Hopefully, at some nice day it will be fully TCI-integrated and SunSDR-compatible remote ATU, produced by EE with love and fair price
Meantime, I am in progress to build a prototype, based on original N7DDC idea, but with completely re-build control, allowing full remote operations from PC and integrated with ESDR via TCI.
If anyone interested (yeah, I know remote ATU mostly in use by field operators and DXpeditioners) - more than welcome to discuss. I will also appreciate any suggestions of feature list.
Коротко для русского коммьюнити:
Удаленно управляемый антенный тюнер, основанный на известной идее N7DDC c новым блоком управления, позволяющим интегрировать ATU с ESDR через TCI. Описание по ссылочке выше. Приветствуются идеи и пожелания.
Because I continue receiving questions in my email, I think it is good idea to give you an extended update regarding this project.
1. There will be two different HW versions of ATU in focus and support
- 7x7 L-network (based on M7DDC original design, ATU produced by UR3IQH). Here is control board schematics (for 7x7 board please google, it is well known),
- 8x8x8 T-network (new design)
(Sorry for low picture quality, it is the best you can get from KiCad export. For better picture quality please wait untill I upload KiCad files to GitHub in Jan'21)
2. Both ATU versions will be supported:
- single software line for ESP32.
- universal ATU control application (cross platform)
3. Every ATU version will need different physical board design due to different number of relays to be connected.
ESP32 software will have integration with ESDR via TCI:
- use of VFO freq data to simplify ATU initial settings per band
- use of DRIVE and TUNE for single- or multi-stage tuning
- ATU connect application will allow to set custom DRIVE values
Often Q: will it work with non-EE (non-TCI) transeiver platforms?
A: In principle - yes. Obviously while using with non-TCI device, you will loose automation and have to manually set the DRIVE. You will also lose all future automation, which is going to be embedded into tuning algorithms with TCI.
Q: What? Algorithms? It will be more than one tuning algorithm?
A: Yes. It is less applicable to L-network and more fun for T-network type. Anyway - yes it will be at least three tuning algorithms available. Rationale - while in most cases you will select your favourite (or "classic") algorithm, however. According to experience, it is often that more than one algorithm needed for different kind of antennas.
Q: What will be typica ms range for relays?
A: Not applicable for our design. The method we are embedding is targeted to perform best available match within minimum possible steps/time. Generally speaking, ATU will tune within minimum possible time and with best available SWR - for given setup. You do not need to worry about mseconds or other stuff.
Q: but will be those settings (like ms) still available?
A: Yes. Not preciselly ms. Different kind of settings, which - in principle - will provide you with same kind of settings, which you typically achieve with tuning ms delay for relays.
Q: What kind of algorithm does that?
A: Our own. Nothing secret. Read the code, it is open-source. In a nut shell - we are using high resolution ADC with mutli-core processor threading and less-step tuning approach combined with real-time (approx 100 samples/sec) data from FWD/REV readings.
Q: What is min/max power will be required for tuning?
A: Anything between 1W and 100W. Theoretically it can be even as minimum as 100mW.
Q: What is power limit?
A: For L-match you can ask UR3IQH, he is making 500/1000W versions.
T-match is targeted is targeted to support 500W with minimum double safety factor (simply means that we are using 1-1.5KW components)
Q: What are physical dimensions?
A: L-network will be dictated predominantly of what current producers are doing.
T-network will fit into plastic tube of 150mm diameter and 220mm length. Why tube? Because it is remote ATU, it is designed to be mounted on mast. And we are trying to achieve less wind resisting design.
Q: When KiCad and code will be available?
A: End of Jan'21
Q: When T-match will be tested and photos/videos available?
A: End of Jan'21. Boards are printed and on the way, all components will arrive by 10th of Jan.
Для русскоязычного комьюнити:
- Это обещает быть интересный АТУ. Ну а если не получится - то хотя бы развлечемся
А вообще запихните страничку в гугл-транслейт ((примерно так), должно перевести более-менее нормально. Ну а если понадобится (и после того как заработает) - откроем русский тред.
Последнее редактирование: 31 Дек 2020 04:48 от VK6NX.
- WiFi. ESP32U with external antenna will provide up-to 50 meters (typically). It is preferred media type for control-plane design. However there are situations when WiFi is not suitable (i.e. distance, interference, etc). Hence:
- RS485. TX2/RX2 from ESP will be connected to RS485-embedded module and utilise special-brew network protocol (tested and working on version 1 of P140-coil based ATU). It is, obviously, slower than WiFi, but still does the job.
- via both above the connection will go to MQTT server (this can run on Linux/Mac/Win). For those you, who do not know what MQTT is - just google. It is very cool and practical protocol. Fully tested in previous applications.
Construction parts arrival update:
Printed boards for 8x8x8 (DHL is promising 11.01.21 by end of business day)
Components for 8x8x (Fedex is promising 06.01.21 by end of business day)
7x7 ATU board (EMS is not promising anything) - unknown, may be up to 60 days.
Hence, I will be working on T-match after 11th. As I am bad solder, it is expected a week or so before I start field testing.
- 7x7 ATU (from UR3IQH) is still "somewhere". Tracking shows it was left RU on 7th of Jan'21 (this is last tracking record), hence I have absolutely no idea when EMS is going to deliver. Control board for it is ready, I am just waiting for ATU to arrive to continue testing.
- 8x8x8 complete went through initial testing for 2 weeks and:
1. We have found several construction issues, mainly referred to RF interference.
2. We have tested 4 different options for relays control (MOSFET+Driver / Logical Level MOSFET / MOSFET Optocoupler / H-bridge)
3. As the result of 1 and 2 above we have new schematics (based on H-bridge) and new boards are ordered. As the boards came from same factory (according to experience) I am expecting them to arrive by end of Jan. Hence (hopefully) in first week of Feb'21 I will give you more expanded update with all details.
HW version 2 was successfully tested over last weekend. Fully on 40-30-20-17-15-10 bands. Partially (LOL) tested on 80m - still tuned to 1:1.3.
Tested on one "good" (AV-640) and two "bad" antennas (MFJ-1796 and "made from trash" 1/4 40m shortened dipole, placed 3 m above the ground). Those last 2 - to test how this ATU will work with really shitty antennas. Typical SWR on every band was 1:1.05 - 1:1.2
ATU currently was tested to 100W only. High power tests are upcoming in couple weeks.
Unfortunately is was rainy weekend on my QTH, so I did not manage to get photos and video, however, those are upcoming this week.
Please note, that there are significant schematics and construction changes, since version 1. (Obviously because version 1 test ws not really successful).
L-network constriction, locations of C1 and C2 networks on board were changed. Also control board drivers were replaced with different HW version.
I am currently working on polishing ESP32 code and adapting ATUconnect SW with all necessary user features.
Future plans for HW v3.
T-network will become "3-in1". By adding 1 more relay, current HW version will turn in new one, where networks will be selectable by user - either "original" T-network, or L-network with C-L, or L-network with L-C.
Для русскоязычного комъюнити:
ATU версии 2 протестировано на 40-30-20-17-15-10. Работает. КСВ (даже на весьма хреновых антеннах) строится в промежутке от 1:1.05 до 1:1.2 (в зависимости от того, что за тип и резонанс у антенны). Типичные "стартовые" условия были такие: КСВ >> 1:5 (между 1:5 и 1:10), в точке резонанса и 100-400 КГц от точки резонанса. Чуть попозже я выложу фото, как настраивалась антенна MFJ1796 на 40 метрах - вот с такими начальными параметрами:
Как видно на скриншоте с анализатора у нее минимальный стартовый КСВ около 1:1.2 в диапазоне oколо 10КГц (от 7000 до 1010), а далее ... 4-5-10-20. На 7050 у данной "антенны" вот такие параметры - настраивается в 1:1.05
На 7100 там уже 1:10 стартовый, ну и так далее.
В итоге во всем диапазоне 40м на данной антенне были получены КСВ не хуже 1:1.3
В целом получается не сильно хуже, чем мой полностью рабочий remote ATU на вариометре от P-140
Да, ну и полное описание тут - vk6nx.net/RATU_T_v0.2-en.html
Последнее редактирование: 08 Март 2021 14:19 от VK6NX. Причина: description added
Equipment used: MFJ-1796 (modified) antenna. This is my backup antenna for IOTA DXpeditions and it is set to work at sea shore, not backyard . I even found the shittiest place for it on my backyard - to make things really hard for the ATU and perform its stress test.
Static details about the antenna in this test:
Test band 40m, antenna statically tuned at 7009 KHz
Bandwidth with SWR from 1:1 to 1:2 - approx 10KHz (Yes, this is very narrow band on 40 meters, because it is extremelly shortened antenna. Not really recommended for 40 meter band, but really valid for ATU stress test).
8x8x8 ATU (this one you are, perhaps, familliar with from my previous posts)
KXPA100 (with internal ATU set to bypass)
RigExpert AA-600 antenna analyser
Disclaimer: sorry for photos quality, have to test after dusk due to obvious 40 meter band reasons.
1. Measure antenna parameters at given frequency with no ATU connected (AA600 connected directly to antenna via short coax).
2. Connect ATU. Tune. Measure whole setup (antenna+ATU+50 meter coax cable) parameters at "PA end" of coax.
3. CQ CW and monitor public CW skimmers
4. Move to next freq
Here are the results. 7008 KHz
Antenna parameters before tuning
"PA end" after tuning. Interesting how ATU "fixed" shitty initial resistance and reactance, though.
CW skimmers result (I have used my VK3 call sign, as I am in VIC now, not in WA)
Antenna parameters before tuning. Still within "reasonable" range
"PA end" after tuning... Hmm, looks like I need to polish "fine tuning" in the algorithm
CW skimmers result
Antenna parameters before tuning. That is really... bad ... Do not use that antenna without ATU
Lets see what tuner can do with it:
And CW skimmers result
And now to real extreme ... 7100 KHz
Antenna parameters before tuning. Obviously, completely out.
And Skimmers... Really, I do not expect anyone to hear, usually CW skimmers set to check CW only part of band. And antennas of profy-skimmers usually tuned to CW range too. But, anyway, it worth to try
And here we go:
Sure, Almost 10dB less in compare with other freq, but still really surprising assuming initial parameters.
Many thanks KA1W, KO4AXD, K0ZR, who made QSOs with me today. I do not know those guys, and really not expecting them to read this thread, but they helped to prove the concept
Quick update about project status and recent tests/findings.
1. Changes iin SWR detector circuit:
- we are completely changing detector circuit, moving away from diodes. Initially it was a hope to "keep it cheap and simple". However, measurements and findings clearly show that is very hard to compensate known diodes issues with software, specifically on low powers values. Hence, we are changing to AD8307-based detector circuit. 16-bit ADC will remain. BN43-4312 will remain with slight change in number of turns, it is more than sufficient for current schematics. Full set of ATU with new circuit expected to be completed in next 2.3 weeks.
2. ESP32 software is practically finished. Working on extending few parameters into User Interface. Max distance WiFi tests will be completed a bit later,
3. New version of ATUconnect is in next 2-3 weeks. I am waiting for ESDR3 beta release to complete set of TCI tests.
1. AD8307-based detector circuit was successfully tested. Working very stable for power from 0.5W to 100W. (500W/1KW testing is still pending, I am waiting for PA to arrive).
2. Two types of Tandem-match were tested together with AD8307:
- on FT82-43 (-29.54dB and -20dB) and -25dB "PI" Att.
- on BN43-3312 (-29.54dB and -20dB) and -25dB "PI" Att.
both versions should be capable to handle up to 2800W, however currently used hot R in Tandem is 3W only, hence the limitation of Tandem will be around 1800W PEP.
3. Field tests are currently continuing for AD8307, mostly to finalise changes in algorithms. However, final HW version will be based on AD8310 (as it is more stable version with less errors).
Tandem-match together with AD8310 detector circuit and ADS1115/ESP32 is designed to be used without changes in:
- variable coil-based remote ATU
- 8x8x8 remote ATU, and
- standalone SWR/Power meter.,
Последнее редактирование: 09 Апр 2021 15:06 от VK6NX.
We have tested two types of AD8310:
- "white" AD8310, purchased from US via Newark Electronics (Element14 in AU) (former Farnell) on $12 USD/pcs
- "clone" AD8310, purchased from Ebay of chinese seller on $4 USD/pcs
Testing and measurements been performed on each (FWD and REF) chip separately with combined attenuation of -54.54dB
Results: "White" FWD
[50 Ohm -54.54dB: Fwd] Voltage at 1kW: 2569.747403
[50 Ohm -54.54dB: Fwd] Voltage at 0.9kW: 2558.31317
[50 Ohm -54.54dB: Fwd] Voltage at 100W: 2319.859725
[50 Ohm -54.54dB: Fwd] Voltage at 10W: 2069.972046
[50 Ohm -54.54dB: Fwd] Voltage at 1W: 1820.084367
[50 Ohm -54.54dB: Fwd] Voltage at 0.1W: 1570.196688
[50 Ohm -54.54dB: Fwd] Voltage at 0.001W: 1070.421331
[50 Ohm -54.54dB: Fwd] Voltage at 0.0001W: 820.533652
[50 Ohm -54.54dB: Fwd] Slope: 12494.38 mV/dB
[50 Ohm -54.54dB: Fwd] Intercept: 790.453 V
[50 Ohm -54.54dB: Fwd] Function dir: "y = 1820.084367 + 108.52484 ln(x)"
[50 Ohm -54.54dB: Fwd] Function inv: "y = exp((x - 1820.084367)/108.52484)"
[50 Ohm -54.54dB: Rfl] Voltage at 1kW: 2573.87638
[50 Ohm -54.54dB: Rfl] Voltage at 0.9kW: 2562.545191
[50 Ohm -54.54dB: Rfl] Voltage at 100W: 2326.240662
[50 Ohm -54.54dB: Rfl] Voltage at 10W: 2078.604943
[50 Ohm -54.54dB: Rfl] Voltage at 1W: 1830.969225
[50 Ohm -54.54dB: Rfl] Voltage at 0.1W: 1583.333507
[50 Ohm -54.54dB: Rfl] Voltage at 0.001W: 1088.06207
[50 Ohm -54.54dB: Rfl] Voltage at 0.0001W: 840.426352
[50 Ohm -54.54dB: Rfl] Slope: 12381.79 mV/dB
[50 Ohm -54.54dB: Rfl] Intercept: 795.18 V
[50 Ohm -54.54dB: Rfl] Function dir: "y = 1830.969225 + 107.546826 ln(x)"
[50 Ohm -54.54dB: Rfl] Function inv: "y = exp((x - 1830.969225)/107.546826)"
[50 Ohm -54.54dB: Fwd] Voltage at 1kW: 2604.423559
[50 Ohm -54.54dB: Fwd] Voltage at 0.9kW: 2593.157442
[50 Ohm -54.54dB: Fwd] Voltage at 100W: 2358.209943
[50 Ohm -54.54dB: Fwd] Voltage at 10W: 2111.996326
[50 Ohm -54.54dB: Fwd] Voltage at 1W: 1865.78271
[50 Ohm -54.54dB: Fwd] Voltage at 0.1W: 1619.569094
[50 Ohm -54.54dB: Fwd] Voltage at 0.001W: 1127.141861
[50 Ohm -54.54dB: Fwd] Voltage at 0.0001W: 880.928244
[50 Ohm -54.54dB: Fwd] Slope: 12310.68 mV/dB
[50 Ohm -54.54dB: Fwd] Intercept: 810.299 V
[50 Ohm -54.54dB: Fwd] Function dir: "y = 1865.78271 + 106.929215 ln(x)"
[50 Ohm -54.54dB: Fwd] Function inv: "y = exp((x - 1865.78271)/106.929215)"
[50 Ohm -54.54dB: Rfl] Voltage at 1kW: 2603.740694
[50 Ohm -54.54dB: Rfl] Voltage at 0.9kW: 2592.428629
[50 Ohm -54.54dB: Rfl] Voltage at 100W: 2356.522918
[50 Ohm -54.54dB: Rfl] Voltage at 10W: 2109.305142
[50 Ohm -54.54dB: Rfl] Voltage at 1W: 1862.087366
[50 Ohm -54.54dB: Rfl] Voltage at 0.1W: 1614.86959
[50 Ohm -54.54dB: Rfl] Voltage at 0.001W: 1120.434038
[50 Ohm -54.54dB: Rfl] Voltage at 0.0001W: 873.216261
[50 Ohm -54.54dB: Rfl] Slope: 12360.89 mV/dB
[50 Ohm -54.54dB: Rfl] Intercept: 808.694 V
[50 Ohm -54.54dB: Rfl] Function dir: "y = 1862.087366 + 107.365316 ln(x)"
[50 Ohm -54.54dB: Rfl] Function inv: "y = exp((x - 1862.087366)/107.365316)"
1. AD8310 is definitely recommended over AD8307 (we cannot get Slope more than 8-9V/dB on all tested AD8307). Both "clone" and "white" AD8310 have demostrated Slope and Intercept parameters very close to datasheet (please note that one input out of avaliable two is used in described application, hence Slope of +24 mV/dB as per datasheet has to be devided by 2).
2. "Clone" AD8310 definitely can be used, it is quite efficient in price for money value as well.
3. It is recommended to use the "pair from one reel". Much easier to calibrate. More possibility that chips from one reel were produced with close conditions and have close parameters.
4. Calibration is still needed for every chip to get as much as possible close log function.
5. No difference found between "shielded" (with AD8310 ground shield covered) and "unshielded" construction. We will be using "shielded" version in 8x8x8, as we are making it for 500W+
Anyway, we are back on track now, finalising the 8x8x8 ATU construction with AD8310-based tandem.
Hopefully I will be able to post some test outcomes by end of the month.
Последнее редактирование: 21 Апр 2021 05:54 от VK6NX.
I have received number of emails asking to expand step-by-step calibration for AD8310. Rather than answer each email privately, I think it is better idea just to capture it here for tracking records.
Ad8310 calibration methodology Disclaimer: Below guide is not "best of breed", not even "recommended". This is just the easy way our team found quick and simple. I am sure there are better ways and would be happy if someone will share their own.
1. Step 1 - calibrating REF.
All we want to do in this step is to make sure we capture how REF AD8310 performs if connected as FWD.
Important - do NOT solder to Tandem board initially: FWD AD8310 chip, and C1
Use the transceiver with gradually controlled PA output of 1-100 watts. If you have any power output source with <1W (something like RF generator, which can produce output of (example) 10mV, 100mW - this will make calibration more accurate). Also use any SWR/Power meter (better with good precision).
1.1 Solder REF AD8310 chip and R2, R4, R7-8 (those can be single resistor) and C2
1.2 Run temporary wire between C2 and pin 8 (INH) of REF AD8310.
1.3 Connect tranceiver, Tandem, Power meter and 50Ohms dummy load and do set of tests: run the carrier with 1W power output, control the output level on Power meter and record the voltage value, measured with multimeter on the output pin 4 (VOUT) of REF AD8310 (alternatively voltage can also be measuread at the output of L2 by miltimeter or even controlled by ADS1115).
1.4 Repeat step 1.3 for different power outputs. In our tests we did run output in watts 1-2-3-4-5-6-7-8-9-10-20-30-40-50-60-70-80=90-100.
1.5 Once completed, remove temporary wire between C2 and REF AD8310. Solder FWD AD8310 and C1
2. Step 2 - calibrate FWD.
In this step all we do is repeating step 1 for FWD AD8310.
2.1 Connect transceiver, Tandem, Power meter and 50Ohms dummy load.
2.2 Run the carrier with 1W and record the record the voltage value, measured with multimeter on the output pin 4 (VOUT) of FWD AD8310.
2.3 Repeat step 2.2, try to keep carrier output levels as much close as possible to your tests for REF.
At the end of step 2 you should have the voltage outputs for FWD and REF recorded. We have used simple Excel file to record the outputs and get ready to next step. Example is attached.
3. Step 3 - calculate the log function for FWD and REF.
AD8310 is the logarithmic amp. Obviously, the voltage outputs you have recorded in Step 1 and 2 are representing AD8310 logarithm. To get further use of the outputs, we need to convert them into function.
There are number of ways how you can do that (surely you remember that from your school and uni ). Because we have very lazy people in our team, we (again) have chosed "quick workaround". Thanks to Dmirty (VK3FDMI), we are using his tool for VSCode: github.com/dkaukov/log-power-meter
3.1 Replace the voltage (and if needed - power) values in index.js
"const tmV01" and "const tm50ohm54dbFwd" is FWD
"const tmV02" and "const tm50ohm54dbRfl" is REF
3.2 Run program via "Run and Debug" in VSCode. Look at the output in Debug Console and scroll down to the section
If you will be using Excel example from above for the records, the formatting for VSCode pretty much ddone in OUT tab.
Please note that AD8310 does not care much about the Tandem attenuation. You can change transformer attenuation and Pi-att velues as you wish. The only thing to care while adjisting attenuator is to not overload AD8310 Inputs (INHI, INLO) - which is according to datasheet is ±2.0 var ±2.2 and IMPORTANT in dBV!. It is better to look at datasheet anyway if you will be modifying Tandem attenuator.
In our case we have used 30:1 transformer based on FT50-43 (-29,54dB) and Pi-att (-25dB) of 56 Ohm-442 Ohm-56 Ohm. Hence, the combined attenuation is -54.54dB
With this att we should end in mid of Slope (according to Figure 3 at page 6 of the datasheet) and we are assuming the max power values will end up in a range where AD8310 has -10dB att.
Hope this helps.
Последнее редактирование: 29 Апр 2021 05:29 от VK6NX.