Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

   Home   Help Search  
Pages: [1] 2 Next   Go Down
  Print  
Author Topic: Switchable High Value Attenuators  (Read 5863 times)
TANAKASAN
Member

Posts: 933




Ignore
« on: July 25, 2011, 06:21:30 AM »

I want to build a switchable attenuator for receiver measurements, 0dB to 150dB in 10dB steps. Assuming a 10, 20, 40, 80 sequence I can switch it in binary but the problems are the 40dB and especially 80dB attenuators.

Building an 80 dB attenuator is easy (50.1 ohm to deck at each end and a 250K in series) but switching it in and out is proving to be impossible. Most relays only have 60dB or so of isolation and even specialist RF relays only give 75dB to 80dB with even this tailing off the higher you go in frequency.

So, any ideas? How would you build a variable high value attenuator? Working frequency is 0.5 MHz to 500 MHz.

Tanakasan

Logged
M0HCN
Member

Posts: 473




Ignore
« Reply #1 on: July 25, 2011, 07:39:37 AM »

It would cost you a few extra relays, but how about splitting the 80 as two 40db stages.

so you would have 10,20,40,40,40, or even something like 10,20,30,30,30,30.

Place a relay each end of each section rather then using both contact pairs on a DPST relay, and arrange things to short the bypass line to ground at both ends when the attenuator is switched in.

Decoupling the control lines will be critical as will screening between sections.

Now making the thing flat over 0.5 - 500Mhz, that is the trick.
Logged
WB6BYU
Member

Posts: 13336




Ignore
« Reply #2 on: July 25, 2011, 09:50:19 AM »

Actually I'd limit each stage to about 20dB and make sure you have a shield across the CENTER of each
stage as well as between stages.  (The former is more important than the latter.)

I have a commercial switched attenuator built this way: there is a shield that fits tight against the back
of the switch with one lead through it for the BYPASS and another for the attenuator (the middle resistor
of a PI network can be placed in hole if desired.)  Then the wire between sections also passes through
another shield wall.  Construction may be easier if you used separate SPDT relays at each end of each
attenuator section and possibly grounded the bypass line when using that attenuator stage.

If accuracy is important, the lower values of attenuation are better because the attenuation value
isn't as dependent on the resistor tolerance.

I use my attenuator for transmitter hunting on 2m FM to reduce the input signal to my receiver to keep
the S-meter on scale (they usually have a range of only about 12dB.)  The steps total over 100dB, but I
find around 60 to 70dB is the maximum usable value - beyond that the signal picked up on the coax (double
shielded) and directly through the case of the radio exceeds the attenuated signal from the antenna.

So while an 80dB pad is easy to design, getting anywhere near that much attenuation in the real world
is very difficult, even WITHOUT a switch, because all the rest of your equipment needs to be completely
shielded as well.
Logged
W8JI
Member

Posts: 9296


WWW

Ignore
« Reply #3 on: July 25, 2011, 12:12:43 PM »

I've successfully built relay and manual switched 30 dB pads, but it takes some care with relay or switch selection and groundplane layouts. Shielding is NOT necessary with careful layout. For example, I easily get 60 dB or more port isolation under worse case conditions with open boards in antenna switches. These are not even groundplane backed layouts.

I would not try to have larger than 30 dB pads though if you want UHF.

I've looked at this stuff very carefully in the past, and I'm sure of this because of many commercial products that have resulted in very low leakage (including attenuator pads).

73 Tom
Logged
W6RMK
Member

Posts: 656




Ignore
« Reply #4 on: July 25, 2011, 10:24:57 PM »

If you can find them, look for the HP ap notes on attenuators and RF switches.

High attenuation with good leakage performance is all about attention to detail.
Logged
WD4HXG
Member

Posts: 186




Ignore
« Reply #5 on: July 26, 2011, 03:05:07 AM »

One thing that will bite you when building and using
attenuators with HIGH VALUES of Attenuation is attention
to clean connections, clean PC Boards, and clean grounds.

You can not believe the mischief that a little bit of dust can
cause when across the Teflon insulation in the connector.
Keep any connector clean. I use EverClear (195 Proof)
grain alcohol ( available at your local state beverage
controlled retail sales outlet - ABC Store for us on the
East Coast) with tightly wrapped cotton tipped wood
stem swabs. I wash the PC Boards in alcohol and then
rinse in distilled water when working with high levels
of attenuation.

Ground Planes should be kept meticulously clean and free
from any oxidation or sulfating. I would clean the PCB surfaces
so they shine and then spray with a protective coating such
as Krylon clear.

At the risk of speaking of the obvious, an item that is
often overlooked is that 150 dB of attenuation is tremendously
large amounts of attenuation. Consider if you had a 1 watt desired
signal level with 150 dB of attenuation in front of it you would need
a signal source amplitude of 1,000,000,000,000,000 watts. Can you
imagine the efforts you would have to go through
to keep that signal level out of everything in your lab? Obviously
you are working with signal levels in the other direction but the
principle is the same.

One of the things I have done in the past is to design the signal
source to generate levels no higher than -10, -20 or -30 dBm.
Oscillators designed with supply currents in the uA range can
be snitty to get started but it is a lot easier to use less attenuation
then it is to try to ditch signal amplitudes 15 orders of magnitude
above what you are wanting. 120 dB is much easier to accomplish
than that elusive last 30 dB. 
Logged
WB2WIK
Member

Posts: 20611




Ignore
« Reply #6 on: July 26, 2011, 09:12:22 AM »

150 dB in a single enclosure at VHF-UHF is damned close to impossible. Cheesy

I worked at HP (now Agilent) during the development phase of several step attenuator products.  One very popular one is the model 355D, which is 0-120 dB in 10 dB steps and good through 1000 MHz.  One that is very similar to that is used in the output of the model 8640 signal generator, which is a project I worked on.

Today these would use chip resistors for sure, but back "then," we used very small axial leaded 1/4W carbon film resistors with the leads cut off to about nothing, soldered to a ceramic substrate with gold plated lines and pads, making pretty much a "thick film hybrid circuit."  The enclosure was an aluminum casting with very good tooling and isolation chambers between attenuator sections, with a machined flat surface where the cover attached with lots of screws to make a very RF tight bond even at 1000 MHz.  The switching was done by insulated cams mounted to a camshaft which was the rotary switch, pressing contacts onto various gold plated metallized pads to make or break connections that stepped the attenuator sections into the line in sequence.

No relays involved.

Those attenuators had true 120 dB isolation at 1000 MHz, and actually tested pretty good at 1300 MHz.  And down to DC, of course.  The 355D looks like this: http://www.alliancetesteq.com/node/17477

We did try to add sections to that design to make a 140 dB attenuator with two more steps, but couldn't achieve enough isolation in a single package with any reasonable design.

Some of the guys on this project were former Radio Frequency Labs engineers who had a very deep background in similar designs and there was always some argument about using stepped (resistive) attenuators vs. waveguide-below-cutoff attenuators which can achieve 140 dB okay but obviously can't work down to "DC," and aren't much good down below a few MHz.  The resistors won on our project, since the generator was intended to dial down to 500 kHz. Wink
Logged
AA4HA
Member

Posts: 1482




Ignore
« Reply #7 on: July 26, 2011, 10:00:15 AM »

I cannot speak highly enough about the HP attenuators. I have one rotary attenuator that is part of my lab setup and another that is BCD controlled with SMA connections on either end and a ribbon cable out of the side. It is about 4 inches long, by 1" x 1" in size. I think it came out of an old spectrum analyzer. It is a great little device and if I recall correctly I paid something like $30 on eBay.

A few models I use right now are;
HP 8496H
Hp 8496G
a bunch of fixed attenuators from 10 to 40 dB each with APC7 connectors.

It sounds like you are going to use the attenuator for measuring/aligning receivers. You may find that the incidental leakage from the test equipment may make measurements way down in the -110 to -150 dBm range very difficult to do as even a small, imperfect seal on a RF gasket is enough to cause problems.


Tisha Hayes, AA4HA
« Last Edit: July 26, 2011, 10:12:06 AM by AA4HA » Logged

Ms. Tisha Hayes, AA4HA
Lookout Mountain, Alabama
TANAKASAN
Member

Posts: 933




Ignore
« Reply #8 on: July 26, 2011, 11:00:43 AM »

To WB2WIK

I think I have seen the attenuator design you talk about during my research. Each attenuator was a tiny square ceramic element and (I think) they moved up or down depending on the attenuation requirement. It's an interesting piece of design but I have yet to find a detailed drawing. Certainly I am willing to adopt a mechanical approach rather than an electromechanical system using relays but more technical details would be appreciated.

As for the 150 dB maximum attenuation requirement, I'll take the best I can get and I have the time to fiddle around. -121 dB is the same as wiring up a 0 dB oscillator and seeing an S1 signal on the S Meter so this is an easy check to make. -127 dB is an S0 signal and -137 dB is an S0 signal with the preamplifier switched in. After this things get silly with signals buried in the noise but a waterfall display may 'see' them.

It's an interesting engineering challenge and I find this much more fun than sitting in front of a rig swapping numbers in a contest. Opinions of course may differ.

Tanakasan
Logged
M0HCN
Member

Posts: 473




Ignore
« Reply #9 on: July 26, 2011, 11:37:11 AM »

One other critical detail if using relays:

Make sure there is a few mA of DC wetting current flowing in both attenuation in and the shorting position, relay contacts without wetting current are trouble.

Regards, Dan.
Logged
WB2WIK
Member

Posts: 20611




Ignore
« Reply #10 on: July 26, 2011, 12:02:41 PM »

To WB2WIK

I think I have seen the attenuator design you talk about during my research. Each attenuator was a tiny square ceramic element and (I think) they moved up or down depending on the attenuation requirement. It's an interesting piece of design but I have yet to find a detailed drawing.

I doubt you will.  That design is covered by numerous patents and there are some drawings in those but not enough information to reproduce the same design.  There's a lot of magic in these things.

RLC Electronics in NY made something similar for a while (about 30 years ago) but they don't anymore; possibly they were hit with a lawsuit (I don't know) for violating the HP patents.

I could breadboard the whole design from memory, but without a properly machined enclosure with all the interstage shielding and tight fitting/gasketed cover it would be impossible to achieve anything close to 120 dB, especially in the VHF range.  The "BNC" model had trouble achieving the 120 dB due to connector issues.  The "TNC" model could nail it every time.
Logged
WB2WIK
Member

Posts: 20611




Ignore
« Reply #11 on: July 26, 2011, 06:35:40 PM »

To WB2WIK

I think I have seen the attenuator design you talk about during my research. Each attenuator was a tiny square ceramic element and (I think) they moved up or down depending on the attenuation requirement. It's an interesting piece of design but I have yet to find a detailed drawing. Certainly I am willing to adopt a mechanical approach rather than an electromechanical system using relays but more technical details would be appreciated.


If you can deal with frequencies above 10 MHz or so (into UHF), a waveguide below cutoff attenuator can be very good, and very accurate, and completely adjustable (not in 10 dB steps, but continuously, with about 0.05 dB resolution) over a range of 140 dB or so.

They were commonly used in signal generators for 50 years.  Problem is, you need a more powerful source because they have about 20 dB insertion loss minimum, when cranked "all the way up."  If you need a 0 dBm output signal, you need to start with a +20dBm source.

Logged
WB6BYU
Member

Posts: 13336




Ignore
« Reply #12 on: July 27, 2011, 08:46:11 AM »

I think I have one of those in my junkbox from an old HP-608.

Might be difficult to control it remotely in 10dB steps, however.
Logged
W6RMK
Member

Posts: 656




Ignore
« Reply #13 on: July 29, 2011, 12:25:28 PM »

re controlling vane/wg below cutoff attenuators... stepper motor with microstepping controller

But AA4HA's comment about leakage from the box is right on.  At work, we test deep space transponders at threshold (-150 to -160dBm at 7GHz), and it's challenging, even in a screen room with the generators outside.  And, you have to pay careful attention to how you get your test signal so that intermediate frequencies in the synthesis chain don't happen to have harmonics in the wrong places.  You could have the best output attenuator in the world, but if the +10dBm you're tripling up to get the output leaks, and happens to have a -60dBc harmonic, you're out of luck.
Logged
W8JI
Member

Posts: 9296


WWW

Ignore
« Reply #14 on: July 30, 2011, 01:46:46 AM »

I think he was talking about 70 cm wavelength minimum wl, not 7 cm. :-)
He also wanted steps.

Quote
Insert Quote
I want to build a switchable attenuator for receiver measurements, 0dB to 150dB in 10dB steps. Assuming a 10, 20, 40, 80 sequence I can switch it in binary but the problems are the 40dB and especially 80dB attenuators.

So, any ideas? How would you build a variable high value attenuator? Working frequency is 0.5 MHz to 500 MHz

With a careful layout on lower frequencies, which is pretty much what he wanted, the problem is mostly ground loops and switch. So he has it assessed correctly.

The cure is to use multiple steps and to pick the relay carefully. I'd limit attenuation to 20-30 dB per stage.

With surface mount parts and groundplane construction most of his problems will be the switch or relay. I'd avoid waveguide below cutoff attenuators. It sounds like he wants a stepped pad for calibrating or testing things. The biggest problem is the switch or relay needed and so far no help on that.

http://www.teledynerelays.com/rfrelays.asp


I've used these at 20 dB up to 250 MHz and they work fine. The coils are only about 15 mA current, so they are low power consumption:

http://www.mouser.com/ProductDetail/NEC/UB2-12NU/?qs=sGAEpiMZZMs3UE%252bXNiFaVF%2fhK9L3NJJAJ3f6tF9MgDk%3d

or

http://www.mouser.com/ProductDetail/NEC/EB2-12SNU/?qs=sGAEpiMZZMs3UE%252bXNiFaVF%2fhK9L3NJJAuC9wx8qo7Qg%3d

I suspect they will be good even at 500 MHz.

73 Tom
Logged
Pages: [1] 2 Next   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!