NPR New Packet Radio

IP over 430MHz Ham Radio, up to 500kbps, 20W RF.
Extension for HSMM-Hamnet-AREDN. 100% open-source.

Public Chat
Similar projects worth following
NPR (New Packet Radio) is a custom radio protocol, designed to transport bidirectional IP trafic over 430MHz radio links (ham radio frequencies 420-450MHz). This protocol is optimized for "point to multipoint" topology, with the help of managed-TDMA.
Bitrate is up to 500kbps (net, effective bitrate).
The radio modem is cheap (~80$) and home-made, with a 433MHz ISM module inside. The modem is connected locally with Ethernet, therefore no specific software is needed on PC.
The modem is easy to build and reproduce.
You can add a DMR radio amplifier in order to achieve 20W radio power or more.
The main usage is an extension of HSMM - Hamnet - AREDN networks.
All the project is open-source : hardware, software, protocol specification.

NPR kits FOR SALE: sale temporarilly stopped. A new sale will start before end of october, probably world-wide. 

If you want to discuss about NPR:

watch full screen.


Introduction and
quick start guide
et guide de
démarrage rapide
Advanced user guide (EN & FR)
Modem assembly and programing guide (EN & FR)
Spectrum Occupancy
NPR protocol Specification (EN)

PCB Design (Gerbers, BOM, etc...)

PCB v02
BOM Easy to provision
Total cost ~80 euros
per modem
PCB v03
Cheaper BOM
Total cost ~65 euros
per modem
PCB v04
Version sold by
Fits its enclosure.
(Bill Of Materials)
(Bill Of Materials)
PCB files PCB files PCB files

Modem Firmware : binaries and sources

Last stable version
Beta version
Binary file, band 70cm Binary file, band 70cm
Binary file, band 2m Binary file, band 2m
Release note Release note
Source code Source code

x-zip-compressed - 5.31 MB - 10/20/2019 at 12:50


octet-stream - 174.73 kB - 10/20/2019 at 12:46


octet-stream - 174.73 kB - 10/20/2019 at 12:46


text/plain - 2.75 kB - 10/20/2019 at 12:46


plain - 2.54 kB - 10/20/2019 at 12:46


View all 19 files

  • 6- Last news, updates

    f4hdk05/12/2019 at 06:28 0 comments

    29th Sept 2019:

    • I have tested the PA (power amplifier) P8X from DF2FQ. Not recommended for use with NPR, because it does not integrate an RX-preamplifier, which is needed in order to compensate poor RF4463 sensitivity. Refer to "Advanced user guide" annex 4.

    15th Sept 2019:

    • Sale is temporarily stopped. A new sale will start end of october, shipping world-wide.
    • New modulations parameters. I switch to GFSK instead of GMSK, with double deviation (compared to GMSK), because radio sensitivity is greatly improved (4dB or more). The radio chip that I use is too bad for GMSK, much better for GFSK, refer to (french) discussion here. WARNING: the new modulation parameters are not compatible with previous ones. Radio bandwidth can be increased, refer to new documentation.
    • New firmware released with lots of features : new modulation parameters, frequency resolution improved to 1kHz, Frequency shift feature (currently useless), and new command "version". Refer to release notes.
    • A new dedicated firmware is released for 2m band, with the same source code. (mainly for UK). 
    • For your information, the next part of the development will be the "Frequency Division Duplex" feature, with 2 modems at Master side (one for RX, the other for TX), plus one RF duplexer.

    9th July 2019:

    16th June 2019:

    New firmware with 3 new features. Documentation updated accordingly.

    • New modulations (11, 12, 20, 21) for lower datarates, lower bandwidth. Symbol rates: 56kS/s (complies with FCC) and 120kS/s.
    • Frequency range extended to 420-450MHz instead of previously 430-440MHz
    • Periodic SI4463 temperature check, and chip re-calibration if delta temperature above 15°C.

    2nd June 2019:

    • Compatibility of DMR RF amplifier "" set to "NO" inside "advanced user guide".
    • 1st tests of VoIP over NPR. Good in most condition, surely because I have a router with QoS at Master side. Would be better with true "QoS" inside NPR modems.

    14 may 2019 :

    • Horizontal polarization recommendation removed from documentation (because IARU do not recommend anymore particular polarization)

    12 may 2019 :

    • introduction of new PCB design v03, cheaper. (The version 02 can still be used)
    • New blog post "planned and possible evoltutions"

  • 5- Planned or possible evolutions

    f4hdk05/12/2019 at 06:26 0 comments

    Please feel free to comment these lists of evolution, and express your requests, or your priorities. Either here or in the public chat.

    Planned evolutions

    • Better FEC algorithm, because currently it is very poor. If you can give me some help, I don't understand a lot standard FEC algo (with all these Maths).
    • Regular temperature check and re-calibration of SI4463 if necessary -> done! (firmware 2019_06_16 and above)
    • Increase to 15 clients instead of currently 7 maxi. → Probably not feasible with current hardware at Master side, not enough RAM (new hardware under study)

    Possible evolutions, if people are interested (raw order, no priority order):

    • Static IP management of some modems → Removed because you can use future "L2 pure ethernet transport feature" which will be more flexible
    • Management via pure serial port (instead of USB)
    • L2 Ethernet transport configuration (instead of currently L3 IPv4); could enable more flexibles topologies, and IPv6 compatibility. But a little bit less efficient.
    • Extension of frequency range to 420-450MHz instead of currently 430-440MHz (IARU region 1 limits) -> done! (firmware 2019_06_16 and above)
    • Other modulation parameters, with lower datarates, in order to decrease RF bandwidth -> Done (firmware 2019_06_16 and above)
    • QoS necessary in order to transport VoIP data over an NPR network

    Very hypothetical evolution:

    • Totally new design, microcontroller with more RAM, PCB optimized for automatic PCB assembly line. Goal would be to mass produce fully assembled modems. Probability low, only if enough people are interested.
    • Frequencies split between uplink and downlink (in case you use NPR Master in a tower where a UHF FM repeater is already present). Additionally, the Master could be used in "FDD" (Frequency Division Duplex) mode in order to use these 2 frequencies simultaneously, like 2G or 3G mobile phones. 2 modems required at Master side. Complex!

  • 4- FOR SALE! Kits (New Packet Radio modems).

    f4hdk04/22/2019 at 06:34 0 comments

    Sale is temporarily stopped.

    A new sale will start before end of october (2019).

    If you want to sell kits from/to a particular region (either Europe or US or another), then I can add your contact e-mail inside this blog post.

  • 3- Firmware history

    f4hdk04/22/2019 at 06:21 0 comments

    for later usage

  • 2- Deployment, NPR networks, tests

    f4hdk04/22/2019 at 06:20 0 comments
  • 1- Pictures

    f4hdk04/22/2019 at 06:19 0 comments

    For later usage.

View all 6 project logs

Enjoy this project?



harzi wrote 07/19/2019 at 06:58 point

Dear Guillaume
I am using FW 2019_04_06 and have one master and two clients working without booster properly.

The gotechnique-booster weren't switching fast enough from RX to TX, but are transmitting also by the clients (anyhow, they have been withdrawn in newer guides as working compatible with NPR).

The booster VR-P25D (UHF) works only on the master. 'set RF_power 8' and higher let the VR-P25D transmit. Lower values 'RF_power 7' or less turn the booster TX off. The level is similar as the gotechnique-booster.

Unfortunate the two clients aren't able to switch the VR-P25D booster to TX. I also tried 'radio off' and 'TX_test', without success.

Do you have any advise? Does the new beta FW help and is it stable enough?

Best regards, HB9GVD

  Are you sure? yes | no

IZ7BOJ wrote 04/13/2019 at 12:22 point

Hi Guillaume, I’m following NPR since the beginning, but I didn’t find any useful detailed study about that before your nice article.

I practiced packet radio for many years, I’ve built several S53MV wide-band radios, experimented TCP-IP over ax25 and I regret that period!

Now I’m very excited about NPR. I’m trying to involve other local OMs in order to build a small experimental network.

Thinking in a “daydreaming way”, IMO one limitation in the protocol could be the maximum number of clients (7).

What appens to the 8th client which send a connection request? Just receive a reject like the AX25 “BUSY”? Then the user should periodically retry until one of the clients disconnects from the master?

Best 73’s

Alfredo IZ7BOJ

  Are you sure? yes | no

f4hdk wrote 04/14/2019 at 06:54 point

Thanks a lot Alfredo for your feedback. 

Answer to your question : if a 8th client tries to connect, then it receives a "NACK" (reject from the Master), and the client modem tries periodically (every 15 sec) to send new connection request, and therefore it will automatically connect as soon as a "client slot" is free.

Initially, my idea was that each client should use its connection for a short period of time (several dozens of minutes) each day. It is currently not designed for 24H/24 connection.

But if there are lots of demand, I will think about increasing this limit to 16 or 32 clients. I will need to make software modification at Master side only.

  Are you sure? yes | no

stevewilliam007 wrote 04/12/2019 at 07:36 point

I have been searching for the packet radio and recently heard from about the newly releasing packet radio which has helped me a lt. There are several latest technologies which are trending in the market.

  Are you sure? yes | no

f1rzn wrote 04/05/2019 at 08:32 point

Ca fait des années que j'espère quelquechose dans ce style, qui nous redonne les mêmes possibilités que le packet radio mais en version moderne. Je me suis permis de partager sur le groupe AREDN (le successeur de BBHN) et sur 2 ou 3 autres pages. Ca me parait très complémentaire de l'AREDN. Franchement bravo et merci. J'espère qu'un jour on aura un satellite geostationnaire qui viendra compléter les possibilités terrestres que cela offre. Ca manque un réseau mondial de données radioamateur.

  Are you sure? yes | no

f4hdk wrote 04/05/2019 at 12:27 point

Merci beaucoup pour ton retour. Préviens-moi si tu veux tester cette solution. Tu as mon e-mail si besoin : f4hdk[at]free[dot]fr

  Are you sure? yes | no

F4HZO wrote 04/03/2019 at 20:06 point

Super ce système cela offre des possibilités inédites en matière de couverture Hamnet surtout pour les communications d'urgences !

Merci de partager la connaissance !!

Meilleurs 73 de Florent F4HZO

  Are you sure? yes | no

f4hdk wrote 04/04/2019 at 17:17 point


  Are you sure? yes | no

bferrell wrote 04/03/2019 at 19:58 point

In regards the 433Mhz ISM module, Adafruit has one for just a bit under $10 US

  Are you sure? yes | no

DL4APT wrote 04/03/2019 at 12:06 point

if i find the time i will look into modifying the design to use this much cheaper Ethernet module (same chipset but less then 4€ compared to the 20€)

  Are you sure? yes | no

Jackson wrote 04/03/2019 at 03:56 point

The DMR radio amplifier is available here:

Up to 40W output, really powerful

  Are you sure? yes | no

Richard Green wrote 04/02/2019 at 21:45 point

I'd like to experiment with this in the USA.  I can't seem to find a source for several of the critical parts.  Farnell has a subsidiary nearby, called Newark Element14, and all the small parts are available there.  The PCB I can get manufactured locally.  But two of the modules seem to be unavailable.  The itead LM2596 voltage regulator module is listed as 'Retired' on their website, and I have not found any source for the RF4463 F30 module at all.  aliexpress carries a 'demo board' carrier for it and similar, but not the modules themselves.

If I can find sources for the parts, I'd consider doing a mass order, and making up kits to sell at the big Hamvention next month.

I do have one concern about the protocol itself.  For the most part, NPR concerns itself only with the Physical and link layers, but why does it also act as a DHCP server?  Why not keep it pure and leave that to the IPv4 router?  And allow other protocols (ICMP, IPv6) as well?

  Are you sure? yes | no

f4hdk wrote 04/03/2019 at 03:23 point

LM2596 module is available here (and many other places) :

RF4463F30 :


About your last question: a NPR client acts as a DHCP server, because inside a NPR network, IP are assigned dynamically among clients, and the protocol do not carry Ethernet overhead but diretcly carry IP packets (therefore DHCP is not relevant in the radio part). I plan to allow "static IP" assignment for clients, but I don't know when I will be able to program it., and/or if the small microconrtoller will have enough ressources. 

  Are you sure? yes | no

Richard Green wrote 04/03/2019 at 13:25 point

OK, I'm beginning to get it.  I ran into this once before with an external Wifi-Ethernet bridge called a 'game adapter' which did not carry ethernet MAC of the client across the Wifi link, but chose proxy ARP with its own MAC instead.  Therefore, DHCP server on router would only hand out a single IP lease. Not suitable for multiple client machines at remote end.

  Better to carry full ethernet header within NPR.  More overhead, but more universal application.

  Are you sure? yes | no

f4hdk wrote 04/03/2019 at 16:16 point

@Richard Green With NPR,  can of course put several IP machines (PCs) behind one single client modem. Just set 'client_req_size' to more than one. Take a look at "advanced user guide", especially 'annex 5' and 'chapter 6.2'.

  Are you sure? yes | no

Richard Green wrote 04/03/2019 at 16:29 point

Yes, I saw that.  I still see that as a workaround, a hack, rather than a simple, standards-compliant solution. It would be far better, more adaptable, more universal, if NPR functioned as a fully transparent network bridge in the ISO sense.

  Are you sure? yes | no

RipperFox wrote 04/08/2019 at 02:14 point

About the microcontroller: Ever thought about using an ESP32? I think they are  cheaper and even faster + come with WiFi, wired ethernet can be added with a 1€ board (see e.g.
Thanks for your project!

  Are you sure? yes | no

DL4APT wrote 04/03/2019 at 10:40 point

Nice RF has an aliexpress store: and they also have the RF4463F30 on sale (a lot of them just don't use a space in the name an the search will find a lot or go on the nicerf store to the menue point silconlabs)

  Are you sure? yes | no

Daniel Roseman wrote 04/01/2019 at 01:11 point

Question for you. Does this protocol pause to identify itself? If so, how? What happens if I don't 'set' the callsign in the firmware? I read all of the docs, and I can't find where it identifies itself to the world at large.

  Are you sure? yes | no

f4hdk wrote 04/01/2019 at 04:37 point

Sorry, I'm not sure that I understand the question "Does this protocol pause to identify itself?" English is not my native language.

You can find some answers in the "protocol specification", paragraph 6.3.

  Are you sure? yes | no

Daniel Roseman wrote 04/01/2019 at 20:59 point

Under American law, Amateur radio stations have to pause and identify themselves at least every ten minutes. 

47 CFR § 97.119 - Station identification.

  Are you sure? yes | no

f4hdk wrote 04/01/2019 at 21:04 point

OK Daniel. The NPR modems transmit their callsign (plus the callsign of the peer modem) every 2 seconds. The callsign can contain up to 13 characters.

Do you think that it's enough reguargind the US law that you mention?

  Are you sure? yes | no

robert.ehresman wrote 04/01/2019 at 21:47 point

I do not believe the Who message as described in 6.3 would be compliant station identification in the US. Station identification requires inclusion of the Call Sign. While quite useful, the Who message information would also not be legally necessary here.

Sorry I was typing this while you answered Daniel. 13 characters would be sufficient.

But it sounds like the specification needs an edit.

  Are you sure? yes | no

f4hdk wrote 04/02/2019 at 03:19 point

@robert.ehresman I do not understand your point. Why do you think that the WHO feature of my "New Packet Radio" is not compliant? It already includes the callsign. Refer to §6.2 of the spec also.

  Are you sure? yes | no

fabian wrote 03/31/2019 at 18:19 point

please add linux. And mobile power 24h/365 day

  Are you sure? yes | no

f4hdk wrote 03/31/2019 at 18:46 point

The modem is OS independant. It connects to a host (PC) via Ethernet. Therefore, it is already compatible with Linux.

Have you read all the documentation?

About "mobile power 24h/365d", I have not understood what you expect. Please clarify that.

  Are you sure? yes | no

fabian wrote 04/01/2019 at 11:06 point

comaptible != linux inside and power offgrid

  Are you sure? yes | no

Ken N8KH wrote 03/27/2019 at 02:47 point
I have not yet received an answer to my question: are kits available. Are the motherboard (and daughterboards) available for purchase? How much??

  Are you sure? yes | no

f4hdk wrote 03/27/2019 at 04:23 point

Hello Ken. I'm only a hobbyist, I have other things to do in my life, that's why I don't reply rapidly. Please be patient.

About kits, I don't know if I will be able to ship outside Western-Europe. If someone in the US can organize that for US people, it would be great. 

The price for one modem kit is about 80€ here, plus shipping.

Please send me an e-mail to f4hdk[at]free[dot]fr and I will keep you informed. But once again, you will have to be patient, it will take weeks, months to organize that.

  Are you sure? yes | no

Ken N8KH wrote 03/28/2019 at 02:24 point

I'm very interested in this project, and also helping you to write software and/or develop hardware.  I've been interested in high-speed data radio for about 2 decades or more.

I've also looked into the Silicon Labs chip-sets in the past, and found them very interesting.  I'm glad to see that you've taken my ideas with Silicon Labs chips farther than I ever did.

  Are you sure? yes | no

k0elr1 wrote 03/28/2019 at 00:03 point

I wouldn't get too excited about this. By my reading of the US rules (specifically, parts 97.305(c) and 97.307(6), data transmissions are limited to 56 kilobaud and 100 kHz bandwidth in the 70 cm band. I could be wrong, but I'd check before I ordered anything.

  Are you sure? yes | no

Ken N8KH wrote 03/28/2019 at 02:21 point

> data transmissions are limited to 56 kilobaud

> and 100 kHz bandwidth in the 70 cm band

You are correct.

But it is a stupid rule.  It should be rescinded.

And I understand that it is fairly easy to get an experimental license to operate in (or out of) the amateur bands.  So it could be used under an experimental license.

  Are you sure? yes | no

robert.ehresman wrote 04/01/2019 at 21:02 point

Gentlemen, I direct you here for the 2005 ARRL legal brief position on broadband data transmissions in 70cm as they interpret CFR 47 Part 97.3 FCC regulations.

Read it yourself, of course, but the summary conclusion by the League legal team is that if a data transmission method MAY be used to carry encoded video combined with other information (whether the video is actually present or not) the FCC would therefore classify the emission as Amateur Television, and the allowable transmission bandwidth can be at least 9 MHz on amateur allocations above 420 MHz. I am aware there are some groups and individuals in the US that have been operating HSMM or Broadband Hamnet style 802.11 based mesh radios in the 70cm band (radios by Xagyl or Doodle Labs) for years. This has probably never been tested in court or as far as I know any sort of enforcement action mind you, but that is why we have the League and why they established such a position at the beginnings of HSMM. Regionally there may be a conflict with common practice such as properly coordinated ATV repeater coverage. I believe effective video teleconferencing has been accomplished at rates well below 500 kbps for decades. 

  Are you sure? yes | no

cottsak wrote 03/26/2019 at 06:10 point

What's the point-to-point tested range and some throughput data for said test? Also do you have images of the antennas?

  Are you sure? yes | no

f4hdk wrote 03/27/2019 at 05:45 point

We have not tested maximum distance for point to point links. But you can achieve big distances with high gain Yagi antennas at both sides of the link. 

  Are you sure? yes | no

Ken N8KH wrote 03/28/2019 at 02:22 point

My napkin calculations about a decade ago resulted in about a 25 mile range for two stations with 200 kHz bandwidth, 20 W each, and a vertical omni-directional antenna with about 6 dB of gain, and some moderate feedline loss.

  Are you sure? yes | no

f4hdk wrote 03/28/2019 at 13:04 point

@Ken N8KH  about your distance calculation...

Warning, omnidirectional antennas are hard to use at such high Symbol Rates (up to 500kS/s), due to "multipath" issues. You should prefer directional (Yagi) antennas, at least on one side of the link (client side). I will probably add explaination about that inside the documentation. 

If you want to discuss more about that, do not hesitate to switch to the "public chat room" of my project. 

  Are you sure? yes | no

Ken N8KH wrote 03/26/2019 at 03:35 point

Are kits available for the modem?

  Are you sure? yes | no

shawn wrote 03/25/2019 at 16:38 point

I see you mention a kit in your presentation.  Count me in if you are trying to gauge overall interest.

  Are you sure? yes | no

Stuart Longland wrote 03/24/2019 at 09:05 point

How wide are the channels to produce 500kbps data rates?

  Are you sure? yes | no

f4hdk wrote 03/24/2019 at 09:22 point

Hello, you can refer to the table on page 18 of the "introduction document": RF bandwidth is 750kHz for 500kbps of net-effective datarate, over a 1Mbps channel. It is similar to DATV SR-500kS/s. Like for DATV, we use directional antennas.

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates