[Syzygies]

Thanks. I just reread this thread, and I think I have a handle on this now. I ordered one of these beasts.

People should look at How Stepped Attenuators Work.

Do the back two sections come off? Could one jury-rig a second switch from them? For economy, one might want to consider shunt or series attenuators, using two sections of this beast and half as many resistors. Would also save space.

In another thread, amb and I discussed the DACT atten with the M³ amp. He raised concerns about interconnect length, interference, working against benefits of the atten. More of an issue with a monster like this.

I see this primarily as the "control" in audio experiments. Can one populate it with $20 to $30 of resistors, and get better sound than using a pot? (Probably; hard to beat the ladder design.) If I were going to spring for $1 resistors, I'd also spring for a better, smaller switch.

As for the boards, shouldn't they be two-sided with a ground plane? This whole idea sinks or swims on beating the issues that concern amb. We'd also want to shield the interconnect...

How do people want to use this? It would be nice if the finished product was interchangeable with an Alps Blue pot, dropping into the same holes in an amp board (and leaning way forward!).

 

[Syzygies]

I'd argue that the resistors in a stepped attenuator should be the same as the resistors used elsewhere in an amp. If better resistors improve the atten, then, gee, we're 80% of the way there, why not upgrade the rest of the amp, too?

Whatever two resistors per channel get used in the ladder, they're playing a very similar role to, say, R1 and R2 in the PIMETA and similarly labeled amps.

I'll ask it, why not have a ladder stepped attenuator replace R1 and R2?

 

[Porksoda]

Quote:

Originally Posted by Syzygies I'll ask it, why not have a ladder stepped attenuator replace R1 and R2?

Because you'll end up with either varying input impedence to the op-amp, or varying pass frequency and amplifier input impedence.

If you can live with varying op-amp input impedence, then there's no reason not to...

 

[Syzygies]

Quote:

Originally Posted by Porksoda Because you'll end up with either varying input impedence to the op-amp, or varying pass frequency and amplifier input impedence.

Let me put this better: There is a Thevenin-equivalent circuit to each ladder step followed by R1, R2. One could design this circuit into the ladder steps, skipping R1, R2. Doing this sort of thing is a luxury that designing a ladder presents, that a standard log pot doesn't present. Each ladder step can be anything we like, taking into account exactly your concerns. This is a luxury I haven't seen anyone take, so I was asking why.

Thevenin-equivalent circuits have different current patterns, however. In the standard log pot, R1, R2 arrangement, only a trickle of current is passing near the op amp input. The Thevinin-equivalent circuit would route more current near the op amp input. I'm not sure which the physics of resistor noise prefers, if this matters.

Or one could do better, knowing one was skipping R1, R2. Their values only approximately stay out of the kind of trouble you suggest. Or am I wrong?

 

[Porksoda]

I'll have to sit down and draw out some diagrams and figure out how it would work out as far as keeping pass frequency, input impedence, etc. constant.

As far as the switch itself goes, mine arrived today. Looking at it, I don't really think the boards are necessary. Each switch position has a through-hole in the lug. It seems to me it wouldn't be all that bad to solder the resistors to the through-holes between the decks. This is how normal ladder stepped attenuators are done, yes? The only problem is the ground end of each channel will be all of the free ends of the "bottom" resistors soldered together. Not exceptionally neat or sturdy, but it will work

The other benefit of this is that you could cut/sand off most of the boards, and make the attenuator considerably smaller.

 

[motherone]

I bought one of these switches off of ebay for a steal a few months ago (I think $12 or so shipped). I've got my resistors as well (based on the goldpoint curve), all RN60 Dale's.

I just whipped out the multimeter and it looks like it's wired like this (looking at the switch from the front, with the "large" holed board (8 pos) on top and the small holed board (18 pos) on bottom).

Bottom left: "input" to switch
Top left to right: pos1-17
Bottom right to left: pos18-23

So you multiply this by 4 decks. From left to right, it would correspond to the first deck being Rg for channel 1, then the second being Rin for channel 1, the third being Rg for channel 2 and the fourth being Rin for channel 2.

On the "large hole board", looking at it with the large holes facing you and the switch sideways, it looks like (from right to left): input, output, ground, input, output, ground.

I'm not 100% sure on these.. I'm going to start playing with mine soon and see how it turns out

Please feel free to correct me if any of this is wrong.

 

[motherone]

By the way, I would go with RN55 resistors if you haven't purchased yours yet. The RN60's won't fit without creative mounting on the smaller holed board.

 

[Syzygies]

So my switch finally came. It'd been so long, I forgot what the box was.

After a few moments of "What was I thinking?!" I took it apart.

Vishay Dale RN55 resistors fit beautifully through the holes in the boards that come with the switch.

There are various stepped attenuator designs that could take advantage of this observation. I'm still waiting for the other shoe to drop on which one to go with...

The boards are 1 3/4" square. I'd desolder the now useless pins extending from the board, and reassemble the 3" deep switch into a shorter configuration. One could imagine cutting the boards down in size, but it'd only save 1/4" each way, not worth it unless one truly needs the space.

 

[Edwood]

LOL, what are you doing?

-Ed

 

[ericj]

I have just (nearly) finished some work with my Alps rotary switch, so, I'm bumping this thread, and have some comments.

Long & short: very easy to build these without the little boards that Triode doesn't have for sale anymore.

There is no need to go printing any boards.

I used $1.60 worth of phenolic perfboard (pad per hole) and some ingenuity.

The "in" signal goes to the wipers on the first and last decks, outs from the inner wipers, and ground is in the center.

If you keep the ground lines separated, you can potentially use this attenuator in a dual mono or "balanced" input amp in the future.

I used the Goldpoint 50k values, leaving out the 1st value since this switch has only 23 positions, and the very bottom notch seemed the most expendable.

I'm not quite finished because I'm missing two values of resistors, so i haven't soldered anything up, but I thought I'd share my (near) success with anyone who bought one of these and ended up not using it.

First, to dispel some concerns:

Quote:

Originally Posted by z2trillion lol, me too. I'm thinking about using the 3 cent koa speer resistors mouser offers.

Unfortunately you can't. There isn't enough granularity in the values of KOA resistors available from mouser.

The cheapest you can do for many of the needed values in metal film from Mouser is Xicon, and then you're buying baggies of 10 for 90 cents, and a lot of those values you need only two of.

That's what i did, and it sucked. I spent around $25 on relatively cheap resistors.

And even then, I'm sitting here with an unfinished attenuator in a state of total disbelief that i somehow failed to order 16.0k and 4.99k.

It may turn out that Dale RN55's bought one at a time wouldn't have been substantially more expensive. I'm not going to look them all up to find out.


Quote:

Originally Posted by MisterX It would be a Major PITA to build one one without the boards. It wouldn't be so bad if the fiber board sections (or decks if you will) were not so big but they are right right in the way if you want to jumper across the solder rings.

It turns out it's not so bad.

The pin spacing crossways is slightly off of .1", which is only a little annoying. I hardly notice it on the side with only 6 pins on each deck, and on the other side they only fan out a little.

It's the spacing between the decks that causes problems.

The gap in the center is too narrow, and the gaps between the 1st and 2nd and 3rd and 4th decks are slightly too large.

From the factory, the longer plastic spacers are approximately 14.9mm according to my cheezy digital calipers.

I filed all four of them down to approx. 14.0mm and now the inner and outer pairs of decks can straddle a normal .1 spaced perfboard.

If you have a Harbor Freight Tools store in your town, watch their advertisments - the Centech 6" digital calipers are not fabulous but they frequently go on sale for $16 and beat the hell out of the manual kind.

You could potentially keep going another .1" to make the frame a whole .2" shorter than mine is, if you're really cramped.

The gap in the center is not quite wide enough to allow end-to-end placement of the RG resistors I used two small nuts that were rolling around the bottom of my parts bin to expand the center gap to about 21.3mm iirc.

After adjusting these gaps, the attenuator can now be fitted with standard perfboard.

I used two pieces, one 17x22 pads, the other 11x22 pads.

It would have been better to use an 18x22 board for the busier side, as this would give a more clear place to jumper the ground lines.

The smaller board leaves the wiper pins free. I would rather connect my signal wires to the pins themselves than to perfboard.

It would also be possible to solder the attenuator directly to a large chunk of perfboard. In that case, you would solder the final 6 sets of resistors to the perfboard and then settle in the less-busy side of the attenuator.

It may be possible to forego filing down the plastic spacers, instead widening the outer gaps slightly with nonconductive washers. You would have to be careful to leave enough thread available on the end of the attenuator to attach the end deck.


Quote:

Originally Posted by MisterX Next question that comes to mind is what you like to "fix" on the old boards?

Having populated my perfboards, I don't understand why the included boards were as complex as they are.

Since my way of building this attenuator selects 1 resistor at a time on each input between each set of wipers, the only bussed signal is ground. For the rest of it, it's just connecting resistors to the pins.

I plan to strip two lengths of solid copper wire to run the breadth of both boards (along the inner pins of the RG resistors) and jumper between them.

Anyway, this is what I've got.

Eric's Gallery :: Hacks :: img_0526
Eric's Gallery :: Hacks :: img_0527