SUPER-AMP Mastvorverstarker fiir 2 m, 70 cm, 23 cm und 13 em 1 DX-Ertolge dem Zutall iberiassen? Wenn es um aie Aufnahme schwacher DX-Signale geht, ist der Vorverstirker der wichtigsie Baustoin in der Emptangsantage. Unsere SUPER-AMP Vorverstarker sind zuverlassige und unentbehriiche Werkzeuge fir den ertolgreichen UKW-Amateur er DX-Erfolg ist kalkulerbar geworden! 1 Warum iiberhaupt ein Vorverstirker? Addieren Sie die uavermeidichen Verluste des Koaxkabels zu der Rauscheahl Ihres Transceivers. Das Ergebnis ist oft iberraschend schlecht: is 2u 648 aut 2-m und 8 4B auf 70-cm sind durchaus ,normal". Im 23-cm Band wird die 10-dB-Marke oft dberschritten. Der Einsatz unsarer Mastvorverstirker macht diese Rechnung ibertlissig. Mach Einsckalten des Vorverstirkers wird die Gesamt-Rausch- zah! Ihrer Anlage schlagartig aut ca. 1 dB reduziert, eine Verbes- serung, die deutiich horbar it Der echte Signal-Gewinn betragt also viele dB ~ zuviel, um daraut 2u verzichten! 1 Warum ein SUPER-AMP von SSB-Electronic? Unsere langjahrige Erfahrung in Konstruktion und Autbau von Mastvorverstérkern stand Pate bei der Konzeption der SUPER- AMP Serio, Modernste MeGgerate und verbesserte MeSvertahren standen 2ur Vertugung um 2u entwickeln und zu optimueren. ‘Das Ergebnis kann sich sehen lassen, Die Verstarker besitzen nicht nur eine sehr gute Rauschzahi sondern zeichnen sich auch durch eine hohe Betriebssicherneit lund eine problemiose Handhabung aus. Auch Conteststatianen vertraven auf die Leistung und Zuverléssigkeit der SUPER-AMP Verstarker! Noch Fragen zur Technik? GaAsFots, Koaxrolais, HFVOX, Fernspeisung, Direktspeisung, N- Buchsen, Tetlon-Leiterplatte (SP-23 u. SP-13) einstellbare Durchgangsverstarkung (SP-2, SP-70), fankensteile Banctiter, 2-schalige Gehausokonstruktion, Edelstahlschrauben usw. sind ‘nur einige markante Stichworte, um die Technik der SUPER- AMP 2u beschreiven, Austuhriche Informationen uber diese Gerate senden wir Innen gerne kostenlos 2u ~ oder bestellen Sie einfach unseren um: fangreichen Gesamt-Katzlog: Componenten & Systeme, den Sie durch Einsendung von OM 5. in Briefmarken erhalten Leistungstahige Fernspeiseweichen mit elektronischer Verzégerung fir PA's OCW 158 tr 2-m, 70-cm DM 162, DCW 15-SHE ftir 23.u, 13cm DM 199, SUPER-AMI UTA ETT SP-2 SP-70 SP-23 SP-13 Frequenzbereich 144 - 146, 430-440 1250-1300 __2300-2400 Miz Rauschzahl, typ. 08 og 09 area 2: OB Verstarkung. typ. 20 20 20 25 ab ichaltleistung, HF - VOX 100 10 10 Watt Schaitleistung mit DCW 15 500 100 50 Watt Betriebsspannung 136 198 «13.8 Voit Stromavinahme, ci 350 350, 400 120 mA ‘Mastdurchmesser, max. 58 BB 58 mn Prei DM 374, DM 374- DM 55. DM 395 (SSB: Ingenieurburo fur Nachrichtentechnik Panzermacherstr. 5 Tel.: 02371/6454 58664 Iseriohn Fax: 02371/67593Front Page/Titel: TF/GADHF Expedition |.to.1: G4AODA, GADHF, GOULF, GAVXE (See p. 98) DUBUS-Informationen erscheinen vierteljihr- lichim Marz, Juni, Septemberund Dezember. Der Abonnementspreis betrigt in der BRD DM 29,- . Durch Uberweisung des Abonnementspreises aut unser Postgiro-Konto werden Sie fir ein Jahr Bezicher der DUBUS. DUBUS-Information is published quartely in March, June, September and December. For sub- scription please contact your national repre= sentative or transfer the appropriate annual rate, which can be seen below, by international postal money order to DUBUS Postgiro account. Also EURO-Cheques are acceptable alternatively, if they are issued in DM. No other cheques can be accepted. Abo-Verwaltung/Circulation: Norbert Goettsche, DL8LAQ, Tulpenweg 3, D-24568 Kaltenkirchen, & (++49)4191/88138 Abo-Konto/Subser.-Account: Postgiro Hamburg, Kto.-Nr. 266292202, BLZ 20010020 Herausgeber/Publisher: DUBUS Verlag GbR, Postfach 500368, D-22703 Hamburg 50, FRG Chefredakteur/Editor-in-Chief: Rainer Bertelsmeier, DJOBV, Gluecksburger Str. 20, D-22769 Hamburg, Geschdftsfuhrer/ Manager: Joachim Kraft, DL8HCZ, Gruetzmuehlenweg 23, D-22339 Hamburg, Urheberrechte/Copyright: Allle Beitriige unterliegen dem Urheberrecht. Ab- druck, ganz oder teilweise, sowie kommerzielle Nutzung nur mit schriftlicher Genehmigung der Verfasser oder des Verlages. All articles are protected by Copyright. Repro duction or commercial use only with written per mission from the author, International Representatives Austria: Reinhilde Riml, OEY TY, Markistr. 35, A-6971 Hard Australia: Doug McArthur, VK3UM, 30 Rolloway Rise, Chimside, Victoria 3116 Belgium: Luc Selis, ON4ASL, 13 Heikestrat, B-3190 Boortmeerbeek Bulgaria: C. Mintehev, LZIDP, Ul Boruigrad 8 AP 10, 6004 Stara Zagora Cazch Republik: Zdeno Sterbacck, OK2PZW, Dvorska 16, 67801 Blansko Denmark: Finland; ‘Gert Rahbek-Udengaard, OZIFKZ, Risvangen 2, DK-8362 Horning, Irja-Lisa Mansnerus, Skillnadsgatan 37, 22100 Matickamn AE. Latham, EIGAS, 226 Belgard Heights, Dublin 4 Patrick Magnin, F6HYE, Marcorens, F-74140 Ballaison “Tony Fumagali, I2FUM, Via Masia 34, 1-22100 Como “Toshihiko Takamizawa, JELAAH, Parkiown 21-502 946-16, Kitahassaku-cho, Midori-ku, Yokohama 226 Carlo Diedert, LXICD, 15 Op det Hobuch, L-$832 Fentange Joe Apap, SHE, 34 Msida Seafront, Msida Netherlands: Veron Verkoop Burcau, P.O. Box 1166, NL-6801 BD Arnhem Norway: Jan-Martin Noding, LASAK, Voidlien 39 B, N-4623 Kristiaansand S Portugal: Tiago J. Frederico, CTIWW, Nagoselo do Douro, 5130 S. Joao Pesqueira Andrezj Wiodarczyk, SPX, PO, Box 67, 76-270 Ustka Sli Iulius, YOUIS, Ste. Iasi 1, 1900 Timisoara IR. Daglio Accunzi, EA2LU, Manual Irribarren 2-5 D, E-31006 Pamplona Croatia& Hungary: Miba Habic, SS1FB, Vodnikova 8, 61000 Lubljona Sweden: Lars Petterson, SMSIVE, Smegarden, Talby, S-71594 Odemsbacken Switzerland; Ferdinand Stimpfli, HB9MIO, Exzelkofenstr.90, CH-3308 Grafentied United Kingdom: Roger Blackwell, G4PMK, 57 Station Road, Scholes, Leeds LS15 4BY usa: Russel L. Miller, NTART, 12041 S.W. Peninsula Drive, Crooked River Ranch, OR 97760 Other Countries: Contact DUBUS Verlag, Subscription Rates: Europe DM 29,-; Overseas Air Mail DM 40,- (USS 25.00) DUBUS 4/1993 3Editorial ...... ‘Technical Reports... sass a sa HEMT Preamp for 24GHz by JEIAAH ..... « = * seep see 6 Power Amplifier for 24GHz by JELAAH « 3 * oe . sees oe Modern 10GHz Transverier System (Part II) by G3WDG . HEMT Preamps for 23cm by DJ9BV Baugruppen fur 24GHz by DB6NT Microwave News. . EME News Tropo News ... MS News... 6m News . ES News ... Aurora News ....---00+55 News & Comments . Beacon List... Top List. be published in DUBUS 1/1994 - -. Will be published in DUBUS 1/1994 Redaktion/Editorial Staff Technical Reports, Microwave Parts & Kitchens Comer: Rainer Bertelsmeier, DJOBV, Gluecksburger Str. 20, D-22769 Hamburg Hints& Kinks: Gerold Bichle, DESGX, Elberfelder Ste. 12, D-1055S Berlin Computer News: Hellmuth Fischer, DF7VX, Am Holtsaut 2, D-33178 Borchen Microwave News: John Sorensen, OZIPU, Rosenvej 49, DK-9300 Sarby, Denmark EME-News: Bernd Wille, DL7APV, Amsterdamer Sir. 9, D-13347 Berlin, D.22339 Hamburg Tropo News & 6 m News & Beacon List: Joachim Kraft, DLSHCZ, Gruetemueblens Meteor Scatter News & Top Norbert Gattsche, DLSLAQ, Tulpenweg 3, D-24568 Kaltenkirchen ‘Aurora News & ES News: Edmund Ramm, DK3UZ, Anderheitsallee 24, D-22175 Hamburg FAL News: Marco Pavia, IK2CFR, Via Sala 3, 1-22060 Vigano, Italy News & Comments: Klaus Tiedemann, DLAEBY, Halskestr 35, D-12167 Berlin Electronic Mail & AX25 & BBS een dk3uz@[Link] or kiaus@[Link] AX25 Co UN ero oa Ree ao Peer PTI ee eA un RC UAC DUBUS 4/1993 4Editorial Liebe Leser! Diese Ausgabe der DUBUS erscheint verspatet Leider hat mich ein Unfall, dessen Folgen inzwi- schen behoben sind, dazu gezwungen, die Endre- daktion dieser Ausgabe zu verschieben. Diese Ausgabe ist wieder voll mit technischen Artikeln. Daher muBten cinige Rubriken wie z.B. die Bakenliste auf die nfichste Ausgabe verschoben, werden. Wir hoffen auf Ihr Verstiindnis, daB regel- méfige Rubriken aktuellen Artikeln Platz machen, missen. Damit wird letztlich die international be- kannte, hohe Qualitit unserer Zeitschrift gewahr- leistet. ‘Nun noch einige Anderungen: Der Bezugspreis fir das Inland ist auf DM 29,- angehoben worden. Weiterhin wird fir saumige Zahler das Heft 1 des Folgejahrgangs nicht mehr wie bisher gratis gelie- fer. Achten Sie daher bitte auf eine piinkiliche Uberweisung. Die Auflage ist im letzten Jahr wieder um ca. 10% gestiegen, was wirals ein positives Zeichen fiir die weltweite Geltung der DUBUS ansehen. Viel Glick fiir 1994 und 73 Rainer, DJ9BV Dear Reader This issue of DUBUS appears about 1 month late. We apologise for this inconvenience. An accident three months ago and a subsequent surgical operation forced me to postpone the final editing of this issue. We will be back in the normal time schedule for the next issue. Because of the large amount of technical articles in this issue some periodicals like the beacon list have to be delayed for the next issue. We hope for your understanding, because advanced technical arti- cles about VHF and microwave amateur radio are the backbone of our intemationally respected and world-wide distributed ma Some changes: From next year on we are not longer able to deliver a gratis issue No.1 for those who don’t renew their subscription. Therefore don’t forget to renew at your distributor. Last year the circulation of DUBUS has increased by about 10 percent especially in countries outside of Germany. We take this as a positive feedback for the quality of our work and the ever increasing, world-wide reputation of DUBUS which is dedi- cated to the publication of advanced techniques for narrow band DX-operationin VHF and microwave amateur radio, Happy new Year 1994 and 73 Rainer, DJ9BV DUBUS 4/1993Technical Reports: HEMT Preamp for 24GHz by Toshihiko Takamizawa, JEIAAH HEMT Preamp for 24GHz Toshihiko Takamizawa, JETAAH Parktown 21-502, 946-16 Kitahassaku, Midori-ku, Yokohama 226 Japan Abstract: A two stage preamp with MGF4917C HEMTS achieves more than 174B gain and less than 2.5dB noise figure on 24GHz. It uses SMA- connectors for input and output. The problems of the selection of suitable connectors and of suitable monolayer chip capacitors for DC-separation are discussed with comparative measurements. Kurzfassung: Der 2-stufige Vorverstirker ist mit, MGF4917C HEMT’s bestiickt und erreicht eine Rauschzahl von weniger als 2,5dB bei mehr als Bild/Figure 1: Overall View on 24GHz LNA DUBUS 4/1993Technical Reports: HEMT Preamp for 24GHz by Toshihiko Takamizawa, JETAAH Boay Cov 3 a Sicle plare Bild/Figure 2: Case 174B Verstiirkung. Eingang und Ausgang sind mit SMA-Buchsen ausgeristet. Die Probleme eines méglichst verlustarmen Ubergangs von SMA auf Microstrip und der verlustarmen Kopplung dber einschichtige Chip-Kondensatoren werden im Zu- sammenhang mit Mefwerten diskutiert. 1. Introduction I have developed an LNA on 24GHz using a MGF4917C. The project was inspired by one pro- fessional article on a conference held in Finland several months ago. This March when | visited UK, I met G4KNZ and he passed me this article titled "22GHz Band downconverters for DBS’ by Thom- son Consumer Electronics R&D France. HB9MIN also has this article and asked Thomson what de- vice has been used. Thomson people used the MGF4317C (another package of MGF4917C) and itisa2-stage configuration. The original workused 0.4mm thickness PTFE board and K connector. Designed frequency is 22GHz band, which is new DBS frequency. This experimental 22GHz LNA is claimed to perform with 22dB of gain and 1.7 10 1.84B of NF. I was very encouraged by the article and started the project last April. 1. Einfuhrung Das hier vorgestellte Projekt eines LNA’s fir 24GHz in koaxialer Technik wurde durch einen Artikel von dem Forschungslabor von Thomson CSF inspiriert. G4KNZ hat mich dankenswerter- weise darauf aufmerksam gemacht. Erich, HB9MIN, hat bei Thomson recherchiert und erfah- ren, daB in diesem LNA MGF4317C HEMT’s von Mitsubishi verwendet wurden. Die MGF4917C_ sind 4317C mit kurzen Anschliissen. Der Thomson LNA arbeitet auf 22GHz und hat eine angegebene Rauschzah! von 1,7 bis 1,8dB bei 22dB Verstir- kung. 2. Design The board used is 0.25mm thickness Cu-clad 250GT with 18um copper clad from Arlon. Size of the PCB (Figure 3) is 19mmx 18mm(WxL) which is just the same length of the 22GHz LNA. | used a radial stub for bias feed line to decouple RF energy. An interstage coupling capacitor used is a mono layer micro chip capacitor from Johanson dielectric capacitor type 101UDDQORSC94 0.5x. 0.5mm size. This capacitor improves VSWR and insertion loss of mainline tremendously (Figure 2). When using usual 1005 or 1608 size multi layer DUBUS 4/1993Technical Reports: HEMT Preamp for 24GHz by Toshihiko Takamizawa, JEIAAH ceramic chip capacitor, an intemal parasitic reac- tance should be taken in account when tune the matching circuit. 2. Design Das Material der Platine ist 0,25mm dickes PTFE- Substrat 250GT von ARLON. Die GroBe der Pla- tine ist 19x18mm (Bild 3). Die Koppelkondensa toren sind Einschicht-Kondensatoren von JO- HANSON, Typ 101UDDQORSC94 mit einer GrdBe von 0,5x0,5mm. Diese Kondensatoren ha- ben niedrige Einfigungsdimpfung und niedriges ‘VSWR auf 24GHz. Normale Kondensatoren sind nicht geeignet. 3. Box From this project on, I have changed the housing case for the amplifier. It is machine milled alumi- num block construction case. See fig. 2 for the detail. This type of case is easy to build when mounting the PCB inside of the case. Two side plates where an input and output SMA connectors are mounted can be separated from main body. It is very convenient to change the connector if needed. 3. Gehdause Bild 2 zeigt die Gehausezeichnung. Es i die Buchsen auszuwechseln, da das Gehiiuse getrennte Sei- tenwiinde hat. einfach, 4. Construction 1. First, cut PCB to fit appro- priate case. 2. Make ground contact for source using very thin(0.03 to 0.05mm) copper foil. 3. Make ’U shape’ of Imm leg + Imm bottom + Imm leg by 1.5mm width ribbon. 4, Cut two [Link] slits in par- allel of Imm space at source area of the board. Production version of PCB has via hole at source pad. You do not need this work. 5. Make "U shape’ ribbon upside-down and feed the legs through the board. 6. Bend the legs right angle and make sure the legs are completely flat on the surface of the board 7. Solder the legs with very small amount of solder. If too much, remove solder by desoldering tool or *solderwick’ (TM). 8. Solder chip parts. 9. Mount the PCB with silver paste or bonding glue 10. Make gate and drain wiring by very thin wire. T used one piece of peeled usual wire. It is about 0.1mm diameter. 11. Ground gate and drain terminals to case. [Link] shorten HEMT’s gate, drain and source legs to 0.3mm length. No leg for source is okay because around the package, you can see [Link] metalization area for source which can be soldered 13. Solder HEMT and after that remove ground wire on gate and drain terminals 14, Attach both side plates with appropriate SMA connector. I strongly recommend to use Hirose HRM300-110S or equivalent which has thin(2.2mm) PTFE extension with 3.2mm of ex- Bild/Figure 3: PCB DUBUS 4/1993Technical Reports: HEMT Preamp for 24GHz by Toshihiko Takamizawa, JEIAAH OSE sma SMA HRM300-1108 feHIDO- HS Hse “ 259-5074-00 052.561 ~00 or -O2Z s tk ‘yp com Ty TT ost PF ”. PP IP 9K URE mona lager” eo Johanson Dielectric. | zl {O1UDD@ORSCH ote Vqz DEVICE: MAR44ITC x2 Bild/Figure 4: Circuit Diagram/Schaltung tension length. Usual 4.1 7mm extension SMA con- nector never work well above 20GHz according to ‘experimental work. M/A com type 2052-5674-00 or -02 works well with tenth of dB degradation from Hirose. Cut and shorten a center conductor of SMA to 0.5 to 0.7mm. DO NOT SOLDER CEN- TER CONDUCTOR OF THE SMA. It is quite enough electrical connection to fit the center con- Guctor to the stripline track by light pressure when, attaching the side plate to main body. 15. Tune the LNA. My experience indicates there will be 13 to 14dB of gain on 23 to 24GHz band without tuning stubs. 2 to 2.5mm away from gate is stub position and stub length is about 2mm. It can be applied to both stage. No drain tuning is required. I am using 0.05mm thickness 0.6mm width copper ribbon or gate leg of FET for stub 16. Additional small foil on the bias line will boost the gain about 0.5 to 1.0dB. DUBUS 4/1993 17. There is no decoupling chip capacitor at the drain bias pad area. Before I didit, butthe amplifier did work as an oscillator. Be careful for decoupling, work. 4. Konstruktion 1. Die Platine wird auf MaB geschnitten. 2. Die Durchkontaktierung an der Source mit 0,05mm dicker Kupferfolie ausgeftihrt. 3. Fertige ein U mit Imm Schenkellinge aus einem 1,5mm breitem Cu-Band, 4-7. In die Platine werden dort, wo der FET sitzt, zwei 1.5mm breite Schlitze in Imm Abstand ge- schnitten. Das U-Blech wird mit den Beinen nach unten durchgesteckt, flach auf die Platine gebogen und mit wenig Lotzinn verlétet. (In der Produk- tions-Version der Platine gibtes dort Durchkontak- tierungen und diese Arbeit entfllt).Technical Reports: HEMT Preamp for 24GHz by Toshihiko Takamizawa, JELAAH. Bild/Figure 5: Close View on 24GHz LNA Nose Fqur ard Gain plot of 28GHe ataga LNA using MGFAITCEIAAM Bild/Figure 6: NF Measurement of LNA DUBUS 4/1993, 10Technical Reports: HEMT Preamp for 24GHz by Toshihiko Takamizawa, JEIAAH 8. Alle SMD-Teile werden aufgelétet. 9. Die Platine wird mit Silberpaste cingebaut. 10-11. Die Gate- und Drain-Anschliisse der HEMT’s werden mit diinnem Draht mit den Sour- cen verbunden. 12. Alle Beinchen der HEMT’s werden auf 0,3mm gekiirat. 13, HEMT’s einléten und nachher die Kurz- schluBdriht von Gate und Drain entfemen. 14. Beide Seitenplatten des Gehviuses werden mit den SMA-Buchsen versehen. Empfehlenswert ist Chie A 12-0 ai ai -We 4 18.68 dB REF - .00 2 07-07'93 CH2: B 18-0 a2/ pa ued SIRI 42.00006H2 —CRSR 440.2076H2 STOP +26. 500GHE 24GHz 2stage LNA using MGF4917C Wide band analysis with min, noise setting JELAAH Sia and set by HP6350B, 835944 Marker HP8757A. 2.4/5.7/10.2/24.0CHz ‘HP85027B fe Si2GHe 10.2 GHe HP11664E AF zzde Sd OID Bild/Figure 7: $11 and $21 (Broadband) DUBUS 4/1993 VWTechnical Reports: HEMT Preamp for 24GHz by Toshihiko Takamizawa, JETAAH der Typ HRM300-1018 von Hirose, der cine 3.2mm lange und 2,2mm dicke PTFE-Muffe hat. Als Ersatz kann man auch die M/A-COM type 2052-5674-00 nehmen. Normale SMA-Buchsen_ mit 4,7mm dicker PTFE-Muffe funktionieren nicht als Koax-Microstrip Ubergang. Den Innenleiter der Buchsen auf 0,5 bis 0,7mm kiirzen, Die Innen- leiter werden nicht aufgelotet, sondem haben nur Kontakt durch Pressung! Darauf ist bei der Mon- tage der Seitenplatien zu achten. 15. Ohne Abstimmfihnchen werden 13 bis 14dB Verstirkung erzielt. Durch Anbringen von 2mm Jangen und 0,6mm breiten Abstimmfihnchen ca. | | | STRT 422.005H2 CRSP 24Giz 2stage LNA using MGF4917C CH2s Bo -Me + 17-69 da 10-0 B/ REF - 00 cE 07-07-"93 1724.0206H2 STOP +76-0006Hz JELAAH Gain: 17.648 by HP8350B, 835944 NF: 2,0dB at 26.0GH2 HP8757A Vds : 2.0 HP850273 HP11664E Ids: 10mA Min, noise figure setting Deanna HP3G6C Bild/Figure 8: $11 and $21 from 22 to 26GHz DUBUS 4/1993 12Technical Reports: HEMT Preamp for 24GHz by Toshihiko Takamizawa, JEIAAH. 2,Smm vom Gate entfernt kanndie Verstdrkung auf 16. Kleine Zusatzfihnchen auf der Bias-Leitung ber 174B und die Rauschzahl auf unter 2,5dB konnen die Verstarkung nochmals um 0,5 bis 14B gebracht werden. atheben, HRM303S with 18mm length 50ohm line 6-2-9 Cu-clad 0.25mm thickness 0.6mm width CHit A -He =~ 3.34 dB CH2: Bo -Me = 8.95 di 10.0 dB/ REF - .00 dB 10.0 dB/ REF - — .00 di MILLIMETER UAVE LABDRATORY, | i | | L i L | STRT +2.@0Q0GH2 —CRSR 474.050GH2 STOP +26. 5006Hz Bild/Figure 9: Insertion Loss of HRM303S SMA DUBUS 4/1993 13Technical Report EMT Preamp for 24GHz by Toshihiko Takamizawa, JEIAAH 17. Die Bias-Leitung wird nicht ber Kondensa- + HP346C, EATON7380 for NF measurment toren ‘abgeklatscht’. Das fihrte zu Oszillationen. * HP346C,8970B for NF plot 5. Results Atypical LNA performs as follows: Thave measured the LNA on my bench using the following instruments: . NF: 2.2dB . Gain: 17.5dB ‘+ HP8350B+83594A,8757A,85027B,11664E © Vos: 2v for $11,S21,$12,S22 measurement * Ips: 10mA HRM300-110S 6-2-93 CHin A -Me = 20.86 dB cH: B -Me = - 1.07 dB 10.0 ¢8/ REF - .00 a8 10.@ 68/ REF - -00 dB T wiuckmerer dave LABORATORY =P \e™ a 7 4: STRT 42-0000Giz CRSR +24.05@6Hz STOP +25-S@@GHz Bild/Figure 10: Insertion Loss of HRM300-110S SMA DUBUS 4/1993 14Technical Reports: Power Amplifier for 24GHz by Toshihiko Tokamizawa, JE|AAH 5. Ergebnisse Die Prototypen wurden mit folgender Ausriistung, gemessen: © HP8350B+83594A,8757A,85027B,11664E fiir die Messung von S11,$21,S12,S22 © HP346C,EATON7320 fiir Rauschzahlmes- sung * HP346C, HP8970B fir Plot der Rauschzahl Ein typischer LNA hat folgende Werte: . NF: 2.24B * Verstirkung: 17.5dB . Ups: 2v : Ips: 10mA 6, Conclusion This project was very successful indeed. Even though comparing Thomson’s work, difference of NF is about 0.5dB and it is well catching up the work. They used K connector and my LNA isusing SMA. [Link] change is well acceptable. I strongly wish to thank to G4KNZ and GBKMH for their introduction of the Thomson article. 6. SchluB Ich glaube, daB das Projekt ein Erfolg w: man die Ergebnisse von Thomson verg! die auf 24GHzerreichten Werte fiirdie Rauschzahl nur 0,5dB schlechter, was auch an den bei Thom- son verwendeten besseren K-Konnektoren liegen kann. Ich méchte G4KNZ und G8KMH herzlich fiir die Information ber den Thomson Artikel danken. Power Amplifier for 24GHz Toshihiko Takamizawa, JETAAH 1. Introduction I have developed a real HPA on 24GHz using Fujitsu FLR devices. With great success of LNA, next step to overcome was a real HPA. The HPA employs exactly the same construction of LNA. Difference is only drain bias circuitry using a fer- Tite bead instead of a 1000hm chip resistor. First I have developed a 2 stage HPA using FLRO16FH and FLRO26FH, then a single stage HPA using FLROS6FH was developed. [have tried FLRO26 + FLR056 combination at first. However it allowed only 8-94B of gainand about +18dBm of Podriven by +10dBm input. I wonder if there is more driving power needed for FLR0S6. Then I have decided to develop those two HPAs. 1. Einfihrung Um einen Leistungsverstirker fiir 24GHz (HPA) zu konstruieren, habe ich exakt die gleiche Platine und das gleiche Gehiiuse verwendet. Die Transisto- ren sind allerdings Leistungs-GaAs-FET’s Fujitsu FLRO16FH und FLRO26FH. Nur die Stromversor- gung wurde geiindert, indem man statt des 100 Ohm Widerstandes eine Ferritperle einsetzt. Versuche mit einem einstufigen Verstirker be- stiickt mit dem FLROS6FH und einem zweistufi- gen Verstirker mit FLRO26/0S6 blieben ohne Er- folg. Es wurde nur eine Ausgangsleistung von 18dBm bei schlechter Verstirkung erreicht. 2. Design, Case, Construction As mentioned above, Please refer LNA part for these items. 2. Design, Gehduse, Aufbau Wie gesagt, kann man alle Details dem vorigen Artikel ber den LNA entnehmen, DUBUS 4/1993 18Technical Reports: Power Amy ier for 24GHz by Toshi ko Takamizawa, JE1AAH Bild/Figure 1: Overall View on 24GHz HPA 3. Construction Hints Only one thing what has to be concerned about this 2-stage amplifier is to ensure drain bias decou- pling. At first, I do not use a ferrite bead for each drain bias line. Each FET’s drain seems almost directly connected as DC view point which allows only few decoupling value at lower microwave frequency region. It resulted self oscillation!!! I have put a ferrite bead at drain bias line instead of ajumperwire. About a single stage amplifier, there is no coupling capacitor used for gate and drain DC-cut purpose. It provides better VSWR (because straight S0ohm line only) and easy construction. If critical depends on your application, put a mono layer microwave capacitor at usual position. Tam using a floating type Coax-WG transition for operation, it is possible to use. If you use HP K281C type grounded center conductor transition, you need a drain DC-cut capacitor at least. 3. Aufbauhinweise Wichtig fir die Stabilitit ist eine gute Entkopplung der Drain-Zufdhrungen. Eine Ferritperle auf den Drain-Leitungen léste das Problem. Man kann auf die Koppelkondensatoren im Ein- und Ausgang verzichten, wenn man erdfreie SMA-Hohlleiter Uberginge benutzt. Der HP K281C hat allerdings einen geerdeten Stift. Also Vorsicht! 4. Results Using a lot of diagrams, I can show the extremely good performance of the 2 stage HPA. When I was measuring an HPA, some characteristics depends ‘on DC biasing condition were realized which are: + For maximum gain, Vps is about 5V for each device and Vps is almost OV for each device. Since there is one Vps terminal for two devices used, two separate Vps sup- plies would provide better result, For maximum saturated power out put(Psat), Vos is 6.8V. Vo, does not affect a lot on Psa but affects total amplifier gain at lower input level and power added effi- ciency. Total gain goes up with lower VG, at lower input level and it causes more drain current for FLRO16FH (efficiency goes worse). At higher output level(+20dBm), as gain is already com- DUBUS 4/1993, 16Technical Reports: Power Ampiifier for 24GHz by Toshihiko Takamizawa, JEIAAH pressed at this output level, varying Vo, * For higher gain: Lower Vps and higher cur- does not change a lot on total amplifier rent. Setting! For higher Psat and good ef- gain. Vo, does not affect a lot on both total ficiency: High Vps and lower current but gain(liner gain) and Psa, at this Vps setting lower liner gain. (Setting2) of 6.8V. This is trade off of biasing what + For everything: High Vps and high current you request to this amplifier. (Settingl) but lower power added efficiency. (Set- ting3) CH2: 8 me + 23.01 3B 10.0 d8/ REF -~ .@@ dB 08-20-'93 MILLJMETER HAVE LAB} RATORY! | | | [a : | 4 git STRT +22-0@@GH2 —-CRSR +24.170GHz STOP +26. 00062 24GHe Qstage HPA using FLROLGFH + FLRO2OFH JELAAH Gain: 23.0143 by HPB3SOB,835948 Vds: 5.0¥ HPB757A Vgl: OV at 24. 170GH2 HP85027B Ve2: OV HPLL664E Pin: -8dBm Max gain bias condition Bild/Figure 2: HPA: $11 and $21 from 22 to 26GHz DUBUS 4/1993 7Technical Reports: Power Amplifier for 24GHz by Toshihiko Takamizawa, JELAAH + Setting! I have measured the HPA on my bench using following instruments: + HP8350B483594A,8757A 85027B,11664E for $11,S21,S12,S22 measurement © Wilton 43KC-20 (40GHz 20dB attenuator) Vps: SV Vo, and Vo, = 0V Linear gain 234B at an input level of ~8dBm The 2 stage HPA performs as follows: = Compressed gain 19.34B at input level of OdBm with Vps=5.5V,Vo,=0V and ches B 420-14 din . 1.0 dB/ REF # 20.09 dim 08-20-'93 r ' — = | NTLUEMETER WAVE LagpRATORY | (rout) | f " * | | | \ 421d + | i 2 120380 | | | | | +18dBm | a | | | i | j | | +17¢am | 4 ‘16000 | | | j | 1 | #15d8a |} = - i | | | | | | | Poy | | #14eBa |» ! | | __| _ “2 0 1 + Qt stat #[Link]: “thon .osadltt itp +748 enact! poe 24GHz HPA using FLROL6FH + FLRO26FH Pin/Pout characteristics JELAAR Liner gain: 17.843 by P8350B,8359 Vds: 6.75V HPB757A Vgl: -0.01V HP85027B Vg2: -0.78V HPLL664E Psat: 421.04din Max. Psat plus more gain at low input level setting Bild/Figure 3: HPA: Pout versus Pin DUBUS 4/1993 18Technical Reports: Power Amplifier for 24GHz by Toshihiko Takamizawa, JEIAAH Voy=-0.26V. This means at 1mW Pry 85mW Pour. However Psa: never exceed +20dBm. Setting? Vos 68V Ve, -0.29V Vo, -0.80V Linear gain:16.4dB (x 44) Psqi +21.06dBm (128mW) 421.30dBm was observed later mW Pr ~> 46mW Pour 2mW Pry -> 91mW Pout + Setting’ - Vos 6.75V = Ve, -0.01V - Vez -0.78V — Linear gain: 17.84B (x 60) = Psa¢+21.04dBm (127mW) imW Pi ~> 60mW Pour 2mW Pry -> 103mW Pour Fujitsu specifies Vos of 8V for FLR series. How- ever my experience shows it can not be applied for over specification use on far above 18GH7z, espe- cially on 24GHz like this. Vps of 8V should work well but with no improvement on Psat and Gain. Bad efficiency isthe only result. You need to select what bias condition you use. 4. Ergebnisse Bei den Messungen ergaben sich folgende Ab! gigkeiten von den Arbeitspunkten der FET’s: Fir maximale Verstérkung soll die Drainspannung bei ca.5V liegen und die Gatevorspannung ist nahe OV. Fir maximalen Output ist eine Drainspannung von 6,8V ginstiger. Bei hohem Output (20dBm) ndem Variation von Ug, und Ug; nicht viel bei der Verstirkung. Folgende Bias-Einstellungen ha- ben sich als ginstig erwiesen: + Einstellung 1 Ups SV UG, and UG, 0V Verstirkung 23dB bei einem Eingangs- pegel von -84Bm Verstérkung 19.3dB bei einem Ein- gangspegel von 0dBm und Ups=[Link], Ug=0V und Ugy=-0.26V. Das bedeutet 85mW Ausgangsleistungt bei 1mW Steuerleistung. Psar erreichte nur 100mW + Einstellung 2 Ups 68V Ug, -0.29V Ug: -0.80V Verstiirkung 16.4dB (x 44) Psa: +21dBm (128mW) mW Pry ~> 46mW Pour — 2mW Pry -> 91mW Pour * Einstellung 3 - Ups 6.75V Uc, -0.01V Ug, -0.78V Verstirkung 17.8dB (x 60) Psat +21dBm (128mW) ImW Pry —> 60mW Pour 2mW Pry > 103mW Pour Die MeSwerte wurden mit folgender Ausristung gemessen: + HP8350B+83594A 8757A,85027B,11664E, fiir $11,S21,S12,S22 Messung, * Wilton 43KC-20 (40GHz , 200B Ab- schwacher) Fujitsu erlaubt 8V Drainspannung fir die FET’s Die Anwendung dieser Spannung engibt keine Ver- besserung in Leistung und Verstarkung, sondem cemiedrigt nurden Wirkungsgrad und ist thermisch problematisch. 5. Trial: A single stage FLROS6FH amplifier [can not report a good result on FLROS6FH. I have tried two chips on my test jig. They performed about 34B of gain and 15 to 17dBm of Po. Ihave three chips but it can be estimated that FH package is not suitable for FLROS6. I need further investi- gation on the device and the application. | will ask Fujitsu to mount TLROS6 in an another package. WG or WF is nice for higher power chips. As Idss is exceeding 220mA, I believe the chip inside is FLROSG but performance is ... DUBUS 4/1993 19Technical Reports: Power Amplifier tor 24GHz by Toshihiko Takamizawa, JEIAAH Bild/Figure 4: Close View on 24GHz HPA 5. Ein Versuch: Einstufiger Ver- starker mit FLRO56 Einstufige Verstirker mit dem FLROS6FH funktio- nieren nicht gut. Testwerte waren 34B Verstirkung bei max. 174Bm Ausgangsleistung, Das FH-Ge- hiuse ist offensichtlich nicht geeignet. Ich werde meine Freunde bei Fujitsu fragen, ob sie mir den ‘TLROS6 Chip in ein WG oder WF Gehiiuse ein- bauen kinnen. Dann werde ich das nochmal unter suchen, 6. Conclusion This HPA. project was very successful except FLROS6. As a result of these projects, PCB size should be as small as possible and using mono- layer microwave chip capacitor is a key to success if you want to catch up with state-of-the-art per- formance. I could get previously reported 3 stage HPA performance by 2 stage HPA. I hope this amticle will be helpful for 24GHzers in the world and be much appreciated to hear from you regard- ing obtaining devices and PCBs, QAs and so on, Please feel free to call me, Fax to me or write to me. [ hope to hear from 24GHzers. Special Thanks Istrongly wish to thank to Fujitsu for their special offering of FLROS6FH in stead of usual chip form supply, but not succeeded hi! 6. SchiuB Das HPA-Projekt war ebenfals erfolgreich. Kleine Platinen und Mono-Schicht-Kondensatoren sind der Schliissel zum Erfolg, So konnt e die Leistung des alten dreistufigen Verstirkers mit einem zwei stufigen ({2}) erreicht werden. Falls Teile beschaff werden missen, knnen 24GHz DXer mich anruf: wen oder ein Fax schicken. wiirde mich freu- en. Besonderen Dank gebihrt der Fa. Fujitsu, die mir die FLROS6 Chips in ein FH Gehiuse montiert hat Leider funktionierte dann der Verstirker doch nicht References [1] T. Takamizawa, JELAAH, '24GHz Trans verter with HEMT", DUBUS 2/1991, pp.37-48 [2] T. Takamizawa, JELAAH, *24GHz Trans- verter with HEMTs’, DUBUS 3/1991, pp.24-32 DUBUS 4/1993 20Technical Reports: Modern 10GHz Transverter System (Part Il) by C. & P. Suckling Modern 10GHz Transverter System (Part Il) (Modemes 10GH7z-Transverter System - Teil Il) Charles Suckling, G3WDG & Petra Suckling, G4KGC ction: The second part of this series covers the description and alignment procedures for the various modules which make up the G3WDG 10GH7z transverter. These are: + G3WDGO01 Signal Source (Multi- plier/Amplifier) G3WDG002 10GHz - 144MHz Receive Converter G3WDG003 144MHz - 10GHz Transmit Converter G3WDG004 10GHz HEMT Preamplifier Part III in DUBUS 1/1994 will describe G3WDGO00S 2-stage Amplifier G3WDGO006 1-stage, 250mW TX-Ampli- fier (MGF1801) G3WDG007 1-stage, 1WTX-Amplifier (MGF2430A) G4FRE Power Supply G4NJTOO] Transverter Interface Unit Bild/Figure 1: G3WDGOOI (Top View) DUBUS 4/1993 21Technical Reports: Modern 10GH2 Transvertet For an operational 10GHzstation with SOmW out- put you need a G4DDK0O1 LO, a 002 RX-con- verter, a 003 TX-converter, and the G4FRE power supply with the GaNJT TIU. To achieve 250mW output power you have to add a 006 PA, for 1W output another 007 PA. To lower the standard noise figure of the RX from 2.54B downto 1.04B you have to add a 004 HEMT-preamp. The 10GHz transverter system is a fully modular 50Q system. Einfiahrung: Der zweite Teil der Artikelserie be- schreibt die verschieden Module des G3WDG Transvertersystems. Es sind im einzelnen: G3WDGO001 Signalquelle (Vervielfa- cher/Verstirker) + G3WDG002 10GHz - 144MHz RX-Kon- verter © G3WDG003 144MHz - 10GHz TX-Kon- verter G3WDGO04 10GHz HEMT Vorverstitker ‘Teil II] in DUBUS 1/1994 wird folgende Module bringen + G3WDGO00S 2-stufiger Verstiirker + G3WDGO006 1-stufiger 250mW Leistungs- verstiirker (MGF1801) f System (Part Il) by C. & P. Suckling * G3WDGOO7 L-stufiger 1000mW Lei- stungsverstirker (MGF2430A) © G4FRE Stromversorgung * G4NJTOOL Transverter Interface Einheit Fir eine funktionierende 10GHz Station braucht man einen G4DDK001 LO, einen 002 RX-Kon- verter, cinen 003 TX-Konverter, eine G4FRE Stromversorgung und eine G4NJT Kontrollein- heit. Will man die Empfiingerrauschzah! auf ca. 1dBemiedrigen, kannmannoch einen 004 HEMT- Vorverstirker wahlweise in Koax- oder Hohlleiter- technik einfigen. Zur Leistungserhohung sindein- mal der 006 Leistungsverstirker mit 250mW Lei- stung und der G07 Leistungsverstirker mit 1000mW Leistung geeignet. Das ganze System ist modular in SOQ-Technik realisiert 1. The G3WDGO001 multiplier/ amplifier The G3WDG001 was the first module in the series to be developed and proved that GaAs FET multi- pliers and amplifiers could be constructed repro- ducibly in microstrip form by amateurs with no access to specialist facilities or test equipment. Bild/Figure 2: G3WDGO001 (Bottom View) DUBUS 4/1993 22Technical Reports: Modern 10GHz Transverter System (Part Ii) by C. & P. Suckling Bild/Figure 3: Circuit Diagram of G3WDG001 0.5GHz Output Issue © STI ov ety DUBUS 4/1993 23Technical Reports: Modein 10GHz Transverter System (Part Il) by C. & P. Suckling While not part of the transvertersystem, it isavery useful module in its own right. It represents a very simple way of generating a narrow-band signal on 10GHz, when used with the G4DDK004 local oscillator. The main application for this design has. been forbeacons, but it could also be used for ATV, and as a packet link transmitter. The module is capable of operating anywhere in the 10 - 10.5GHz band. Usually it is used to gen- erate 10,368MHz, when a drive frequency of 2592MHz is required (at a power level of about 10mW). A suitable oscillator design in the G4DDK004, described in DUBUS 3/93. ‘The circuit diagram of the G3WDGOOL is shown in Fig. 3, and the component layout in Fig. 4. The input stage is an MAV11 MMIC amplifier with a shunt capacitor C16 on its input to improve in the input VSWR. The output from the MAV11 drives the FET multiplier via a lumped element matching circuit L2/L3. This circuit has sufficient bandwidth to accommodate the required range of drive fre- quencies. The FET is normally biased close to pinch-off with the drive signal driving the drain current up to 10-15mA. This has been found to produce the best conversion loss as a multiplier. A series tuned circuit on the drain of the FET (L4/printed capacitor) forms a short circuit at the drive frequency, which improves conversion loss bya few dB. Drain bias is fed via LS and decoupled with a short circuit stub, The output multiplier is fed via some matching stubs to the filter FL1. The idea for this filter was borrowed from previous designs [1]. Small probes made from circuit pins (the same type that are used for grounding the chip components) couple in and out of the filter. The output of the filter is fed to a2 stage amplifier, which typically produces some SOmW output power. The amplitier was designed to cover the full 10-10.5GHz band and has a gain of 19-20dB. Special attention was paid to the grounding of the FET sources to guarantee reproducible perform- ance. Experiments with passing the source leads through the board or using copper foil showed that these methods are not reproducible. The method which was adopted uses small circuit pins ("veropins") to ground the sources, which behave very similarly to plated through holes. Good qual- ity coupling capacitors (ATC series A type) were chosen because of their guaranteed performance. ‘They are a little more expensive than conventional chip capacitors but have extremely low loss and behave the same every time. These factors allow the design to be built with the expectation that it should work first time with no need for tuning! The bias feeds for the amplifier use quarter wave high impedance lines made from silver plated wire- ‘wrap wite and radial stub low impedance decou- plingelements. Low frequency decouplingistaken cate of by series resistors and decoupling capaci- tors. Alignment Alignment of the unit is straightforward if the following sequence is adopted: 1, Connect a current meter between the +8V supply and the +8V rail on the board. Set RV1,2 and3 so that full negative bias will be applied to the FET gates. Attach a suitable load/power indicator to the output socket. This might consist of an SMA to WGI16 transition, attenuator and diode de- tector, or an SMA attenuator and coaxial power meter. Apply DC power, and with no oscillator drive applied, the current shown onthe metershould be no more than a few microamps. If the current shown is higher than this look for mis-connected or damaged components. Adjust RV1 so that the current rises to about 2mA. Apply 2.5GHz drive and the current should tise (typically by 10mA). Set RV2 to increase the current drawn by 20mA (total now about 30mA). Set RV3 to increase the current drawn by a further 20mA (total now about S0mA). Asmall amount of RF output may now be seen on the power meter. Ifnot, do not worry at this stage. The filter is now tuned for maximum ‘output, Assemble the locknut on to the screw so that approximately 7.5mm of screw thread protrudes from the nut, Without moving the nut, insert the screw into the cavity until the nut touches the cavity. Slightly tighten the nut with a spanner. Keeping the nut slightly tight with a spanner, tum the screw a few turns either side of the initial position looking for a peak in output power. Once the peak has been found tighten the locknut and if necessary readjust the screw for maximum power. For those who like to do their own car mainte- nance, this operation is very similar to adjust- 2, 3. DUBUS 4/1993 24Technical Reports: Modein 10GHz Transverter System (Part Il) by C. & P. Suckling Bild/Figure 4: Parts Layout of G3WDGO001 sogord sovevesos Kune © ‘su Bpuno® prog UBnonL @ xq veri © PUB ZEAL bx indy otsowves so 452: pioR BACH zHoszoz +B sto xd 5% Ag'z- age io a 10 ors suse oa 4 Foti pected © csc ZHOGO! - Ob ris 4 ons) 25 DUBUS 4/1993Technical Reports: Modern 10GHz Transverter System (Part Il) by C. & P. Suckling ing tappets! With the screw supplied, it should not be possible to find any other resonances, but of course a check should be made with a wave meter or spectrum analyser before put- ting the uniton the air!! It should be noted that, with the screws supplied, the heads may pre- vent the bottom lid of the box from fitting correctly. If this is a problem then after align- ment the screw should be removed and the head sawn off. A screwdriver slot is then cut in the end of the screw with a small hacksaw to permit the screw the be adjusted. The screw is refitted to the cavity and the alignment procedure repeated. Parts List G3WDG0O01 MAV11* Parts List G3\WDGOO1 Item Straight length 0.2mm silver plated wire (CuAg), tinned Imm lends, mounted flat on board Part-No. FL Silver plated cavity resonator, M4 screw and nut, probe length 4.7mm Veropins Imm single ended (25 required) Sockets SMA 2-hole or 4-hole, gold plated (2 required) Tinplate Box. [37x111x30mm (Schu- berth Size 3) Feedthrough |100pF not critical BiasParts |G 4FREO23PCB and paris PCB G3WDGOO1 Issue B IBirkett ‘Black Dot’ ‘180pF 0805 [2.2pF ATC100 100uF Tantalum ‘pF 0805 1OOR Wire Ended 0.5W l47R 0805 |47R or 10R 0805 220R Wire Ended 0.5W 10k horizontal skeleton carbon preset (Philios lOCP 10H) lu 61 0.3mm enamelled copper wire (ecw) close wound, 2mm dia., 2mm lead length, mounted Imm above board 12 ‘16mm length ecw. tinned Imm ends, mounted fiat on board 9.5mm length as L2 [straight length 0.3mm lecw, tinned Imm ends, Imounted flat on board, * Items in kit are in bold facet 10. Adjust the gate bias settings (RV1,2.and3) to optimise power output. 11. This completes the alignment of the unit. The power output should be between 50 and 100mW. If a well matched load can be guar- anteed, itis possible to increase power output by a small amount by changing R6 to 10R. However, this is pushing the transistor ratings alittle and is not recommended where the unit is intended to be operated continuously, for example as a beacon. I power output is below SOmW, it may be possible to increase this by adding small tuning tabs to the microstrip circuitry. This is done by cutting a small piece of copper foi! (2x2mm to start with) and slid- ing this around with a wooden cocktail stick while monitoring power output. The tab is then soldered in position (with the power off). As many tabs as are needed may be added, The output spectrum of a typical unit is shown in Fig.5 1. G3WDGO001 Vervielfacher/ Verstarker Der G3WDG001 wurde als erstes Modul ent- wickelt und kann als Bake, ATV-Sender oder als Packet Radio Sender dienen. Er ist im Bereich von DUBUS 4/1993 26Technical Reports: Modern 10GHz Transverter System (Part Il) by C. & P. Suckling 10 bis 10,SGHz abstimmbar, Als Oszillator dient zweck- maBig der G4DDK001 (DU- BUS 3/93). Bild 3 zeigt das Schaltbild und Bild 4 zeigt die Teileanordnung. Die Eingangstufe besteht aus einem MAV11 MMIC. Der Ausgang treibt einen FET Ver- vierfacher. Dieser wird mit ge- ringen Ruhestrom (‘Pinch- Off") betrieben und zieht bei Ansteuerung 10...15mA. Ein Serienkreis (L4/Gedr. Kap.) am Ausgang auf der Eingangsfre- quenz verbessert den Wir- kungsgrad. Uber ein Anpa sungsglied und das Filter FL1 wird das Signal ausgekoppelt. Die Ein- und Auskopplung des '———— Filters geschieht uber Veropins. Nach Verstirkung durch zwei weitere Stufen liegt ein Pegel von SOmW vor. Der Verstirker ist breitbandig und hat ca, 20dB Verstirkung. Die Methode, die Sourcen der FET’s mit Veropins zu erden, hat sich sehr bewaihrt und ist beinahe equivalent zu professio- nellen ’Through Holes’. Die ansonsten gebriiucl liche Methode , die Source-Beine durehzustecken oder Kupferfolien zu verwenden, hat sich als nicht so reproduzierbar erwiesen. Die Koppel-C’s sind ATC-Chips. Sie sind zwar teurer, aber sind villig reproduzierbar und verlustarm und damit ihr Geld wert, da sie nicht zu unvorhergesehen Schwierig- keiten beim Betrieb fihren. Die Spannungszufiih- rungen an Gate und Drain werden durch versilber- te, diinne Drihte (’Wire-Wrap’) mit hoher Impe- danz realisiert. Radiale "Stubs’ blocken fiir die Betriebsfrequenz ab. Fuirniedrige Frequenzen sind Scrien-Widerstinde und Chip-Kondensatoren vor- gesehen. Abgleich Die Abstimmung ist einfach, wenn man die folgen- de Reihenfolge beachtet 1, Das Modul wird aber einen StrommeBgerit an das ausgeschaltete Netzgerit mit 8V ang schlossen. 2. RV1,2 und 3 werden auf volle negative Span- nung an den Gates eingestellt Bild/Figure 5: Output Spectrum G3WDGOO01 Corl cag a ETS cat Cn ‘ 3. An den Ausgang LeistungsmeSgerit an- schlieBen, 4. Gleichspannung anlegen und verifizieren, da der Strom nur einige A betrigt. Wenn der Strom hdher ist,nach defekten Bauteilen su- chen oder nach Fehibestiickungen 5. RVI so einstellen, daB der Strom ca. 2mA betrigt 6. Ein Signal auf 2,SGHz anlegen und Stroman stieg auf 10mA beobachten RV 2soeinstellen, da der Stromaufca.30mA ansteigt. 8. RV3soeinstellen, daB der Strom aufca. 50mA ansteigt. 9. Schraube am Filter ca. 7,5mm eindrehen, die Kontermutter leicht anspannen und fiir cin Maximum an Leistung suchen- Schraube hin- und herdrehen!, Die Abstimmposition ist ein- deutig, 10, RV12 und 3 kénnen nun auf maximale Lei- stung nachgestellt werden, 11. Die Leistung sollte ea. 5 100mW betragen. Das Ausgangsspektrum ist in Bild 5 zu sehen, 2. G3WDGO02 10GHz - 144MHz Receive Converter The circuit diagram of the G]kWDG002 receive converter is shown in Fig. 8, and an artist’s impres- sion of the board layout is given in Fig. 9. Fig. 10 DUBUS 4/1993 27Technical Reports: Modern 10GHz Transverter System (Part Il) by C. & P. Suckling shows the connections to the reverse side of the peb. Referring to Fig. 8, the 25S6MHz LO input signal is fed to Fl which acts a quadmupler to 10224MHz. The multiplier is exactly the same design as [Link] the G3WDGOOL. The output from FL1 at 10224ME1z is fed to the LO portof the diode mixer via a microstrip impedance matching net- work. The mixer uses a series diode pair, D1, connected between the ends of a folded three-quar- ter wavelength line. This configuration gives good rejection of the LO signal at the RF port, and vice-versa. L9 is a shorted quarter-wave line to provide the required low impedance IF return path, while having no effect at the LO frequency. The conversion loss ofthe mixer including its matching networks has been measured as 6-74B. A two-stage low noise amplifier is provided to reduce the noise figure ofthe unit toa more accept- able level. The amplifier design is very similar to the amplifier used in the G3WDGOO1, except that a low noise FET can be used in the input stage as an option. F2 can be either a "red dot" or "black dot" FET according to the level of performance required, Noise figures of 2.5dB with areddot FET and 3.2dB with a black dot FET are typical. Nor- mally the amplifier is perfectly stable, but occa- sionally some red dot FETs have shown signs of instability, and this has been cured by substituting a black dot FET. The input circuit of F2 has an optional stub which in some cases has been found to lower the noise figure. It is normally not re- quired. The overall gain of the LNA section is about 20dB with a noise figure of around 2dB. ‘The output of the LNA passes to the mixer via a bandpass filter FL2 and a high-pass filter (C11), The image rejection with a 144MHz IF is typically better than 25dB. Note that it is possible to tune FL2 to the image frequency, so care needs to be taken when tuning up (See below). The IF output from the mixer is fed to a low noise amplifier at 144MHz via a low pass filter consisting of L10, a quarter-wave line and L11, to prevent any 10GHz signals reaching the IF amp! Alignment 1. Temporarily bridge out the image filter FL2 as shown in Fig, 11. The wires are 0.3mm diame- ter spaced one wire diameter apart and overlap by Smm. They are mounted flat on the board, ‘The wires form a crude side-coupled wide- band filter section with the properties of DC Bild/Figure 6: G3WDG002 (Top View) DUBUS 4/1993 28Technical Reports: Modern 10GHz Transverter System (Part Il) by C. & P. Suckling isolation and low RF loss at 10GHz, Preset FL1 and FL2 as described above for the G3WDGOO1, with 7.5mm of screw thread protruding from the nuts, and the nuts lightly tightened on the cavity. Tum RV1, 2 and 3 so that full negative gate bias will be applied to the gates of the FETs on power up. Insert a multimeter in series with the +5V supply and the +5V rail on the board. Connect a load to the 10GHz input socket, such as an SMA termination, attenuator or an SMA - waveguide transition feeding a horn antenna or other well-matched load. Apply DC power, and the indicated current on the multimeter should be a few microamps only. If not, look for shorted or damaged components. Adjust RV1 to give an indicated current of about ImA on the meter. Apply 25S6MHz drive (SmW min) and the current should rise (to 5-15mA). The absolute level is not critical se 8. 12, 13, Connect a 144MHz ssb receiver to the IF output socket and tune C1S for a noise peak. This should be a fairly significant peak if the IF amplifier is working properly. Set RV2, 3 to give OV gate bias to F2 and F3. Adjust FL1 and when complete, RV1_ for maximum noise. Use the procedure described in step #9 in the alignment procedure of the G3WDGO01 to adjust the filter tuning screw. Power off and remove the wires "bridging’ FL2. Power onagain and adjust FL2 for maxi- mum noise. Two noise peaks should be found, about 0.4 turn apart. The correct peak is with the screw furthest out, ie smaller penetration. Reset RV2, 3 to give full negative bias to F2 and 3. Adjust RV2 to cause the indicated cur- rent to rise by approximately 12mA and RV3 to cause a further increase of 15mA. The total indicated current should be in the region of 35-40mA. If a signal source is available, check this can be heard satisfactorily. It is worth reconfirm= ing that FL2 is tuned to the correct frequency, and not the image! If no such signal is avail- Bild/Figure 7: G3WDG002 (Bottom View) DUBUS 4/1993 99Technical Reports: Modern 10GHz Transverter System (Part Il) by C. & P. Suckling Bild/Figure 8: Circuit Diagram of G3WDG002 NS = zr2 = OSE 5 ess e25 : aos 8°a of. a5 350 Sos ore 5 258M LO Input 10.3606Ht2 Inout DUBUS 4/1993 30Technical Reports: Modern 10GHz Transverter System (Part Il) by C. & P. Suckling able, try listening for local signals on the band, or perhaps harmonics from lower frequency ‘equipment (eg 1152MHz x 9 =10368). . Remove the test meter and make good the connection. . Final alignment can be done with a noise figure meter if available, or a weak signal Adjust RV1, 2 and3, FL1 and 2, and C15 for best results. After alignment the converter should have a noise figure of 2.5 to 3.5dB and a conversion gain of 27-30dB. Check that the performance is not degraded when the RF lid is in position. If signs of instability are present, the piece of lossy rub- bersupplied with the kit should be glued to the inside of the RF lid, above the low noise amplifier section, This material is very effec- tive at curing box-induced oscillations, and is the kind of material fitted to LNBs for the same purpose! Parts List G3WDG002 1 a item. F1LF3 [Birkett ‘Black Spot’ Birkett ‘Red Spot’ DI DMF3909-99 Series Diode Pair TT BFR9O/91 [c1-3,C6-9 —_|180pF 0805 €13,14,16-19 |10nF 0805 C4511 1 00uF Tantalum ci2 22-47uF Tantalum 1C10,19 }2.2-10uF Tantalum cis 30pF Trimmer R1,3,45,6 _|47R 0805 RZ 220R Wire Ended 0.5W RP 15k 0805 RE 4k7 0805 Ro 220R 0805 RIO 100R 0805 RN 470R 0805 | RVT-3 2k2 horizontal skeleton carbon preset (Philips OCP 10H) LT Témm length 0.3mm lecw, tinned Imm ends, formed into loop, mounted flat on board _ L2 19.5mm length as L2 Parts List GS3WDGOO2 Part-No. Item Straight length 0.3mm jecw, tinned 1mm ends, mounted fiat on board. |The wire should be lsoldered to the edge of the stub and as close as possible to the drain of F1 Straignt length 0.3mm lecw, tinned Imm ends, Imounted fiat on board [Straight length 0.2mm silver plated wire (CuAg), tinned Imm lends, mounted flat on jooard |Combinded total length lof 0.2mm silver plated lwire mounted flat omn board as shown in figure. Solder at both lends of central stub section 4tumns 0.6mm copper wire, 5mm L.D., turns spaced 0.3mm, mounted Imm above board, center tapped Silver plated cavity resonator, M4 screw and nut, probe length 4.7mm. ‘Imm single ended (33 equired) SMA 2-hole or 4-hole, gold plated (3 required) 37x111x30mm (Schu- berth Size 3) 100pF not critical Bias Parts G4FREO23PCB and parts PCB G3WDG002 Issue B * Items in kit are in bold face! 13. 15-9 L10,11 L12 FLI2. Veropins Sockets Tinplate Box Feedthrough DUBUS 4/1993 3Technical Reports: Modern 10GHz Transverter System (Part Il) by C. & P. Suckling Bild/Figure 9: Parts Layout of G3WDG002 du se Aeyiaag yuauodwog da4anUuOD 9Aja9ay ZHDO! 200 DGMED Od OU Jo opis eueIdpUNEI6 ey) LO ZI9 Supunow woyIUBIOg X. ‘suogeuuos Aiddng —o 809013 O puno.6 ud oe, © nzbe 32 DUBUS 4/1993Technical Reports: Modern 10GHz Transverter System (Part Il) by C. & P. Suckling Bild/Figure 10: Wiring Arrangement tor G3WDG002 +12V im IF Output LO Input RF Input 2. G3WDGO002 10GHz - 144MHz Empfangs-Konverter Das Schaltbilddes G3WDG002 Empfangskonver- ters istin Bild 8 zu sehen. Die Teileanordnung folgt aus Bild 9 und die Verdrahtung auf der Platinen- riickseite aus Bild 10. Das Eingangssignal auf 2556MHz wird vom Vervierfacher F1 auf das Fil- ter FL1 gefliht. Das ausgesiebte Signal auf 10224MHz steuert den Mixer mit dem Diodenpaar D1. Die Seriendioden sind mit einer gefalteten Y%)-Leitung verbunden, die fiir eine gute Isolation zwischen LO- und HF-Port sorgt. L9 ist eine kurz- geschlossenen /¥4-Leitung, die als eine nieder- Bild/Figure 11: DC Block ohmige ZF-Riickfuhrung wirkt und auf der Be- triebsfrequenz als Parallelkreis wirkt. Der Misch- verlust dieser Anordnung betrigt nur 6...74B in- Klusive der Anpalinetzwerke. Ein zweistufiger Vorverstrker mit ca. 20dB Ver- stétkung und 2dB Rauschzahl ergibt eine Gesamt- rauschzahl des Konverters von ca. 2,5dB, wenn in der Eingangsstufe cin "Red Dot’ FET verwendet wird, Der Ausgang des Vorverstiirkers gelangt iber das Spiegel requenzfilter FL2 und den Hochpass C11 auf den Mischer. Die Spiegel frequenzunter- driickung ist bei 144MHz ZF ca, 25dB. Der ZF- Ausgang des Mischers gelangt iiber einen TiefpaS, (LAO, Viertelwellenleitung und L11) an den rau- scharmen ZF-Verstirker mit einem BFR90. Abgleich 1. Man iiberbriickt zwecks Abstimmung das Fil- ter FL2 mit Hilfe der Anordnung in Bild 11. Die beiden Parallel-Leitungen werden durch 0.3mm dicke Leitungen gebildet, die einen Drahidurchmeser voneinander entfernt liegen. FL1 und FL2 werden vorabgestimmt, indem man die Schrauben soweit eindreht, daB ca. 7,Smm Linge des Gewindes auBen zu sehen ist. DUBUS 4/1993 33