Archives de la catégorie ‘Filter mobo’

Hermes Lite by KF7O

Publié: 24 novembre 2016 dans Filter mobo, SDR

I never mentionned it on this blog or on the Electrolab’s Wiki, but some time ago, I built an Hermes Lite SDR. It’s a nice piece of hardware that Steve KF7O gave to the community. I said “gave”, because it’s a full open hardware, open software project.

This rig is a true DDC/DUC SDR, with a 12 bits/60 Megasamples ADC (an Analog Device AD9866 transceiver on a chip). It gaves me hours of fun, and could be used as a serious baseband bedrock for many extensions (transverters, power amps etc).

From a software point of view, this small and low cost SDR is fully compatible with the TAPR’s Hermes high-end system. It could be followed with any kind of Hermes client software, like OpenHPSDR GNUradio, GHPSDR3-Alex….

double réception hl deca

This is a snapshot of a dual RX config. of OpenHPSDR behind the Hermes Lite

The total cost of this system is around 100 to 150 $/€. The version I built (the one here described ) is allready deprecated, and will be soon replaced by a compact, monoboard 2.0 edition.

Anyhow, let’s take a short tour of this project


cote hermes

On top, the BeMicro SDK “usb key”, supporting the FPGA. The Hermes lite itself is the small green board located on the right of the BeMicro, followed by a simple output board (two transformers winded on binocular).

On the lower side, a bandpass filter (tcheb) originally designed for the Softrock Mobo 6.3 project. This bandpass filter is used in transmit mode only. Different receive tests show no difference between an “barefoot” Hermes Lite and a filtered one.

The bandwidth of each filter has been enlarged to lower insertion losses

Hermes Filter bw 1-8 to 24-3 MHZ

On the right of the case, a 10 W power amplivier (a TAPR’s original Pennywhisle).

As the Hermes Lite is note able to directly drive the Pennywhisle, a 20 dB/1W driver, using a single operational amplifier (OPA2677)  is inserted between the rig and the amp

amp recto

This home made amp has been designed to fit into low profile machined cases (former MCL 3dB hybrid couplers sold a few $ on eBay)


The pennywhisltle itself is mounted on the rear panel. A better amp, like the 20 W G6ALU (particularly VK3PE Glenn’s version) would have been more suited… but I keep this amp for the H.L. 2.0

So far, this combination gives good results and can drive a 300 W PEP power amp without problems



On the other side of the case, a CW Curtiss Keyer (former MFJ keyer), the linear power supply (as Steve said : “lot’o heat”… but with a good and silent fan, it works as advertized.

The brown board with 5 connectors is the switching interface between the Hermes Lite and the filter board

ULN sw

This is a quick and dirty hack with an ULN relay driver.


Filter mobo phase One : over

Publié: 3 juillet 2009 dans Filter mobo

The design of the 8/16 elts filter mobo has been completed. Now, we can launch a proto and see if what’s on the paper will work as advertised.

I still have to finish the “BCD command board” (this one was not urgent and mandatory for Alex’s project) and a fully new switching board for the high power version of the filter.

Anyway, here is the full and final production of the different elements of the filter.



This card decodes an I2C bandswitching command comming from Alex’s microcontroler, and sends the corresponding signal to one of the multiple input of an ULN2803 (relay driver)

Each I2C interface has a selectable adress, determined by a matrix of 2×3 0 Ohms pullup resistors.



Quite a lot of “invisible changes” have been made in the position of the traces, tuning of each component size and position and so on… Enough space has been preserved to solder a “U” shaped shielding located all over the central bus. This will probably enhance input/output isolation and minimize all RF re-injection risks.

The FILTER DAUGHTER CARD remains unchanged…


… excepted some angle stripline to try to keep the input impedance as close as possible to 50 ohm. But the angle connectors binding this daughter board to the mobo will certainly destroy the quality of the line. Anyhow, it’s not  a reason to have “ugly traces”. I hope the next “filter” bloging wil have some real picture to illustrate the story.

Back to work

Publié: 21 juin 2009 dans Filter mobo, SDR

After a looooooong stand by and quite a lot of miles by plane, car and train, I’m back to work.

Some minor modifications have been made to the filter mobo (larger and adapted strip line at the input and output).


The mobo has heavily been redesigned : for a better ground plane and better coupling, pass-through relays have been replaced with gull wing components. This way, a unique 2,9 mm HF bus (50 ohms strip line) goes from relay to relay. RF flow on the upper surface has very limited accidents.The I2C command board will probably be redesigned within this week, to comply with Alex’s wishes (better integration, no address assignation by default).

A Minimelf Diode (4148) pad has been added close from each relay. But as most of the “form C dual circuit/dual pos” DIL relays are polarised; it won’t be necessary to add this component most of the time.

The central bus is far from perfect an will be optimized all along this week.

The 100 W++ filter mobo will probably have a different design : same command daughter board, same filter daughter boards, but 4 separated relay boards (pin to pin compatible with the command board) will concentrate the switching stage. Wiring between filters and relays will be made with Teflon coax cable. I’ve try to use the pcb design with higher power rated relays, but in vain.

CAD re-design of the whole project

Publié: 13 avril 2009 dans Filter mobo

The mobo and daughter cards are slowly redesigned under Kicad. The final result will probably be simpler and a little bit more optimized, as the software is surprisingly powerful.

The daughter card is almost finished. I must solve some problems concerning the grounding zone and learn how to correctly print the postscript file to etch a prototype, but it’s a matter of days.

The main mobo is far more complex and Kicad is trying to play with my nerves, inverting some components without logical reason. I’m sure I’ll have the last word J

The 3D result is a little bit smarter than the qwick and dirty ExpressPCB, isn’t it ? It’s anyhow a marvelous function to hunt misconceptions and wrong placements.

Filter mobo, first prototype

Publié: 4 avril 2009 dans Filter mobo

Some component are missing… I cannot go further.  But








As I was out of eyelets, I couldn’t have metalized holes… that’s why relays are soldered high above the pcb. Surface mount devices and/or flater relays should be used on the final mobo. I’m loosing too much space with those one.

Dauthter cards are fitting perfectly (excepted that they are a little bit too short… something lile a thenth of an inch). I’ve completely messed up with the Command daughter boards, inverting the printout without noticing it…. this is why so many wires are running on the back side…

So far, bands are switching correctly. I have to order a handfull of ULN2803, PCF8574, 4028 and other things next monday

New batch of PCB

Publié: 30 mars 2009 dans Filter mobo

The first proto showed some misconceptions and a lot of cabling mistakes. The humain brain … well, MY brain is not able to “mirror” the pinout of a simple relay…

So I had to etch a new mobo again. As the former daughter cards were also wrong, I draw a batch of daughter boards.

From left to right :

First column : a set of filter daughter cards (one “backside”, four “components side”)

Second column, from top to bottom :

– Command daughter board, BCD to 2xOctal

backside and component side

– Command daughter board, I2C to 2xOctal

backside and component side

Third column : the debuged new mobo (hope so)

Fourth column : the wrong one with some components.

…. time consuming 😦

Nothing will be sodered tlll next month.

It fits !

Publié: 12 mars 2009 dans Filter mobo

This first picture is for…. some snowboarder living in Singapore

…. softrock on the rocks of the french Alps


the local altitude is 2800 feet (that’s low for us), mountains around reach 6500 feet to 13 000.ft The QTH is located approximately 45 miles away from the Mont Blanc, the highest mountain in Europe (15767 ft)

the UWB/level converter/audio input is installed… and it fits perfectly


A close-up on the knitting


and a general overview of the SR, Alex’s Mobo, the frequency display and the PCB of the filter’s prototype … all that stuff fits in a 5’1/4 drive enclosure.

I did’t allready tested the UWB… the pic is not yet programmed.

Filter mobo, first prototype

Publié: 9 mars 2009 dans Filter mobo

At last, I found some time to play with expoxy.

As I still cannot make a real “black” film, I’m still unable to use photo processing to print my pcb. For this reason, I used the “laser toner transfer” method. And, after several unsucsessful tests, I found the good temperature, the right paper, the good pressure, the best toothbrush (to scrub the paper)…

… And I got this. It’s my first pcb done with this process.

Phase I : printing, transfering (with a laminator), scrubing and etching

The “white dust” covering the surface is the paper left on the toner

to reach this quality, I had to sacrifice about 10 sheets of paper. Borders are absolutely clean. The only drawback of this method is that important flat surfaces are far from perfect. The toner deposit is not constant.

Phase II : cleaning the first side

After cleaning the board… it seems to be usable. I hope the irregularity of the main ground plane won’t affect too much the quality of the circuit

the following picture shows a filter daugter card and the interface described some time ago (level interface TTL to 3.3v, and Audio interface with a pair of Triad transformers). Laser transfer technique works better on small boards. The trace between the two circuits is a 0.010 inch line (0.25 mm), without any cut.

Phase III : printing the other side (ground plane and vias)


positioning the other side after drilling 6 marks

drilling the board

I broke my last 0,9 drill, which is absolutely necessary for the HE14 right angle connectors. I’ve no more 1 and 1.1 mm drill…

The first daughter card was bad (connectors where not in front of mobo’s connectors). Dimensions on the cad software are right. It’s probably a wrong printing ratio or something like that.

Next step : tinning the board to protect it against oxidation, plating the vias with silver compound… and looking for drill on ebay

The small company I spotted to print my filter mobo is unable to draw .010 traces. I had to partially redesign the data bus on the central “island”.

As I was under my CAD software (still the same… I know, it’s a shame) I took some time to make a I2C command board (that’s Mike’s idea)

above, the redisigned BCD daughter board (I made some mistake on the former design, the central connector had a quarter inch ofset)

the lower board is the I2C one. The electronic diagram has been “pirated” from the superb CDG2000 schematic

Still the same size, still the same footprint. I think I will draw a simple “parallel hard wired daughter card” for people wishing to use a basic rotating switch, just to keep the design universal and adaptative “on the go”.


Meanwhile, to test my first “home made pcb etching lab”, I’ve printed the I0CG “SSB K1 mod” . First results are “just usable”… I’m far from a real “professional looking” board. But it’s encouraging.

Anyhow, this is another story.. let’s focus on the filter. Next time, I hope I’ll be able to post some picture of the filter mobo

Non, je n’étais pas mort

Publié: 6 février 2009 dans Filter mobo

Cette longue période de silence était provoquée par… le boulot et la conception (l’adaptation serait un mot plus juste) d’un filtre plus ou moins universel, pouvant jouer le rôle de passe bande ou passe bas, à la réception ou à l’émission-réception (tant que la puissance de sortie ne dépasse pas 30 W)

Cela faisait déjà longtemps que je ruminais un tel projet. Plusieurs schémas de principe avaient été jeté sur papier, sans grand résultat jusqu’à présent. Puis, sous l’impulsion et l’énergie d’Alex et Mike, j’ai repris le collier, sous leurs indications et leurs conseils techniques.

Le but était surtout de pouvoir proposer un filtre simple à construire, offrant si possible un “gain” (terme impropre, je sais) de 60 à 80dB et commuté de manière la plus propre possible. Il devait être  adaptable tant aux récepteurs “general coverage” qu’aux systèmes amateurs plus “pointus”, pouvant accepter un pilotage

– soit par I2C,

– soit par bus de 4 bits sortant d’un microcontroleur (pour Rocky ou PowerSDR),

– soit par adressage parallèle direct (commutateur rotatif).

Egalement au cahier des charges, la possibilité de monter une version “de base” couvrant de 3,5 à 30 MHz en 8 segments de bande, ou une version “étendue” acceptant 16 filtres, de 160 m à 6 m (1,8 à 54 MHz), avec les gadgets habituels (dont un “passe tout”, un commutateur d’antenne H/V et, par la suite, les accessoires indispensables tels que : préampli réception, atténuateur programmable…).

Le choix des filtres est laissé à l’utilisateur : soit un passe bas à réjection de type Wetherhold, soit un passe bande 3 bobines copié à PA3AKE. En outre, les filtres étant montés sur des cartes filles purement passives (pas de relais comme c’est le cas sur le filtre PA3AKE), rien n’interdit à chacun d’y monter le type d’elliptique  ou de Cheb qui lui plaira. La base est assez saine pour être utilisée comme plateforme de test et d’expérimentation “in vivo”.


Description rapide :

La carte de commande. Elle est amovible optionnelle, et modulaire Sans elle, il est possible d’utiliser le filtre “à l’ancienne”, en commutant les bandes avec un selecteur 8 ou 16 position (ou moins… ou plus, il n’y a pas de véritable limite si ce n’est la place que tout çà peut prendre).

la carte de commande reçoit l’ordre 4 bits du microcontrôleur (à l’origine du projet, un UWB programmé par 9V1AL,compatible avec les softs Rocky et PowerSDR). Ce mot est directement converti par un décodeur octal (4028) , puis “bufferisé” par  un uln2803. Le choix d’un cicuit cMos permet d’accepter des data au niveau Cmos ou TTL (voir 3,3V même).

La sortie octal amplifiée est utilisée pour commuter les relais qui sélectionneront le filtre de bande adéquat.

La version “de base 8 filtres” n’utilise bien entendu qu’un seul 4028 et un seul uln2803. La version 16 filtres multiplie tout par deux… les circuits de commande ET la carte mère supportant les filtres (c’est la même carte mère qui est utilisé, conçue pour jouer à la fois le rôle de carte “stand alone”  “master” ou “slave”)


Le filtre :

A gauche, le bus de commutation provenant de la carte contrôleur. Chaque section du filtre est une copie servile du pa3ake, mais miniaturisée avec des tores T50 ou T68. Les capas sont soit, dans le cas d’un récepteur, des cms format 1206 en NP0, soit, pour les chasseurs de dB, des styroflex, soit, si le filtre sert également à l’émission, des mica argenté spécifiées à 50 V.


Les filtres sont physiquement montés sur une “carte fille” qui se branche entre les deux relais. La carte mère n’est donc qu’une série de bus (de commande, HF, de signalisation, d’alimentation…)

Aperçu d’un des filtres et de la carte de commande

aperçu de la carte mère : (les cartes filles sont connectées perpendiculairement à la carte mère, la carte de commande est située sur la droite)



On remarque que l’entrée HF est à droite.

A gauche, les SMA de sortie servent à cascader deux “cartes mères” (ou plus)  pour disposer de 15 bandes filtrées et d’un “passe tout”.

Si la face supérieure est pratiquement aveugle, afin de présenter un plan de masse le plus continu possible et ainsi améliorer le facteur de forme des filtres , la “face cuivre” est un peu plus travaillée

A vue de nez, la carte mère est proche de l’échelle 1 (les cartes fille “filtre” et “commande” sont agrandies.. l’on peut se repérer grâce aux connecteurs latéraux, qui sont des HE14 au pas de 2,54.

Le soft de cao utilisé étant très limité, il faut savoir que les relais sont situé SOUS la carte mère, ainsi que les diodes “antiretour”? Sur la face supérieure, les connecteurs des filtres et carte de commande, les prises d’extension , d’alim et  entrée/sortie HF, les “”straps” chokés avec des ferrites, et un “gadget” : un couple résistance/diode LED pour indiquer directement le filtre en fonctionnement.

les capas de découplage utilisées sont  au format 1206, les diodes au format 1206/miniself, les relais se trouvent pour quelques centimes chez Selectronic, les connecteurs soit sur ebay, soit chez Gotronic.


A noter que le “chenal central” devrait pouvoir être enfermé dans un blindage en “U” afin d’améliorer le niveau d’isolation d’entrée/sortie et minimiser les risques de ré-injection dans les lignes d’alim (qui, chokées et découplées, on déjà peu de chance d’accrocher).

Une semaine de boulot acharné m’attend. Le proto ne sera pas tiré avant une dizaine de jours.

Toute remarque sur la conception générale est la bienvenue… cela va sans dire.