Hi all,
I believe I am also being affected by this. I regret that I did not
investigate more deeply early in testing of the new modes to see why my
amplifier often goes into some anomalous condition at the start of a
FST4/W transmission and does not always recover. I have also been seeing
evidence of significant RF output waveform anomalies at the end of all
FST4/W transmissions. I did not realize these modes apparently use
amplitude shaping at the start and end of a transmission until others
brought it up. If I had realized this early on I would have immediately
raised a red flag. Although I routinely monitor my RF output with an
oscilloscope (that's how I know there are anomalies), I unfortunately
don't have a storage scope to capture and examine the events in detail.
The class D and E amplifiers widely used on LF and MF cannot reproduce
amplitude variations. The FETs are run as switches with two valid
states: on and off. Many of the FET driver chips can improperly clock
the FETs when the amplitude of the input waveform is below a certain
design level. This can cause several issues, including damaged FETs.
Also note that in some cases these amplifiers will produce output
spectra that is not clean when in these abnormal drive states. If the
shaped waveform was intended to produce a cleaner signal, it is likely
having the opposite effect in some cases (almost certainly including my
signal based on the anomalies I am seeing).
I realize we are in a late stage of development of these modes now,
possibly about to see a GA release. But is there any way this shaped
envelope can be changed (perhaps as a user option) such that the
waveform will start and stop instantly (or nearly so, as other nodes do)
without amplitude shaping?
73 and Happy New Year,
Paul N1BUG
On 1/4/2021 9:59 PM, Peter Hall wrote:
Hi WSJT Developers,
I’ve not submitted a bug report before and hope that this email find its
way to the appropriate team members.
Recently, in the RSGB low frequency and WSJT-X user groups I’ve raised
the issue of the slow envelope rise times on FST4 and FST4W causing
problems with the non-linear power amplifier designs (Class D or E)
common among LF/MF users. For your information, I’ve reproduced the
original post in the WSJT-X group below, but note that the thread has
developed as other users have injected their own experiences. The
thread is at:
https://wsjtx.groups.io/g/main/topic/79269122?p=Created,,,20,2,0,0
<https://wsjtx.groups.io/g/main/topic/79269122?p=Created,,,20,2,0,0> In
particular, please note that ON7YD has retracted his first measurements
(referenced in my original post, below) and has supplied an update which
agrees well with my own and other envelope risetime characterizations.
If you’d like any additional information or assistance, I’d be very
happy to help.
73, and thanks for your assistance, Peter (VK6HP)
Original WSJT-X user group post containing details of issue and
configuration:
Hello everyone,
I've just joined this group and would be interested to know if anyone
can shed some light on an issue I'm experiencing. I do quite a bit of
work on 630 m using a TS-890S exciter and a homebrew 300 W Class D
amplifier. The 0 dBm drive for the PA is taken from the Kenwood "DRV"
output. For those interested, the transmit antenna is a ZS6BKW HF
doublet loaded as a Marconi T and the most usual receive antenna is a 1
m diameter active loop. I'm running Windows 10 on a Microsoft
SurfaceBook and a USB connection between the PC and the radio's internal
audio interface ("sound card"). The software version is WSJT-X
v2.3.0-rc2, although the -rc1 version behaves identically. I use PTT/CAT
control but have experimented with VOX with no resolution of the central
issue.
I've had no problems running any JT modes on any bands until I began
trying out the FST4 and 4W modes on 630 m a few months ago. Both WSPR
and JT9 work very well on 630 m and the new modes also work, but with a
small twist: the risetime of the FST4 envelope is so slow that it upsets
the clocking arrangement in the push-pull PA. This causes the PA
protection to activate, with a consequent small loss of signal at the
start of the transmission. It's not a particularly serious problem, but
is annoying. I do have a hardware fix in the form of what is
essentially a noise gate circuit between the radio and PA; the drive
signal has to be above a set level (chosen to reliably clock the PA)
before it is passed to the amplifier. However, after some recent
investigation I can quite clearly see a difference in the transmit
delays and transceiver waveforms when doing nothing else but changing
from e.g. WSPR to FST4W-120.
In brief, the WSPR envelope rise time is of order a few milliseconds and
the delays are well-behaved, with a zero transmit delay actually
corresponding to a few ms delay in seeing the TS-890S drive waveform.
With FST4W, the RF envelope rise time is very slow indeed, with many
tens of ms observed. The transceiver delay is also much longer -
typically more than 100 ms.
I should add that I've successfully sent and received FST4W signals,
including some 4-300 and 4-900 contacts.
I'm unable to spot any transceiver setting differences between old and
new modes and have not yet experimented with other transceivers.
However one fellow member of the RSGB LF Group was kind enough to show
that, using a simple sound card, he was unable to see any difference
between the envelope characteristics of various modes. It does look
like a software issue to me and I'd appreciate comments from anyone
running a similar setup, or with relevant suggestions.
73, Peter (VK6HP)
Peter J Hall
Emeritus Professor of Radio Astronomy Engineering, Curtin University
ICRAR/CIRA,
Building 610,
1 Turner Avenue
Bentley Technology Park, WA 6102.
Ph: + 61 400 801 531
Post: GPO Box U1987, Perth WA 6845, Australia.
www.icrar.org <http://www.icrar.org/>
_______________________________________________
wsjt-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
_______________________________________________
wsjt-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
