Topics

#u3s QCXPTXCO install #u3s


BrianB
 

Hi,

I finally got around to modifying my U3S radios by removing the 27mhz crystal and installing Han’s QCX+TXCO.

The first one took me a couple evenings to decide how to tackle it. The second one took about 10-min.  Frequency stability now is incredible and it totally resolved a drift issue I had on one transmitter.

By clipping off a corner of the tiny board it left a U shape that fits nicely on the Si5351A board. One offcut lead through the Out of the board to the empty crystal hole holds it in place. Two additional power leads to the TXCO pads and it’s done. See pictures.

73,
BrianB
N6CVO


Dave
 

Nice.  Very nice.  That is why I ordered four TCXOs.

Dave


On Sep 1, 2020, at 15:59, BrianB <brianb@...> wrote:

Hi,

I finally got around to modifying my U3S radios by removing the 27mhz crystal and installing Han’s QCX+TXCO.

The first one took me a couple evenings to decide how to tackle it. The second one took about 10-min.  Frequency stability now is incredible and it totally resolved a drift issue I had on one transmitter.

By clipping off a corner of the tiny board it left a U shape that fits nicely on the Si5351A board. One offcut lead through the Out of the board to the empty crystal hole holds it in place. Two additional power leads to the TXCO pads and it’s done. See pictures.

73,
BrianB
N6CVO

<C9EB0B13-F8AC-45AC-ACDB-B617D1D8D1E5.jpeg>
<8F55562C-B805-4571-B036-0460FBB64D80.jpeg>


BrianB
 

Here is a link to them

https://shop.qrp-labs.com/qcxptcxo


BrianB
 

All my comments should read "QCXPTCXO" or "QCX+TCXO". 

Dyslexia is a bitch. Sorry for any confusion.

73,
BrianB
N6CVO


KEN G4APB
 

Hi Brian,
the tcxo is listed as 25Mhz, not 27Mhz, so did you just change the reference frequency in the menu list? I thought 27Mhz was chosen to get the WSPR tones correct. I would be interested to know what is the highest band and mode you have this working on please?


73 Ken G4APB


Dean Smith
 

Hi,Brian , Ken,Dave et all,
I'm just waiting for my TCXO to show up, to do exactly the same.
From what i was informed by Hans it would work ok, but don't expect much at 144mhz.
Ive used previous tcxo's by fox without any success through jitter. Hopefully these will shine!!
I love the pics BrianB... perfect!
Dean (G7EOB)


Dean Smith
 

And here is what happened using 27Mhz Fox TCXO with QRSS over various frequencies...
https://sites.google.com/site/g7eob1/radio/ultimate/si5351a-jitter


Hans Summers
 

Hi all

In the U3S I have an algorithm for choosing the Si5351A configuration parameters so that the WSPR tone spacing is very precisely 1.46Hz as it should be. 

With a 27MHz reference my algorithm works on all bands from 2200m to 2m. 

With a 25MHz reference I believe my WSPR tone spacing algorithm would not work. This is based on modeling but I haven't actually tried it on 2m. Other modes on 2m would be fine. 

73 Hans G0UPL 

On Fri, Sep 4, 2020, 20:15 Dean Smith <bardezbiker@...> wrote:
Hi,Brian , Ken,Dave et all,
I'm just waiting for my TCXO to show up, to do exactly the same.
From what i was informed by Hans it would work ok, but don't expect much at 144mhz.
Ive used previous tcxo's by fox without any success through jitter. Hopefully these will shine!!
I love the pics BrianB... perfect!
Dean (G7EOB)


Alan G4ZFQ
 

With a 27MHz reference my algorithm works on all bands from 2200m to 2m.
Hans,

I tested at 2m with a 26MHz reference, it seemed OK.
And, has anyone any idea just how close to 1.46Hz the spacing needs to be? There must be some latitude built in.
(For example some insist upon mS time accuracy for WSPR when several seconds does not seem to make any real difference.)

73 Alan G4ZFQ


Hans Summers
 

Hi Alan

I don't think I ever modeled my algorithm at 26MHz! 

Some time back I did study the WSPR decoding process and according to my understanding, as timing and tone spacing deviate from their nominal values, decode is still possible but the probability of decoding decreases. It isn't a hard cut-off point. 

The same applies to drift but here the situation is more clear. Drifr if +/-4Hz during the transmission is tolerated, no more than that. Less well known is that the drift correction models a linear drift vs time elapsed since the start of the transmission. Again, deviation from linear drift will decode but with impaired probability. Unfortunately in the real world where drift is often caused by heating, an exponential drift curve is a closer approximation than a straight line. I've often wondered, given the immense computing power available on today's machines compared to even a few years ago, if the decoder could be enhanced by adding things like different shapes of drift curve.

At some point the probability does reach zero. I couldn't say quantitatively at what deviation the decode probability would suffer say, a 50% penalty...

73 Hans G0UPL 


On Sat, Sep 5, 2020, 10:31 Alan G4ZFQ <alan4alan@...> wrote:
>
> With a 27MHz reference my algorithm works on all bands from 2200m to 2m.
>

Hans,

I tested at 2m with a 26MHz reference, it seemed OK.
And, has anyone any idea just how close to 1.46Hz the spacing needs to
be? There must be some latitude built in.
(For example some insist upon mS time accuracy for WSPR when several
seconds does not seem to make any real difference.)

73 Alan G4ZFQ




KEN G4APB
 

Hi Hans, so do you use a different algorithm on the QCX vs the U3S?, as the qcx claims WSPR up to 17m band. I just bought a couple of these tcxos for my drifting U3Ss for WSPR use up to 10m. Do you think this may not work as well as I hoped?
73 Ken G4APB


Hans Summers
 

Hi Ken

QCX uses the same WSPR algorithm as U3S. 

U3S will work fine with either 25MHz or 27MHz, up to 4m. So you will be perfectly fine on 10m. 

Using 25MHz, I'm just not convinced it will work on 2m for WSPR (other modes should he fine). I remember modeling the algorithm for 25MHz and it didn't iteratively find a solution for 2m. I'm not sure what it actually will do! Maybe it will work but just as lower precision!

73 Hans G0UPL 

On Sat, Sep 5, 2020, 18:01 KEN G4APB via groups.io <lfoofui.nbz42=ntlworld.com@groups.io> wrote:
Hi Hans, so do you use a different algorithm on the QCX vs the U3S?, as the qcx claims WSPR up to 17m band. I just bought a couple of these tcxos for my drifting U3Ss for WSPR use up to 10m. Do you think this may not work as well as I hoped?
73 Ken G4APB


KEN G4APB
 

Hi Hans,

on the QCX+, you say to remove C2 before fitting this tcxo module but it appears to be just supply decoupling. Does any similar capacitor need to be removed when modifying the Si5351A module?

73 Ken G4APB


Jim Mcilroy
 

Hi

C2 is removed so the TCXO can be a close fit onto the board basically.


On 06/09/2020 18:32, KEN G4APB via groups.io wrote:

Hi Hans,

on the QCX+, you say to remove C2 before fitting this tcxo module but it appears to be just supply decoupling. Does any similar capacitor need to be removed when modifying the Si5351A module?

73 Ken G4APB


Hans Summers
 

Hello Ken

The reason C2 is removed on the QCX+ is simply because of the way I designed the TCXO module (which was an afterthought); the Gnd and +3.3V pads of the TCXO module line up with the C2 pads on the QCX+ PCB so that the TCXO PCB can sit snug on the QCX+ PCB and power can be connected simply by soldering wires through the C2 component holes. Similarly the output is wired through the appropriate crystal pad which is connected to Si5351A pin 2. 

In order to replace the supply decoupling function of C2, I put an SMD capacitor on the TCXO module for this, which is C101. 


If modifying the Si5351A, you just need to find somewhere to connect the Gnd and +3.3V supply lines to the TCXO module. If you find it convenient to remove one of the supply decoupling capacitors on the Synth board, so as to make a convenient connection, then there is no harm in that. But neither is it NECESSARY to remove a capacitor. 

73 Hans G0UPL
http://qrp-labs.com

On Sun, Sep 6, 2020 at 8:32 PM KEN G4APB via groups.io <lfoofui.nbz42=ntlworld.com@groups.io> wrote:

Hi Hans,

on the QCX+, you say to remove C2 before fitting this tcxo module but it appears to be just supply decoupling. Does any similar capacitor need to be removed when modifying the Si5351A module?

73 Ken G4APB


Dean Smith
 

just been playing with the 25mhz tcxo for q"C"x+ in a U3s.
QRSS is good to 2 meters! a little drift to start due to other components, BS170's ect, then it settles down nicely!
Much better than the fox type tcxo i tried in the past.
Something strange happens to the height of FSK at around 6M it seems to double in height?? maybe me...
Been playing WSPR on 30m with it, and all seems ok. No drift and consistant frequency reports.
i wonder if this TCXO will fit on the Pads on the later revision Si5351a board? would look a lot tidier.
Be nice if you could buy the TCXO on it's own and solder up, or ready tacked to the board?
Certainly a lot less scary than the OXCO option, for those with less experience.
impressed with the results.
Got 10m decoding with my own system on wspr. just wonder how far we can go :)
Dean (G7EOB)


Allan Nelsson
 

Dean I don't know the details but you may take a look at the RFzero https://www.rfzero.net/ its stability is so good it can be used for WSPR on 23 cm.

73 Allan OZ5XN


KEN G4APB
 

Hi Brian,
thanks to your pictures, I modified my multiband U3S to take this QRPLabs 25MHz TCXO module, however, I connected GND to pin 1 and Vdd to pin 18 on the Si module. This worked very well and it was very satisfying to see a line of 0Hz drift results on all bands from 160m to 10m, something I never achieved before even with the QRPLabs OCXO.
Annoyingly, the latest version of wsjt-x v2.2.2 has a ‘sub process error’ bug when bandhopping. Has anyone found a fix for this? I had to revert to v2.2.1 to test my U3S WSPR decoding.

73 Ken G4APB


Alan G4ZFQ
 

error’ bug when bandhopping. Has anyone found a fix for this? I had to revert to v2.2.1 to test my U3S WSPR decoding.
Ken,

As I said on WSPRnet that IS the (temporary for how long?) fix.

73 Alan G4ZFQ