The Khozmo Mkii Stepped attenuator

[Bigman80]

I don't understand, why is no one recommending the Khozmo Ladder stepped attenuators?

Is it just cost, they are 40 t0 50% more expensive than the Shunt or Series, as you need 96 resistors per channel, not 48/49.

The problem with the Shunt has been explained, you have different impedance for every setting.

For me the problem with the Series type, is just that, you have 48 resistors in series with the output seeing the number of resistors dictated by the rotary control position and the remaining number going to ground, the sum of the two equaling 48. At low volume there will be many resistors in series for the signal path, is this not a bad thing?

Surely the Ladder type is the best, where you have pairs of resistors for each volume position, one for the signal and one (matched) to maintain impedance. There is only ever one resistor in the signal path and constant impedance is maintained across the controls range.

The only difference is that the Series uses multiples of resistors for each value, the Ladder uses a single resistor.

I would have thought the Ladder was the better choice?
You still get +/- 0.1dB channel matching if you want a stereo version.

So why not the Ladder?

I went with the advice given to me by Rothwell and Firebottle and bought the Khozmo Series.

I couldn't tell you why the Ladder would or wouldn't be better but if it's better than the Series, it's worth every penny of that extra 40-50%

[Qwin]

I went with the advice given to me by Rothwell and Firebottle and bought the Khozmo Series.

I couldn't tell you why the Ladder would or wouldn't be better but if it's better than the Series, it's worth every penny of that extra 40-50%

The Shunt/Series use one brushing contact, the Ladder uses two, which is a contributing factor I suppose, but I wouldn't have thought this was as detrimental as a hole bunch of components strung together. The old adage, less is more comes to mind.

I know you are happy with the results Oli, we are probably talking very minor differences here anyway.

[Qwin]

There are 48 resistors in the circuit regardless of the volume setting. The idea that some are "in the signal path" and the rest are "just going to ground, so they don't matter" is totally and fundamentally wrong. All 48 resistors are doing the same job regardless of the volume control setting.

Is using 48 resistors inferior to using 2 resistors? In my opinion, no, it's superior. Resistors have a voltage coefficient http://www.resistor.com/tech_voltagecoefficient.htm If you use 48 resistors instead of 2 the voltage across each resistor is lower and therefore any distortion is lower too. Also, the power handing capacity of 48 resistors is much higher than 2 resistors so any heating effects will be much less with 48 resistors. Admittedly the power dissipation is microscopic so the benefits are probably insignificant, but it's better to have higher capacity than lower imho.
Also, with 48 resistors instead of 2, each resistor is a low value and any stray capacitance between switch contacts will have a much smaller shunting effect across a small resistor compared to a big one.
They are my reasons for advocating the use of "series" stepped attenuators rather than so-called "ladder" stepped attenuators. BTW, I think the term "switched pair" is preferable to "ladder" but nobody else seems to think so

Your point about individual, smaller value resistors, having lower distortion is right I think, but it is additive across the number used, so you end up with more distortion overall. This is marketed as the rational for using the Ladder type, which is just a load of individual voltage divider networks.
I just looked at who is making/selling what and there are not many Series Type made compared with Shunt or Ladder. There are some Ladder types selling at around a Grand or more, if simpler, cheaper to make alternatives gave better results, I don't think they would bother.
On the other hand Khozmo make all three types, but give no indication as to which is preferable.
Some of the Ladder type using surface mount devices, with their shorter signal paths are incredibly compact.
Ladder attenuators were all the rage a while back, had one retro fitted in my Croft Pre, still plenty available, just wondering why they had not been considered in this thread.

[sq225917]

I use the SMT ladder version of the Khozmo 10k with relay switching and optical encoder. It's a very nice pot. Not quite sure it's as clean as a similar value Goldpoint but it does have a remote control in the mix.

In a ladder arrangement isn't the total resistor noise the total resistance / the number of resistors, or thereabouts

[Qwin]

I use the SMT ladder version of the Khozmo 10k with relay switching and optical encoder. It's a very nice pot. Not quite sure it's as clean as a similar value Goldpoint but it does have a remote control in the mix.

In a ladder arrangement isn't the total resistor noise the total resistance / the number of resistors, or thereabouts

I think that would give you an average for the noise across the settings.
In a Ladder there is only ever one resistor connected in series with the signal and one to ground.
When you move up one notch it connects a different resistor in series with the signal and a different one to ground, which maintains impedance and so on for each step.

Those motorised units with digital display look really nice, but ouch, are they expensive.

[sq225917]

They are expensive, and mine had a sticky relay so from step 24 to step 32 one channel was down. Not so noticeable for the first few clicks up from 24, but stepping from 32 to 31 was an 8db drop. It's off for repairs. He was super quick in responding to my email and no questions asked. It's a beautiful unit and sounds great, above 32.... ;-)