That extend to the Connecten of DX-Clusters, a Convers-round or a Mailbox with an ASCII-terminal at present, transfer-speeds would use mostly from. However for many possible new applications, entries are too slow, also 9600 bauds about the factor of 10.
Examples for such applications here are Mailboxes with HTML-format, digital language-transfer or also real-time picture-transfer with ISDN-Quality, named. The construction of Link paths, that suffices these requests, already strides ahead. What is still fully missing until now, is efficient user-accesses.
The Tx/Rx switch-time should be smaller than 1ms in order to give not away the advantages of the high transfer-speed through long TX-Delay times again.
If comparable hinterlands should be reached entries with 9600 bauds, the S/N must be, because of the bigger width of the receiver, approximately about 10dB higher. This can be reached by increase of the transmission-power about 10dB or through the application of scanners.
For the application with unfavorable location-circumstances and at Digipeaters, we recommend a final stage with approximately 20W output-power for the moment. First experiences show that is usually also sufficient more inferior power. Therefore, we introduce also a smaller final stage with 2-7W output-power here.
The Transceiver should suffice following requests:
|Baud rate||76,8 to 115,2 kBauds|
|Modulation-width||100Hz - 80kHz|
|Channels||1, switchable for Digi or User|
|HF-output-power||20W, (perhaps reducible on 2W)|
It turned out with the prototypes that the internal Reset-logic only had the desired effect on very deep tension-being-breaks. The PIC-Processor works reliably until under 3V, however the PLL-IC loses the programming. In order to prevent that the transmitter-control is driven with wrong programming then, an external Reset-switch, that responds certainly also with low tension-being-breaks and introduces the new-programmingg of the PLL, was integrated.
With a Reset, four frequency-telegrams are sent by the processor with four different chip-addresses over the I2C-Bus. Hereby, all combinations are covered from RX or TX and User or Digi. With a Jumper at the Adress-Select-Pin of the respective SDA3302, his chip-address is put in and so that selected the matching frequency-telegram.
At the VCO is connected a buffer-stage BF960, with a Dual-Gate Mosfet. This became chosen because of the very low repercussions from output to the input (S|12|^2 from < -40 dB) and provide a good decoupling with it to the oscillator. A power of 3dBm is available at the output of the BF960.
The increase of the frequency-stability of the oscillator with load-alterations at the output still follows BFR90 two more buffer-stages with ever one. Each stage has a reinforcement of 11dB. In order to reach an output-power of 10dBm and another decoupling, two 7dB pads are still connected in series the transistor-stages. These provide the transistor-stages a good adaptation completely additionally as well.
In order to reach a good decoupling of the tension-care for the single stages, 3 tension-regulators find application. A 78L05 for the PLL and the processor and per a 78L09 for oscillator and first buffers and for the two BFR90.
As important for low phase-flush of the oscillator, an additional filtration of the care-tension has itself (R10, C22, C7, C29) proved which sinks the intoxication-side-ribbons in 100kHz spaces of the bearer about more 20dB.
As output-terminal, a BNC-connector can be used. One installs C13 between method-pin of the socket with the circuit board "externally" directly then. A SMC or SMA-Beech, that is soldered directly on the leader-track, is also possible for example - electrically sure favorable, but these sockets are not so current.
As next, the application of the transmitter definitely turns JP1 with Jumper for User or Digipeater. For the operation with the User (434,900 MHzes) must be pulled the Jumper. The care-tension, on which the programming of the PLL-IC takes place through the PIC, is positioned now first. With faulty programming, LED D2 shines. If everything is in order, it flashes only shortly and stays away then.
At TP2, the phase-comparison-frequency of the PLL is to be done as TTL rectangle-signal. With Trimmer C24 is balanced the frequency on 12,5kHz. On no account, one should try to measure the frequency of the 6,4 MHz quartz directly. This swings would be detuned also with a good touch-head too much into series-resonance and its vibration-frequency.
The avc voltage of the PLL is available at TP1. When the slow spinning the Sky-Trimmers C10 should have gotten ready these for 4 to 5V. The Trimmer on this occasion virtually stands in middle-position. If the avc voltage lies about 9V, so the frequency of the VCO, that Trimmer must, become out-turn too deeply further. If she/it lies over 0V barely, so the capacity of the Trimmers must be increased. The tuneup of the transmitter is already finished with it.
The VCO-Frequenz divided through the PLL-IC can be removed at TP3. With gerasteter PLL, 12,5kHz are also to be done here. The test-point is not necessary to the tuneup, with a possibly necessary debugging can be checked here however also without 500 MHz counters, whether and where the VCO swings. The division-factor of the PLL with User-operation amounts to 34792.
In order to reach a mould of the output-spectrum of at least 40dB at the channel-borders, the lifting may not be bigger than 25-30kHz with a baud rate of 76,8kBaud. In order to avoid disturbances in the neighbour-channel, the modulation-tension should be put in at the modem-output smaller than 500mVss.
With our pattern-constructions, we have measured following collector - as well as Drain-tensions in the method: T1 6V, T2 8V, T3 5,2V, T4 4,2V. The output-power lay with all based prototypes over 10mW barely.
From reasons of the übersteuerung stability, we wanted to put in a Fernsehtuner-IC for mixers and oscillator. That in  would use TDA5030 is manufactured no longer of Philips meanwhile. First election would have been the TDA5630. Unfortunately, this IC is not beschaffbar at this time, so that is used a TDA5331T. This has a mixer-input with a Impedanz close to 50Ohm and a low-ohm-y SAW-Filter driver. Both fits into the receiver-concept well.
A bigger problem tossed up the procurement of a suitable filter for the first ZF with approximately 200kHz width. We have a SAW-Filter with 41.7MHz in the middle-frequencies after longer search and a width of 300kHz found. Unfortunately, this is filter relatively expensive. Differently to that in [2, 76800 bauds in the 70cm-Band] and  published circuits is now put in the clearly more affordable filter TFS80B with 80MHz in the middle-frequencies. Through the flexible concept of the receiver, the necessary adaptations last in borders.
As frequency for the second ZF doesn't come because of the required big width 455kHz in consideration. With 10,7MHz, ceramics-filters stand for radio-receiver by the disposal, that a matching width has and is sufficient their group-term for data-application. As IC for the second ZF, one uses the SA626 of Philips. It is intended for the direct connection of ceramics-filtering with 330Ohm Impedanzes and is contained as well as oscillator, mixers also as limitter-transistor, FM-Demodulator and an extremely fast RSSI (Received signal strength indicator - an output for a logarithmic field-strength-ad) -output. This is used with a Komparator with adjustable threshold also as DCD together.
Like also in the transmitter is looked after the single modules with own tension-regulators in order to avoid uncoupling.
The Demodulatorkreis Fi5 is balanced without at the beginning of-signal with an oscilloscope at the NF-output on maximum tension and best " symmetry " of the flush. Fi1 and L2 become with weak at the beginning of-signal (for example from corresponding gejumperten transmitter-control without antenna) on maximum tension at the RSSI-output put in. Possibly, the own-flush of the receiver for this also already suffices. Without at the beginning of-signal, 350 to 450mV should lie flat after taken place tuneup on the RSSI-output - the copy-scatters with the SA626 is rather high here. One still balances completing R23 so that exactly " low " is without at the beginning of-signal of the DCD-output.
For the application in Duplex-systems without antenna-switches, one should equip the transmission-stage low-passport-filter, since it contributes to the suppression of upper-waves. The diode D1 becomes through a matching condenser (approximately 1nF) replaces and the diodes D2 and D3, the condensers C17 to C20 as well as R6, R7, L1 and C9 can remain unequipped.
Since hybrid-modules have a bad efficiency, one should keep an eye on a good refrigeration. One must heed with the dimension of a cooler for a Duplex-Digipeater furthermore, that these Digis have a transmitting-time of near 100%. The application of heat-conductor-paste only recommends itself, if the module possesses mass-connections. Lead the mass against it across the cool-flange, this is direct, therefore without heat-conductor-paste, to connect with the circuit board-mass.
Since drills are not possible on the circuit board through the mechanical construction, the few conventional prefabricated parts must (T1, IC2, L2, L3 and R9) on the prefabricated part-side is soldered. This first appears somewhat strange, no big problem represents with the few prefabricated parts however. C15 and R8 don't become put on the circuit board but externally-supported. One installs C15 between the method-pin of the antenna-socket and the circuit board. R8 serves the drainage of statics on the antenna-management and is soldered by the method-pin of the antenna-socket to the tinplate-casing.
For the spool L1 is wound 9 meanders 1mm silver-wires about the shaft of a 5mm driller. Afterwards the spool is being pulled apart carefully now generally that the meanders touch themselves at no position. The labeling of the BFR91 points in direction of the equipment-side. The application of heat-conductor-paste between module and cool-bodies doesn't recommend itself, since the module is connected only over the cool-flange with mass.
For Fullduplex-operation, the modules should preferably be connected with double shielded coax-cable together, in order to avoid Übersprechen. RG223 or RG214 offer themselves here.
To the completion of the system on the User-side, the connection is still missing between Transceiver and PC - the modem as well as the TNC. The PR-standard-equipment - a TNC2 with a TF in the Hostmode - is already desperately overtaxed also with at most possible tuneup with a throughput >19k2. Plugin-modems like PAR96 or PICPAR behave similarly. At this time probably only relatively expensive hardware-solutions, like RMNC or TNC3, in question, come here. That offers a base for a software-solution modem-adapter for the EPP described in . It is probably available in easily changed form from March '98 with Baycom.
 Liebeck, M., DL2ZBN; Kurpiers, A., DL8AAU: High-Speed Packet-Radio - a transceiver concept for the 70cm-Band, Skriptum 13. International Packet-Radio Convention, Darmstadt 1997, S. 35-45.
 Liebeck, M., DL2ZBN; Kurpiers, A., DL8AAU: High-Speed Packet-Radio - modules for the 70cm-Band, ADACOM magazine 10, S. 7-22, 1997.
 Rech, W. -H., DF9IC, et al,: A modem-adapter for the EPP, Skriptum 13. International Packet-Radio Convention, Darmstadt 1997, S. 46-51.
Copyright © 1997
This page was created by DL2ZBN, Sunday, 28. December 1997
Most recent revision Sunday, 22. February 1998
Translated by EB7CJO. Latest revision 10 July 1998