Progress towards using RTLSDR dongles for interferometry

1544 views
Skip to first unread message

Marcus D. Leech

unread,
Sep 2, 2013, 8:51:34 PM9/2/13
to sara...@googlegroups.com
Made some some progress today towards my goal of being able to use these
cheap dongles for interferometry. I acquired a number of these
Meiden 14.4MHz TCXO oscillators--they're available cheaply on eBay,
and are good to about 2.5PPM. Which is not bad. I constructed a doubler
circuit, and then filtered the output with the two crystals that I
removed from two of these TV28T dongle units (which are 28.8MHz). Amplified
with an ERA-3, and I can drive two dongles from the resulting common
clock without issue. 28.8MHz is an odd-ball frequency, and fortuitously,
there are a lot of these Meiden 14.4MHz TCXOs floating around, so it
seemed like the obvious way to go...

So, when I go to visit DRAO in 10 days or so, one of the plans is to try
out some interferometry with these dongles, and now we have a good method
for producing the necessary common master clock.

http://www.sbrac.org/files/28Mhz_osc.jpg

and

http://www.sbrac.org/files/28_mhz_doubler.pdf

That's before I put an ERA-3 on it as a buffer-amplifier.


Jim Sky

unread,
Sep 3, 2013, 11:14:27 PM9/3/13
to sara...@googlegroups.com
Brilliant to use the old xtals as filters!

Jim S

Marcus D. Leech

unread,
Sep 3, 2013, 11:23:19 PM9/3/13
to sara...@googlegroups.com
On 09/03/2013 11:14 PM, Jim Sky wrote:
Brilliant to use the old xtals as filters!

Jim S

Thanks.

Yeah, they're kinda-crappy as crystals go (about 100PPM or worse) for making master clocks out of.  But just fine as filters for a better clock circuit :)

A factor of 40 improvement in accuracy  means you could take reasonably-good doppler measurements with these cheap dongles. I picked up
  50 of these Meiden oscillators for about $0.50 apiece, so adding decent clocking doesn't raise "project costs" too much.

WIth the improvement in accuracy, there's a corresponding improvement in phase-noise as well, although I haven't measured it, these TCXOs tend
  to have excellent phase noise, compared to a "naked" crappy crystal.

I also picked up some surplus 10MHz OCXOs on eBay, which will make excellent reference clocks for a couple of my USRPs for pulsar work.


--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Den Koawl

unread,
Sep 4, 2013, 5:08:25 AM9/4/13
to sara...@googlegroups.com
Very Good Marcus !  Well done !
 
Den

Steve Olney - VK2XV

unread,
Oct 27, 2013, 3:08:52 PM10/27/13
to sara...@googlegroups.com
G'day Marcus,

In a past life (when I was working for a living in valve days...) I had some experience with injection locking of crystal oscillators.  Injection locking can actually be a design problem where interfering signals related harmonically to a crystal oscillator frequency can "pull" the oscillator.

Given this I wonder if it might be simpler to just couple some of the 14.4 MHz TCXO oscillator energy to the dongle 28.8 MHz crystal in-situ ?

I might get one of those TCXOs and give it a try.

Cheers

Steve, VK2XV

Marcus D. Leech

unread,
Oct 27, 2013, 5:14:39 PM10/27/13
to sara...@googlegroups.com
> G'day Marcus,
>
> In a past life (when I was working for a living in valve days...) I
> had some experience with injection locking of crystal oscillators.
> Injection locking can actually be a design problem where interfering
> signals related harmonically to a crystal oscillator frequency can
> "pull" the oscillator.
>
> Given this I wonder if it might be simpler to just couple some of the
> 14.4 MHz TCXO oscillator energy to the dongle 28.8 MHz crystal in-situ ?
>
> I might get one of those TCXOs and give it a try.
>
> Cheers
>
> Steve, VK2XV
The problem appears to be that despite the two dongles getting a common
clock, the resulting baseband signals aren't always coherent. Tests
continue.
Some people have some success, others, not. It's puzzling.


--
Marcus Leech
Principal Investigator
Shirleys Bay Radio Astronomy Consortium
http://www.sbrac.org

Steve Olney - VK2XV

unread,
Oct 27, 2013, 5:32:13 PM10/27/13
to sara...@googlegroups.com
Marcus,

Sorry - I didn't make myself clear - I was merely talking about the method of of getting a better 28.8 MHz clock.   That is, instead of going to the trouble of building a doubler, etc, just feeding a small amount (via small caps ?) of 14.4 MHz energy directly from the TCXO into the 28.8 MHz crystal in-situ. If this works then this would be a simpler solution usable by a wider audience as use as an SDR.

The narrower issue of coherence for interferometry is a separate matter to investigate as you indicate.

Steve

Marcus D. Leech

unread,
Oct 27, 2013, 5:52:56 PM10/27/13
to sara...@googlegroups.com
> Marcus,
>
> Sorry - I didn't make myself clear - I was merely talking about the
> method of of getting a better 28.8 MHz clock. That is, instead of
> going to the trouble of building a doubler, etc, just feeding a small
> amount (via small caps ?) of 14.4 MHz energy directly from the TCXO
> into the 28.8 MHz crystal in-situ. If this works then this would be a
> simpler solution usable by a wider audience as use as an SDR.
>
> The narrower issue of coherence for interferometry is a separate
> matter to investigate as you indicate.
>
> Steve
>
Yup. Never tried it. Somebody could, and report on the results. For
my own work, I'm happy with what I have, for sure.

Steve Olney - VK2XV

unread,
Oct 27, 2013, 6:03:21 PM10/27/13
to sara...@googlegroups.com
I have ordered some 14.4 MHz TCXOs.  I'll report back in a few weeks successful or not.

Raydel Abreu (CM2ESP)

unread,
Oct 28, 2013, 9:02:17 AM10/28/13
to sara-list
Marcus,

It may seems trivial, but did you tried with a "Delay" Block. I have two dongles which have not been connected to a common source. But I have seen that one dongles usually begins to send samples about two seconds later than the other.

I add a delay in other to attempt synchronize the internals delays, of course as there is no common source it is still without coherence.

Perhaps, using an slider to adjust the delay until archive coherence could help?

One more question unrelated, It doesn't matter which computer I use, both the Waterfall and the FFT sinks runs very very slow when connected to a RTL-SDR source in all my experiments. However with your programs (simple-ra) it seems to run normal (unless high integration values are used of course....)

Raydel


2013/10/27 Marcus D. Leech <patchv...@gmail.com>

Marcus Leech

unread,
Oct 28, 2013, 2:41:13 PM10/28/13
to sara...@googlegroups.com
What sample rate are you using?

Also, turn-down the frame-rate on your FFTs -- the default is 15FPS, which is usually too high.



Marcus Leech

unread,
Oct 28, 2013, 2:42:15 PM10/28/13
to sara...@googlegroups.com
Yes, I've used a delay block.  At higher sample rates, the startup delay is smaller, on the order of 400ms.

I think what I might end up doing is writing a correlation-search function to establish the precise delay required to optimize correlation.



On Mon, Oct 28, 2013 at 9:02 AM, Raydel Abreu (CM2ESP) <cm2...@gmail.com> wrote:

Raydel Abreu (CM2ESP)

unread,
Oct 28, 2013, 3:16:17 PM10/28/13
to sara-list
Oh I see...

I didn't pay attention to the startup delay with different sample rates. I am looking forward to hear from your next attempts.

Regarding the other question, of the slowness of computer with my experiments it is quite estrange because GNU-Radio behaves slower when using 0.250MSPS instead of 1.0MSPS. I believe it should be faster with less sample rate at the USB dongle but it doesn't. About the FFT sink now I see I was using the wrong update cycle. So it is FPS (Frames Per Second), a small value less updated, I wrongly though it was update cycle in miliSeconds, so I increased the value. How silly!!!

Now, back to RadioAstronomy topic, my dish for neutral hydrogen observation is almost ready, I'm looking forward to use more frequently your Simple-RA tool!!!

Thank you for the help!!!

Raydel


2013/10/28 Marcus Leech <patchv...@gmail.com>

Marcus D. Leech

unread,
Oct 28, 2013, 6:03:01 PM10/28/13
to sara...@googlegroups.com
On 10/28/2013 03:16 PM, Raydel Abreu (CM2ESP) wrote:
> Oh I see...
>
> I didn't pay attention to the startup delay with different sample
> rates. I am looking forward to hear from your next attempts.
>
> Regarding the other question, of the slowness of computer with my
> experiments it is quite estrange because GNU-Radio behaves slower when
> using 0.250MSPS instead of 1.0MSPS. I believe it should be faster with
> less sample rate at the USB dongle but it doesn't. About the FFT sink
> now I see I was using the wrong update cycle. So it is FPS (Frames Per
> Second), a small value less updated, I wrongly though it was update
> cycle in miliSeconds, so I increased the value. How silly!!!
>
> Now, back to RadioAstronomy topic, my dish for neutral hydrogen
> observation is almost ready, I'm looking forward to use more
> frequently your Simple-RA tool!!!
>
> Thank you for the help!!!
>
> Raydel
>
At 250ksps, the default gr-osmosdr buffer size is too large, so data
doesn't get delivered continuously to the flow-graph. I can't remember
the name
of the parameter, but if you shrink down to about 25% of the default
size, it works out well.



Raydel Abreu (CM2ESP)

unread,
Oct 29, 2013, 9:04:19 AM10/29/13
to sara-list
I will do that. Thanks Marcus!!!


2013/10/28 Marcus D. Leech <patchv...@gmail.com>
--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to

For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
--- You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+unsubscribe@googlegroups.com.

Dan Peabody

unread,
Jul 28, 2014, 4:38:48 AM7/28/14
to sara...@googlegroups.com
Hi Steve,

I was wondering if you had had any success in injection locking your 28.8MHz crystal with the 14.4MHz TCXO.

I was going to attempt this myself but then i found this on eBay

http://www.ebay.co.uk/itm/171354014902

But it would still be interesting to know if the 14.4MHz clock modules can be used in this manner.

Kind regards

Dan

Steve Olney

unread,
Jul 28, 2014, 8:20:40 AM7/28/14
to sara...@googlegroups.com
G'day Dan,

I actually used a 9.6MHz TCXO. because a tripler is easier courtesy of
the higher 3rd harmonic content of a square wave.

I am ashamed to say I didn't document it, so I have no details to give.

Steve

Dan Peabody

unread,
Jul 28, 2014, 9:55:32 AM7/28/14
to sara...@googlegroups.com

Hi Steve,

Thanks for the quick response. I'm not really to fussed with the details - i just wanted to see if this approach works and is worth experimenting with.

so did you just inject the 9.6MHz into one side of the 28.8MHz crystal and rely on the 3rd harmonic to give you the injection lock?

Cheers

Dan

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-unsubscribe@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
--- You received this message because you are subscribed to a topic in the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/sara-list/gg4WjwDB7Pw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to sara-list+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Marcus D. Leech

unread,
Jul 28, 2014, 10:00:02 AM7/28/14
to sara...@googlegroups.com
On 07/28/2014 09:55 AM, Dan Peabody wrote:
>
> Hi Steve,
>
> Thanks for the quick response. I'm not really to fussed with the
> details - i just wanted to see if this approach works and is worth
> experimenting with.
>
> so did you just inject the 9.6MHz into one side of the 28.8MHz crystal
> and rely on the 3rd harmonic to give you the injection lock?
>
> Cheers
>
> Dan
>
I'm currently using one of those 28.8MHz TCXOs, but still, no luck in
getting meaningful coherence except at harmonics of the master clock
frequency (I suspect because the synthesizer goes into integer-N mode
when that happens).

There's a register in the R820T synthesizer called "DITHER" that some
have experimented with turning OFF, and this seems to help with
coherence in situations where the desired frequency isn't an integer
multiple of the reference frequency.

Steve Olney

unread,
Jul 28, 2014, 9:08:31 PM7/28/14
to sara...@googlegroups.com
Dan,

What I did was remove the crystal completely.  I constructed a separate 9.6MHz oscillator/tripler chain and fed the 28.8Mhz output of that into one of the crystal input pins (I cannot remember which one).

I would not recommend relying on injection locking (even though I may have suggested it in the past) as this would introduce another uncertainly (phase jitter/jumps) into the mix.  Already it has been observed by others that even a clean 28.8MHz signal has not been successful for coherent operation.

Cheers

Steve
Reply all
Reply to author
Forward
0 new messages