digital attenuator

[grawk]

I'd definitely be out at the RK50 price point.

[Kerry]

The plan is to run the primary assembly using the reasonably priced resistors. It will likely be possible to also have a portion done with a different BOM.

Edited April 19, 2011 by Kerry

[Kerry]

PLEASE SEE POST FOR THE LATEST COUNTS.

Here is an updated list. Seems to be stabilizing now.

Name - Assembled - Bare Boards

grawk - 4 - 0

Driftwood - 4 - 2

luvdunhill - 0 - 2

ujamerstand - 4 - 0

deepak - 2 - 2

cetoole - 0 - 10

manaox2 - 2 - 0

el_doug - 2 - 0

DouglasQuaid - 2 - 2

jgazal - 2 - 0

Lil Knight - 4 - 0

nattonrice - 0 - 4

kevin gilmore - 4 - 4

looser101 - 4 - 0

chinsettawong - 2 - 0

Kerry - 2 - 4

Fing - 4 - 0

blubliss - 4 - 0

n3rdling - 4 - 0

guzziguy - 2 - 0

Pars - 0 - 2

Nebby - 4 - 0

Flyingsparks - 0 - 4

morphsci - 4 - 0

dwhat - 4 - 0

minivan - 2 - 0

Craig Sawyers - 2 - 4

MASantos - 2 - 0

Vortex - 0 - 10

Beefy - 0 - 4

Samuel - 2 - 0

deadlylover - 0 - 6

MrMajestic2 - 0 - 4

tcpoint - 0 - 2

spritzer - 4 - 0

Shaman - 0 - 2

Horio - 0 - 2

TOTAL - 76 - 70

Edited May 11, 2011 by Kerry

[kevin gilmore]

success writing 8 bit words to the relays.

be careful that there are 2 different kinds of pcf8574 with different addresses

i ordered pcf8754 (address 64) but was shipped pcf8574A (address 112 )

[Craig Sawyers]

I've been working with some z-foils as part of another project - free issued by the company I'm working with on this one. Very interesting. I've been using both SM and leaded versions.

These are nominally 1% tolerance, but actually measure better than 0.1% (ie better than my Fluke 87V can resolve) - which is good. These came from Charcroft, from whom I am still awaiting a quote for the attenuator resistors.

The SM ones have the resistance value printed on the top of the expoxy case. The leaded ones have absolutely no markings at all - the only way to check value is with a DVM.

I must admit it is a bit daunting to look at a very small pile of resistors and think of the monetary value of them....

[kevin gilmore]

added resistor pack to increase drive current to relays for 3.3v

tested with 3.3 volts drive and 5v and 12v relays

tested with 5 volts drive and 5v and 12v relays

would definitely work with 24v relays.

three versions now due to possible difficulties in getting

surface mount resistors all from one manufacturer.

2 different sizes of smt and .4 inch thru hole

http://gilmore.chem.northwestern.edu/attenuatorsmt16net.pdf

2 different sizes of thru hole only, .2 and .4 inch

on this one, the 2 extra ground layers are probably unnecessary

http://gilmore.chem.northwestern.edu/attenuatorsmt17net.pdf

edit: middle one probably unnecessary

but wait: there is MOAR

3 different sizes of thru hole only, .15, .2 and .4 inch

this runs the complete list of thru hole resistors from 6 cents each all the way to $36 each.

32 resistors needed for stereo, 64 for balanced

http://gilmore.chem.northwestern.edu/attenuatorsmt18net.pdf

actual working touchpanel controling the relays

(and yes i know now that one of the buttons is 1 pixel off)

had to turn it sideways because if you have a volume control with resolution of 256,

then you NEED 256 pixels wide

http://gilmore.chem.northwestern.edu/touchpanel2.jpg

Edited April 23, 2011 by kevin gilmore

[ujamerstand]

Love the touchscreen. Is it capacitive type or resistive type touchscreen? Will it still work if you place it behind a thin slab of plexiglass? (let's say 1~2mm?)

[kevin gilmore]

resistive! Nope, will not work behind plexiglass. Lets be serious here...

resolution of one pixel. 320 x 240. Not gonna happen behind glass or plexiglass.

You find me the apple display and touchscreen as a seperately purchasable item, and

we will talk about it. This display and the cpu that drives it is a one piece item

that is <$100.

NOW MOAR MOAR BETTER. but 6 layers better idea for ground under the smt.

http://gilmore.chem.northwestern.edu/attenuatorsmt18net.pdf

kerry wanted the 1206 smt resistors in addition, and i found a way to

cram those in too. So this should be it.

Edited April 24, 2011 by kevin gilmore

[luvdunhill]

Let's talk!

http://www.sparkfun.com/products/9188

Edited April 24, 2011 by luvdunhill

[kevin gilmore]

i think you should handle the programming of this one.

Whatever cpu you come up with just make sure it has an i2c

output to drive the attenuator.

gotta go watch Dr. Who premiere. will fix too small silkscreen

tomorrow.

Edited April 24, 2011 by kevin gilmore

[luvdunhill]

i think you should handle the programming of this one.

Whatever cpu you come up with just make sure it has an i2c

output to drive the attenuator.

I was just being an ass. I think the part you choose is really the best option out there.

[Craig Sawyers]

gotta go watch Dr. Who premiere.

Good man! It sure is a helluva first (part one of two) episode for this series.

[Craig Sawyers]

kerry wanted the 1206 smt resistors in addition, and i found a way to

cram those in too. So this should be it.

OK - Charcroft said that they had good stock on the untrimmed Vishay 1206 Z-foil. I've just mailed them to ask for the quote to include 1206 too. Larger physical size resistors would have a lead time. After talking to them, it turns out that a major driver of cost is tolerance - so I've asked for 1% - the 1% ones that I have measured are all better than 0.1% (0.009dB nominal, not counting statistics. Even if you do, it will be a rootN thing. So maximum number of resistors = 17, so the maximum error due to 0.1% tolerance is <0.04dB. At -128dB, so who cares.

[Craig Sawyers]

Hokay - I got some numbers from Charcroft. Note that 6R34 and 169K are not available, since Vishay do not supply untrimmed stock to cover those values. The costs below refer to the VSMP series http://www.charcroft.com/site/pdf/data/Charcroft.VSMP.04.09.Is4.pdf and the SCAR series http://www.charcroft.com/site/pdf/data/Charcroft.SCAR.04.09.Is2.pdf both in 2512 size. Tolerance is 1%, and MOQ is 10 off each value. Tax at 20% would have to be added too.

Delivery is around 10-12 weeks.

255R to 16K9, £9.88 each

38K3, £10.52 each

82K5, £11.16 each

The 1206 size VSMP are cheaper, but restricted to <25K

255R to 3K65, £6.57

6k04 to 16K9, £8.30

That puts the total cost (including tax) for a stereo attenuator using 2512 size at £360, or balanced at twice that. The 6R34 and 196K would have to come from elsewhere.

The 1206 option is cheaper at £235 for stereo, but the 6R34, 38K3, 82K5 and 169K would have be alternative resistors.

Multiply by whatever factor is appropriate for local currencies.

What is the interest in this level of crazyness? Either on list or via PM.

[luvdunhill]

Craig, sorry for my laziness, but what is the resulting impedance of the config you are proposing?

[digger945]

10k.

169k for the half db step and 6R34 for the 64db step.

[Craig Sawyers]

10k.

169k for the half db step and 6R34 for the 64db step.

Check.

[MASantos]

That's quite a sum for resistors! Is there anything not as crazy as z foils but above the "regular" vishay dales?

[Shaman]

I'd say the Vishay MELF SMDs that Max suggested are our best bet.

We can only go as low as 24k though.

[Kerry]

It looks like the the assembled boards will come in between $76 (1% MELF) and $95 (.1% IRC Thin Film). Both of these boards are the 24K versions.

Right now I am leaning to the .1% IRC, but I'm open. I'd also have to check if PCBNet can do the assembly with the MELF resistors onto the 1206 pads.

I sill need to to get some final pricing based on current quantities, but I think I am close.

Regarding the foil resistors that Craig is looking into, there may be an option to have a small number of boards assembled with this BOM. I will need to check this once resistor choice is finalized.

@Craig - You might want to look into 24K version. There seems to be better availability, even over a 25K version.

I am going to review the traces on the board to make sure it looks OK, but would like someone else to jump in here and lend a hand. Can't be too careful

Edited April 30, 2011 by Kerry

[Beefy]

$95 (.1% IRC Thin Film). Both of these boards are the 24K versions.

Sounds like an excellent option for me, as a board-only participant.

[El_Doug]

Regarding the foil resistors that Craig is looking into, there may be an option to have a small number of boards assembled with this BOM. I will need to check this once resistor choice is finalized.

That would make my day

[deepak]

I would be interested in a foil version with 1 db steps to reduce the blow to the wallet if anyone else was thinking about madness

[Craig Sawyers]

@Craig - You might want to look into 24K version. There seems to be better availability, even over a 25K version.

I am going to review the traces on the board to make sure it looks OK, but would like someone else to jump in here and lend a hand. Can't be too careful

Problem is that the higest value in z-foil of any description is 50k - so even a 10k version of the attenuator is missing the 169k. Upping the attenuator resistance just makes the availability of z-foils worse. I suspect that <50k is due to the combination of resistive material and the complexity of the metal pattern, and it does seem a hard limit for z-foil.

[Kerry]

Sorry Craig. I saw that in your earlier post, but my mind isn't the steel trap it used to be