DSP Audio Filters
Date: Wed, 24 May 95 08:40:37 EDT
From:  
Michael Owen W9IP
To: 
CC: 
Subject: DSP digest of answers

Friends-
This is a digest of the responses received to my DSP question. Interesting
reading!   - W9IP

Mike, W9IP wrote:
: I wonder who among us has had an opportunity to A-B compare the
: various DSP audio filters (TimeWave, W9GR, MFJ, what-was-the-other-one?)
: I'm interested in both super-weak EME performance as well as contest
: performance/QRM.  Is any brand deficient in dynamic range, or have a
: hiss or crummy audio perfomance?

Yes - here's the short version of our article in RadCom (RSGB) for
September 1994.

**************

G3HCT and G3SEK were able to compare five DSP audio filters, from
the separate viewpoints of the HF DXer and of the VHF/UHF DXer. Although
some our criteria were quite different, we reached very similar
conclusions in independent tests. We also compared the DSP filters with
one of the best analog audio filters, the Datong FL3. Our review
concentrated on real-life use rather than paper specs or bench tests.

DSP filters can do remarkable things with HF CW and SSB, but you can read
all about that in the magazines. These notes are about how they work with
weak CW.

W9GR: CW note always clear and clean, with no noticeable ringing. On very
weak CW there is some high-frequency hiss due to the dithering technique
used to reduce 8-bit quantization noise. Although the 30Hz bandwidth is
suggested as "ideal for EME and weak signal work", it is only available at
800Hz; the one bandwidth available at 400Hz center frequency is 100Hz. The
12-way rotary switch is only marked in modes and we found the settings
difficult to remember. No bypass to hear your CW sidetone on TX.

NIR-10: On CW the NIR-10 has 600 and 250Hz fixed bandwidths, fully tunable
from 300-3200Hz center frequencies. At 250Hz BW the CW note was not as
clean as the W9GR, and with very weak signals there was a noticeable
'ticking' noise like ignition interference. We strongly disliked the mode
controls, two 3-way toggle switches which are misleadingly marked.
Sidetone bypass on TX.

DSP-9 / 9+: Audio quality sounded more 'reconstituted' than the others,
and the CW note a little more 'ringy', though still much better than
narrow xtal filters. [Later: the new Version 3 firmware is said to have
improved the audio quality, but we haven't tested it thoroughly yet.]
On very weak CW the random-noise suppression function
helped to reduce the filter's own ringing. We liked the simple push-button
controls, which made it very quick and easy to find the best mode
settings. CW bandwidths are 500/200/100Hz at either of two center
frequencies, selectable by push-button: in the DSP-9+ you can set jumpers
for any two from 400, 500, 600 or 800Hz; in the DSP-9 the only possible
combinations are (600+750Hz) or (400+500Hz). The 9+ has sidetone bypass
but the DSP-9 doesn't.

The DSP filters performed much better than I expected on very weak CW,
although we were unable to test them on aurora or other modes in which
phase coherence is partly lost - for example 10GHz EME. They all worked
well in combination with the 500Hz IF CW filter, and always reduced the
ringing. In narrow audio bandwidths, using the 500Hz IF filter reduced the
background noise compared with an SSB IF bandwidth, but an additional
250Hz IF filter didnt make any further improvement. The flat-topped
passbands are useful at UHF and above, where signals sometimes drift a
little. Making switched A/B comparisons with a Datong FL3 (at 100Hz BW,
400Hz center), both the W9GR and the DSP-9+ were almost as good as the
FL3, which rang more than the DSP filters but still pulled the signals
through. At least, that's how it sounded to my ears - yours might be
different!

I have not been an audio-filter fan in the past, but found that I could
copy EME signals with the DSP-9+ that would have been very difficult
without it. All the other filters have gone back to the distributors, but
somehow the DSP-9+ stayed behind. It might even help with some of the weak
SSB signals on the 20m net...

**************

UPDATE: G3HCT and I will shortly be reviewing two tunable filters, the
Timewave DSP-59+ and the MFJ. We'll be comparing these with the DSP-9+,
our previous "best buy", upgraded with the new Version 3 firmware.

As before, I'll try to produce a short summary with any extra details
for VHFDX users.

HELP NEEDED, PLEASE!

1. We'd still like to hear from anyone who has had experience
   using DSP filters with auroral CW and SSB, or 10GHz EME.

2. Has anyone compared DSP filters with the JM1MCF-type "noise
   smoothing" filters using bucket-brigade audio delay lines?

--
73 from Ian G3SEK            Editor, 'The VHF/UHF DX Book'
                            'In Practice' columnist for RadCom (RSGB)
Professionally:
IFW Technical Services       Clear technical English - anywhere.
---------------------------------------------------------------------

Hi Michael,

I have exactely the same interest as you, contest and EME. I have just
aquired a MFJ DSP, but have no comparison and only little experience
yet. Its 30 Hz bandwith is effective pulling weak cw out of the noise,
but naturally it is very critical. Takes too much time during
contests. Its noise reduction does really reduce noise, but I am not
yet sure if it improves readability. However it is more pleasent to
listen to. Only at 144 MHz I have RF running into its audio, but
muting the unit should take care of that. The MJF has an AGC, but no
indication if the input audio level is correct. I think I will wire it
to the line (phone patch/modem) output of my radio (IC-736).

If you get any response on DSP's I would like to know.

What VHF-net did you crosspost to? I am on one at stanford and one at
icon-fonts-xerox. Only the last sometimes sends a few mails. Perhaps
they work together, but I have only seen your posting on moon-net.

73, Palle, OZ1RH (OZ9EDR contest team)  Net: i0987@dc.dk

PS. OZ9EDR has 20 kw erp on 50 MHz and 50 kw erp on 144 MHZ, still a
little short of eme performance, but we work on it. I calculate on a
17 dBd monster for 50 MHz...


_______________________________ Svaradskiller
__________________________________
Emne: DSP filters
Forfatter:mowe@slumus.stlawu.edu pe internet
Dato:    22-05-1995 17.55


Mike

I have had no experience with the real good add DSPs but I found the Radio
Shack one cheap as it is to be very good on c.w. - at least as good as the
500 Hz filter in the FT736 in the absence of strong nearby QRM. Better yet
, tho I have no converter to try it, my new FT1000 has a DSP that sounds
even better on c.w.  Does not appear to overlaod even in the presence of
strong signals (but am also using the 500 Hz filter in that unit.

Will be interested to see the results of your survey.  Particularly
interested in DSP ability to reduce noise - the Radio Shck one does a
very poor job of that - the Yaesu does not purport to reduce noise.

73 and hope to hear you in the June contest

Gene  W3ZZ
=======================================================WED MAY 24, 1995
08.30.52


Mike:
  I did an A/B compare between the DSP-9+ (Timewave) and the  NIR-10 for
about two weeks.  My *only* interest was noise reduction.  The DSP-9+ had
a more agressive noise reduction setting (that could also be changed -
jumpers) than the NIR-10 (fixed noise reduction).  I told by the JPS folks
(at Dayton) that their *new* NIR-12 has better noise reduction than the
-10 did.......So I like the 9+ (although their hardware is pretty poor -
a mini plug for a phone jack?!) over the NIR-10.
Those are the only two that I've tried - but will have an opportunity to
"road test" MFJ's (Mississippi's Finest Junk) in the near future.
Good luck.               73      Dick K3MQH/V26T
---------------------------------------------------------------------
Hi Mike,


G3SEK has compared a few DSP filters and written a few words about his
experiences in the K2UYH EME News Letter. That was a few month ago, but
since I can't find it so quickly to confirm, I didn't follow-up instead
of reply..

Take care, Klaus.
---------------------------------------------------------------------

Hi Mike,
My experience has been only with the DSP-59+ Timewave and the W9GR
original ala QST of a few years ago. And, even that was actually version 2
. The original article used proms, where the version I got had memory on
the dsp chip and was programmed to handle ten combinations of functions.

The W9GR had some audio problems, mostly quantizing noise appearing on the
output. Several fixes were suggested by several people. I was able to
reduce, but not eliminate, the noise. Adding a SCAF after the W9GR took
care of that. I used the W9GR's 30Hz CW filter for EME and it was an
improvement on any other combination I had then. On SSB, noise reduction
was effective, tho most apparent on OSCAR 13. My buddy, WB4MJE, found it
beneficial mobiling on 20 meters. The notch worked well, and may be faster
than my DSP59+. I guess there was some distortion introduced on SSB when
both noise & notch filtering. But, if the band required their services,
the resultant signal-to-noise was usually more useful than without.

As far as the DSP59+ is concerned, there is just no comparison.I primarily
use it at 25Hz on EME, and after A/B'ing against the W9GR, took the W9GR
out of line. I decided the 59+ was on the order of at least 3 db better
most of the time, listening to EME and weak tropo signals.
(I wish I'd had AF9Y's current FFTDSP version, which has a calibrated
signal indicator.)

For my money, both the filters need more audio output.I think they're both
rated at 1 watt, but my receiver's 2 watts makes tons more noise. The 59+
also seems, near its highest output level, to show some noise coupling or
maybe feedback developing. As I seldom run it at that level, it's seldom a
problem. The 59+ also has AGC, which the W9GR does not.

I see an add in the latest QST for a new W9GR filter, but don't know
anything about it. Other than I recall it to be a lot less expensive that
the 59+.  For the moment, I stick with the 59+.

73, Dave
=======================================================WED MAY 24, 1995
08.31.06

Mike:
Several months ago, the Satellite Operator had a comparasion between the
TimeWave, and the JPS NIR-10.  I wrote the part on the NIR-10.  You might
read these remark and make a decision.  I am really impressed with the
NIR-10's performance. I recently loaned mine to another operator in Seoul,
HL9RFT as he is in the RF nightmare of Korea.  He tells me he is highly
impressed with it, and plans on getting one of his own. I see JPS has come
out with "12 model" with is a dual DSP unit. Don't know much about it, but
you might contact them for the specs.


73, Bob HL9WH
=======================================================WED MAY 24, 1995
08.31.09

I have the NIR-10, which was probably the other one you referred to. They
have the NIR-12 out now too.  I haven't used it extensively, but I have
used it on 6M at W2SZ/1.  I usually keep it bypassed until I need it. The
couple hundred ms delay makes quick replies awkward.  The automatic notch
works great, but you don't have too many tuner-uppers on 6. When the rates
really drop off in the middle of the night, I tend to put it on to cut
down on the background noise. I am sure that I don't use it as effectively
as possible, but it does help at times.

Ed
WA2SCA
=======================================================WED MAY 24, 1995
08.31.13

In my experience on weak signal vhf (25 years), I have found that excess
selectivity is over rated as an aid to reception.  I know that the charts
and equations will say that narrower BW improves SNR and it does. However,
the trained  ear/brain combination is an excellent narrow band tracking
filter.  I believe that the major benefit of selectivity is the reduction
in fatigue that occurs during long operating periods.

 IMHO, the elimination of distortion, hum and high frequency hiss is the
key.  In my experience, this almost immediately rules out active filters.
I have yet to buy or build one that would adequately drive a pair of hi-fi
headphones without introducting some crud.  The best I did in one homebrew
receiver was a class A transistor with about 300 mA standing current. Most
of the IC amps all suffer from crossover distortion at headphone levels
and tubes, well.... they're real humdingers.

I 've had best luck with passive LC filters of modest selectivity (200 to
1000 Hz BW).  I have a TimeWave DSP-9 and find it very annoying.  The
sampling noise is atrocious. I don't recommend it for weak sig work.

Note, the foregoing applies to weak sig work without QRM.  HF DXing is
another story.

Regards, Wes Stewart, N7WS
---------------------------------------------------------------------

Date: Fri, 26 May 1995 09:39:04 EDT
From: Larry Kayser 
Subject: DSP Experience with MFJ 784, Timewave

I was unable to get this down before Mike circulated his summary.

My experience is with the MFJ 784 ( which I own) and observations of a 
Timewave.

Observations on Timewave 59+.  This was observed in the Noise Reduction
mode during an evening of listening to AO-13 SSB contacts and the
performance was impressive. The Noise Reduction was quite impressive in
that one was not as tired of listening to the raw background noise we have
all come to know and love associated with the AMSAT Satellites.  There
certainly was some of what I call pumping, some distortion as well, but
this was for me completely in line with the normally highly distorted
audio one hears via AO-13 and in fact on most HF stations as well.

Conclusion, If I ever went back to Satellite or 6 Meter dead band
monitoring (waiting for openings) I would sure want one of these devices.

MFJ 784.  This unit for my purposes has a single unique useable feature -
the Frequency of operation is tuneable. This makes the unit very useful
for me on HF, were almost no one can get really zero beat anymore and I
can use this feature along with the continuously variable band width to
usually get rid of the QRM I am seeking to get rid of.  At 30 Hz I do have
some pulse stretching, which I think is called ringing by others, but only
at the weakest signal levels. (I think the pulse stretching becomes
objectionable when one shrinks the bandwidth below that required to
accomodate the signal)  I find the unit very useful for general HF
operation, the unit is NOT good, and I do not exepect it to be, on summer
QRN in my experinece.  I especially enjoy the use of the filter on 40 WPM
CW and higher, I can really knock the noise down and the QRM out.  My
regular HF receiver (whole station in fact) is remote controlled over an
FM Link and I enjoy taking the extra garbage the link introduces out with
the MFJ.

Headphones. My good friend, and QRQ CW buddy, Irv Wineman, W9GA, gave me a
pair of old TELEX headphones and as a result I have sent my HiFi Stereo
specials back to the HiFi set.  What a magical difference these old
headphones make on HF!  The old TELEX headphones have a natural limited
bandpass, the lows are gone, and so are the highs.  The next best thing is
what happens when I plug these headphones into the MFJ. I do not hear much
of the noise and stuff that so upsets some of those who wear HiFi
Headphones.  I have done a little testing of what has to be done to a set
of HiFi headphones to make them approximate a set of old TELEX headphones,
I put a lumped constant CW filter I built 30+ years ago out of Telephone
Loading coils in the circuit, that made the HiFi phones basically usefull
on the '784.  Remember please this is for my ears and my experience.

MFJ-784 in Weak Signal work.  Last year I was one of the Ops at VE3ONT and
I had my first playing with a FT1000D and its audio filters.  I was in
general impressed by the filter, I was not impressed with the way it
worked, I had nothing to tune with.  Somewhat later I took delivery of the
'784 and I wanted to compare my experience.  I used the VE3TEN 5 Watt ten
meter beacon on 28175 Mhz FSK over a 70 mile path.  The receiver was a Ten
Tec CORSAIR 2 which is an excellent weak signal RX.  I use NO AGC, Manual
RF gain, and the MFJ 784. The antenna, a small vertical, was fed to the RX
through a commercial 50 ohm step attenuator in 1 db steps to 60 db.  This
attenuator was previously carefully calibrated as it was used in various
AMSAT work over the years.  With this setup the '784 let me go 16 db
further into the noise over the same setup without the MFJ in the circuit.
This was an average over a dozen observations over a two weekend period.
The goal was to copy the call sign of the beacon and sync into the
interval timing period.  Observations which were impacted by Meteor Pings
or by Airplane Reflections (small doppler shifts) were discarded. The
method of operation I use for finding very weak signals is to sweep the
filter over the target frequency range and to listen for the signal to pop
out of the noise as one sweeps the edge of the filter pass band by the
desired signal frequency.

Conclusions MFJ-784.  This unit works and works well for me.  The tuneable
Frequency and Bandpass feature is of critical importance for me.  Will it
let me hear further into the noise than without it, my answer is yes. This
is conditional for me on using Headphones which do not pass on the Highs
and Lows that are not needed for CW work.  I have never had any problem
finding weak signals, copying them has for me been somewhat harder.  This
comes from actual comparison of my weak signal copying skills with the
others during the last VE3ONT weekend in '94.   The '784 is just one more
tool in the box that I will use, I am impressed and for me the filter is
better than anything else I have ever had.

Comments, discussion to myself please.

Larry VA3LK / WA3ZIA
---------------------------------------------------------------------

Date: 25 May 95 22:30:18 EDT
From: "Hank  M. Meyer, Jr." <75013.164@compuserve.com>
Subject: NIR-10

Dear Mike:
Thanks for the data on DSP's. Have used an NIR-10 here for 1.5 years on
near earth tropo. Upgraded to 3.0 firmware about a year ago, made a
siginificant improvement in ability to extract weak signals from noise.
Several cases where signals were not copyable without the DSP, but
perfectly copyable with it engaged. Have since acquired an NIR-12, and it
works much better than the 10, In automatic mode the radio sounds like it
is on squelch, and I had to put noise back in to make sure the radio was
working. The 12 has provision to disable operation on xmit so radio
sidetone on cw is not impaired by audio delay. The dual processors take a
big chunk out of all noise including line noise. All in all an improvement
over other filters. The lack of ringing in narrow settings is great. The
ability to set the lower and upper cutoff frequencies independently is a
singular improvement over the 10. I dont like the quality of the audio
when the unit is bypassed, and it could stand more audio output. Over all
it's a great improvement over what was available previously. I cant wait
to see it appear at IF.
73
Hank W6GGV 

------

From ve7bqh@wimsey.com Fri Dec  8 20:46 EST 1995
Date: Fri, 8 Dec 95 17:45 PST
To: Charles R MacCluer 
From: "Lionel H. Edwards" 
Subject: Re: Texas Instrument DSP board

Hi Chuck:

I talked with you along with Dave,W5UN after your presentation at the last
CSVHF meeting in Colorado Springs. You ended up loaning the Texas
Instrument DSP board to Dave along with the books and your custom software

After Dave had the filter for a short while,he suggested I use it as he
was going to be away on an extended vacation. So on Dave's last visit with
me in September,the filter arrived at VE7BQH.

I have been using the filter with a 450 cycle tuned frequency since
October. Like all new filters I have had,it takes some time to get fully
accustomed and confident with them (months!).

To get the very most out of the filter,I found I needed to add an small
external audio amp to allow the input and output levels to be set to the
optimum points. This is,of course,typical of any digital filter.

With the operating conditions set as above,I have been getting some very
good results. Clearly this filter produces a better signal noise and copy
ability than any other filter I have or tried. This statement includes a
Time Wave 59+. I have built to many filters to describe. suffice to say
the last three have been relatively simple digital switched cap types. My
prime filter was last of these. This is not to say the improvement with
the TI DSP was large. It was not. However,any noticeable difference is
significant when listening to signals in the area of 0 dB S/N!

The "bonus",particularly at this QTH (in the city),is the noise reduction
provided by the DSP technology. I have made several contacts in high noise
conditions which would have been difficult if not impossible otherwise.

The weakest link I still have to deal with is the tuning rate of the RIT
on digital VFO used with my TS830. I need to slow down the rate either
electrically or mechanically. It is a bit to touchy for a filter of this
bandwidth and signals as weak as I get!

So,once again I thank you for you presentation.

73,
Lionel,VE7BQH
ve7bqh@wimsey.com


Back to N.E.W.S. Home Page