Support QRP CoA Everywhere

Monday, 13 July 2020

N7DCC ATU



This had created some interest on GQRP list so I ordered it again.
7x7 version arrived... smt PIC this time (1698 derivative), not soldered to the board yet and not programmed either.

I'll include build notes here.

Discussions from GQRP Groups list:

Auto ATU Kit from N7DDC ? anyone tried it?


 m0bmn
Jun 29   

Hi Guys,
Just spent £35 or so on a auto atu kit by N7DCC, it was a bit of a impulse buy which I am already wondering if I have done the right thing.
Anyone built one?, I only just noticed that it as a lot of SMD parts including the microcontroller. The controller isn’t programmed either having had a read of the notes, the description of the item (eBay ) is a bit poor as it says no need to burn firmware but later (I didn’t read that did I!) say that the chip is supplied blank.
My eye sight is as not as good as it once was and I dont think I could solder the controller chip anyway.
If you have tried this kit please let me know.
Thanks
Paul M0BMN

 robert miles
Jun 29   

Hi Paul
Finished mine a couple of weeks ago.
My first attempt at SMT and got it working after reprogramming the PIC and replacing a relay I burned with the soldering iron.
I will pm you some lessons learned if you want them.

Cheers Bob
GM4CAQ

 Tim Ostley
Jun 29   

Hi,

I bought mine from a Russian Amateur who offers these in all stages from a bare PC to ready built and tested. The difference in cost between the full kit and one ready built and tested was only $20 which seemed a good deal with a transformer, 6 or so toroids and about 40 sm components to solder.

It took nearly four weeks to arrive (as indicated in the advert) and came to about £48 including postage.

I did need to make two changes to the configuration bytes in the software, one for the display I chose to use and one to reduce the minimum power for tuning from the default 5 watts to 1 watt. I might also make one other change which is to the swr for auto tuning. As standard it will auto tune if the swr drops below 1.6 and on one band the swr was around that value. Whilst talking I found the tuner going into retune mode and then getting confused as the power was going up and down with my speech. You can of course take it out of auto mode but I had not done so! The leads for the bypass and auto buttons have to be soldered on to two small pads on the underside of the board (with the version I received) which were not very well labelled.

I am quite pleased with it, especially for the price, it works well within the limits of its L and C maxima. It has 7 L and 7 C. It is not as good as the auto atu in my K2 which has 8L and 8C but it does tune faster.

Best 73’s

Tim

M0CZP

 robert miles
Jun 29   

Thanks for the tip about the pads on the reverse of the board.
I found the experience of building the unit (with magnifying glass and iphone in magnify mode) invaluable.
Interesting for this group is the built in SWR power meter that uses a tandem match. Not come across this before and he provides info on making FSD of 40W as well as 150w and 1500w!

I would like to build another with the 5:1 turns ratio to see if it will work better at qrp.

I built mine with the 10:1 ratio and it only tunes with 4w into a nearly resonant antenna.. much better with 10w. I hope/think for the Flex 1500 the 5:1 ratio will be right on.

I did manage to melt the case on a relay with my iron and ordered 10 from China .. twice.. so I have 19 spare..

Terry VK5TM
Jun 30   

I've also built one but not had the chance to hook it up properly and test it.

Went together ok, but then I am used to working with smd components, bench tests show no problems.
--
Terry VK5TM
 1 person liked this

 Barry - GM4TOE
Jun 30   

Our club has been building a number of these and I know there are a few boards (5 ?) fitted with the programmed controller (and possibly a lot more).
Any interest then PM me

73 Barry GM4TOE
 1 person liked this

 Tim Ostley
Jun 30   

Hi,

As a followup from my last post.

I didn’t make it clear that I bought the ready built and tested option.

If you need to program it or change some of the operating parameters then the cheap after market Picit3 works well together with the free download of MPLAB. Bear in mind that you will need the MPLAB IPE (integrated program environment) NOT the IDE (integrated development environment) unless you are going to modify the code rather than the set up parameters. You need 4 wires to connect the Picit to the target micro and you also need to put power into the ATU independently.

The unit does have auto tune as well as manual tune and a bypass, using 3 buttons. It can be configured in auto mode with no display in order to use it remotely. The relays are not usually latching ones so it needs power all the time it operates but it does remember the last settings after power off / on.

There are no other band memories although there is some room left on the PIC chip EEPROM. As the software is open source it would be possible to set up band related memories but some sort of band input or frequency counter cct would be needed.

Hope this is of some use.

Tim
M0CZP
(in a damp Isle of Wight)
 2 people liked this

 Paul Bowen
Jun 30   

Ge

I am putting together 10 for a club project. I have soldered in the pic ic,
header pins, and a cap and written the hex file to 10 boards. I soldered the
pic ic by lining up the ic pins hold with blue tack. Then tack one end check
it then flood the pins with solder and draw off excess solder with good
quality solder wick.
There's not much interest in construction so there will be some full kits
available with detailed instructions and all the surface mount parts fitted.
I was hoping to sell them at cost price on the TDARS stand in September just
to get my money back.

Cheers

Paul

M0pnn
 1 person liked this

 Paul Bowen
Jun 30   

Ge

Thanks everybody who has asked the list is full now I will email off list
when all the parts get here etc.

Cheers

Paul

M0pnn
 1 person liked this


Looks interesting for sure.
Might make for an interesting kit build if you wanted to tinker too as you could program your own ATU as well with specific options if you managed to wade through the code or pick alternate options for L&C combinations that might suit YOUR antenna better especially on the low bands. The uPC source is 'out there' for the 5x5 and 7x7 options.

Similar you could include different displays.

I think, might be wrong,, lot depends on your antenna but it is more suited to 40m to 10m, might struggle finding a match on 60m or 80m and definitely not suitable on 160m or 472KHz in most situations.

72

Dom
M1KTA
 1 person liked this

 Tim Ostley
Jul 1   

Dom,

It is interesting and indeed you can hack the software. The software is not too bad to follow - there are comments in the code! One of the reasons I got one was because I could make some changes. Thoughts so far include latching relays, 8x8 rather than 7x7 to increase the range and band sensing to provide at least one memory per band to speed up tuning - there should be plenty of EEPROM space. I haven’t looked at the tuning algorithm used but it does work quite quickly. One thing that does need fixing is that the unit can get hung up in the tuning mode when using auto if it can’t get a match and the power is not high enough - if the match is poor the minimum 1W tuning power is not low enough if you only start with 5W or less, depending on the rig. The K2 for instance seems to fold back the power in a mismatch to save the finals. A good idea but it means that the tuner doesn’t get enough to tune properly. Changing the sensing transformer may improve things for qrp but may need changes in the code for the A/D scaling.

One thing that I couldn’t find is any details of changes / improvements that have been made by other amateurs.

Best 73’s

Tim
M0CZP
 1 person liked this

 Terry VK5TM
Jul 1   

The manual has details of the changes needed for QRP use, along with a fair number of other options, including different displays.

--
Terry VK5TM
 1 person liked this


Tim,

All praise the N7DCC for releasing the code in the first place, so many manufacturers have kept this all secret squirrel for ages.

Latching relays have to be done really for /P use otherwise will have to remain energised but of course that means more code munging (I just don't have the time right now).

Code comments... wasn't an "Agile" project though (a dig at some dev's at work who ignore comments if reference makes no sense). I haven't seen a "don't know what this does but it works so leave it alone" comment yet which is usually a good sign :-)

A mono OLED display (Nokia 5110 maybe?) is a must too for lower current draw.

I have not flow charted the algorithm yet, I have looked at a few of these and they seem to all go for a brute force lowest VSWR choice hence the bridge included and relaying on the uPC AD inputs.
Not sure if it does the memory thing either on frequency.

Indeed no extra github changes etc.

I don't have a board infront of me right now but there seemed to be two sources Russia and China. One silkscreen in Blue other Green I don't think it matters either way.

72
Dom
M1KTA
 1 person liked this

 Tim Ostley
Jul 1   

Dom,

The design was done by N7DCC who got it working and promptly put it in the public domain with no restrictions. I bought my ready made board from Russia. The board is nicely made but could be improved by bringing the Auto and Bypass button lands out to the edge of the board for connections rather than leaving VERY small pads on the bottom of the board with no through holes and a slightly broken up silk screen so it is not clear which land is which - traced it back to the chip in the end and had to be very careful soldering on the wires to the switches.

I have tried both a 0.91 OLED and a standard 16x2 lcd and both work well. You change a couple of configuration bytes and it sorts itself out. Out of the box default is 16x2 lcd and minimum 5w for tuning.

It is amazing the number of programmers who think code comments are for other people and forget that three months after they have written the code and they have done at least one other project they won’t understand their own code, especially the nuances where there was a very good reason for adopting a particular approach for part of it. I think agile development is write it and ship it fast before anyone asks questions …..

In terms of the algorithms there are some useful sort / search approaches that can be significantly quicker than plain old sequential.

It would be nice to change the rig side so that it saw mostly 50 ohms whilst trying to tune.

To get anything better you are looking at circa £200 for either 8x8 or memories and a lot more if you want both, so £48 including shipping is a good deal.

Best 73’s

Tim
M0CZP

 Terry VK5TM
Jul 2   

Just an update to the QRP mods I mentioned as being in the manual - the manual on Github doesn't have as much info, the manual I had was off a Russian forum and I appear to have overwritten it when I downloaded the latest software update file from Github.

I'll see if I can find the forum entry again.

In the meantime, this is what I could find - for QRP, the winding ratio for the transformer was set at 1:5 rather than 1:10 and you need to change the setting in cell 31 to match (that info is on page 13 of the latest github manual).

You may even be able to go to 1:4, but be vary careful of the power you put through it or you could end up with excess voltage on the inputs of the PIC which will do nasty things to it.

--
Terry VK5TM


Terry,

You could play about with Cell 31 but only on the 7x7 version not the 5x5?.... The transformer ratio and resulting voltage in the FWD&REV is always an issue when using a uPC/ADC in this way though isn't it? I am not a prof electronics engineer but thought you could clamp the maximum voltage into the ADC perhaps to VCC if this were a risk but I don't think this has been done, it could though? You use a couple schottky diodes on the input lines before the ADC don't you? RA0 and RA1 on the circuit diagram. The Schottky diodes are chosen for their low forward voltage drop, such that they turn on before the internal protection diodes in the ADC lines, I expect will have to do some data sheet study. Either way the max possible input in this case is 4.096V exceed it and you would probably let the magic smoke out of the ADC but I expect that could be taken up a bit.

If you look in the code main.h and main.c (version I have anyway) you'll see the code does have an expectation for power> 1500W (assuming whatever physical board it is put on can support that, this one couldn't). Look at void show_pwr(int Power, int SWR) in main.c and void get_pwr() in main.h. The code is being updated though so those references to QRO levels might change soon?

Could you tweak both the circuit and code to allow for a known lower maximum power level through the transformer possibly for the Vmax so you get a better resolution within that 4.096V maximum? I only know this having done it for early 500KHz and then 472KHz operation having to pay careful attention to both you learnt a bit about FWD/REV transformers. I think Drew Diamonds books, I forget which one even walks you through this in some detail. I think he mentioned phase in one too.

I admit most probably will not want to recode but then you have the option as the pin headers for the ICSP are there.

72
Dom
M1KTA

 Terry VK5TM
Jul 2   

Dom.

Yes, cell 31 only appears in the 7x7 version at the moment and the software for the 5x5 version only goes to 2.7 at last check (7x7 is version 3.0).

The voltage on the PIC adc pins is a combination of the turns ratio and the power you are feeding through it so if you are aware of your Tx power and have the transformer turns ratio set up correctly, you will be pretty safe.

PIC's already have the diodes internally on the input pins (but only suitable for overvoltage excursions at low current), so I would be more inclined to fit zener diodes as protection devices. Probably 5.1V types in this case.

The 4.096 ref voltage is internal to the ADC module and not connected to the input pins from my reading of the datasheet and as such feeding 5V to the input pins shouldn't do any damage. All you are going to get in that case is a maximum reading from the ADC module for the extra 1V applied overvoltage.

Reading through all the electrical specs shows the max input voltage on all pins is 5.3V except the Cap Sense pins, so again, I couldn't see it causing a problem.

The version 3 software for the 7x7 board has the facilities to set it up for whatever maximum power level is expected and the 1500w level appears to be staying, as this unit has been used as the basis for other high power tuners.

Also, the 5x5 tuner is somewhat out of favour according to what I read in the Russian forum some time ago and you will be hard pushed to find any for sale on the regular auction sites that I can find.

I would have to dig through the code to see if the extra functions could be included in the 5x5 version but I personally couldn't see the point - 5 C and 5 L are somewhat limiting if you are expecting an antenna to cover a few bands.
--
Terry VK5TM

3 comments:

M1KTA said...
This comment has been removed by the author.
M1KTA said...

Second one just turned up... (blue PCB now green) from Aliexpress instead, this one had the PIC soldered on and an OLED display.
Much cheaper https://www.aliexpress.com/item/4001115906991.html

M1KTA said...

The T68-2 torroids for the aliexpress version seem a tad dodgey.
I replaced with 6 known good ones bought from W8DIZ (https://www.kitsandparts.com/toroids.php) a while back.