200 W) with type-N connectors and an N-connector meter CN650 for 1.2-2.5 GHz! While I have not tested the latter three meters, it is extremely interesting to note the frequency range of the CN-650 (which has 2/20 W forward power ranges) and the recent advertisements for a certain foreign-made 1200-MHz mobile transceiver (and a repeater to work through on this band!). Note that neither the Daiwa meters nor the Bird model 43 presently offer a power metering capability for more than 1 kW in the 220-MHz range and above which may become necessary if the proposed 1500-W output limit goes into effect.

### Ftt I-F Unit

The 1983 Spring Sprints, the June VHF QSO Party and the spring round of VHF+ conferences and conventions will be history. I do not want to prophesy as to what new VHF+ technology or operating advances have occurred at these events, but I will include anything of interest reported to or observed by me in my next column. I will particularly be interested in activity levels on

220 MHz (traditionally lower than either 144 or 432 MHz because of less variety of commercially available equipment), any impact on 1296 or 2304 MHz on account of recent articles in and other amateur magazines and especially on 10 and 24 GHz.

With respect to the latter bands, a number of VHF+ers have changed to a 10.7-MHz i-f for their FM equipment. It is cheaper to build a 10.7-MHz i-f receiver, relative to a 30- to 10.7-MHz or a

110- to 10.7-MHz superhet i-f receiver, if several must be built (one for each station in each section — remember the rules). A solution is based on a small 10.7 fm i-f unit available from MTS Electronics, P. 0. Box 1164, Round Rock, TX 78664. The MTS EK101 kit is priced at under \$10. It has a measured sensitivity (12 dB SINAD) of 25 uV for 75-kHz-deviation signals and an audio output of 400 mV rms. Full usable sensitivity is achieved by adding a good preamplifier, which is necessary with a Gunnplexer • or Mitsubishi F0-UP11KF receiver (the latter unit to be covered in greater detail in a future column) for the 10-GHz band.

24-GHz Gunnplexer?

The 24-GHz band is still a problem. M/A-C0M had advertised and demonstrated a 24-GHz Gunnplexer ® at Dayton in 1981 which was "soon to be" available. I have heard nothing further about a 24-GHz Gunnplexer in the two years since. Does anyone have further info? I'm sure all VHF+ers would like to know the latest news about this 24GHz equipment.

BER Performance of TAPR TNC Modem (continued from page 4)

### TAPR or other modems with fm receivers.

This data also raises questions such as can the TAPR TNC sensitivity be improved? Would alternate modulation techniques provide even greater improvement and would it justify changing the modem standard? Experiments will have to be made to answer these questions. I am interested in hearing from anyone who does on-the-air tests or has done bench tests similar to these on the

[1] David Borden, K8MM0, and Paul Rinaldo, W4RI, "The Making of an Amateur Packet Radio Network," qST, October 1981, p. 28-30.

[2] Tucson Amateur Packet Radio, P. 0. Box 22888, Tucson, AZ 85734

RX DATA

-125

10

f 2

TAPR TNC 1200 BPS

---

\

MOTOROLA SYNT0R 20 dRq = -121 dBm

r

\

\

S

\

- —

\

\

\

\

\

20

IH

\

r

\

\

-121

-115

-121

-115

 TAPR TNC 300 BPS MOTOROLA SYNT0R 20 dBq u -121 dBm \ \ \ \ \ 7,0 1B% \ \
0 0