[jules]

oh forum,

I'm working on populating (I understand this is the right term) an Elma attenuator with the 90+ resistors needed for a 20k switch. I have a very workable spreadsheet that makes the task easy but there is some room for interpretation of its use. My own feeling is that the aim should be to get as close as possible to even (~2dB) per step regardless of the combination of resistors [which may or may not total exactly 20K]. Has anyone else worked through this problem?

A secondary question is; IS it ideal to have uniform 2dB steps throughout the range (particularly in the most used section). Resistor values tend to lead to larger steps at the low end though it is posible to taper these from about 6dB down to 2dB over three or four steps.

Jules

ps spreadsheet available if you are interested

[SamL]

Have a look at Goldpoint diy reference web site.
http://www.goldpt.com/diy.html

They have value for Series, Ladder and Shunt Attenuators.

Have fun,
Sam

[jules]

Thanks Sam,

yes, they (Goldpoint) do have the resistor values but they don't show how close these vlaues are to giving a smooth progression. Comparing the Goldpoint  values to a set from the spreadsheet I mentioned (which shows ideal values, actual values and actual steps achieved with vishay resistors available from Mouser) the Goldpoint set looks a bit rough although I can't really be sure without seeing their values on the spreadsheet.


The spreadsheet makes it possible to specify desired levels of attenuation for each step. It automatically gives the actual resistor values/part numbers in response.

Jules

[AKSA]

Yes Jules,

Go for 2dB steps nominal across the entire range.

It is ideal to have 1.5dB steps in the most used part of the scale, but the problem is you never quite know where this point will be, so using 2dB - just discernible in a blind listening test - is the best option.

This will of course vary slightly up and down from 2dB steps because of the E24 (or E96) value range of resistors you are using.

Cheers,

Hugh

[jules]

thanks Hugh,

I might try putting a few larger (tapering up) values in the 3 oclock to 6 oclock sector with a view to lower values between 8 and 2 oclock particularly but you have answered the key part of my question as cogently as ever.

Jules

[Greg Erskine]

Hey jules,

I've just gone through this same exercise. I'd be a little careful of the spreadsheet if its the one I looked at. It worked on a preferrd value list of Vishay Dale resistors that caused some nasty hickups in the steps. Mouser actually have much bigger range of resistors than indicated in the spreadsheet.

I worked off the Goldpoint values, but I didn't actually verify these values were correct.  They do document their standard jump between each step in numerous locations on their web site.

I can email you the 20K and 25K Mouser Vishay Dale RN60D BOMs that can be easierly imported into Mouser projects.

BTW: If you find a problem with Goldpoint's 20K and 25K values please let me know.

Thanks
Greg

[jules]

why do these things get more complicated?

I've checked the Goldpoint site and from what I can see their "goldpoint taper" has to be an ideal rather than an actuality. It's an interesting taper but it's so regular that I can't see how it can be possible particularly when it apparently applies to 7 different ladders (from 5k to 250k). To be that precise would seem to call for a vast range of custom made resistors.

I've found that it is possible to devise a series of values that should be fine sonically (using Vishays from the mouser catalogue) but it's a bit like juggling 24 balls at once to get there.

Anyone own a Goldpoint? It looks to me as though they have a fairly rapid initial [vol] increase over the first third of the scale (low vol and where the steps might  just be discernable), which then stays at a very regular increase till near the end of the scale (high vol)

Jules

[kyrill]

hi Jules
What I would do is follow a choosen theory but let the final attenuator be shaped by your ear. So buy the R's from a factory with a broader range than you need, build the thing but keep the project open in your mind when you finished building it.
The proof is listening not theory. You will find one or more discernable "hickups" but at precisely known points. The points show you which resistor need to be replaced or when it is a whole range, you can play with the shunt R
To make it perfect in the cheapest way buy cheap 1% resistors and build the anticipated critical range with those. When your ears are satisfied replace them with the better sounding ones. a lot of work, but the result is for life.

Hugh is right that 1,5 db would be better. But not for the whole range. In the most critical range 1db is even better than 1,5.

I learned with my digital attenuator you can hear 0.2  0.3 db (!)  differences
In the critical range steps of 2 db too are too huge, at least for my ears.
Succes
Sounds easier than playing with metal-leaves

[JohnR]

I skimmed thread, so pardon if this was said. It's often considered a good idea to have big steps near the bottom of the range. DACT do that, and I think their system works fairly well. Here are the attenuation steps they use:

0,2,4,6,8,10,12,14,16,18,20,22,24,26,28, 30,32,34,38,42,46,50,60,infinite

(From: http://www.dact.com/html/attenuator_datasheet.html)

Personally I've not found 2dB to be too large a step. To each etc

[jules]

many thanks to everyone for the help ....

The DACT range is interesting and similar to the Goldpoint in that they both use larger steps at low volume and even 2 dB steps through most of the range. I still can't see how they can achieve such precision given that there can be gaps in the range of available resistors. I suppose that at least with pre-assembled types the volume of production allows them to order a number of one-off values.

As Greg says, Mouser has more Vishays than shown in their catalogue (the basis of the spreadsheet) but the unlisted ones are problematic in that they might need to be specially ordered/in lots of 100/expensive/etc.

Roderstein resistors from M. Percy have a much bigger range of values and make it easier to create something with desirable steps.

Step choice is interesting in itself. Given that the total has to come to 60dB and that it is desirable to have some steps of less than 2 dB it's necessary to have some bigger steps somewhere in the range. My own choice has been to arrange these at either end of the scale thus 4.8, 3.1, 2.9, 1.9, 1.3, 1.9, 1.6, 1.9, 2.1, 2.1, 2.1, 2.1, 2.1, 2.1, 2.1, 2.1, 2.9, 3.9, 4.0, 4.2, 6.0. [the starting point here is zero volume]. As you can see it is not smooth and I might still fiddle with it a bit both for overall style and for individual steps. This is a Roderstein set, a Vishay set would be more lumpy. I did think about a Frankenstein using mixed resistors and it might be a great way to compare their influence but sanity prevailed on that one and it stayed as an idea only.

Kyrill, thanks for the specific advice based on experience. Six months from now when you look at your wood work I'm sure it will look fine as an overall picture where now you are immersed in the detail

regards

Jules

[stvnharr]

Jules,
You are making this all too complicated, complex, and hand wringing!
I use a goldpoint 20k shunt attenuator, with the nominal 2db. steps.  Everything is just fine!!!!!  
I personally am just fine with the goldpoint steps.
What you really need to do is just make your attenuator and find your most usable range.   I think you will find that you likely will only be using about half of the steps, at most.   This is where you can tweak a little and maybe lower some steps if you wish.  
Just remember, with the Elma switch you only have 23 positions, no matter what.  Larger than normal steps, you could go 5db quite easily, on the lower 3-4 steps, that will hardly ever be used, would be a good way to gain some additional flexibility.

[jules]

Thanks stvnharr and again thanks to everyone else,

there have been some moments of hand wringing here but only small ones. On the whole it's been an interesting learning experience and I can strongly recommend the spreadsheet as a means to selecting resistors.

I will report back when it's all working but I repeat the offer of the spreadsheet to anyone who is trying to assemble a set of resistors for an attenuator. I now have two forms of it. One uses all commonly available Vishay-Dale resistors for switches of any value. The second uses Roderstein resistors and is only useful up to 24k ohm. [self modified and I'm not a spreadsheet expert]

Jules

[ergo22]

but it's somewhat relevant question....

Do you guys know any solutions for stepped attenuator remote control function. There are two that I'm aware of but both have a quite high price

http://www.bentaudio.com/index2.html

http://www.thel-audioworld.de/bauteile/bauteile.htm
Hmm they had a kit but I can't see it on their page anymore...

Is there any fully DIY solutions out there? I would not mind building the whole thing

Regards,
Ergo

[stvnharr]

Ergo,
I was all set to get the Bent unit when I was contemplating the stepped attenuator question as I was rather spoiled by having remote control.
However, once I built the attenuator and put it into use, I found I didn't change the volume very much, and it was a quite simple matter to just get up and walk a few feet to do so when necessary.
Thus, the remote became a rather expensive, and less necessary proposition than originally thought.
Also, I wasn't sure if the Bent couplers would really work.  Though my attenuator uses an Elma switch, the Goldpoint pc boards are different from the Dact boards.
However, the Bent unit is a good solution to having a remote, and coupling issues can likely be resolved.

Steve

[SamL]

Hay Jules,

Have you finish populating your Elma stepped attenuator? Well, I will be receiving mine soon. Someone ordered from the group buy but did not paid up so I took it.

Jules & Steve,
How does the Elma sound? What brand of resistor and what attenuation steps did you all end up using? A lot of people prefer Dale resistor but I come across others who think the cheaper BC resistor is better. I currently havd a 10K DACT and this Elma better be good if not better. Will go for 20 or 25K since this is the ideal value for GK1.

Cheers,
Sam

[jules]

SamL,

I'm literally in the middle of populating my ladder right now so I can't report on the sound. I understand that the Vishay Dale are highly recommended and Hugh has also backed them somewhere. I used Roderstein partly because they have a huge range of values [which made setting the stpes easier] and partly because negotiatiing with Michael Percy Audio is slightly more personal than Mouser [who are rather larger] although Mouser does have a good "paste" system for ordering and I think Greg Erskine got his resistors from there.  For a report on Dale you could pm Greg.

If you need the spreadsheet I can send it to you. It's specific for Dale values but can be changed if you have the patience.


jules

[stvnharr]

The Elma switch had better NOT have a sound as it's just metal.  The signal is in the pc traces and the resistors.

I have Vishay resistors, Vishay Dale in the ground positions and Vishay S102 in the shunt position.  Maybe some day I'll put S102's in the most used positions on the ground, but it's an expensive proposition.
I don't think it really matters too much what kind of resistors one uses, Dale, Holco, Roederstein, etc.  The benefit is in the design itself of the attenuator.

[SamL]

SamL,

I'm literally in the middle of populating my ladder right now so I can't report on the sound. I understand that the Vishay Dale are highly recommended and Hugh has also backed them somewhere. I used Roderstein partly because they have a huge range of values [which made setting the stpes easier] and partly because negotiatiing with Michael Percy Audio is slightly more personal than Mouser [who are rather larger] although Mouser does have a good "paste" system for ordering.

Most appriciated if you can send me the spreadsheet.
email = samshlim at yahoo.co.nz
replace at with @ and removed all space.
I might go with roderstein too as I have a few bits and piece to get from MPercy for my aksa55. Will see

Cheers,
Sam

[jules]

SamL,

hope you'll excuse a slight delay on this. I'm not at home at the moment and don't have the info on this computer. I'll send you the original form of the sheet and a 20k Roderstein version. I might as well send you a list of the resistors that I extracted from the spreadsheet. You do have options on the way you balance the thing  and you can figure out what suits you when you get it.

jules

ps I'll be interested if you achieve a GK-1, Aksa 55 combined as I think you said you might.

[SamL]

Hi Jules,

No hurry here. I've ruin my soldering iron and yet to get a replacement. It will 2005 before I can do anything.

Don't think I can fit aksa55 and GK1M into a single case. It will be a very large enclosure and become impractical. I was thinking of having aksa55 & TLP as integrated to drive my rear chn. The GK1 is to go with my aksa100 for the front chn. That was the plan but right now I might spend the TLP money on sub amp.
Thinking of using the aksa100 + Elma for the front, aksa55 + DACT for the center and I have a useable 3.1HT system. Build the rest like GK1, intergrated, & maybe Hugh future HT later. Well, I am juggling beteen $/time/audio quality/WAF. Not easy but that's the fun of diy

Have fun,
Sam