NinoTNC N9600A Info&Ordering
page modified December 12, 2022
Specifications may change without notice.
Do not use this device in any situation where the loss of life or property would be the result of the device misbehaving or failing.
If you go beyond this rule, you are self-certifying this device.
This device is built by hobbyists for a hobby project.
We do not certify that this design, or any particular unit, is sound.
No UL or CE listing or testing!
Use it at your own risk.
9600 baud packet radio TNC
The NinoTNC N9600A is a very good TNC.
The N9600A is a USB interfaced 4-bit-rate USB KISS TNC, supporting 1200 baud (Bell 202), 2400 baud APSK, 4800 baud GFSK, and 9600 baud (G3RUH).
The NinoTNC is designed by Nino KK4HEJ for use in Amateur Radio packet data networks.
The 2400 baud APSK modem supports many microphone-audio transceivers, and 4800 baud GFSK is great for use in data-radio/9600 transceivers which don't quite cut it for 9600 baud.
On top of that, it supports both AX.25 for compatability with legacy TNCs, and a new link-layer encoding scheme called IL2P which is a Forward Error Correction equipped protocol designed to be more efficient than AX.25.
The N9600A NinoTNC provides a USB serial interface at 57600 baud.
Brand new TNC
As of this writing, over 1500 of the NinoTNC printed circuit board and CPU have been shipped, starting in March 2020, and over 120 of the units are in service in the NCPACKET off-the-grid social-media network in North Carolina, USA.
See the complete history of the NinoTNC project.
It is going to be up to the cadre of volunteers (that means you) to figure out how to hook this up to each and every 9600 capable transceiver.
We created a g3ruh-modem email reflector to try to accumulate knowledge about interfacing G3RUH TNCs to every radio it will work with.
We'll harvest the best stories from the g3ruh-modem reflector and put it on a web page linked to this website.
We're all volunteers and doing this as a hobby.
9600 baud has been tested and working with Yeasu FT-817, Kenwood TK8180, Tait TM8105, Kenwood TK760, Kenwood TK862G.
In some testing we have used a reference Kenwood 9600 equipped HT, but in other testing we only used pairs of matching radios.
The Kenwood TK8180 required a very strong signal to work at 9600 baud, but it seems to do 4800 pretty well.
NinoTNC did NOT work with Maxon SD125 in our 2020 tests, though we have a suggested work-around/mod, we haven't tested it yet.
We had mixed results using 9600 baud with the Alinco DR235mkIII.
The packets passed through a pair of DR235mkIII at short range, but the TXDELAY had to be attrociously long, longer than a 240 byte long packet by a factor of at least 2:1.
So far we haven't tested the Alinco at long range and probably won't bother because of the rediculous TXDELAY requirement.
Perhaps the Alincos have a future in one-way packet channels?
If you get results from testing the NinoTNC, please post on the email reflector.
Funding, Support, Warranty (not) and our organization
Our plan is to keep the storefront open for pre-orders for the current A4 version of the board but we'll only be ordering boards four hundred at a time.
We're doing this all out of pocket.
As of April 2021 we are caught up with demand but shipping is a volunteer effort so bear with us.
At the worst of times we'll be back-ordered a month or less.
There is no money built in for warranty service or returns so please don't count on any.
Our purpose is to streamline the process so there won't be any mistakes and we're doing this for the glory of the hobby.
That said, if you can participate in giving quality feedback which manifestly advances the project, we'll be doubly
interested in keeping you, particularly, supplied with new and improved firmware and/or board modification instructions.
Please join the TARPN groups.io or the NinoTNC Applications reflector
to be sent updated info.
If you feel that factory tech support and warranty replacement is in your future, I suggest teaming up with some other packeteers in your area.
With a group working on the project, the talent and capability scale well.
If you aren't the best solution-finder for tech issues, perhaps team management and project director is a missing part in the local ham scene.
There is nothing better in Amateur Radio than collaborating on a big project.
Updates and News
Check back here for updates. Also, check the fora
for the latest info.
See the news down below!
How to buy
You'll do two purchases.
TARPN is buying, programming, and then shipping the CPU with the PCB.
Buy these two items at
for $12USD plus tax and shipping.
( if that link doesn't work, try ETSY for N9600A4 )
You'll get the remainder of the parts from Mouser for about $26USD plus shipping for a single board.
For 100 boards the price is down to closer to $18 each.
Find the Mouser ordering steps on the assembly page.
Find our assembly pages on the NinoTNC link page
Test Results and Reviews
The 9600baud G3RUH NinoTNC, with an appropriately fast switching transceiver, achieves 250 to 300 bytes per second on a TARPN link (maxframe=1, ppersist=210).
TARPN puts a priority on low latency, more than high throughput.
We'd love to hear test results using the TNC in a more aggressive mode.
Please post to our reflectors
If you have good luck with the unit, we'd appreciate a review in the forums including
QRZ Homebrew and Kit Projects
If there is no review of the N9600A in the Eham reviews under PACKET TNC, feel free to start one.
It is certainly worth mentioning in the first review:
- the unit is a kit
- sold by a bunch of hams for cost,
- the link to the instructions is on the board,
- the parts mostly get bought by the user directly from Mouser,
- and the total of CPU, PCB and rest of the parts is something under $40.
That would be great!
Transfer rate testing
The TARPN networks use MAXFRAME of 1 for site to site links.
This is done to keep the latency as short as possible which is appropriate for our CHAT network.
Because the testing recorded here was done that way, high speed throughput might not be as good as achievable in BBS forwarding networks.
Test setup: TARPN stations using Raspberry Pi and N9600A3 at each end.
Persist 210, FRACK 4000, MAXFRAME 1, PACLEN 236, SLOTTIME 20.
Network connection made from A to B, B to A.
Elapsed time to transfer bpq32.txt in the BBS.
These links are extremely high SNR (radios at same location).
|tnc to tnc||Mod-ulation||L2|
|Radio A||Radio B||Notes||
|Wire ||AFSK ||AX.25 ||1200 ||181 ||5992 ||33.10 || || |
|Wire ||AFSK ||IL2P ||1200 ||178 ||5992 ||33.66 || || |
|Radio ||AFSK ||IL2P ||1200 ||196 ||5992 ||30.57 ||TM8105 ||TK760 |
|Wire ||APSK ||AX.25 ||2400 ||89 ||5992 ||67.33 || || |
|Wire ||APSK ||IL2P ||2400 ||89 ||5992 ||67.33 || || |
|Radio ||APSK ||IL2P ||2400 ||111 ||5992 ||53.98 ||TM8105 ||TK760 |
|Radio ||APSK ||IL2P ||2400 ||99 ||5992 ||60.53 ||TM8105 ||TM8105 |
|Wire ||GFSK ||AX.25 ||4800 ||59 ||5992 ||101.56 || || |
|Wire ||GFSK ||IL2P ||4800 ||61 ||5992 ||98.23 || || |
|Wire ||GFSK ||AX.25 ||9600 ||48 ||5992 ||124.83 || || |
|Wire ||GFSK ||IL2P ||9600 ||49 ||5992 ||122.29 || || |
|Radio ||GFSK ||AX.25 ||9600 ||49 ||5992 ||122.29 ||TM8105 ||TM8105 |
|Radio ||GFSK ||IL2P ||9600 ||49 ||5992 ||122.29 ||TM8105 ||TM8105 |
News: February, 2022
We did a few things to fix the IRFD123 issue.
Nino has parts for anybody who needs one and can't get it from Mouser.
We're moving our BOM and ordering site for US customers to Mouser instead of Digikey.
Check for postings with "PTT Q1" in the title for latest info.
News January 24, 2022
We made a mistake in October 2021.
We had revised the BOM and substituted the IRFD123 FET for Q1 (PTT transistor) when the IRLD120 went out of stock.
It turns out that this was a less than perfect substitution. However, the IRLD120 was apparently discontinued!
It looks like the new part doesn't key up as many radios as the original part.
We're scrambling to obtain old-stock quantity of the IRLD120 which we'll probably
start shipping with the CPU+PCB order until we come up with a new plan.
As of today the 3oo BOM has an intentional obviously bad part for Q1 to direct users of the BOM to our Groups io page.
Check the NinoTNC groups.io email reflector in the mean time for updates and for suggestions.
Look for postings with "PTT Q1" in the title.
Thanks for any help.
News November 21, 2021
Another order of 200 boards and CPUs is on order.
We have 81 in stock from the Sept/Oct order/delivery.
We're selling about 3.3 per day over the past month.
News Oct 27 2021
RV1 is back in stock.
News October 2021
Another order of 200 boards and CPUs is available.
The price went up a little from $9.29 to $9.59.
For the first time since we started the project, we absolutely could not find a substitute part at Digikey and so kit assembly is hung up.
The problem is RV1 which is the transmit audio volume control.
RV1 should be 10 K ohms.
A lower value is useable for many radios but we picked 10 K ohms to give a usefully large range of adjuatment.
We expect this shortage to last 2 months, but it is halfway over as I write this, I hope.
News September 2021
The May order 200 boards have all been sold and delivered.
Another order of 200 is on the way.
The ETSY store is marked "sold out" until they arrive.
Assuming no plague related hold-ups, these should be for sale in mid October.
News July 12, 2021
Nino is in the process of selling and buying a house and moving the family.
Availability will be intermittent but we expect the PCB and CPU volumes to be ok all year.
The listing on ETSY is up as of today at NinoTNC PCB and CPU on ETSY for $9.29
News June 24, 2021
Sorry for the long gap in news.
In early May we ordered a batch of 200 boards and "fat" CPUs, and there is some of that stock remaining.
As of right now, the ordering at ETSY is on hold.
Nino is in the process of moving from Texas to Virginia.
That move is staged and will only be complete in September.
It is definitely a time of change.
The process is going to be intermittently consuming Nino's time all the way through September.
We expect the ETSY hold to last until shortly after Independence Day, July 4.
We're still calling the NinoTNC A4r2 boards "current," and we don't have plans right now to make any hardware changes.
This seems to be a good version and it is well featured.
Version 4.06 is shipping with these boards.
News April 23, 2021
The NinoTNC A4r2 order of 400 from March 1 has all been sold.
As I type this, there is 1 left available at ETSY.
We'll be ordering another batch.
Please stand by.
News April 22, 2021
A problem has developed that hopefully doesn’t impact too many people.
Those who have boot loaded from firmware 2.89 or before to firmware 3.04 or earlier will have a situation where they cannot boatload beyond what they have now.
The bug will have already hit your CPU and we don’t have a scheme to fix your CPU besides refreshing it using the PICkit and MPLAB-IPE
The bug does not change the on-the-air functionality of the NinoTNC, just boot-loading.
The problem is instantiated because of a compatibility issue between the original ‘a’ bootloader that came with the Version 2 Firmware, and the Version 3 Firmware.
This bug does not affect version 4 CPUs (comes with the’B’ bootloader,) or CPUs which shipped with version 3.00 (comes with the ‘b’ bootloader), or CPUs which still have version 2.90 or earlier.
The bottom line is, if you have version 2.89 or before running on your NinoTNC, don’t attempt to bootload it to version 3.00, 3.01 or 3.04.
While you will be able to boot the NinoTNC and use it after boot loading to it, the bootloader firmware on your NInoTNC will fail to upgrade during subsequent boatload/flash attempts.
We have a fix for this which will show up in the next version 3 release.
What Nino is doing to fix this is including a feature which will REPLACE the ‘a’ bootloader with the ‘b’ bootloader, so if you boatload to this new version using the built in bootloader ‘a’, after you reboot you’ll now have a bootloader ‘b’ in the NinoTNC.
If you have already gone from 2.XX to 3.XX, then in order to boatload further, you’ll need a reFLASH the CPU.
We can either sell you a replacement CPU at cost + shipping, (please email firstname.lastname@example.org) or we can recommend you to get the Microchip PICKIT programming tool which can over-write the CPUs.
The cost of a PICkit runs between $30 and $100. (Cheapest on ebay?).
It [seems to] require a MSWindows PC to run it.
A third option is to find somebody who is local who has the PICkit device to flash your CPU for you.
If you have a pile of CPUs which show the problem, the shipping won’t be so bad to mail them to one of us and then we can mail them back.
If you are anywhere near Raleigh, let’s have lunch.
News March 1, 2021
The 400 A4r2boards are arrived and Nino is getting read to ship to the 200 or so ETSY purchasers.
Nino has 400 chips in stock and they are programmed with v4.00.
News February 13, 2021
The 400 A4r2 boards are held in limbo at customs.
The FedEx tracking says Delivery Pending, Clearance Delay.
It's been stuck in FedEx since Jan 21st.
News January 7th, 2021
The first batch of 200 A4r2 PCBs and chips sold in about a month, so I just placed production orders for 400 more of each.
I expect to receive them by the end of this month.
Hopefully I'll get to the point where I can have some stock on hand at all times available to ship immediately.
I've switched from the dsPIC33EP256GP502 microcontroller to the dsPIC33EP512GP502.
This was motivated primarily by Microchip's stock shortage for the former part.
The 512GP version has a bit more RAM and ROM, which presents opportunities and challenges.
I thought it would be simple to port firmware from 256GP to 512GP, but the RAM memory map is significantly different between the two parts.
The Extended Data Section (EDS) boundary addresses, and the addresses of the Y Memory section (part of the DSP features) are both different.
I've got that all sorted now, and the firmware on the 512GP chips that just shipped has the same functionality as the newest version of 256GP firmware.
Due to the subtle differences in memory architecture between the chips, there will be separate firmware versions for each.
I've updated the flashtnc.py programming script to both chip variants.
The script now checks the hex file and the part for compatibility to prevent mismatches.
I've also updated the bootloader firmware on the latest chips to be much more resilient to interrupted flash update attempts.
Aborted updates can be restarted and continue to completion. The new bootloader firmware identifies itself as 'B' to the programming script on the 512GP chip, and 'b' on the 256GP chip.
The old version is 'a', and only exists on the 256GP chip. For now, the bootloader firmware cannot update itself, though I have a plan in mind to make that possible in the future.
Since there will be separate firmware versions for the different chip variants, the firmware versioning convention will change slightly as well.
3.xx will be used for 256GP chips and 4.xx will be used for 512GP chips until the next major change.
The caveat to this rule is that I shipped the first batch of 512GP chips with a version that identifies as 2.90.
That version will not exist for 256GP chips, though it's basically equivalent to version 2.89.
These are the final versions of the 2.xx series.
The next batch of 512GP chips will ship with 4.00, which differs primarily in memory management and the size of some buffers.
Firmware update script is available here:
News December 24, 2020
The larger/upgrade-for-us dsPIC is not as far back as it looked. Now looking at first week in January.
The PCBs for the A4r2 are now on hand and we have one built up and tested.
It all looks good.
In other news.. we've sold and shipped over 1000 of the NinoTNC CPU+PCB including the A2 and A3.
Also, 200 A4 pre-orders have almost sold out and that will bring us to over 1200.
We'll be doing another 200 really soon but we'd like to get some experience with kit-builders in the field first, in case there are some tweaks we need to make.
News December 16, 2020
The upgraded dsPIC is now backordered as well as the original dsPIC. At least it is not as backordered as before. Now we have to wait until January 18.
200 CPUs are ordered. Boards are going to be here long before the CPUs.
The good news is that we can get the assembly instructions and operation instructions polished up before the boards ship.
The bad news is that people who were depending on this product are waiting. tap tap tap. We're not happy.
News December 12, 2020
Update on N9600A4r2 shipping timeline for the first batch of 200.
I still haven't received verification samples from Microchip (the carrier has had the shipment somewhere in the USA for the last 5 days).
I decided to place a dsPIC order from the factory today anyhow, and Microchip's part stock has already evaporated, with estimated ship dates in mid January.
So I've messaged all preorder purchasers with an apology and updated ship time, with an offer fully refund the preorder if they chose.
I've also updated the preorder listing on Etsy to a 1 Feb 2021 ship date.
PCB manufacturing completed this week and shipped from China yesterday.
My order of 200 parts from Microchip is placed, with an estimated ship date of 18 Jan 2021 from Thailand.
I'll be flashing the parts here myself, so programming and envelope stuffing will commence as soon as they arrive.
Sorry about the delay, there must be lots of dsPICs stuffing stockings this year!
News December 2, 2020
I just activated the N9600A4 preorder listing on Etsy.
PCBs are being manufactured now, and I'll place a production order for programmed dsPIC 33EP512GP502-E/SP parts as soon as I get the verification samples delivered and tested.
The listing cost increased to $9.29 due to the increased cost of the upgraded dsPIC, and some Etsy fees I mistakenly wasn't accounting for on earlier listings.
This listing is for one PCB and dsPIC:
I'll also list stand-alone dsPICs once I place the production order for those parts.
Expect the price for chip-only to be a little less than $7.
News November 28, 2020
We had a minor crisis where it seems that the CPU we are using for the NinoTNC is backordered until June 2021.
That could slow us up a little. The resolution for the crisis is that we'll move to a bigger RAM chip and turn this into an opportunity.
The downside of that is we probably won't be offering newer firmware (eventually) for the older smaller chip.
The good news is that we can program a bootloader on the new chip and sell it to anybody who wants to update.
If we come up with a real important fix for the older firmware, we might do it onto the old chip's bootloader for the people who never bought the bigger chip.
News November 26, 2020
I'm working on the assembly instructions for the A4 RC10 board shown in the nov 19 news item below.
The assembly instructions are linked here: A4-assembly Instructions
They are about 1/4 done as of Thanksgiving day, and they are my first cut based on the A4RC10 boards.
I will be working on them in-place so come back and see the changes.
The plan right now is to order 200 of this board, changing the label to something other than RC10 and fixing a couple of silk-screen issues.
When we get the 200 boards, I'll get one and reshoot many of the photos and also change the assembly order to something which is easier.
News November 19, 2020
We're expecting to ship this before end of the year but this latest prototype board has so far only been built twice and has not seen a field installation as of today.
It passed all 4 bit-rates and both protocols on the bench. The SIGNALS switches all have the desired effect.
This is the latest prototype board.
News November 7, 2020
We've run out of the A3r3 boards at ETSY.
We've been working on a new version of the TNC, called the A4, for a while.
Hopefully we're at the end of the process and will be making the A4 available for sale via ETSY in a few weeks.
Since October 4th we've changed the DIP-switch allocation so all of the audio signal features are brought together on their own 4-bit DIP-switch.
There will be two DIP switches, each 4-bit, on the A4 board.
News October 4, 2020
We're working on a new version of the TNC which will be called A4 for sale in 2021.
The changes we're adding to the new version are:
- Support for very low RX audio signal, sufficient for TM221's microphone Rx audio port.
- Support for external-transmit-inhibit signal.
- Modification to the firmware removes the need for one of the resistor networks (RN2).
- Modifications to one of the filter circuits removes the need for resistor network (RN1).
- Move the output level jumper to a switch.
Now we have a 5-bit dip-switch.
So far this version would be $1 cheaper than the A3 if the optional RX audio low-signal components are not installed.
We're targetting end of December (or later) for release to customers of the A4.
We've just ordered the 5th prototype PCBs for the A4.
Since this isn't fixing any important bugs in the A3r3 board, we're trying to polish this unit and take care of even minor issues before releasing it.
The new version is expected to run firmware which is also compatible with all of the A2 and A3 versions.
In other news: Versions 2.73 of the firmware is out for A2, A3 and A4, and it pushes the capabilities of the 1200 AX.25 receiver another step.
There is now a USB bootloader for the NinoTNC A2, A3 and A4.
We've been testing it on the Raspberry PI.
Bootloader instructions are in the Operation web page
This is an early prototype of the A4 board:
News Sept 27, 2020
Less than 100 boards left in stock.
News Sept 1, 2020
The 200 more A3r3 boards arrived along with the programmed ICs.
Pre-orders have been fullfilled.
We have 166 A3r3 boards in stock.
News Aug 23, 2020
We're ordering CPUs with version 2.51 which does some substantial improvements to IL2P.
200 more A3r3 boards are on order. Expected to be available to ship via ETSY in early September.
Programmed CPU ICs are expected around September 10th.
News Aug 21, 2020
We received the June 28th order of 200 A3r3 boards and fullfilled ETSY orders with thosw from mid July until mid August.
We suddenly had a much accellerated sales rate, possibly since July CQ magazine article on FEC came out?
A3r3 boards sold out.
That's 200 A2 units sold, plus 200 A3r2 units, plus 400 A3r3 units.
We sold 800 units between March 7th and August 21st.
800 in 24 weeks is 33 per week.
News Jun 28, 2020
The June A3r3 boards are running out.
There are 18 left in stock as of today.
We've sold 200 A2 units, 200 A3r2 units, and almost 200 A3r3 units.
TARPN ordered 200 more of the same version (A3 r3) and those should be available for delivery by July 15 or so.
As of now we've managed to use up 600 NinoTNC boards.
This doesn't count some A2 boards which were discarded, as well as some prototype boards.
It'd be neat to know how many of them actually went on the air.
I built a 6-port NinoTNC node at my house linked to KN4ORB, my office, N3LTV, NC4FG, KM4EP, K4RGN.
My office node has four TNC-PI, and I intend to replace all of those with NinoTNC this week.
Video showing 3 NinoTNCs pounding away at 9600 baud
Some issues with deploying IL2P have been found.
Mostly they can be worked around with "Interesting" KISS sequences pasted into a terminal window.
The latest version firmware fixes these issues, we think.
No progress on bootloader. My fault. Been busy. I'll get to it.
The new version of firmware changes the USB test message sent when TEST-TX is released so that the Raspberry PI receives a dump of diagnostics including the Firmware version number.
News Jun 13, 2020
The June A3 boards (marked r3) have been shipped. There are 60 boards left in stock after the pre-orders were filled.
News Jun 7, 2020
The A3 boards have arrived in San Antonio and look good.
The CPUs are still on-order.
The expected ship date is June 15th or maybe June 13th.
News Jun 1, 2020
TARPN has placed orders for 200 CPUs.
Expected ship date is June 15th.
As of today about 100 of the 200 A3r3 boards are spoken for.
Pre-orders should be done on ETSY.com
We have plans to continue this process.
We'll probably be doing another 200 boards for delivery in mid July.
News May 19, 2020
All of the 187 A3r3 boards for May delivery have been shipped.
We ordered 200 bare A3r3 boards to fullfill May 14 thru June ETSY pre-orders.
We expect to ship this group in early June.
The June board, r3 version, is slightly different from the May shipment in that it has a 2-pin jumper and header "j6" to short out R3.
The jumper would be put in place if using the NinoTNC on a data radio, and removed if using it with a microphone-input.
J6 pins and jumper will be in the Bill Of Materials starting today, for the A3 instructions.
This is only 30 cents worth of components and that will keep hams making June orders from getting a DigiKey kit which is missing those parts.
This minorly revised A3 board has a "r3" mark next to the lower left mounting hole and a 2-pin header labelled "j6" to the upper right of the USB-B socket.
I'll be modifying the A3 instructions to populate J6 for those boards which have that socket.
The instructions for shorting R3 on the bottom of the board (step 12 as they sit today) will be modified to only short R3 if J6 header is not present.
In early July we'll start delivering a new PCB where the 2-pin jumper is replaced by a switch labelled Mic/Data.
That board is going to be called the A4, or more exactly, NinoTNC N9600A4.
The A2, A3 and A4 boards will all run the same firmware.
At some point this year we'll have a program which does a field upgrade of that firmware.
Any firmware after v2.2.0 (shipped starting with the first ETSY orders of A2 boards) should be upgradable over USB.
We'll probably support that software for Raspberry PI only, but we'll work with people to make an MSWindows or other version.
We'll also discuss open sourcing the program or at least the generic parts.
Maybe we'll just write up an algorithm which can be rewritten into whatever programming language is convenient.
It's not too complicated.
News May 17, 2020
The pre-order has run out after selling 187 pre-orders.
The really good news is that the 187 pre-orders for May delivery will start shipping tomorrow, Monday the 18th of May.
R3, 22 KΩ is confusing on the A3 model because it needs to be present if using the TNC with an FM voice radio, but should be shorted out if using the NinoTNC with a 9600 baud capable radio.
In the next spin of the board, which we're going to call the A3 r3 model, we're going to put a 2 pin jumper on the board to let the operator short out the resistor on demand, after assembly.
We're already taking pre-orders for the A3 r3 model for June shipment.
News May 10, 2020
The A3r2 boards have arrived and a few have been built up.
The programmed CPUs are on order.
I'm thinking it is still about a week before shipping to ETSY pre-orders.
Some parts are getting backordered.
When that happens I'm adjusting the BOM to select a different manufactured part.
Previously we'd selected parts for lowest price and with what we thought were adaquate quantities in stock at DigiKey.
Once the parts run out we can select another manufacture, still shipped by Digikey, to let the orders flow again.
Please let us know at email@example.com
and we'll take action.
There's no need for you to pay Digikey twice for a part.
There is a minor problem with the output TXA signal level to the transmitter.
There is a 22K resistor called out for R3 which is the last part before the scope-loop and the DE9 connector.
This part can't be in circuit if the TNC is to be hooked to a data radio at 4800 or 9600 baud.
A wire can be used for R3 and the TNC will work well at the higher speeds.
However, when the TNC is attached to the microphone input of a radio, for lower speed poeration, the transmit audio level may overdrive the radio, causing distortion.
We want the resistor to be in circuit for these applications.
What we need is a removable jumper or a minor redesign in the TXA deviation control.
The kit-builder of the A3 unit has a couple of choices.
- You can put a wire in place, instead of the resistor, and use the TNC with the data input on 9600 radio.
- You could put the resistor in place, to use the TNC with the microphone input of a lower speed radio.
- Or you can put the resistor in place and a wire soldered onto the resistor to make it easier to go back and forth.
I'm sorry about the problem.
We didn't have quite enough testing.
By the way... this problem existed with the A2 board as well.
It's just that most of our testing was at 9600 baud or short range where the microphone-input distortion wasn't a problem.
If you have an A2 board and you find the TXDEV pot doesn't go low, and you are using the TNC for 1200 baud, you can add a 22K to 82K resistor in place of L3, or put the resistor in the TNC-mike cable in line with the Mic-Audio wire.
Any 1/4 watt or smaller and any tolerance resistor will do.
Maybe a larger resistor will do.
If the gain can't be brought up enough, then the resistor is too big.
We still need more testing of microphone-audio radios at 2400 baud.
The TK8180 radios work ok at 9600 baud but only if the TXDEV is adjusted precisely and if the signals are really strong. S9 or better.
I don't know why this is true.
The Tait TM8105 radios work very well at 9600, even relatively weak signals and the audio adjustment doesn't seem very finicky.
SD125 data radios seem to work ok at 2400 but not faster. I didn't get any info about precision audio adjustment.
I'm working on a detail about how to use an SDR and Bessel curves to adjust the transmit audio level an an objective/absolution fashion.
Personally I'm not clear that this is absolute. The TK8105 radios kind of shook my confidence.
The NCPACKET network has five 9600 links and one 2400 link using NinoTNCs, plus 25 or so 1200 baud links which are all TNC-PI or half TNC-PI, half NinoTNC.
The Tait TM8105 radios get 1% retries. I did a walkthrough of the network looking for the ends of those links to show you the statistics. ncpacket-ninotnc-retries
News April 30, 2020
Boards still not in Texas, but it's not late yet. I just wanted to post an update before I forget.
DHL says the boards may be delivered to Nino @ home today.
We're still a week away from sending them to the ETSY buyers.
The firmware is now at v2.3.5.
1200 and 9600 modes
This past weekend we did some field testing of the IL2P modes at 1200 and 9600 baud. That testing is not yet complete.
As of v2.3.4 the AX.25 and IL2P transfers at both bauds were very good in bench and field testing.
The only bug we know of in 1200, 9600, AX25 and IL2P from our testing was that in IL2P mode, there is supposed to be a beacon every 9.5 minutes which
describes to any listening station that IL2P is being used.
That beacon is supposed to be sent in AX.25 mode so an uninformed listening station knows what is going on with all the non-decoded packets.
The beacon gives the URL of the IL2P website, https://tarpn.net/il2p.
The beacon is defective in v2.3.4.
Fixing this specific bug is the only change v2.3.5.
Field testing has not yet verified this fix.
It worked in the lab.
2400 has been problematic so far on voice radios, only working with specific modems for specific radios.
We're contemplating how we're going to remedy that problem.
Vertex VX-4000 radio (model# VX-4000v FCCID: K66VX-4000VE), which is a VHF-HI (aka 2m) rig, was shown to at 2400 in both IL2P and 2400 AX.25 using v2.3.5.
However, that radio is reputed to work at 9600 baud. It didn't. It also didn't work at 4800 with NinoTNC.
4800 has been bench testing with good results.
4800 is expected to work on radios made for 9600 but which don't quite work at 9600.
Bench testing of the USB bootloader has been successful with a kludge method of accessing that bootload sequence.
We don't have a ready application to do bootloading, or source code to distribute to enable multiple platform operation.
That capability is still to be completed.
We have a TARPN configuration solution for creating the node.ini file (and thus the bpq32.cfg file) which supports up to 8 NinoTNC at a time.
The TNC-PI support is being dropped from 6 TNCs to 4 TNCs.
Ports 1 through 4 are reserved for TNC-PI.
Ports 5 through 10 are reserved for NinoTNC.
Ports 11 and 12 are for generic USB TNCs but can also support NinoTNC.
The TARPN config is on a private web site and not yet in the general distribution.
If anybody wants to play with it early, send me a private email (firstname.lastname@example.org) and I'll show you how to get to it.
My intention is to switch it on after getting another 8 hours of work on it.
Specifically the problem is that if I put it in the general distribution, it would supplant the existing system and will REQUIRE that the operator runs tarpn-config to get the node to start.
Additionally, there are some features I'd like to add which are purely optional, but which some advocates have been looking for.
News April 28, 2020
DHL reports having the package of 200 A3r2 boards on the way back to Texas from China.
News April 21, 2020
200 A3r2 boards are ordered.
have been taken for 100 of the units.
We ordered only 200 this time because we wanted to see the minor changes (one new resistor, electrically moved both TXA and RXA test-points) before committing to a larger number.
Ordering 500 boards gets another price break, but at $2.50 each@200, this is already pretty cheap.
At least I think that's the price-per-PCB.
News April 20, 2020
The A3 assembly page is 99% done. It still needs to be tested by somebody following along to build a kit but the link is now up on the NinoTNC root page.
Production boards are not yet ordered. Soon.
The code is complete for 2400 baud and 4800 baud modes and they have shown to be very interesting but still experimental.
We don't have complete performance comparisons yet, against 1200 and 9600, and also no field testing of 2400 and 4800.
We hope to have both modes on the air in the next week or two.
The new FM modes will be available when the A3 ships but we'll be looking for feedback.
The four A2 modes work well.
We don't have any time-line for the SSB modes. Call it indefinite for the moment.
Still to be done is the Raspberry PI NinoTNC USB bootloader utility.
News April 1, 2020
The A3 NinoTNC will ship in early May.
I'll make an Assembly page and will post the Bill Of Materials after a couple of the prototypes are built.
Hopefully this will happen by April 19th.
I'll also publish Bill Of Materials for Update info to tell you want parts to buy to convert an A2 parts kit to an A3 parts kit.
I expect the delta to be about $4.
The ETSY page is open for pre-orders. ETSY page for N9600A3 pre-order
News March 28, 2020
The first ten blue A3r1 boards are in.
Click to enlarge
The most important new feature of the Blue A3 boards is that the FTDI module is eliminated in favor of a USB-B
socket and a through-hole Microchip USB-ttlSerial adapter chip.
These parts actually cost less than the A2 board's FTDI module so the parts cost is going to go down!
Also, the parts are stocked for production at DigiKey so we won't run out of them as rapidly and they will be refreshed.
Because of the FTDI chip, the A2 boards show up to the Raspberry PI Linux OS as /dev/ttyUSB0 then 1, 2, 3.
The A3 boards will show up as /dev/ttyAMA0 then 1, 2, 3.
See the mode switch?
We decided 2 switches wasn't enough. A2 has the modes 1200 AX.25, 1200 IL2P, 9600 AX.25 and 9600 IL2P.
The A3 will those same 4 modes and room for more.
There a firmware change required to support the new 4-bit switch.
The switch will allow for more modes to be hardware selectable.
Ideally there won't be any need for software bit rate and protocol selection.
As a teaser... you can imagine that with modern digital modulation and modem emulation that we can do faster than 1200 on an FM channel.
Furthermore, while some of the 9600-ready ham rigs are clearly not capable of using G3RUH @ 9600, what would happen to those radios if they got 4800 bits per second instead?
We'll find out soon enough.
Bug in V2.20 firmware with 1200 IL2P
V2.20, which shipped with the A2 production boards from ETSY, has a bug that keeps IL2P at 1200 baud from working properly.
This is fixed in v2.22 which will be available via bootloading once we get the USB bootloader for Raspberry PI working.
V2.20 works ok at 9600 baud AX.25 and IL2P and also at 1200 baud AX.25.
Our plan is to release the source code for the bootloader once it is working and documented.
It may be a little while. Sorry about this.
News March 14, 2020
Minor redesign is done. Sample boards ordered. This will be A3 and at least the sample boards are blue. .
Edited the n9600-A2 BOM, now version 9, to add the parts designators to the comments on the 0.1uF capactors (4 of them) which are C8 C12 C13 C14.
Formerly those parts were in the BOM with no comment line for parts designators.
Edited the assembly web page
to correct the 0.1uF step so it now shows four 0.1uF capacitors.
News March 10, 2020
I think we're going to go with a minor redesign for future production.
The N9600A3 board will have an integrated USB solution with a type B socket.
We'll do an engineering run of 10 boards and then a run of 200.
Expect delivery of the production run sometime at end of April.
There are 20 boards from the A2 run still available for pre-order on ETSY and for delivery on March 18th.
Pre-orders after those 20 will be A3 boards. I think. That's the plan.
Other news. If you have an A2 board and can't get the FTDI eval board header, there are many other USB interface boards which can be hand wired to the pins on the A2 board.
Here is a diagram of where the pins are connected:
Click to enlarge
News March 8, 2020 afternoon
We're pondering what to do next and when.
We want everybody who wants one of these boards to get one as soon as practical.
As I type this there are 20 of the first 200 left.
The FTDI evaluation module is sold out at Digikey.
[qrz.com], found a Microchip through-hole DIP USB solution we can use and Nino is looking into making an A3 version of the board with the USB solution on the board.
Link to K4DBZ's PCB project solution
Integrating the Microchip MCP2221A would saves us a couple of dollars vs the FTDI module.
Builders of the N9600A2 who could not get the FTDI module can build David's project buying the Microchip MCP2221A from Digikey.
There are two vendors selling the FTDI module we call out.
One is FTDI in UI direct
and the other is
NC4RA tried to order from both of these shops. The FTDI shop would not ship to the US, and NeutronUSA was backordered until May, they said.
Beware that I don't know anybody who has tested these to prove they are the right item.
The next orders may be delayed 3 weeks or 2 months.
If we get enough orders of the A2 board in the next few days, knowing that the FTDI module is hard to get, then we may order another run of the A2 board, needing the FTDI module, sooner.
We could also see major delays again due to COVID-19
Stay tuned to the email reflector
News March 8, 2020
The ETSY page
was turned on again to accept pre-orders.
The first 105 units are being shipped as a total of 17 shipments, mostly USA customers, but 8 units are going to Netherlands.
We're now selling into a pre-order of the 95 remaining PCBs from the first order plus an additional 500.
Pre-orders for the 95 units will be fullfilled after March 18th.
Pre-orders for the next 500 will either be fullfilled in late April, or later depending on if we make minor changes to the PCB layout.
The changes to the PCB would substitute a USB-B socket and Microchip USB chip for the FTDI module.
This would simplify assembly, reduce the cost of the parts, improve the robustness of the USB connection, and fix a big problem where the FTDI eval module called out in the A2 version of the board is not available in the quantities we need.
At this point, by the time 150 more Bills Of Materials are ordered, we'll have used up every FTDI eval board on multiple continents out to late July.
This may have been a mistake?
News March 7, 2020
The boards went on sale at ETSY at ETSY - n9600a2-circuit-board-and-cpu-kit-for
at just before midnight March 6th and we sold 64 of them in 6 hours between midnight EST and 6 am EST. There are 96 boards left and 0 CPUs. 100 more CPUs are on order.
The ETSY page will remain there but we're not taking orders until the CPUs come in. Expcted March 17th.
We have a volunteer to run the shipping and receiving operation and we may go for higher quantities.
There is a discussion about spinning the board again because the FTDI module is single sourced and expensive, and the micro-USB connector is crap.
There are many ways to do USB to TTL Serial. We also want 2 more switches on the board to permit more data rates and other features.
The BOM was changed from #6 to #8.
The only difference is different Digikey part#s for some line items as some of our bom#6 parts went out of stock and there were easy substitutes with more quantity.
News March 5, 2020
The boards are in the USA! Nino is verifying stuff and will post the ETSY page by March 8th.
Video of two NinoTNCs going back and forth at 9600 baud: youtube TARPN NinoTNCs talking at 9600 baud
Here's the bare board.
News February 29, 2020
The boards still haven't arrived. They were delayed by closure of the factory due to the Wuhan coronavirus.
Apparently the factory is back in business as another project we're running got its PCBs.
The factory has changed the status of our NinoTNC order recently and they should be close to them shipping our order.
I expect the boards by March 5 or so.
We've decided to go with Etsy
as our storefront and the price is $7.57 each plus shipping.
We're going to have a means of ordering 12 of them for under $100 including shipping.
The v6 Bill Of MaterialsOLD news!
from late January is still correct.
The TNCs will ship with version 2.20 which includes a USB bootloader.
We have tested the bootloader but the software to run it isn't ready.
The bootloader software will initially be written for Raspberry PI and will run as a stand-alone application and as part of the TARPN updateapps facility.
We're talking about what comes next. Keep an eye on email@example.com
Now sucessfully tested @9600 with Kenwood TK8180, Tait TM8105, Yeasu FT-817. No modifications required. Wiring diagrams coming soon.
We had mixed results with Alinco DR235mkIII.
Bad results with Yeasu FT8900R
We have yet to write an operators manual. Busy busy busy.
News January 27, 2020
200 of the A2 black boards have been ordered, delivery expected by February 12 or so.
We're working on HOW to order the units.
We may be using ETSY instead of PayPal.
We plan to take orders starting mid February.
There may be a quantity order option to save postage for the boards and CPUs.
There is a new v6 Bill Of Materials OLD news!
(which adds three 0.1 capacitors for C12, C13, C14. Manuf: K104K15X7RF5TL2
The 1200 baud receive wasn't as good as I thought it was due to the possability of different equalization on the receivers.
We were getting 944 track 1 and 633 track 2.
This has been fixed.
The new numbers appear to be 948 track 1 and 948 track 2.
The Nino TNC now figures out which equalization curve works best and switches to it.
There are some minor improvements yet to be done and we have yet to write an operators manual. Busy busy busy.
News January 14, 2020
The IL2P protocol is published. Link to IL2P document.
News January 13, 2020
The assembly instructions for n9600a2 are complete. Link to n9600a2 assembly instructions
Nino has been improving the firmware for 1200 baud.
He's been running the TNC's AX.25 1200 baud receiver against WA8LMF's test CD
This is a collection of packet braps recorded on 144.39 in California on a heavily overloaded channel.
The goal is to get your receiver to receive as many of them as you can, or at least to use it as a reference for comparison.
It is not possible to receive all of the packets, because many of them are broken and missing starts, or ends, or middles.
It's a benchmark for good TNC performance.
Here's some results which are published, against our test of NinoTNC A2:
- TNC-X gets 818
- MFJ-1274 gets 884
- NinoTNC A2 gets 940
- KPC-3 gets 967
- Direwolf gets 1000 or slightly better (according to Direwolf's github page)
News December 31, 2019
We have our first A2 units in light green. (released versions will be black)
Two of these are assembled and talking across the bench.
AX.25 @ 1200 baud is working between the two NinoTNCs and with the Kenwood APRS HT.
AX.25 @ 9600 baud also working between the units and with the Kenwood APRS HT.
News December 3, 2019
There are 10 of this model in existence.
These are being tested in North Carolina on the NCPACKET
Our plan is to spin the board one more time to create the N9600A2 with a few fixes.
This board should be ready to sell at the end of January 2020.
Changes expected in N9600A2 - January
- The A1 board uses 7 separate resistors for LED current limiters.
In the A2 version, we're going to replace all those with a single package, and 4 other resistors will be combined into a single package.
- 2 configuration switches are provided to select future (vaporware) features.
- The firmware will have multiple sets of specific drivers, one for each board-ID, and the firmware may expect slightly different wiring or hardware for different versions or designs.
We'll be able to ship one CPU chip which will support several different board-versions.
- The next spin of the N9600A will also delete two of the LEDs and add a potentiometer for TXDELAY setting.
- The two potentiometers will be right-angle parts with the adjustments facing out the end of the PCB next to the LEDs.
- The push-button PTT-TXTEST will be right angle facing out next to the LEDs.
- All the LEDs will be facing out the end. Now we can make a 3D printed chassis and wrap it around the thing.
On a TARPN node it will be assigned to port 7, 8, 9 or 10 for BPQnode as part of thetarpn config
process, and in node.ini.
Interesting Reads from G3RUH and others
AMSAT article 109 by G3RUH
WD6EHR 9600 baud handbook