Mobo noise performance

[Alex Lee]

Hi all,

I feel compelled to say something about Mobo noise performance that
some have commented negatively on.

It was only with the extremely low noise floor soundcard such as the
sdr-widget being available that we could see the "spurs" in the
spectrum. With sdr-widget, you need to reduce the opamp gain of the
softrock 6.3 by 15-20dB to optimize the dynamic range.

Also, the i2c noise when rapidly tuning around is minimized with a
simple hardware mod to the i2c line to the SR (SMT chokes and caps).

We have made the mobo based hardware performance publicly accessible
with the ghpsdr3-alex based servers that anyone can access with
QtRadio that runs on Linux and Windows.

I have yet to see non-Mobo setups of Softrocks showing equivalent
noise floor performance. The ones accessible with the ghpsrd3-Alex
servers do not show better noise performance (in fact they look
worse).

Most softrock non-mobo users have soundcards that have noise floor
that is very high (by sdr-widget standards) and they just cannot see
the spurs that are virtues in the noise floor.

I am not saying that the Mobo cannot be improved. It can. It is a
open source design (as all my other projects such as sdr-widget) and
anyone is welcome to come up with mobo 5 or SR7 etc.

But in the meantime, show me another full 160-10M rxtx Softrock that
perform better than SR63/Mobo/sdr-widget ;-)

Alex

[W9...@aol.com]

I agree Alex.

 

 To me the tuning noise is minimal. I have not yet bothered to modify my boards. I have a lot of hours on my softrock and few on my MOBO UHFSDR Widget combo as well. It is a affordable kit with a lot of options. I have not had this much fun since the Heathkit days! 

 

If anything were going to be changed I suggest a MOBO that would be more universal with other transceivers.  But then it can modified to be universal anyway. I have posted the info in earlier post. 

 

Larry W9AMR 

[Mike Collins]

On 1/2/2012 9:27 PM, Alex Lee wrote:
> Hi all,
>
> I feel compelled to say something about Mobo noise performance that
> some have commented negatively on.
>
>

> I am not saying that the Mobo cannot be improved. It can. It is a
> open source design (as all my other projects such as sdr-widget) and
> anyone is welcome to come up with mobo 5 or SR7 etc.
>
>
>

Hey Alex,

Happy new year to you and the others on the group !!

I'm not sure that most that made comments understand how hard it is to
get near 100dB isolation on a 2 layer board !!!

With the MOBO having RF, digital, analog and power supply circuits all
closely coupled it has amazing performance to me.

I sure remember the 100's of emails on layout comments during the MOBO
layout revisions. Don't remember seeing anyone
else comment during those 1st five MOBO layouts other than Alex, Loftur
and myself.

Likely would need 4-8 layer board at min to even think you might make
improvements. Also need pretty high end tools to analyze coupling and
radiated effects, or plan on doing several layout, build, test cycles.

I have also seen a lot of comments on noise/lockup problems that are due
to the way the MOBO has been integrated into the radio system. Plan on
using many ferrite cores to control the return currents or you likely
will have trouble. I have run my MOBO's and amps at 1.5KW many times
without problem....of course it took some work to get there. Every
interface you should consider the return currents and adding addition
common mode/RF rejection as needed.

Have fun and 73,

Mike Collins KF4BQ

[applewiz2000]

OK I'll bite...

Two people local to me have scrapped their MoBo systems, and AFAIK
left this group. The I2C noise and general level of spurs makes them
almost useless. Retrospectively it was a bad decision to make the
controller part of the main PCB. But even with the controller
outboard, there are still big problems with spurs. This applies to all
systems I have seen shared on QtRadio.

I made a comment during the layout that the board shape would make
layout difficult, and so it did. But I agree that the MoBo itself has
a reasonable performance for a 2 sided design in terms of adding non-
I2C spurious. The SR V6.3 on it's own is not really much better. Tony
Parks had several iterations of the Ensemble boards.

Going to 4 layers of course makes things far easier and more
predictable. But I predict it will not happen unless the "horses stop
galloping off in different directions."

As for overall sensitivity, I disagree with all of the above.
Connecting my SR6.3 > widget input at 29MHz makes only ~3dB noise
floor rise. For the much vaunted resistor changes, reducing by 20dB is
useless. Going down by 6dB is about as far as it will go before
sensitivity at 29MHz is impaired seriously.

Having a receiver with 15dB+ noise figure will not wash above 20MHz
unless a serious antenna is attached. Rule #1 of receiver design.

~Rob

[Graeme ZL2APV]

Yes I am good for 2 pennys worth too. Not being an original designer but an early beta tester I feel I can be impartial. I agree the design is not perfect and there are some fairly clunky areas in the layout particularly around the PA stage where things are jammed in there and some minor board and software modifications have been necessary but the fact remains that it works well although a few spurs and artifacts pop up. With care, shielding and lead lengths plus ferrite common mode chokes e.g. my USB cable has 6 turns through a toroid etc. will reduce these spurs to a minimal degree.

The Mobo/sr63/sdrWidget is a home brew project and I have done many home brew rigs over the last 55 years. When I compare the Mobo set to the best one I built in the past the Mobo is streets ahead. I remember paying the relative same price for my 9 MHz sideband filter as I did for the whole Mobo kit and spent the best part of a year building it for not much less total cost than a current top of the range transceiver. It was only 80 M to 15 M and still had spots in some places in the ham bands. It did not do all modes, could never have decoded dream and my CW filter was an audio filter. I built the Mobo in 2 days, the sr63 in one and the sdrWidget was already assembled. Took me 2 weeks to get the software going as I was an absolute Linux newbie although I had it running on Windows with Rocky in no time, CW only.

Would I go the Mobo way again? Yes I would as it is the best bang for the buck I have come across and certainly achieves all band all mode operation at the lowest cost I can think of. Of course I am tempted to rebuild around the basic design and one day probably will as that is my nature and has been all my life. I am convinced that a more modular construction in individual shielded enclosures with proper decoupling of non signal lines (power etc.) is the key to spur reduction without the complexity of multilayer boards. The Mobo software could remain a standard for any enhanced design.

The Mobo is a very clever design and would have been even better if it had not been constrained in layout by the requirement to mate to the sr63 board. It was interesting to operate the YU1LM DR2A++ receiver with the Mobo and observe the drop in spurs and improvement in dynamic range. I attributed the physical separation and shielding to this. Getting QtRadio operating well is my first priority but after that I will be trying physically separating the mobo and sr63 into their own shielded enclosures and see what happens to the current not very troublesome spurs as a result.

73 Graeme ZL2APV

[The Silver Fox]

Glad to have the remarks to counter the negative ones.  According to just my statistics there are quite a few boards out there and if they were many flaws or needs for improvements, I would think they would have surfaced by now.  Anyway, I am always happy to hear that the kits are beneficial to the RF community.

73’s

Alan – W6ARH

[John Williams]

An experiment I have yet to see performed is to move as much control
logic over to the widget and leave that portion of the existing mobo
empty and/or grounded. That would be interesting...

John - ke5ssh (who is on the fence on my purchase so pleased to see some
positive comments!) I have my 6.3ng built and I have a widget so the
mobo is next....

--

John Williams

KE5SSH - ham since 2007
WQKA523 - GMRS for family use on the farm

[applewiz2000]

Graeme,

I tried covering nearly the whole MoBo and then the SR V6.3, and
various parts/combinations in copper foil. It made no noticable
difference. There are also the bizarre "howling wind" noise humps. I
am unable to prove, but possibly they are caused by jitter through the
74AC74 circuit.

As for being not a problem... compare the spur performance of the
HPSDR Hermes board and weep! But Hermes is 2x + price and release date
has still not been set.

Bang for buck: compared to other SDR based kit rigs the MoBo V4.3 has
more features than any, and is cheaper than some. The Flex 3000
represents better value, but on economic terms only. Handing money
over a counter of course does not teach anyone anything.

----
Alan,

I think the decision to sell the same board again without any revision
was wrong. If I put out such a badly finished layout at work as a
second iteration the boss would be banging the table. The freeware
KiCAD PCB package didn't help.

The final analysis has to be how many people actively use it as their
primary HF radio. Maybe not many...

~Rob

[Graeme ZL2APV]

Hi Rob

Was very interested in your response. I expected a better result from the shielding than you got and felt that my spurs and noise humps were the result of the mobo being close to the sr63 board so you have burst my bubble. I based my expectation on an experiment I did with what was basically a YU1LM DR2++ design which I had modified to use paralleled 4066's to lower the through impedance and a DDS-60 vfo. I connected with short coax's for the signals and ran it through the Mobo with reduced spurs and no sign of noise humps. My DDS-60 only let me tune to 20 M band so I did not get to use the 10 M band where most of the problems show.

What was wrong with Kicad? I have used it for other projects and it seems to work OK for me but then having been retired a few years now I live in a technical dessert and unless tools and instruments are not dirt cheap I don't have them. The same applies to software. You possibly are comparing it to an industrial strength package which I have never had access to.

Your 74ac74 comments may be on the mark and could be some of the reason for the YU1LM improvements. When I get time to look at the hardware side again I will explore that further. To be realistic I am never going to be able to afford to experiment with a Hermes or other equivalent design and even softrocks stretch the budget a bit but I am determined to keep experimenting as long as I am physically able. I have had a lot of benefit from your posts as you have always been constructive and honest with you opinions and sure look forward to exchanging information in the future.

I am using my Mobo as my primary rig on 80, 40, 20, 15 & 10 metres. I do have a TS-940 and have about 1 qso on that to about 500 on the mobo. Yes a Hermes would be lovely but I do not weep for the unatainable but put my energy into that which I can achieve (I hope)

73 Graeme

[Sid Boyce]

I never expected the Mobo to deliver a bespoke SDR performer.

It does what was intended and that is to provide multi-band solution for
the SR v6.3 as opposed to the armstrong method of band changing, also to
provide reasonable power output as a QRP rig and for driving a higher
power PA.

The SR v6.3 provided an affordable platform that worked very well. The
other add-ons extended its capability but together they don't compete on
performance with Flex or Hermes nor were they intended to.

You can't compare even Flex with Hermes for performance or for that
matter with the HiQSDR.
With Hermes they are still stumph about the likely price but it's a
forerunner of future SDR development.

It sports a 8 layer PCB as opposed to 4 layers for the HiQSDR.
Hermes is definitely a leap ahead in hamradio SDR gear - higher
performance and hence expensive components had to be acquired. It will
still form a platform for future developments.

It's a journey not a race to the tape - it's about the fun derived from
these projects especially at a time when home construction had almost
been limited to very simple circuits that few ever built and used.

Looking back, today I wouldn't attempt to build some of the transceivers
I built decades ago and projects such as QRP2000 were just a good
monthly read.
73 ... Sid.

--
Sid Boyce ... Hamradio License G3VBV, Licensed Private Pilot,
Emeritus IBM/Amdahl Mainframes and Sun/Fujitsu Servers Tech Support
Senior Staff Specialist, Cricket Coach
Microsoft Windows Free Zone - Linux used for all Computing Tasks

[W9...@aol.com]

Hi All

 

My 2 cents: You can't buy a entry level 5 watt 10-160m kit in the same price range. Yeah there are a few quirks. It is not a Flex and it is price accordingly. I have mine paired with widget and homebrew LPF and PA I run it in to a Henry 2k4 and have ran it in to my homebrew 4-1000a amp and run full legal limit (USA limit :-)  and work the world. I run it as my main rig. I get a kick out of knowing I built it USB Jack to antenna jack.

 

As changes are made I will make them. I started with a mobo 3.6 and original 6.3 and a crappy soundcard. I look forward to making updates and bulding new boards. I have a USB2SDR board that I have lost in the shuffle I am looking forward to trying it but until it support the MOBO firmware commands it won't see as much use as the Widget.

 

There are a lot of different kit options out there to try. If the MOBO issues worry you don't by it. pick a different kit. I think there are more happy users then unhappy. 

 

Larry W9AMR    

[Oliver Goldenstein]

Hi All !

> I get a kick out of knowing I built it USB Jack to antenna jack.

That is true :-) the same here

I use the Mobo as a main HF Rig, actually from 40-10 Meters with 5
Watts on a poor balcony antenna. And i have fun with the rig and i am
happy when i make a dx contact and get the qsl later. The first qsl's
are already in from the US. And as Larry said everything built up on
my own and with all of your help of course.

During soldering i did a lot of JT65 contacts on 20 and 10-Meter.

Main problem here is more the noise from lcd televisons in the house.

I do only have diy rig's here. Nothing more.

One year ago i started with the sdr-widget and nothing else. Now i
have the Mobo/Widget und just since a week the uhfsdr/widget.

--
Oliver, DL6KBG, JO61UB
Blog: http://dl6kbg.blogspot.com

[John Williams]

Rob.

If you run a 6.3 (ng) with a 10M plug in BPF/PA and I2C from a wdget vs.
the mobo would it yield a substantive difference in noise and spurs?

John

--

[applewiz2000]

Hi John,

I hope to try that some time. There is a 10m filter board from the old
SR Mk.2 days here, and the widget is on 3v3 logic like the Si570, so
they should connect without any sparks flying.

As for using the widget as an outboard controller, that did make some
improvement. But I can't accurately quantify how much. The biggest
improvement was slugging down the I2C lines with beads and 100pF caps.
Small tuning steps now hardly cause any disturbance. Big tuning steps,
e.g. band change cause the Si570 to produce a big glitch, that throws
the DC balance out. This is the real "Achilles heel" of the Si570.

If I didn't get a reasonable performance from my system I wouldn't be
using it. But almost a year has passed now without major hardware
improvements. If there are to be any, I have to make them off my own
back. Which is quite fair enough.

~Rob