I finally found some time to play radio today and so tried the U3s that I built recently.
I have it built with a 25 turn inductor, three BS170's fed with 9v for the PA, no heatsinks, with 200mw output and a small standoff glued to the crystal surrounded with cotton wool to try and keep drift to a minimum, all fitted into the U3s case. Two DC-DC convertor boards, also inside the case, supply the 9v and 5.5v feeds to the main board. Antenna was a Sotabeams linked dipole as an inverted Vee, 6m high in centre with the legs of the dipole trailing along and through a hedge finishing up around a metre off the floor, not ideal but it'll do for this test. The U3s is connected to the qrp labs gps, by shielded cable, which resides in a metal fudge tin with external antenna.
First transmission on 40m yielded no spots. Then I noticed that I hadn't connected the antenna, ooops. That corrected spots were received in Russia, Scandanavia, all over Europe and the Canary Isles, 2774km. Most impressive was the drift reports, virtually all stations reported zero drift, the cotton wool works well. The radio was outside in cold foggy weather run from a 12v 7Ah sealed lead acid battery.
Kit was pretty straight forward to build although I never got the QRP Labs 5w amp to work with the U3s. I have removed the key shaping circuitry and it is now configured in it's simplest form and will be used with a homebrew cw transceiver. It'll be interesting to see how much power it will give.
73
Andy
M0RON

I'm sort of curious how 3xBS170=200mW on 40m at 9V.? That's about what you should get from one BS170 at ~5v on 40m.

Apparently, using switching transistors (in place of power transistors) in parallel to boost output requires precise matching of something, apparently one of the measurements of capacitance within the transistor, as far as I've been able to follow the explanation.? If they're not matched, then only one of the three winds up doing the work, again according to the explanation I don't claim to fully grasp.

I have sixteen BS170s awaiting a fuller explanation of that matching process, but I'm concerned that my lack of an actual antenna may drive the SWR up into the danger zone for the circuitry if I significantly increase the output of my U3S.? I have one of the 5W PA kits built as well, but haven't hooked it up for the same reason.? I don't understand why the build failure rate seems to be so high for the 5W PA kit, but that's another thing that's kept me from powering it up, too.? I wish this had been a more popular kit so there would be a bigger knowledge base here for it.

Anyway, nothing seems to blow at ~150-200mW with presumably awful SWR.?

Even with an "antenna" that's nothing but a thin, bare wire about 45' long strung down a hallway at about 8' above ground level, I've managed spots at almost 2100 miles (3380km) on 40m and just a hair under 3000 miles (4828km) on 20m.? Of course, much of the time I get exactly nothing.?

If I had a real antenna and a couple of watts I'm sure I could perform better.? Strangely, I've never been spotted in Hawaii.

This business of controlling drift is a bit on a mystery to me as the explanations are if not completely illogical, at least counterintuitive -- you gain stability by either (a) preventing the crystal from cooling down or (b) causing it to cool down.? Huh?? In any case, as I understand it, if you don't drift beyond +/-4, it doesn't make any difference anyway.? The default setting gives me as close to zero drift as external conditions will ever permit, but it generates RFI between transmissions enough to be a nuisance.? Changing the settings stops that, but makes for a little more wobbly drift in some bands.

Good luck with your new U3S!

开云体育

TrueBlue,

I'm sort of curious how 3xBS170=200mW on 40m at 9V.? That's about what you should get from one BS170 at ~5v on 40m.

Han's stated:-

Apparently, using switching transistors (in place of power transistors) in parallel to boost output requires precise matching of something, apparently one of the measurements of capacitance within the transistor, as far as I've been able to follow the explanation.? If they're not matched, then only one of the three winds up doing the work, again according to the explanation I don't claim to fully grasp.

from section 5.10 of the QCX Build Manual.
same devices used in a slightly different PA but Hans spent some time writing up the theory of parallel operation.

"There are always minor variations between device characteristics from one transistor to the next. If these were bipolar NPN transistors, we would not be able to parallel them in this way. If one transistor takes more of the load and starts to heat up, its resistance further decreases and this causes it to get even hotter. This process is known as “thermal runaway” and results (quickly) in destruction of the transistor. Emitter resistors are used to help balance the load. But with MOSFETs, their resistance INCREASES as the temperature goes up – so there is an inherent self-balancing when multiple devices are used in parallel, without any need for additional balancing resistors which would increase component count and waste some power."

I have sixteen BS170s awaiting a fuller explanation of that matching process, but I'm concerned that my lack of an actual antenna may drive the SWR up into the danger zone for the circuitry if I significantly increase the output of my U3S.? I have one of the 5W PA kits built as well, but haven't hooked it up for the same reason.? I don't understand why the build failure rate seems to be so high for the 5W PA kit, but that's another thing that's kept me from powering it up, too.? I wish this had been a more popular kit so there would be a bigger knowledge base here for it.

Don't be a defeatist. If you set out with the mind set I can't do it, chances are....
You will never know if YOU will have any problems until you try.

But one thing at a time, if you are using the U3S as the driver then get that sorted first.

The U3S is a kit, the PA is a kit,
if you have limited access to test equipment perhaps buddy up with an elmer or find somebody at a local club
that does his own construction to help you through it, even better get him to buy one as well., and do one each.
There is a lot of info on the PA web page, build manual & App notes, but the info you need is there.

Anyway, nothing seems to blow at ~150-200mW with presumably awful SWR.?

Even with an "antenna" that's nothing but a thin, bare wire about 45' long strung down a hallway at about 8' above ground level, I've managed spots at almost 2100 miles (3380km) on 40m and just a hair under 3000 miles (4828km) on 20m.? Of course, much of the time I get exactly nothing.?

That's the marvel of WSPR, however any transmitter is only as good as it's antenna. But that's a whole different topic.....

If I had a real antenna and a couple of watts I'm sure I could perform better.? Strangely, I've never been spotted in Hawaii.

This business of controlling drift is a bit on a mystery to me as the explanations are if not completely illogical, at least counterintuitive -- you gain stability by either

There's is also an APP note for that:- AN001?

(a) preventing the crystal from cooling down or

The aim is to minimise temperature change and loading to the 27MHz crystal reference and to a lesser extent the Si5351 IC during the WSPR or more demanding QRSS TX cycle / and post TX calibration cycle.
Put the U3S in a box, draughts is bad.
Try not to put anything that gets warm near the crystal, the PA gets warm... But the cure is easy, make a baffle so the heating convection currents can't reach the crystal.
Cut up one of those pre approved credit cards that drop through the door every week,? or just fold up a piece of card.
If you can find the room, perhaps tape a pad of cotton wool? or Polystyrene over the crystal and perhaps IC, as others have suggested.
(I would try and keep? an air gap between the pad and the baffle if you use one to prevent heat conduction.)

The black art of Park Modes, expect to have to experiment. Everybody's unit will have a different spread of tolerances , what works for one may may make it worse on another.
If you are frequency hopping you may need to compromise.
(Don't forget drift reported by WSPRnet is the sum of your drift (if any), propagation induced, (think Doppler Shift) & drift in the receiving station.
Not all the drift maybe you, some receive stations out there have drifting receivers !

The long TX cycles of QRSS is the most demanding mode, if you believe you have a problem, get yourself seen on a local QRSS grabber. You can see any changes in real time(ish) and have a feel what is making things better or worse.

(b) causing it to cool down.? Huh?

You are thinking about, Han's solution of gluing a coin or other people have reported attaching a stick on heatsink to the crystal?

The aim IS NOT to cool the crystal but to add a thermal? mass to it, to again trying to delay short term temperature changes.
Personally I would look through my junk box for something small and heavy ie: dense.
?A mild steel nut would be my preference over a piece of aluminium, but anything that slows the temperature change works....

In any case, as I understand it, if you don't drift beyond +/-4, it doesn't make any difference anyway.? The default setting gives me as close to zero drift as external conditions will ever permit, but it generates RFI between transmissions enough to be a nuisance.? Changing the settings stops that, but makes for a little more wobbly drift in some bands.

( Use the search function to read up about Huff - Puff theory and recommended calibration starting points, just in case what you are experiencing isn't all thermal. )



Cut & pasted from another 'drift thread' it's about the OCXO version of the synth but a good power supply is equally required for the standard version.

Have you seen Han's & Peters High precision OCXO U3S write up in the hints & kinks:-



(In particular, decoupling capacitors around the Voltage Regulators, mounted as close as you can to the device legs.
As well as hopefully decoupling RF, and general hash they prevent parasitic oscillation that affects regulators under some load conditions.)
Pick your favourite semiconductor manufactures data sheet, and check out their recommendations for values.


Alan

Good luck with your new U3S!

TrueBlue,

I have one U3S with three BS170s putting 1/2 watt into a 50 ohm dummy load at 5 V and 1 watt at 9?V on 40 meters.? I did put tiny self-adhesive heat sinks on the BS170s, but they run very cool into a matched antenna; I do not think the heat sinks are even needed.? There is a lot of builder data on the QRP-Labs website as well as in this forum.

The second U3S uses the PA with one BS170 and puts?out about 4 watts at 475 KHz.? The PA kit works as designed with no issues.? I did put a small cooling fan in the case and I do monitor the temperature with a LM34DZ; this is a bit of an experimentation area as clearly specified in the documentation.? I keep the heat sink temperature below Hans' empirical test data results.

I followed Hans instructions, used conservative biasing (easy way to me by measuring power supply current), and matched antennas.

The instructions and supporting documents are very detailed; read them a few times as it is easy to miss key points.? Hope that provides some feedback on success with BS170s and PA kit.

Good luck!

73,
Dave, W4VU

Hi True Blue,
Later this week I'll measure the output of my U3s and post here, I have 80,40,30,20,10 and 6m LPF's in my kit. As frequency increases power output will decrease so I'm expecting slightly more on 80 and much less on 6.
Also I am using 25 turns on the toroid so power will be less than if I had used the 10 turn bifilar winding, this is pointed out in the build manual in a couple of graphs.
73
Andy

On Sun, Jan 14, 2018 at 10:13 am, M0RON wrote:

I am using 25 turns on the toroid so power will be less than if I had used the 10 turn bifilar winding

I have one of each, as the 25-turn is proper for use with the 5W PA kit, and the bifilar is optimum for other use.? My intentions were to use one U3S for experimenting with the transistors and the other with the 5W kit as an intended CW transmitter.? In the interim, I decided I wasn't going to mess with CW in any capacity, so maybe (if the 5W kit works) it'll be for DX beaconing, I dunno.

I've been licensed for almost a year, and my intentions have changed drastically within that time.? I suspect that beaconing will be all I do on HF as QRM is so nasty here that there's really no point in even trying to RX/QSO on HF.? I may do space stuff as it's on cleaner frequencies.

On Sun, Jan 14, 2018 at 11:42 am, TrueBlue wrote:

I've been licensed for almost a year,

Since you've never published your callsign - how would we know ?

开云体育

I find it very strange that a member of this group is hiding behind a nickname and won’t post his (or her) callsign. ?I specifically asked what it was when the New Year QRSS Party happened but nothing was forthcoming.

--
QSO365 - My amateur radio blog:?

?

True Blue seems to have rather quiet.

Maybe he's keen not to let us know he's really John KM6HUV, who also posts as Hazada Masklo.

I wonder who he'll be next week ? ;-)

73 de Andy

On Sun, Jan 14, 2018 at 01:09 pm, Andy G0FTD wrote:

True Blue seems to have rather quiet.

I don't know what you're on about, but if you have some problem with me, why don't you deal with it like a grownup and send me private mail explaining what's wrong.? OK?? Cool.

In the meantime, here's my reply I spent all afternoon reformatting offline, which was why I was so mysteriously "quiet":

On Fri, Jan 12, 2018 at 05:20 pm, Alan de G1FXB wrote:

>> TrueBlue wrote:

>> I'm sort of curious how 3xBS170=200mW on 40m at 9V.? That's
>> about what you should get from one BS170 at ~5v on 40m.

>

But that still doesn't explain why more voltage and more transistors,
both, wouldn't increase output.? According to this ^, it should.

> Hans spent some time writing up the theory of parallel operation.

??? "There are always minor variations between device characteristics
from one transistor to the next. If these were bipolar NPN transistors,
we would not be able to parallel them in this way. If one transistor
takes more of the load and starts to heat up, its resistance further
decreases and this causes it to get even hotter. This process is known
as “thermal runaway” and results (quickly) in destruction of the
transistor. Emitter resistors are used to help balance the load. But
with MOSFETs, their resistance INCREASES as the temperature goes up – so
there is an inherent self-balancing when multiple devices are used in
parallel, without any need for additional balancing resistors which
would increase component count and waste some power."

This is nice to see, as there have been many dozen prior posts on this
board (and elsewhere) about the necessity for matching these BS170
transistors in this specific use.? I'm glad I haven't wasted even more
time and money in that pursuit.? I was >< this close to building a
MOSFET comparator for the task.? Misinformation is murder.

> if you are using the U3S as the driver then get that sorted first.

I'm OK on the U3Ss.? The build on mine was 100% without problems from
first power-up -- it's a bone-simple kit and I have good skills.? The
problems I did have (and there were many) were purely from misinformation
from a surprising range of sources.

> if you have limited access to test equipment...

Strangely, my expenditures on test and bench equipment have exceeded
those on actual radio gear by a factor of about 8:1, most notably on a
new oscilloscope that I was urgently persuaded to buy and have never
used once in the eight months or so that I've had it, aside from zeroing
the leads and testing its self-calibration.

> perhaps buddy up with an elmer or find somebody at a local club

I just let a membership expire on 1 January from what I believe is the
largest club in my area.? They offer a free calendar year's membership
to new licensees (I passed Tech and General on the same day).? I didn't
find membership productive.? My thoughts on this are another post.

> There is a lot of info on the PA web page, build manual & App notes,
> but the info you need is there.

It is really interesting.? The initial part on the tech background of
the design is mostly over my head and not of practical use, but the
actual construction and testing section is excellent.

> The aim is to minimise temperature change and loading to the 27MHz
> crystal reference and to a lesser extent the Si5351 IC during
> the WSPR or more demanding QRSS TX cycle

> The black art of Park Modes, expect to have to experiment. Everybody's
> unit will have a different spread of tolerances , what works for one
> may may make it worse on another.

> The long TX cycles of QRSS is the most demanding mode, if you believe
> you have a problem, get yourself seen on a local QRSS grabber. You can
> see any changes in real time(ish) and have a feel what is making
> things better or worse.

There's only one grabber in the western US, which never picked me up in
that NYD QRSS thing anyway.? However, I did monitor my own TX and it
showed a very linear (if small) drift over a ~7min. TX, consistent with
crystal heat-up.

Many another QRSS transmission I saw on the grabber pages described the
trajectory of a beanbag tossed by a toddler.? I was amazed.

[But QRSS is over in the US, as is every other mode in the (currently
full) U3S firmware except WSPR and CW.? There's no point beaconing
signals with no one listening -- and they aren't.? Even the most recent
modes are essentially history.? I went from at least fifteen JT65 spots
a day to one or two, and two or three JT9 spots a day to that many in a
month.? Even WSPR is dropping off, it seems.

Hans really needs to provide some of these graveyard modes in the U3S
with the solemn offices of Christian burial and free up some firmware room for
FT8 and future modes that bear some interest.

The object in propagation beaconing is to get the absolute maximum
possible number of spots to see some patterning, and you can't do that
with dud modes.? I also think that a faster mode should provide more spots
during marginal propagation.? The current ubiquity and speed of FT8
seem to be ideal.]

>> (b) causing it to cool down.? Huh?

> You are thinking about, Han's solution of gluing a coin or other
> people have reported attaching a stick on heatsink to the crystal?

> The aim IS NOT to cool the crystal but to add a thermal? mass to it,
> to again trying to delay short term temperature changes.

Now, this actually makes sense - but there are dozens of misinformed
posts here on the subject that refer to using heatsinks on the crystals.
?Heatsinks are not there to provide "thermal mass" (interesting turn of
phrase), but to defeat thermal mass, i.e., to cool.? Thus my confusion.
Thanks for clearing that up.

> A mild steel nut would be my preference over a piece of aluminium,
> but anything that slows the temperature change works....

I think if I were convinced that this was a problem for me, I would just
wall-in the module and fill it with spray can expanding foam insulation.
I doubt one could improve on that.

>> In any case, as I understand it, if you don't drift beyond +/-4, it
>> doesn't make any difference anyway.? The default setting gives me as
>> close to zero drift as external conditions will ever permit

Once again, I'm still not understanding the quest for zero drift, at least with
WSPR.?

> a good power supply is equally required for the standard version.

[...]

> (In particular, decoupling capacitors around the Voltage Regulators,
> mounted as close as you can to the device legs.? As well as hopefully
> decoupling RF, and general hash they prevent parasitic oscillation
> that affects regulators under some load conditions.)

My voltage stability is absolutely rock-solid.? As to noise, I can't
say, but I could probably measure it with my (unused) oscilloscope if
knew how.? I'm unclear on the effects of noise on WSPR and the JT/FT modes.

On Sun, Jan 14, 2018 at 01:09 pm, Andy G0FTD wrote:

True Blue seems to have rather quiet.

I don't know what you're on about, but if you have some problem with me, why don't you deal with it like a grownup and send me private mail explaining what's wrong.? OK?? Cool.

In the meantime, here's my reply I spent all afternoon reformatting offline, which was why I was so mysteriously "quiet":

[edit] This is not cutting and pasting into the editor.? I'll keep trying.



On Fri, Jan 12, 2018 at 05:20 pm, Alan de G1FXB wrote:
On Fri, Jan 12, 2018 at 05:20 pm, Alan de G1FXB wrote:

>> TrueBlue wrote:

>> I'm sort of curious how 3xBS170=200mW on 40m at 9V.? That's
>> about what you should get from one BS170 at ~5v on 40m.

> http://qrp-labs.com/ultimate3/u3mods.html#morepower
But that still doesn't explain why more voltage and more transistors,
both, wouldn't increase output.? According to this, it should.

> Hans spent some time writing up the theory of parallel operation.
??? "There are always minor variations between device characteristics
from one transistor to the next. If these were bipolar NPN transistors,
we would not be able to parallel them in this way. If one transistor
takes more of the load and starts to heat up, its resistance further
decreases and this causes it to get even hotter. This process is known
as “thermal runaway” and results (quickly) in destruction of the
transistor. Emitter resistors are used to help balance the load. But
with MOSFETs, their resistance INCREASES as the temperature goes up – so
there is an inherent self-balancing when multiple devices are used in
parallel, without any need for additional balancing resistors which
would increase component count and waste some power."

This is nice to see, as there have been many dozen prior posts on this
board (and elsewhere) about the necessity for matching these BS170
transistors in this specific use.? I'm glad I haven't wasted even more
time and money in that pursuit.? I was >< this close to building a
MOSFET comparator for the task.

> if you are using the U3S as the driver then get that sorted first.
I'm OK on the U3Ss.? The build on mine was 100% without problems from
first power-up -- it's a bone-simple kit and I have good skills.? The
problems I did have (and there were many) were purely from misinformation.

> if you have limited access to test equipment...
Strangely, my expenditures on test and bench equipment have exceeded
those on actual radio gear by a factor of about 8:1, most notably on a
new oscilloscope that I was urgently persuaded to buy and have never
used once in the eight months or so that I've had it, aside from zeroing
the leads and testing its self-calibration.

> perhaps buddy up with an elmer or find somebody at a local club
I just let a membership expire on 1 January from what I believe is the
largest club in my area.? They offer a free calendar year's membership
to new licensees (I passed Tech and General on the same day).? I didn't
find membership productive.? My thoughts on this are another post.

> There is a lot of info on the PA web page, build manual & App notes,
> but the info you need is there.
It is really interesting.? The initial part on the tech background of
the design is mostly over my head and not of practical use, but the
actual construction and testing section is excellent.

> The aim is to minimise temperature change and loading to the 27MHz
> crystal reference and to a lesser extent the Si5351 IC during
> the WSPR or more demanding QRSS TX cycle

> The black art of Park Modes, expect to have to experiment. Everybody's
> unit will have a different spread of tolerances , what works for one
> may may make it worse on another.

> The long TX cycles of QRSS is the most demanding mode, if you believe
> you have a problem, get yourself seen on a local QRSS grabber. You can
> see any changes in real time(ish) and have a feel what is making
> things better or worse.
There's only one grabber in the western US, which never picked me up in
that NYD QRSS thing anyway.? However, I did monitor my own TX and it
showed a very linear (if small) drift over a ~7min. TX, consistent with
crystal heat-up.

Many another QRSS transmission I saw on the grabber pages described the
trajectory of a beanbag tossed by a toddler.? I was amazed.

[But QRSS is over in the US, as is every other mode in the (currently
full) U3S firmware except WSPR and CW.? There's no point beaconing
signals with no one listening -- and they aren't.? Even the most recent
modes are essentially history.? I went from at least fifteen JT65 spots
a day to one or two, and two or three JT9 spots a day to that many in a
month.? Even WSPR is dropping off, it seems.

Hans really needs to provide these graveyard modes with the solemn
offices of Christian burial and free up some firmware room for FT8 and
future modes that bear some interest.

The object in propagation is to get the absolute maximum possible number
of spots to see some patterning, and you can't do that with dud modes.
I also think that a faster mode will provide more spots during marginal
propagation.? The ubiquity and speed of FT8 seem to be ideal.]

>> (b) causing it to cool down.? Huh?

> You are thinking about, Han's solution of gluing a coin or other
> people have reported attaching a stick on heatsink to the crystal?

> The aim IS NOT to cool the crystal but to add a thermal? mass to it,
> to again trying to delay short term temperature changes.
Now, this actually makes sense - but there are dozens of misinformed
posts here on the subject that refer to using heatsinks on the crystals.
?Heatsinks are not there to provide "thermal mass" (interesting turn of
phrase), but to defeat thermal mass, i.e., to cool.? Thus my confusion.

> A mild steel nut would be my preference over a piece of aluminium,
> but anything that slows the temperature change works....
I think if I were convinced that this was a problem for me, I would just
wall-in the module and fill it with spray can expanding foam insulation.

>> In any case, as I understand it, if you don't drift beyond +/-4, it
>> doesn't make any difference anyway.? The default setting gives me as
>> close to zero drift as external conditions will ever permit
Once again, I'm still not understanding the quest for zero drift.

> a good power supply is equally required for the standard version.
[...]

> (In particular, decoupling capacitors around the Voltage Regulators,
> mounted as close as you can to the device legs.? As well as hopefully
> decoupling RF, and general hash they prevent parasitic oscillation
> that affects regulators under some load conditions.)

My voltage stability is absolutely rock-solid.? As to noise, I can't
say, but I could probably measure it with my (unused) oscilloscope if
knew how.? I'm unclear on the effects of noise on WSPR and the JT/FT modes.

>> TrueBlue wrote:

>> I'm sort of curious how 3xBS170=200mW on 40m at 9V.? That's
>> about what you should get from one BS170 at ~5v on 40m.

>

But that still doesn't explain why more voltage and more transistors,
both, wouldn't increase output.? According to this ^, it should.

> Hans spent some time writing up the theory of parallel operation.

??? "There are always minor variations between device characteristics
from one transistor to the next. If these were bipolar NPN transistors,
we would not be able to parallel them in this way. If one transistor
takes more of the load and starts to heat up, its resistance further
decreases and this causes it to get even hotter. This process is known
as “thermal runaway” and results (quickly) in destruction of the
transistor. Emitter resistors are used to help balance the load. But
with MOSFETs, their resistance INCREASES as the temperature goes up – so
there is an inherent self-balancing when multiple devices are used in
parallel, without any need for additional balancing resistors which
would increase component count and waste some power."

This is nice to see, as there have been many dozen prior posts on this
board (and elsewhere) about the necessity for matching these BS170
transistors in this specific use.? I'm glad I haven't wasted even more
time and money in that pursuit.? I was >< this close to building a
MOSFET comparator for the task.? Misinformation is murder.

> if you are using the U3S as the driver then get that sorted first.

I'm OK on the U3Ss.? The build on mine was 100% without problems from
first power-up -- it's a bone-simple kit and I have good skills.? The
problems I did have (and there were many) were purely from misinformation
from a surprising range of sources.

> if you have limited access to test equipment...

Strangely, my expenditures on test and bench equipment have exceeded
those on actual radio gear by a factor of about 8:1, most notably on a
new oscilloscope that I was urgently persuaded to buy and have never
used once in the eight months or so that I've had it, aside from zeroing
the leads and testing its self-calibration.

> perhaps buddy up with an elmer or find somebody at a local club

I just let a membership expire on 1 January from what I believe is the
largest club in my area.? They offer a free calendar year's membership
to new licensees (I passed Tech and General on the same day).? I didn't
find membership productive.? My thoughts on this are another post.

> There is a lot of info on the PA web page, build manual & App notes,
> but the info you need is there.

It is really interesting.? The initial part on the tech background of
the design is mostly over my head and not of practical use, but the
actual construction and testing section is excellent.

> The aim is to minimise temperature change and loading to the 27MHz
> crystal reference and to a lesser extent the Si5351 IC during
> the WSPR or more demanding QRSS TX cycle

> The black art of Park Modes, expect to have to experiment. Everybody's
> unit will have a different spread of tolerances , what works for one
> may may make it worse on another.

> The long TX cycles of QRSS is the most demanding mode, if you believe
> you have a problem, get yourself seen on a local QRSS grabber. You can
> see any changes in real time(ish) and have a feel what is making
> things better or worse.

There's only one grabber in the western US, which never picked me up in
that NYD QRSS thing anyway.? However, I did monitor my own TX and it
showed a very linear (if small) drift over a ~7min. TX, consistent with
crystal heat-up.

Many another QRSS transmission I saw on the grabber pages described the
trajectory of a beanbag tossed by a toddler.? I was amazed.

[But QRSS is over in the US, as is every other mode in the (currently
full) U3S firmware except WSPR and CW.? There's no point beaconing
signals with no one listening -- and they aren't.? Even the most recent
modes are essentially history.? I went from at least fifteen JT65 spots
a day to one or two, and two or three JT9 spots a day to that many in a
month.? Even WSPR is dropping off, it seems.

Hans really needs to provide some of these graveyard modes in the U3S
with the solemn offices of Christian burial and free up some firmware room for
FT8 and future modes that bear some interest.

The object in propagation beaconing is to get the absolute maximum
possible number of spots to see some patterning, and you can't do that
with dud modes.? I also think that a faster mode should provide more spots
during marginal propagation.? The current ubiquity and speed of FT8
seem to be ideal.]

>> (b) causing it to cool down.? Huh?

> You are thinking about, Han's solution of gluing a coin or other
> people have reported attaching a stick on heatsink to the crystal?

> The aim IS NOT to cool the crystal but to add a thermal? mass to it,
> to again trying to delay short term temperature changes.

Now, this actually makes sense - but there are dozens of misinformed
posts here on the subject that refer to using heatsinks on the crystals.
?Heatsinks are not there to provide "thermal mass" (interesting turn of
phrase), but to defeat thermal mass, i.e., to cool.? Thus my confusion.
Thanks for clearing that up.

> A mild steel nut would be my preference over a piece of aluminium,
> but anything that slows the temperature change works....

I think if I were convinced that this was a problem for me, I would just
wall-in the module and fill it with spray can expanding foam insulation.
I doubt one could improve on that.

>> In any case, as I understand it, if you don't drift beyond +/-4, it
>> doesn't make any difference anyway.? The default setting gives me as
>> close to zero drift as external conditions will ever permit

Once again, I'm still not understanding the quest for zero drift, at least with
WSPR.?

> a good power supply is equally required for the standard version.

[...]

> (In particular, decoupling capacitors around the Voltage Regulators,
> mounted as close as you can to the device legs.? As well as hopefully
> decoupling RF, and general hash they prevent parasitic oscillation
> that affects regulators under some load conditions.)

My voltage stability is absolutely rock-solid.? As to noise, I can't
say, but I could probably measure it with my (unused) oscilloscope if
knew how.? I'm unclear on the effects of noise on WSPR and the JT/FT modes.

Finally, here it is after yet another reformatting