Switched Attenuator

[dominicbeesley]

Hello all,

For testing radios and tellys and for a lot of my other experiments an attenuator is a useful piece of equipment. I used to have a switched attenuator that I made out of pieces of PCB and lots of switches http://dds-60-siggen.blogspot.co.uk/...ttenuator.html. I never got round to putting it in a proper box and it suffered from reliability problems - dodgy soldering and the switches I used were a bit naff. It finally died a death over winter when it got damp in the workshop and has developed verdigris due to my not properly washing the flux off.

A few weeks ago I decided to upgrade and I bought a switched attenuator from ebay. This should allow from 0 to -191dB of attenuation with 8 attenuators of 1,2,4,8,16,32,64 and 64 dB each. It was made by Texscan and seems to have been something to do with cable TV. It has two TNC (thanks to the forum for helping me identify these) connectors and 16 pins for activating the attenuators. The attenuators are in sealed units inside but feature some kind of relay.

After a bit of playing around I found that the advert on ebay was not 100% correct (it said it was TTL compatible) but that the attenuator relays need about 15V to switch over. Unfortunately, I'd planned to run all of this off a 5V supply (I've got an old PC PSU that I plan to use to power my homebrew bench equipment)...Fortunately I'd just bought a hand full of DC to DC converters (xtw-sy-8) which will step up from 3-30V to 5-35V at a decent power.

Instead of buying a bunch of switches I've decided to have a microcontroller, two push buttons and an LED display of the attenuation value. I've designed a microcontroller program that allows you to press up/down or hold for fast increment/decrement. This doesn't switch immediately but waits until no button is pressed for around 1/2 a second before setting the relays. Also when setting the relays it first switches in attenuators before switching them out and it switches the lower valued ones before the others. I've done this to save the relays from excessive switching, to make sure there are fewer power spikes and also to make sure that the attenuation never goes lower than it needs to when switching between two values - to avoid overloading any device under test.

Most of the rest of the circuitry is just dealing with switching the relays. The resistors between collector and load are bypassed with capacitors. This means that when a relay is turned on initially it will see the full 15V supply but this will drop off as the capacitor charges and then the hold current will be through a 2k2 resistor. This should save some current but provide enough of a kick to shift the relay.

I knocked up a circuit diagram and PCB layout in Kicad and then made the PCB this evening. I made another a few evenings ago which was aborted - the tracks were too small and it didn't come out very well with missing portions of track and bridged tracks. This time I've used bigger pads, thicker tracks and not tried to route tracks between pins. It was still a challenge to solder though!

This is the first time I've used my new laser printer to develop a PCB. It still requires several over-prints to get dark enough for a good contrast on the acetates I've used but is a lot less fiddly and less prone to fingers smudging the print!

I drilled the holes and tinned the PCB with plenty of flux then gave it all a thorough wash before fitting the components. It worked almost first time - one LED segment was out but that proved to be a very small solder bridge between one track and the ground.

I'll post up details of the case when I get round to making it....I _will_ get round to making a box for this one!

If anyone's interested I'll post up circuit diagrams and PIC source code.

D

[Wendymott]

Hi Dominic. Looks good to me, as a veteran of home pcb manufacture. Just one tip if I may be so bold. When I print artworks, I always do two onto the same clear acetate. Then lay one on top of the other and tape.. I have tried various other methods over the years and find clear acetate x 2 the best. However there are other threads using a glossy paper and a hot Flat iron, but my tests were not too good to make me change my method.
Nice work though.

[kellys_eye]

A very nice construction indeed.

I would be grateful if you would publish the schematic and code as I have an application for the same.

What impedance is the attenuator? I'm assuming 75ohms given the TV application?

I've also had problems getting accurate resistance values (having to use the E96 range to get any reasonable accuracy) to get up to 16dB which is as high as I want to use practically speaking but I still have four such sections for a maximum of 90dB attenuation.

The digital readout? Is that a device readily available? I have quite a number of four digit parallel-addressed dot-matrix LED devices that could be used with some code modification.

Nice pcb result too. I use the laser-transfer technique and get similar results.

[dominicbeesley]

Thanks Wendy,

I have done similar to that - two passes through the printer onto the same acetate. I tried doing two layers but found that lining them up was difficult and that it tended to blur towards the edges due to the thickness of the acetate. Over printing works ok just leave enough time between passes and make sure its lined up very accurately in the paper guides. I tried an iron - probably cheaper but I just ended up with a mess of broken tracks.

I'd like to make a vacuum bag at some point rather than balancing lots of weights on the acetate to hold it flat to the pcb...

I use some decent ready to use FR4 PCB material with photoresist already on. It's very tough PCB and, though its expensive, is worthwhile as its stronger than the usual yellow stuff and the tracks don't peel off easily which I find a problem if I give it too much welly with the soldering iron. I have a can of spray on resist but not tried it yet.

I use an ultra violet LED and a quick guess of exposure time (5mins30secs) this time. It has never failed and though the LED was expensive (~£20) its worked well and is not too harsh on the eyes like some UV tubes.

Hi Kelly,

I've not checked the impedance properly but it was advertised as 50ohm and has 50ohm connectors so I'm hoping 50. If not I'll add some pads either end.

The display is a NOS one like the ones in old HP calculators - I bought a pack of 5 recently and I suspect I'll get more as they're quite useful and cute! I got them here http://shop.pimoroni.com/products/re...it-led-display there are other places selling them.

I'll gather together the circuits and code and post them up tomorrow.

Cheers for the kind words

D

[kellys_eye]

Dominic - don't expect much from the spray-on UV resist. Getting an even coat is very important but also very difficult to achieve and any variation affects the exposure time for good results.

My own experience of the stuff is more than disappointing - I wouldn't use it again.

Are you using a 'single' UV LED? (hence the high price). I've seen many projects using arrays of 5mm devices but never heard of a single LED being used.

[Wendymott]

Hi Dominic. Ok about registration, I use a Samsung Mono laser printer but the paper guides.. lets say are "variable", thus I do the two layer approach.
I am very fortunate to have a 3M's UV light box, from the old company, when it was regarded as obsolete. My exp time is 45 secs.. I too use FR4 pre coated board from CPC, the "budget" variety, thus if I make an error its not too costly.
Kellys_eye, you must have a "diffused" light source, otherwise boards would be over exposed in some places and under in others..
I bought some GU10 U/V lamps, but discarded them, as the distance from the board was too great for a short exposure time.

[kellys_eye]

Hi Wendy,

I know and understand the principle of UV exposure and use a converted flatbed scanner with UV tubes fitted (homebrew) as my own device however I was intrigued by Dominics assertion that he paid ~£20 for 'an LED' which sort of implies a single device.....

The UV exposure units in recent DIY articles tend to rely on arrays of individual UV LEDs properly spaced to ensure even coverage but I can't imagine anyone paying ~£20 per LED for such a solution!!!!

From exposure times I'm starting to suspect that Dominic does use a single UV LED but at a distance - to avoid the errors you mention - hance his quoted exposure times of 5-1/2 minutes.

Just intrigued by the make/model of UV LED he's using....

[dominicbeesley]

Hi both,

There's a bit more about it here: http://dds-60-siggen.blogspot.co.uk/...rboard_17.html I bought a single 5W luxeon and mounted it on the inside of an old computer power supply housing. It gives a very nice even pool of light and gives off UVA rather than the quite hard UV that some UV tubes give off which is less harmful to skin and eyes though I still avoid looking at it for long periods. I also like the fact that it is a bit slower I quite often don't even bother with a timer and just count the five minutes or so down in my head.

I've just had a look and RS no longer stock the same Luxeon Star but searching "167-8962" brings up a near equivalent. I'm not sure of the exact wavelength or intensity pattern as I've managed to lose the datasheet. However the near equivalent would probably do the same job! I tried the live chat thing on RS to try and find out the old datasheet / spec but they were singularly unhelpful!

I've not found a problem with the single LED approach and have been told by those who know that it won't work...but it does! I've not really experimented with the exposure time but I suspect it's over exposing the bit directly under the light a little but not by a huge amount. I think its a "batwing" led so it is designed to throw a nice even pool of light.

The advantage of a single "point source" is that if the mask is not perfectly placed on the board it still casts a fairly good contrasty shadow and the error is a slight displacement whereas with a diffuse light source any lifting results in blurring.

I did try doing a PCB in my EPROM blower try but that seemed to over expose and break through the print too easily and narrower tracks were washed away - I probably over exposed it though as I gave it 3 minutes in there which may have been way too much.

The limiting factor at the moment for me on tracks is:
- spacing: too close and any dust / muck on the transparency bridges tracks
- spacing: too close and solder bridges are easily formed when tinning
- track width: too narrow and they can lift during soldering
- track width: too narrow and the print tends to flake / get rubbed off the transparencies
- pads: drilling becomes too difficult

I've manage to place tracks between pins / pads in the past on standard dip packages but this requires a steady hand/eye when soldering and very careful drilling.

I use a big power craft drill press with proper pcb bits at the highest speed it will go. The bearing is not bad in the drill press but it does shake a little and lining up can be difficult with very small pads so I tend to take the easy option and stick to single sided, fat pads and tracks...

I've been pleasantly surprised with my cheapo HP mono laser two passes through and if I'm careful with the guides it will overprint with no registration errors that I can see with the eye.

[dominicbeesley]

As promised, I've gathered together the files and corrected the mistakes (or at least the ones I've spotted).

I've added the zips at the end of a blog post here: http://dds-60-siggen.blogspot.co.uk/...ttenuator.html

Dom