UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > General Vintage Technology > Components and Circuits

Notices

Components and Circuits For discussions about component types, alternatives and availability, circuit configurations and modifications etc. Discussions here should be of a general nature and not about specific sets.

Closed Thread
 
Thread Tools
Old 1st Dec 2017, 4:14 am   #21
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,576
Default Re: Attenuators - theory and the design of.

I have attached some notes from Tektronix on how to make flat attenuators in the high impedance environment for scope input/probe circuits, there you also have to deal with cable capacitances as well. The give examples of x10 and x100 probes. It might be of some interest.

The trickiest attenuator I ever had to build was a 1000:1 while still retaining a wide bandwidth and having it 50kV capable as well. The trick in the end was to use a metal rod as a compensation capacitor that ran parallel spacing to the main 50 Meg ohm dropper resistor. It shows the frequency response on a square wave test from a PG506 calibration generator in the article;

http://nebula.wsimg.com/1b62eb4a5f8e...&alloworigin=1
Attached Thumbnails
Click image for larger version

Name:	att1.jpg
Views:	22
Size:	72.8 KB
ID:	153192   Click image for larger version

Name:	Att2.jpg
Views:	21
Size:	61.8 KB
ID:	153193  
Argus25 is offline  
Old 1st Dec 2017, 2:14 pm   #22
G0HZU_JMR
Octode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 1,404
Default Re: Attenuators - theory and the design of.

If it is of any interest I've used the method I outlined in post #19 to design short antennas with wide(r) bandwidth than you would get by using a series inductor to cancel the capacitance of a very short antenna. This is a bit of a strange and dodgy method that dates back >60 years at least. With an active series -capacitor you can get much more bandwidth but this does come with a lot of baggage. The noise floor will be higher and the signal handling will be limited. But I've made active antennas like this that are just a few inches long for use down at just a few kHz. I currently use one to listen to the LW band and also stuff below 60kHz.

One type of -capacitor design is bi directional so it can be used for Tx as well. But only at low power. I'm not sure how well it would work as a pantry Tx but such an antenna would have a very wide bandwidth. Possibly the whole MW band? I've made this type of antenna for use at VLF and up at VHF but never the MW band so I can't be certain. It is a bit of a novelty method to get wider bandwidth and with care it can help with efficiency too (in theory at least).

See below for the general idea when applied as an attenuator. I suspect no one has ever made an attenuator like this and no one ever will but you can see that the attenuation is flat and the phase is flat.
This is for a perfect -50pF capacitor. With real world circuits the response will degrade a bit especially at lower frequencies and it might drift a bit in phase (over temperature) at lower frequencies. But you can see the general idea. You can do the same trick to negate inductive reactance as well but it might be harder to get a flat response. Instability is never far away with this method either so the whole structure can be quite highly strung especially if used with an antenna. Any significant change in self capacitance due to proximity effects can send it into wild oscillation until normality is restored.
Attached Thumbnails
Click image for larger version

Name:	NcapPAD.jpg
Views:	28
Size:	53.5 KB
ID:	153212  
__________________
Regards, Jeremy G0HZU

Last edited by G0HZU_JMR; 1st Dec 2017 at 2:28 pm.
G0HZU_JMR is offline  
Old 2nd Dec 2017, 1:15 am   #23
Wendymott
Heptode
 
Wendymott's Avatar
 
Join Date: Jun 2013
Location: Huddersfield, West Yorkshire, UK.
Posts: 801
Default Re: Attenuators - theory and the design of.

Hi Peeps... not that it is VERY technical... but I just put together a 6 pole switched attenuator for my SSB signal generator project, using Teledyne reed relays to switch the steps. It looks reasonably flat to over 150 Mhz..
The resistor values are "near abouts" but I include the table for correct values. All resistors are SMD, and I have just ordered 4700 devices with small graduated steps... to help me hit the correct values.
Attached Thumbnails
Click image for larger version

Name:	Relay switched atten Ver 2.jpg
Views:	41
Size:	60.3 KB
ID:	153228  
__________________
Should get out more.

Regards
Wendy G8BZY
Wendymott is offline  
Old 2nd Dec 2017, 2:11 am   #24
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,576
Default Re: Attenuators - theory and the design of.

Wendy,

(Love those Teledyne relays)

There is something about your attenuator that I don't understand, maybe its for some reason. Looking at your attenuator sections the input and output impedance starts out for the first section at over 200 Ohms then it drops with each section and the output network is around 50 ohms.

I think it should be that each attentuator section ideally would have a 50 Ohm input and output impedance (but I might need correcting on this)

I have in the past worked out the formula for this myself (it has been checked) attached. The only thing though the formula was configured for "attenuation factor" so a x2 attentuator would drop the level to 1/2 or a x10 attenuator would drop it by 1/10. So to use this formula you would convert your dB values into an attenuation factor A. The formula could be simplified, but it works as it is.

If you insert 50 in for the Z in the formula the results will make sure the input and output Z of each attenutor section remains at 50R
Attached Thumbnails
Click image for larger version

Name:	atx.jpg
Views:	24
Size:	57.6 KB
ID:	153229  
Argus25 is offline  
Old 2nd Dec 2017, 3:39 am   #25
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,576
Default Re: Attenuators - theory and the design of.

.......just another note about those attenuator networks (resistors selected for a 50R environment);

Each one (group of 3 resistors in a ladder attenuator) is deigned so that the attenuation is correct and the input impedance to the group appears as 50R only when its output is terminated by 50R.

So it would be expected that the output of your switched attenuator network (the final group of 3 resistors) would be terminated in 50R, probably at the other end of a cable where you were feeding the signal. If not terminated there in 50R, then the final output attenuator group would not give the correct dB attenuation and its input impedance to that group wouldn't be exactly 50R either.
Argus25 is offline  
Old 2nd Dec 2017, 9:48 am   #26
G0HZU_JMR
Octode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 1,404
Default Re: Attenuators - theory and the design of.

Quote:
There is something about your attenuator that I don't understand, maybe its for some reason.
The attenuator sections look fine to me? The resistor values all look to be close to the classic values for 2, 3, 5, 10 and 20dB sections. I can't see how or why anyone would do it differently unless you went to T sections rather than Pi.

At work we have always used 300R for the shunt resistors in a 3dB pad and I think this is common practice to allow 300R resistors to be used here. But 270R is also used here. Note that there is a decimal point typo in the table because the series resistor should be 248R and not 24.8R for a 20dB attenuator.

For ultra wideband operation (up into the GHz region) the frequency response of a 20dB attenuator can be improved if it is made up from arranging 120R (or 124R) 0805 package SMD resistors in series or parallel. It's possible to make a precision 20dB attenuator by selecting/grading 5% 120R and 124R resistors on a DMM and it should work well up into the GHz region. 0805 SMD resistors with a value in the 82R to 150R region will behave very well over a huge bandwidth. Often you will see commercial attenuators where the designer has tried to use parallel or series resistor combinations to improve the bandwidth in a similar way.

At work I still use an old 'company' DOS program to design Pi or T attenuators. It dates back to the 1980s but it works really well. I can remember the values for the classic attenuators but this program is still very useful for custom values or for different input or output impedances.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 2nd Dec 2017, 10:16 am   #27
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,576
Default Re: Attenuators - theory and the design of.

Quote:
Originally Posted by G0HZU_JMR View Post
Quote:
There is something about your attenuator that I don't understand, maybe its for some reason.
The attenuator sections look fine to me? The resistor values all look to be close to the classic values for 2, 3, 5, 10 and 20dB sections. I can't see how or why anyone would do it differently unless you went to T sections rather than Pi.
The values might be classic, but the input impedance to this attenuator unit is not 50R. Surely it would be better to have this attenuator such that you could plug it in series with any 50R system and not have any impedance bumps ? Or is it regarded as not important that the input of the attenuator represents a 50R load ??
Argus25 is offline  
Old 2nd Dec 2017, 12:17 pm   #28
G0HZU_JMR
Octode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 1,404
Default Re: Attenuators - theory and the design of.

It looks like a classic inline (insertable) 50R attenuator to me. So the input impedance will be 50R if used as intended (i.e. inserted inline) in a 50R system. It has been built using E12 values so it will not be exactly 50R but close enough for most applications I think? It would be fine for use inline for attenuating a signal from a 50R sig gen that will be fed to a 50R receiver or spectrum analyser for example. Do you interpret some other role for it?
__________________
Regards, Jeremy G0HZU

Last edited by G0HZU_JMR; 2nd Dec 2017 at 12:32 pm.
G0HZU_JMR is offline  
Old 2nd Dec 2017, 1:00 pm   #29
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,576
Default Re: Attenuators - theory and the design of.

Quote:
Originally Posted by G0HZU_JMR View Post
It has been built using E12 values so it will not be exactly 50R but close enough for most applications I think?
Yes you are quite right, I went through it with the equations I posted, (attached), it was the rounded E12 values that threw me off I think, but like you say close enough for most applications.
Attached Thumbnails
Click image for larger version

Name:	attt.jpg
Views:	24
Size:	24.6 KB
ID:	153258  
Argus25 is offline  
Old 3rd Dec 2017, 12:39 pm   #30
Skywave
Dekatron
 
Skywave's Avatar
 
Join Date: Jun 2006
Location: Chard, Somerset, UK.
Posts: 5,778
Arrow Re: Attenuators - theory and the design of.

Wendy: I see that your attenuator does not give the full range of attenuation in 1 dB increments. (E.g. 4 dB and 6 dB are not available.) Presumably, for you, that limitation was not relevant? However, if you can provide one more relay and a one more switch contact (therefore total of seven steps), you can still achieve the same max. atten. of 60 dB but in 1 dB increments using the following sequence:
1,2,2,5,10,20,20

OTOH, if that seventh step cannot be provided, then if the last 20 dB step is omitted, perhaps an external 20 dB atten. could be fitted when required, thus providing the full 60 dB max. atten.

Just a couple of ideas.

Al.
__________________
"Your idea is crazy! But the question is this: is it crazy enough . . . . to be true?" (Niels Bohr)
Skywave is offline  
Old 3rd Dec 2017, 12:52 pm   #31
Wendymott
Heptode
 
Wendymott's Avatar
 
Join Date: Jun 2013
Location: Huddersfield, West Yorkshire, UK.
Posts: 801
Default Re: Attenuators - theory and the design of.

Hi peeps.. Sorry to "hijack" the original thread........ Al and Argus... I do understand your comments, slapeds wrists on my behalf...I have a few Relays left, and I have not fully built the pcb yet, as I am waiting for the SMD's from China...Jeremy.... Sorry for the typo..got my "dot" in the wrong place..the nice thing is about my modular construction is that any section may be modified without the full thing being modified, When I checked the "prototype" module it was terminated via the Rogol, so 50 ohms was maintained both in and out.. and I will ensure when it is fitted that this is maintained.
It is a good job that I have not completed drilling the front panel yet, to add the 1 db switch..as it is..the prototype panel was looking a bit shabby.
Aghhhh back to the pcb cad ..
__________________
Should get out more.

Regards
Wendy G8BZY

Last edited by Wendymott; 3rd Dec 2017 at 12:55 pm. Reason: Typo
Wendymott is offline  
Old 3rd Dec 2017, 3:44 pm   #32
Skywave
Dekatron
 
Skywave's Avatar
 
Join Date: Jun 2006
Location: Chard, Somerset, UK.
Posts: 5,778
Arrow Re: Attenuators - theory and the design of.

Quote:
Originally Posted by Wendymott View Post
Sorry to 'hijack' the original thread.
O.K. For me, that was not a problem, since this thread - based on comments in my OP - had started to run dry.

Quote:
Originally Posted by Wendymott View Post
The nice thing is about modular construction is that any section may be modified without the full thing being modified.
Exactly - and each module can be tested to check that it's performance is up to the design spec. Also careful attention usually needs to be made to individual module screening and inter-module connections.

Quote:
Originally Posted by Wendymott View Post
The prototype panel was looking a bit shabby.
Hah! A prototype panel that doesn't look a "bit shabby" usually signifies that the design hasn't yet been fully validated!

Reminds me of a quote I heard somewhere from someone -
"You'll know when the design is almost completed: that's when the drawing has so much Tippex on it, the drawing will stand up on its edge on the bench!"

Al.
__________________
"Your idea is crazy! But the question is this: is it crazy enough . . . . to be true?" (Niels Bohr)

Last edited by Skywave; 3rd Dec 2017 at 3:52 pm.
Skywave is offline  
Old 4th Dec 2017, 7:18 am   #33
julie_m
Dekatron
 
Join Date: May 2008
Location: Derby, UK.
Posts: 6,672
Default Re: Attenuators - theory and the design of.

Is there a good reason for not using steps of 1, 2, 4, 8, 16 and 32dB?
__________________
Julie {formerly known AJS_Derby}
julie_m is online now  
Old 4th Dec 2017, 7:47 am   #34
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,576
Default Re: Attenuators - theory and the design of.

[QUOTE=G0HZU_JMR;996130]
Quote:
At work I still use an old 'company' DOS program to design Pi or T attenuators. It dates back to the 1980s but it works really well.
Do you know what version of DOS that runs on ? I really like vintage DOS programs.

It is a bit time consuming using the equations. I left the equations I worked out myself in the unsimplified form with the 4 inside the square root sign to remind me that the solution was a quadratic ! I guess I could write an assembly language program to run in DOS on my vintage computer would be a bit of fun.
Argus25 is offline  
Old 4th Dec 2017, 1:11 pm   #35
Wendymott
Heptode
 
Wendymott's Avatar
 
Join Date: Jun 2013
Location: Huddersfield, West Yorkshire, UK.
Posts: 801
Default Re: Attenuators - theory and the design of.

Julie...... I think that it is the power / voltage steps.... i.e 20 db = X 10 being more convenient.. but I am sure there is a much better explanation.
__________________
Should get out more.

Regards
Wendy G8BZY
Wendymott is offline  
Old 4th Dec 2017, 1:57 pm   #36
Skywave
Dekatron
 
Skywave's Avatar
 
Join Date: Jun 2006
Location: Chard, Somerset, UK.
Posts: 5,778
Arrow Re: Attenuators - theory and the design of.

Quote:
Originally Posted by julie_m View Post
Is there a good reason for not using steps of 1, 2, 4, 8, 16 and 32dB?
Yes: in practice, attempting to achieve an accurate attenuation of more that 20 dB in a simple switched attenuator design is difficult, on account of signal leakage around that 32 dB section.

Al.
__________________
"Your idea is crazy! But the question is this: is it crazy enough . . . . to be true?" (Niels Bohr)
Skywave is offline  
Old 4th Dec 2017, 8:01 pm   #37
G0HZU_JMR
Octode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 1,404
Default Re: Attenuators - theory and the design of.

Quote:
Do you know what version of DOS that runs on ? I really like vintage DOS programs.
I think it runs OK on all Windows OS from about 1990 through to today. I suspect it will work on DOS versions from around the late 1980s onwards? I use it at work on Win7 and I've used it here at home on everything from DOS to Win8. It runs from the command line. Sadly, I can't distribute it because it has the company name on it on the startup screen.

I think it was written in c by a member of staff who left just before I joined (I joined in 1990) and so I'm using a very 'late' version dated 1990 that may have been tweaked by another member of staff. I keep using it because it is very simple to use. It prints out all the usual values in 1dB steps and then if a custom version is needed then just type in the value. eg 20.6dB. It does have a bug on custom values because it lists the resistor values in the wrong order but this is no big deal. It was a general RF design suite and it also designs inductors, programs old school synthesisers via the parallel port, designs PLL loop filters, programs common DACs and has a classic RF signal path analysis built in for gain/NF/IP3 etc. But I just use it for the attenuator design these days.

I wrote my own suite of RF design programs at around the same time and I did all these in Basic and then converted them to DOS programs using QuickBasic when I joined the company in 1990. So I have a load of command line programs here that I still use every day here at home and at work. They are quick and easy to use for impedance matching or for series/parallel conversion or for general LCRQ calculations.

But see below for a screenshot of the attenuator program. I didn't write this one but I still use it regularly for Pi attenuator design.
Attached Thumbnails
Click image for larger version

Name:	atten.gif
Views:	20
Size:	15.1 KB
ID:	153368  
__________________
Regards, Jeremy G0HZU

Last edited by G0HZU_JMR; 4th Dec 2017 at 8:17 pm.
G0HZU_JMR is offline  
Old 4th Dec 2017, 8:51 pm   #38
julie_m
Dekatron
 
Join Date: May 2008
Location: Derby, UK.
Posts: 6,672
Default Re: Attenuators - theory and the design of.

Quote:
Originally Posted by Skywave View Post
Yes: in practice, attempting to achieve an accurate attenuation of more that 20 dB in a simple switched attenuator design is difficult, on account of signal leakage around that 32 dB section.
I see. I just did the maths and 32 dB means only about 1/1590th of the signal is supposed to be getting through. That would be a bit sensitive to leakage, then .....
__________________
Julie {formerly known AJS_Derby}
julie_m is online now  
Old 4th Dec 2017, 10:33 pm   #39
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,576
Default Re: Attenuators - theory and the design of.

Quote:
Originally Posted by G0HZU_JMR View Post

But see below for a screenshot of the attenuator program. I didn't write this one but I still use it regularly for Pi attenuator design.
Thanks for that screenshot !

After I worked out my own equations for the pi attenuator I found them later on the net in different forms, but they give the same result. I didn't do them for the T version. That is a handy DOS utility, saves playing around on the calculator. I built a lot of pi attenuators into male-female BNC fittings, as these are used to calibrate Tek scopes and the originals are really expensive.

If you scroll to the 3rd and second to last page of this article there are photos of the ones I made:

http://worldphaco.com/uploads/TEKTRO...THE_DS1225.pdf
Argus25 is offline  
Old 5th Dec 2017, 12:54 pm   #40
Wendymott
Heptode
 
Wendymott's Avatar
 
Join Date: Jun 2013
Location: Huddersfield, West Yorkshire, UK.
Posts: 801
Default Re: Attenuators - theory and the design of.

Just to add my "two pennorth"...I attach the pcb artwork for my X 6 relay switched attenuator... I added Back emf diodes pads just in case they are needed if switching with a transistor.
There will be a screen around the pcb for leakage reduction eventually. Just to put minds at rest I have switched one of the 20 dB attens to 1 dB... but thats easy with smd.
Attached Thumbnails
Click image for larger version

Name:	Relay switched atten  1.jpg
Views:	25
Size:	149.4 KB
ID:	153420  
__________________
Should get out more.

Regards
Wendy G8BZY
Wendymott is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 8:41 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2018, vBulletin Solutions, Inc.
Copyright ©2002 - 2018, Paul Stenning.