EMRFD Message Archive 834
Message Date From Subject 834 2007-06-06 20:31:16 Jason Milldrum Step Attenuator Project
I recently completed a 0-80 dB step attenuator project and had an
opportunity to do an evaluation of the circuit. I found some
interesting results about the importance of internal shielding and the
frequency response of the attenuator. If you are thinking about
building one of these for yourself, you might want to take a look at
the article.
<http://www.nt7s.com/index.php5?page=stepatt>
73,
Jason Milldrum, NT7S
=====
Amateur Station NT7S - <http://www.nt7s.com>
QRPedia - <http://www.qrpedia.com>836 2007-06-07 14:40:05 Wes Hayward Re: Step Attenuator Project
Outstanding. Great to see all of the work with the attenuator. I
was especially pleased to see that you were doing some measurements.
I've built several slide switch based step attenuators over the years
and might make a few suggestions regarding construction.
1). Avoid loops of wire between switches. If you have a shield, run
the wire through a hole in the metal rather than up and over, for the
loop can be very inductive.
2) Generally try to maintain a 50 Ohm transmission line at all
points along the path. Generally this means keeping the wiring close
to the ground.
3) Make sure the coax connectors are grounded with extremely short
lines.
4) Don't tolerate any paint on the inside of the boxes. This is a
hard thing these days when the LMB folks and others seem to merely
dip the boxes as a way to paint them. I prefer to go for
performance rather than looks, so I buy the bare aluminum boxes when
I can. If I have to remove paint, I use a wire brush in a 1/4 inch
power hand drill.
5) Remember that the main function of shields in low impedance
circuitry is to control where ground currents flow rather than to
build an electrostatic barrier for electric fields.
George Daughters (then WB6AIG) and Will Alexander (WA6RDZ, now SK)
published a wonderful little paper in 73 magazine back in the 60s
when Jim Fisk was the technical editor. This was before Fisk and
Skip Tenney left to form HR. George and Will did measurements on a
couple of shielding methods and reported on performance up through
450 MHz or so. If any of you want this paper, I have a scanned
version that I can send via email. I do not want to post it, for
that would still violate copyright laws. Sending one copy to an
individual for self study is, however, not a violation. So if anyone
wants it, let me know off list.
Someone on this list ask for a copy of that paper a while ago.
Unfortunately, their email address was hidden so I could not sent
anything of list. Moreover, their email address was not in
QRZ.com.
On a related measurement issue, I got some interesting data a couple
of days ago from Bob Larkin, W7PUA. This regards the performance of
the power meter that Bob and I described in QST in June, 2001. That
meter was the one based on the Analog Device AD8307. The performance
of the meter is compromised at power levels above -10 dBm, especially
at the highest frequencies. No problems were observed up through
300 MHz other than the slow ripple in response that was reported
earlier. I have added information to the section of my web page
that deals with this project. The url is
http://users.easystreet.com/w7zoi/pmeter.html
Anyway, keep up the good work Jason and thanks for sharing your
results with us.
73, Wes
w7zoi837 2007-06-07 15:35:32 rds_6 Re: Step Attenuator Project
> hard thing these days when the LMB folks and others seem to merely
> dip the boxes as a way to paint them. I prefer to go for
> performance rather than looks, so I buy the bare aluminum boxes when
> I can. If I have to remove paint, I use a wire brush in a 1/4 inch
> power hand drill.
spot sandblaster with a built-in sand reservoir above the nozzle. I
think I paid $7 for mine. It gives a whole new meaning to "bare
metal." It takes a lot of air, though, and requires
eye/skin/respiratory protection, but it makes paint disappear. Just
don't over-do it or the roughness might become unevenness that could
interfere with a good, reasonably continuous electrical connection.
Anything desired not to be abraded can be masked with tape.
-Ross839 2007-06-08 00:47:28 Jason Milldrum Re: Step Attenuator Project
Wow, I should have talked to you first before starting the attenuator! I
don't recall seeing much of what you have said anywhere else. It would
have been very useful when I was working on the project. I can see that I
have some more modifications to do to the attenuator. Thanks Wes, you've
given me a lot more to think about. Funny how something which seems so
simple has so many nuances to consider.
73,
Jason NT7S
Wes Hayward wrote:
> Hi Jason, et al,
>
>
> Outstanding. Great to see all of the work with the attenuator. I
> was especially pleased to see that you were doing some measurements.
>
> I've built several slide switch based step attenuators over the years
> and might make a few suggestions regarding construction.
>
> 1). Avoid loops of wire between switches. If you have a shield, run
> the wire through a hole in the metal rather than up and over, for the loop
> can be very inductive. 2) Generally try to maintain a 50 Ohm transmission
> line at all points along the path. Generally this means keeping the
> wiring close to the ground. 3) Make sure the coax connectors are grounded
> with extremely short lines. 4) Don't tolerate any paint on the inside of
> the boxes. This is a hard thing these days when the LMB folks and others
> seem to merely dip the boxes as a way to paint them. I prefer to go for
> performance rather than looks, so I buy the bare aluminum boxes when I
> can. If I have to remove paint, I use a wire brush in a 1/4 inch power
> hand drill. 5) Remember that the main function of shields in low
> impedance circuitry is to control where ground currents flow rather than
> to build an electrostatic barrier for electric fields.
>
> George Daughters (then WB6AIG) and Will Alexander (WA6RDZ, now SK)
> published a wonderful little paper in 73 magazine back in the 60s when Jim
> Fisk was the technical editor. This was before Fisk and
> Skip Tenney left to form HR. George and Will did measurements on a
> couple of shielding methods and reported on performance up through 450 MHz
> or so. If any of you want this paper, I have a scanned version that I
> can send via email. I do not want to post it, for that would still
> violate copyright laws. Sending one copy to an individual for self
> study is, however, not a violation. So if anyone wants it, let me know
> off list.
>
> Someone on this list ask for a copy of that paper a while ago.
> Unfortunately, their email address was hidden so I could not sent
> anything of list. Moreover, their email address was not in QRZ.com.
>
>
> On a related measurement issue, I got some interesting data a couple
> of days ago from Bob Larkin, W7PUA. This regards the performance of the
> power meter that Bob and I described in QST in June, 2001. That meter
> was the one based on the Analog Device AD8307. The performance of the
> meter is compromised at power levels above -10 dBm, especially at the
> highest frequencies. No problems were observed up through 300 MHz other
> than the slow ripple in response that was reported earlier. I have added
> information to the section of my web page that deals with this project.
> The url is
> http://users.easystreet.com/w7zoi/pmeter.html
>
>
> Anyway, keep up the good work Jason and thanks for sharing your
> results with us.
>
> 73, Wes
> w7zoi841 2007-06-08 07:37:36 f2dc2003 Re: Step Attenuator Project 842 2007-06-08 08:20:33 Preston Briggs Re: Step Attenuator Project 843 2007-06-08 08:25:34 michael taylor Re: Step Attenuator Project
> Why do so many people use slide switches?They have a nibbling tool maybe? Then it isn't a big deal to make
> Why not use toggle switches, saving the effort of squaring up several
> round holes?
rectangular holes.
<http://www.adelnibbler.com/> available from Kanga US.
-Michael, VE3TIX844 2007-06-08 09:15:03 Jason Milldrum Re: Step Attenuator Project
sitting around gathering dust, so that's what I used. Yes, it was a bit
more of a pain to cut the holes with a nibbling tool, but it wasn't
incredibly difficult.
Next time I build one, I'm going to get some high-quality toggle switches.
73,
Jason Milldrum, NT7S
=====
Amateur Station NT7S - <http://www.nt7s.com>
QRPedia - <http://www.qrpedia.com>
Prest848 2007-06-09 12:14:59 Roy J. Tellason Re: Step Attenuator Project
> Hi Jason, et al,Yes, that page was rather nicely done...
>
> Outstanding. Great to see all of the work with the attenuator. I
> was especially pleased to see that you were doing some measurements.
> I've built several slide switch based step attenuators over the yearsAgreed.
> and might make a few suggestions regarding construction.
>
> 1). Avoid loops of wire between switches. If you have a shield, run
> the wire through a hole in the metal rather than up and over, for the
> loop can be very inductive.
I remember seeing something similar in a magazine, probably Popular
Electronics but maybe perhaps Radio-Electronics or Electronics World (I never
got the typical ham mags), and in that one the enclosure for the setup as
well as the partitioning between switches was done with bits of double-sided
circuit board material, which strikes me as a fairly easy way to build it,
if you have some convenient way of cutting the stuff. Connections from one
to the next were very direct straight bits of wire that passed through a hole
drilled in the shield/partition.
This is a construction method that I've been meaning to explore...
--
Member of the toughest, meanest, deadliest, most unrelenting -- and
ablest -- form of life in this section of space, a critter that can
be killed but can't be tamed. --Robert A. Heinlein, "The Puppet Masters"
-
Information is more dangerous than cannon to a society ruled by lies. --James
M Dakin849 2007-06-10 05:30:52 Don Kelly Re: Step Attenuator Project
circuit board material, which strikes me as a fairly easy way to build
it..."
Roy,
There is a "Low-Power Step Attenuators" project in the 1994 ARRL
Handbook (p25-37)using double sided PC board for the partitioning and
the enclosure.
I added a 1, 3, 6, 10 dB step attenuator to my HB signal generator
using this PC board method. I used a Hartley oscillator similar to the
EMRFD design. Output is not consistant over the whole tuning range and
the built-in, low value attenuators are convenient.
Don850 2007-06-10 21:07:13 adams Re: Step Attenuator Project
> "the partitioning between switches was done with bits of double-sidedSee 1995 ARRL HB, Chapter 26, 26.40-26.41.
> circuit board material, which strikes me as a fairly easy way to
> build
> it..."
>
> Roy,
>
> There is a "Low-Power Step Attenuators" project in the 1994 ARRL
> Handbook (p25-37)using double sided PC board for the partitioning and
> the enclosure.
>
> I added a 1, 3, 6, 10 dB step attenuator to my HB signal generator
> using this PC board method. I used a Hartley oscillator similar to
> the
> EMRFD design. Output is not consistant over the whole tuning range
> and
> the built-in, low value attenuators are convenient.
>
> Don
>
>
http://www.farcircuits.net/test1.htm
for the board set from Far Circuits for $9. These have
the slide switch holes already cut out to save you a ton
of work.
Hope this helps narrow down the search for information.
>--
Chuck Adams, K7QO k7qo@commspeed.net
http://www.k7qo.net/ personal web page
Moving to Arizona? Bring your own water.7409 2012-04-14 18:05:10 Kerry Step Attenuator Project
http://i43.tinypic.com/1zg3n1d.jpg
It looks good, as I like my projects to look, but performance is terrible! :(
Here are some sweeps; the amount of attenuation switched-in on each is shown; red is attenuation, blue is phase and green is unput VSWR;
http://i41.tinypic.com/1ghr36.jpg
http://i43.tinypic.com/rkuaew.jpg
http://i44.tinypic.com/2vkydet.jpg
Note the"peaks" at intervals of ~300MHz and the phase changes; the body of the unit is 250mm long so I have created a 1/4-wave resonator. :(
Construction is based on SM resistors & 50-ohm microstrip with shielding;
http://i41.tinypic.com/35bch83.jpg
http://i43.tinypic.com/5344zs.jpg
I have added vertical & horizontal shields at the ends since the above photos were taken; the sweeps were done with these shields in place.
I don't think that this layout can be made to work. One option is to halve it to make two units; the resonances would then be at 600MHz intervals but that doesn't seem to be a very satisfactory solution to me.
Any suggestions?
Kerry.7410 2012-04-14 18:10:33 Kerry Re: Step Attenuator Project
Jason's design was the basis for mine; his website description was very helpful. Thanks Jason!
Jason got good results with his uncased design; the results for mine were reasonably-good also until I put it into the case.
Kerry.7411 2012-04-14 18:23:45 Tim Hills Re: Step Attenuator Project
Instead of 7 10dB steps use a 10, 20 and 40 eliminating 4 of the switches.
Boxes within a box construction might help too.
My projects tend to look like crap which probably explains why the XYL
won't let them live anywhere but the shack :(
Tim Hills
Sioux Falls, SD
7412 2012-04-14 18:23:46 Chris Trask Re: Step Attenuator Project
Chris
>Chris
>I have been building a step attenuator; here is the finished product;
>
>http://i43.tinypic.com/1zg3n1d.jpg
>
>It looks good, as I like my projects to look, but performance is terrible! :(
>
>Here are some sweeps; the amount of attenuation switched-in on each is shown; red is attenuation, blue is phase and green is unput VSWR;
>
>http://i41.tinypic.com/1ghr36.jpg
>
>http://i43.tinypic.com/rkuaew.jpg
>
>http://i44.tinypic.com/2vkydet.jpg
>
>Note the"peaks" at intervals of ~300MHz and the phase changes; the body of the unit is 250mm long so I have created a 1/4-wave resonator. :(
>
>Construction is based on SM resistors & 50-ohm microstrip with shielding;
>
>http://i41.tinypic.com/35bch83.jpg
>
>http://i43.tinypic.com/5344zs.jpg
>
>I have added vertical & horizontal shields at the ends since the above photos were taken; the sweeps were done with these shields in place.
>
>I don't think that this layout can be made to work. One option is to halve it to make two units; the resonances would then be at 600MHz intervals but that doesn't seem to be a very satisfactory solution to me.
>
>Any suggestions?
>
7413 2012-04-14 18:26:58 Kerry Re: Step Attenuator Project
I should have said that the basis for my design was the unit built by Paul NA5N;
www.qrpedia.com/sites/default/files/STEP_ATTEN_1.pdf
My apologies to both Jason and Paul.
Kerry.7414 2012-04-14 18:40:59 Kerry Re: Step Attenuator Project
I kept to 10dB sections as 20dB ones can have leakage problems; fewer sections would, at least, shorten the case and that was the basis of my idea to cut it in half to make two separate cases.
But that's a "work-around", not a "solution".
The connectors are BNCs that take RG-316 crimped; they are, therefore, RF-sealed.
The braid of the RG-316 is connected to the PCB groundplane at each end and the centre conductor is connected to the microstrip at that point so RF is delivered to the PCB by the RG-316.
I used 3/64" copper rivets (from a model-makers' supplies shop) as vias. There is one at each pi-section; that's the solder bump that's visible at each switch;
http://i40.tinypic.com/34hw8dk.jpg
The shields have copper shim wrapped-around each top edge and soldered.7415 2012-04-14 18:43:04 Alex P Re: Step Attenuator Project
Very nice mechanical work.
I think that you have some probable cavity resonance going on. This has shown up in the past7416 2012-04-14 18:50:08 Kerry Re: Step Attenuator Project
Yes, obviously a resonance as the cavity is 1/4-wavelength at 300MHz where the "peaks" are in the sweeps.
I did sweep it before casing it up; I didn't keep the results but they were fairly good so I proceeded to case it up to obtain the horrible results.
If you look at the link to NA5N's unit you will see that he got fairly good results un-cased; he intended to put his in a case but it's probably a good thing that he doesn't seem to have got around to it although his unit is much shorter than mine.
Kerry.7417 2012-04-14 18:57:08 Kerry Re: Step Attenuator Project
http://i42.tinypic.com/mojyq.jpg
There's a good deal of work there; that's what makes the results all the more disappointing.
Kerry.7418 2012-04-14 19:21:17 Dave Re: Step Attenuator Project
addition to the metal top would help?
Dave - WB6DHW
<http://wb6dhw.com>
7419 2012-04-14 19:23:47 Kerry Re: Step Attenuator Project
I wondered if something like steel wool might be effective in breaking-up the cavity.
Kerry.7420 2012-04-15 04:28:22 VK1OO Re: Step Attenuator Project
I've been hoping for a while that someone would take a step forward and
incorporate computer control of the switches into such an attenuator
design. This would allow development of software tools to help automate
testing.
Cheers
Mike
VK1OO7421 2012-04-15 05:07:15 dixonglennb Re: Step Attenuator Project
> Why do so many people use slide switches?Since the attenuation occurs between the two poles of a single DPDT switch (and not between the switches), low coupling between the two poles is important (that is where the shielding should really go if there were only some way...).
I suspect the slide switches may be better in this respect, though have no data to support it besides a picture of the switch internals in my head.
73,
Glenn AC7ZN7422 2012-04-15 05:22:27 Ashhar Farhan Re: Step Attenuator Project
have seen this in a qex article of last decade.
- Farhan
On 4/15/12, dixonglennb <glennbdixon@gmail.com> wrote:
>
> Preston asked:
>
>> Why do so many people use slide switches?
>
> Since the attenuation occurs between the two poles of a single DPDT switch
> (and not between the switches), low coupling between the two poles is
> important (that is where the shielding should really go if there were only
> some way...).
>
> I suspect the slide switches may be better in this respect, though have no
> data to support it besides a picture of the switch internals in my head.
>
> 73,
> Glenn AC7ZN
>
>
>
--
Sent from my mobile device7423 2012-04-15 05:53:44 dixonglennb Re: Step Attenuator Project
> I've been hoping for a while that someone would take a step forward andHi Mike,
> incorporate computer control of the switches into such an attenuator
> design.
You may be familiar with the 31 dB step attenuator chips made by Triquint, Hittite, Minicircuits, etc. for cell phones. I think they are used to set the cell phone transmit power. You can get SPI and parallel port versions and some claim to go down to DC. MiniCircuits offers theirs assembled in a box intended to connect to your computer's legacy parallel port, giving automated control. I suspect it does not work very well as the range is just 32 dB or so and parallel ports tend to be pretty noisy from an EMI standpoint.
I had always wanted to try these chips for a step attenuator and have now designed one using three of the Triquint TQP4M9072 chips in series. These chips require a little bit of DC power so my design is battery powered. Attenuation is controlled manually by up/down buttons or automatically through an IrDA port so there is no electrical connection from the controlling computer to the attenuator.
The printed circuit layout is off to the shop (batchpcb.com) and I expect delivery within two weeks. My son Adam and I will then assemble and test it. I will report the results here and if the project is successful I'll give design details.
These chips are very tiny and a lot can go wrong. It is presumptuous to think someone can make a 90 dB attenuator work right on the first try but it will be interesting to see what problems we encounter.
73,
Glenn AC7ZN7424 2012-04-15 08:00:54 Ivan Makarov Re: Step Attenuator Project
state switches:
http://www.makarov.ca/vna_att4230.htm
Regards,
Ivan
VE3iVM
----- Original Message -----
7425 2012-04-15 13:36:23 dualraoul Re: Step Attenuator Project 7426 2012-04-15 15:42:33 kb1gmx Re: Step Attenuator Project 7428 2012-04-22 04:43:58 VK1OO Re: Step Attenuator Project
Thanks for the feedback on computer controlled attenuators. I'm looking
forward to seeing more of the Glenn AC7ZN design.
Mike
VK1OO7430 2012-04-23 23:18:17 Kerry Re: Step Attenuator Project
Results are much better; not as good as I hoped but mostly acceptable.
Insertion Loss;
http://i41.tinypic.com/21oa2s5.png
1 + 2 + 4 + 8 dB switches on;
http://i41.tinypic.com/t8wvvc.png
Now this is interesting; this shot shows 1 + 2 + 4 + 8 + 10 dB of attenuation. The difference is that, in the upper trace, the extra 10 dB was put-in with the leftmost of the six 10 dB switches; in the lower trace the extra 10 dB was put-in with the rightmost 10 dB switch.
http://i41.tinypic.com/344xda9.png
(Note that the references for the traces have been shifted to separate them for clarity; the vertical scale is the same for each).
I think that the unit is just too long for good performance without resonances; that is borne out by other tests.
Performance worsens as more attenuation is switched in until the final horror of full (75 dB) attenuation;
http://i41.tinypic.com/vxd6t.png
I will decide whether to accept it as-is and remember to keep attenuation down to about 35 dB total or whether to shorten it.
One option is to cut it in half and make two out of it; I think that two five-switch units connected by co-ax would behave better than the 11-switch single unit.
It's been an interesting, if frustrating, exercise; I've learned a lot.
It's ironic that my next project might be a resonant line; this attenuator project has taught me a bit about resonant cavities! :)
Kerry.7431 2012-04-24 10:38:42 Andy Re: Step Attenuator Project
inside of the case ALL the way around) is used between every section.
Those little stubby "shield" walls probably do very little to help, if anything.
Maybe you can rescue the current package by adding a few dividers,
with some sort of fingers on their circumference to make contact with
the case. Or solder them to the case with some means to wipe against
one another when the case is fully assembled.
> I've been hoping for a while that someone would take a step forward andYears ago I worked at a company that made relatively inexpensive
> incorporate computer control of the switches into such an attenuator
> design. This would allow development of software tools to help automate
> testing.
relay-switched RF attenuators, not for sale by themselves but used
within their larger products (test equipment). They also had software
calibration to compensate for attenuation errors, I think including
frequency response.
Andy7432 2012-04-24 23:07:15 Kerry Re: Step Attenuator Project
http://i49.tinypic.com/295bpqc.jpg
This was not soldered to the shields or the base PCB; it was made to "grip" them but, of course, it only contacted a few shields.
I don't like to be beaten so today I had another go. I enclosed each switch compartment in its own copper box;
http://i50.tinypic.com/29uvt3l.jpg
I ran copper shim along each side and seam-soldered it to each shield and base joint. I then closed-in every second box top and soldered each lid to its mating shield.
I then soldered-in the "missing" box tops then rolled the side wall shims over and seam-soldered the lot;
http://i45.tinypic.com/nbaq2t.jpg
There are now more shields than were used in Ben Hur and Kester will soon declare a record profit! :)
But it was worthwhile. Here are some sweeps;
This is with 1 + 2 + 4 + 8 dB switched-in;
http://i49.tinypic.com/r72a7b.png
This is with 1 + 2 + 4 + 8 + 10 dB switches on; as in the one I posted yesterday there are two traces, one using the LH 10dB switch and one using the RH 10dB switch;
http://i49.tinypic.com/21edpnd.png
Unlike yesterday's results the two traces are identical.
Results deteriorate as more 10dB steps are switched-in but they are still not too bad until all switches (75dB) are on;
http://i48.tinypic.com/2rh673s.png
Not perfect but not too bad up to about 500 MHz; I think that one or two 20dB steps might have been better.
I'm fairly happy with it now. It's been a good tutorial in the importance of shielding as frequency increases.
Kerry.