Comparing JTDX and WSJT-X

Two different sources have said JTDX decodes more stations than WSJT-X.  Is this true?  That is the theme of this project today.  JTDX states it is “a derivative of work based on WSJT-X”.  There are word of mouth claims that JTDX made improvements to the decoding algorithm.  Today’s project used two identical radios and two similar computers connected to the same antenna through a splitter and the results were compared.  One computer ran JTDX and the other WSJT-X.  All were running the latest versions of software,  identical settings, and levels.  Here is what was found.  

On some passes both applications had the same number of decodes and they were the same stations on both but most of the passes JTDX did indeed have more decodes.  Here is a typical example of a pass taken at the same time slot. Compare the bottom left of the screens.

2020-03-13 (3)-DESKTOP-T04IGTF


2020-03-13 (3)

Observing the pass for time stamp 222930, there are 9 stations decoded by WSJT-X (upper screen), and 12 stations decoded by JTDX (bottom screen).  Digging a little deeper the three stations not decoded by WSJT-X are CO8LY, KC8GCR, and K4TT.  Were those stations too weak for WSJT-X?  Apparently not because weaker stations were decoded by WSJT-X, for example, K2AJK and K1BDC.  Are there clear indications of why those three stations were not decoded by WSJT-X?  It needs deeper research.  It is apparent that JTDX decodes more stations.  The results speak for themselves.

Both radios are Icom IC-7300’s.  The computers are both Intel NUC’s with Windows 10 and the latest updates.  To verify the results the computers were swapped, the tests repeated, then the radios swapped, tests repeated, and finally the applications on the computers were swapped. None of the swapping around showed any changes in the results.  The antenna was a random length of wire vertically oriented.  The mode was FT8.

JTDX is available here:

WSJT-X is available here:

FT8 Shootout

A comparison of FT8 decodes using different radios.


This project is an experiment to find out what settings and which of the models of transceivers we own is better for FT8 decoding.  We will count the number of decodes on each pass using the same antenna through a splitter.
Questions we will attempt to answer:  Is the $3000 IC-7610 worth the extra money over the $1000 IC-7300?  Is the $4000 Flexradio 6600?   How does AGC affect the number of stations decoded?  Should gain be set high to allow the weak stations a chance, or should it be set low to keep the strong stations from overloading the input?
Decoding is done in the computer application (WSJT-X v2.1.2) not the radio.  All models are direct sampling sdr’s.
Preconceived notions are that the more expensive radios will perform better.


Test 1 – 7610 vs 7300 number of decodes per pass

7610     vs    7300

32                 36

33                 32

27                 29

28                 26

26                 25

36                 36

26                 27

Comment: Sometimes one wins. Sometimes the other. Sometimes they tie.  It’s a toss up.

Test 2 – SNR of same stations on each radio (callsign suffix shown)

7610    vs     7300

KV    +6        +4

LXD  -20      -20

ENN -21      -21

ON   +17      +16

Comment: There is more compression with strong signals by the 7300 .  Thus one would conclude there is slightly more dynamic range on the 7610.

Test 3 – Flex  vs   7300 number of decodes per pass, congested band

Flex    vs    7300

33                 33

31                 32

36                 35

30                 31

35                 36

36                  35

Comment: It’s a toss up.

Test 4 – Flex  vs   7300 number of decodes per pass, dead band

Flex    vs    7300

2                    2

2                    2

2                    2

3                    3

2                    2

Comment: Identical

Test 5 – SNR of same stations on each radio

MRB  -18       -19

JCN    -11       -10

RW     0         +1

PCC    -12      -12

Comment: Toss up.

Conclusion:  No substantive difference in the number of decodes among the three radios tested.  There are vast differences in features that may affect other modes or user experience but the  difference is minimal as measured by the yardstick of FT8 decode counts.

Notes:  It was found that the highest number of decodes was always with the rf gain as high as possible just short of the point of overload.  Strong stations were always decoded but if the rf gain was reduced weak stations disappeared.  It was also found that the best setting for AGC was FAST.  Other settings, including the setting recommended in the WSJT-X User Guide of AGC OFF, reduced the number of decodes.


Regard with caution any count of 36.  JTAlert was used as a counting tool and it only counts up to 36.  There could have been more than 36 in a pass and the extras were not counted.

– 30 –

WSPR Station

A Weak Signal Progagation Report (WSPR) station is on the air now from the permanent site at Strasburg.  It runs 1 watt on multiple bands. shows the results.   Hardware is a Yaesu FT-817 transceiver and an Intel Compute Stick computer.  Software is WSJT-X Ver 1.8 running in WSPR mode.  Antenna is a Hustler 6BTV trap vertical.  Ground radials are welded-wire fencing.


KC0RF, John, and I were just finishing up when rain hit.

Control is done over the Internet using a remote access application, Teamviewer.  FCC regulations specify a beacon can only run when an operator is present to control it.  Whenever I leave the house the WSPR station is shut down by using Teamviewer.

Here’s a random screen snap of the results for the last hour.  These stations have “spotted” my 1 watt beacon and posted it to in the last hour.  We’re getting many more spots with the base antenna out here than we did with the mobile antenna at home.

Screenshot 2017-08-11 15.45.08

August 15, 2017 – measuring the swr with a MFJ-259B antenna analyzer showed high swr on 30 meters and 80 meters.  We changed the WSPR configuration to only listen on those two bands.  Hopefully we’ll get time to readjust the 6BTV before the eclipse next Monday.

It’s working now but a few hurdles had to be cleared to get this station on the air, as you’ll see in the previous WSPR post.

WSPR Station

The Solar Eclipse QSO Party urges, “Operate a wide-band automated receiver at your station” like a wspr station.  The station has been in staging at home before it gets deployed to the permanent site in Strasburg but not working correctly.  Finally a breakthrough.   Today we figured out a problem.  Now we’re seeing our wspr station being spotted on, running only 1 watt and feeding a Hustler mobile whip antenna.

Screenshot 2017-07-23 19.54.16

This is the Hustler mobile antenna getting that big 1 watt signal around the country.


The WSPR station being staged before deployment looks like this.

FullSizeRender (1)

The computer is an Intel Compute Stick, on the left side half way up running Windows 10 and WSJT-X version 1.8.  The red box is a West Mountain Radio PlugNPlay Rig Runner to interface to the Yaesu FT-817.  The yellow candy wrapper is shrink wrap around a dc-to-dc converter.  It takes in 12 volts dc and converts it to 5 v dc for the powered usb hub and for the Intel Compute Stick.  Green and blue plugs from the PlugNPlay go into an external usb sound card.  One of the usb dongles is a Yaesu CT-62 CAT interface cable.

Deployment plans include packaging the above apparatus into a fiberglass outdoor enclosure, then adding a solar panel, controller, and batteries.  The permanent antenna will be a Cushcraft AP8A trap vertical picked up at a recent hamfest.  The complete 12 volt station draws 1.3 amps when transmitting (at 1 watt) and draws .6 amps when receiving, including the computer. Can’t use an automatic tuner because wspr changes bands and listens before transmitting.  That means the tuner is still tuned to the previous band instead of the current one.  Must use resonant antennas like trapped vertical.

We’ll turn on band hopping and cover 80 through 10 meter bands.  If we get the ambition we’ll add a 6m vertical stub and cover that band, too.  This project is targeting the August 21, 2017 solar eclipse.  In addition to WSPR we plan to operate the Solar Eclipse QSO Party using weak signal modes JT65, JT9, and FT8 on the main station. We’ll be receiving on 630 meters and transceiving on 160 through 6 meters making as many QSO’s as we can.

First plan went out the window because a trapped vertical needs radials and the weeds  make that impossible.  Plan B is to try a Comet CHA250B because it requires no radials.  It just didn’t get a signal out. Out the window.  Plan C is to receive only at Strasburg and transmit only from home.  Can’t receive from home due to a high noise level.  But at Strasburg  when unchecking all the transmit boxes in wspr it no longer band hops.  Is this a bug?

A solution popped up while reading the manual to try to see why it stopped bandhopping.  There is a “tune” checkbox that sends unmodulated carrier for a few seconds after changing bands.  We just happen to have the ideal tuner — a SGC SG-211.  We’ll try that next.

The SG-211 failed because it needs more than a watt for tuning.  Next we tried a LDG RT-100 and it worked perfectly.

Update, September, 2022 —  Zachtek WSPR Transmitter is discovered and solves the problem.  See QST, September, 2022, ZachTek WSPR Desktop Transmitter, page 44.  “The ZachTek WSPR Desktop Transmitter is a standalone multiband WSPR transmitter that contains its own WSPR firmware, so it does not require an external computer and software to operate.  Although it’s a low-power transmitter, you will be surprised how far you can get with it — even mobile!”.  The transmitter was ordered, arrived a week later, and was installed.      

For an antenna, the Hustler 6BTV that was originally purchased for WSPR in 2017 was called upon. Part of 6BTV installation is carefully tuning each band to the WSPR frequency for that band.

The ZachTek has been in service since early September, 2022 and the results are quite impressive as demonstrated by the late morning screen shot below.

It runs on 5 volts and is very reliable so far. There is one issue, however, and that is rf interaction with the main station. When the main station is transmitting on 40 meters the ZachTek crashes, even with a lot of toroids on the leads. It recovers with a reboot. This issue was solved by installing a relay to turn off the ZachTek whenever the main station is turned on. When the main station is turned off and the ZachTek is turned back on, it reboots automatically and begins transmitting again. All settings are preserved each time and does not require manual intervention. Missing out on WSPR while the main station is running is very frustrating when the main station is running for a long time. An alternate solution is needed.

Commentary. At any given moment in 2022 there are more stations on FT8 than on WSPR or any other beacon network. The shear quantity of stations operating FT8 outnumbers the number of WSPR stations and therefore provides more information. As good as WSPR is, PSKreporter may be better because it shows a broader picture of propagation. Take a look at PSKreporter for the same time and draw your own conclusion. PSKreporter is configured to display the local grid square, all bands, all modes, for the past 15 minutes.