Passive attenuator for monitors

[CurtZHP]

Haven't even finished one project and already thinking ahead to the next! 

I want to build a relatively simple passive attenuator to go between sound card outputs and monitor amp.  All balanced connections.

I started out with this...
https://www.scribd.com/doc/285370206/AT78-OTB

Straight up passive control using a simple double gang 10K pot.  Works.  The only hiccup is that, with the pot all the way down, there's still a little audio there.

I'd like to upgrade this a bit, so I'm kicking around replacing the pot with a stepped attenuator.  Since Goldpoint and/or DACT are WAY to rich for my blood, I was looking at the cheaper imported alternatives on Ebay.  Has anyone tried these, and what are your experiences with them?

Plan B would be a better pot.  The one that's in there now is just something I found in a drawer.  I think it's a Bourns.  Probably came out of a radio console at work.

 

[ruffrecords]

Monitor amps and powered speakers all tend to be more sensitive than they need to be.  This could be part of the reason you can still hear somenting with the control turned right down.At Neve back in the 70s, the monitor outs used to be fed via a 10k:600 transformer which drops the level by 12dB so it might be worth tweaking your existing circuit to do something similar. All you need to do is drop the 10K pot to 2K2 or 2K5 to drop the level by about 12dB.

Cheers

Ian

 

[CurtZHP]

Monitor amps and powered speakers all tend to be more sensitive than they need to be.  This could be part of the reason you can still hear somenting with the control turned right down.At Neve back in the 70s, the monitor outs used to be fed via a 10k:600 transformer which drops the level by 12dB so it might be worth tweaking your existing circuit to do something similar. All you need to do is drop the 10K pot to 2K2 or 2K5 to drop the level by about 12dB.

Cheers

Ian

I was under the impression that's what the 4.7K resistors in series with the inputs did.

Your mention of transformers gives me a Cunning Plan.  I came across a few old UTC transformers.  Maybe put those in the chain?

 

[abbey road d enfer]

Straight up passive control using a simple double gang 10K pot.  Works.  The only hiccup is that, with the pot all the way down, there's still a little audio there.

The problem comes probably from the fact that the CMRR of the connection is not good enough, particularly on the receiver (powered speaker) side. There is not much you can do about it. * Inserting a xfmr there may improve cut-off, but you need expensive xfmrs if you want to maintain fidelity.

I'd like to upgrade this a bit, so I'm kicking around replacing the pot with a stepped attenuator. 

Probably won't help at all (or so little it wouldn't be worth it).

Plan B would be a better pot.  The one that's in there now is just something I found in a drawer.  I think it's a Bourns.

Since you have it, try it, but again, I doubt it would solve your issue. Don't spend any money before finding out what's the real culprit.
I'm waiting to be proved wrong, though...

*Actually, connecting the pot as a standard pot (top to hot of source, wiper to hot of load, and bottom to both colds) may provide better cut-off, depending on the topology of the output stage. Resulting CMRR is very poor, but for short cable length is adequate.

 

[CurtZHP]

The problem comes probably from the fact that the CMRR of the connection is not good enough, particularly on the receiver (powered speaker) side. There is not much you can do about it. inserting a xfmr there may improve cut-off, but you need expensive xfmrs if you want to maintain fidelity.

UTC LS series expensive enough? 

Probably won't help at all (or so little it wouldn't be worth it).

Wasn't thinking in terms of audio quality, but in terms of repeataibility of settings.  But then, that's what panel graphics are for.

Don't spend any money before finding out what's the real culprit.
I'm waiting to be proved wrong, though...

I haven't spent a nickle on it so far.  Will try to maintain that as long as possible.

 

[CurtZHP]

*Actually, connecting the pot as a standard pot (top to hot of source, wiper to hot of load, and bottom to both colds) may provide better cut-off, depending on the topology of the output stage. Resulting CMRR is very poor, but for short cable length is adequate.

It's going directly from the pot to the balanced inputs of a Crown D-75 amp, if that helps.

 

[ruffrecords]

I was under the impression that's what the 4.7K resistors in series with the inputs did.

The attenuator topology used means when the pot is turned right off the two 4K7 resistors are shorted together. Their sum therefore represents the minimum load presented to the driving source and clearly this is designed to be close to10K so it looks like a bridging load. That's the principal reason for them being 4K7. If your source is capable of driving a 600 ohm load then you could divide all the values by 10 and get exactly the same performance plus the ability to drive longer cables.

Having said that, the two 4K7 resistors together with the 10K pot do form an attenuator which drops the level by about 6dB. Changing the pot to 2K5 will drop the level by another 12dB.

The problem with using pots is the off resistance. Most regular pots will will drop the level by about 60dB in the fully off position. If you want a true mute then it would probably better to use a separate mute switch for this to both short the outputs and disconnect them from the inputs.

Cheers

Ian

 

[CurtZHP]

I was under the impression that's what the 4.7K resistors in series with the inputs did.

The attenuator topology used means when the pot is turned right off the two 4K7 resistors are shorted together. Their sum therefore represents the minimum load presented to the driving source and clearly this is designed to be close to10K so it looks like a bridging load. That's the principal reason for them being 4K7. If your source is capable of driving a 600 ohm load then you could divide all the values by 10 and get exactly the same performance plus the ability to drive longer cables.

Having said that, the two 4K7 resistors together with the 10K pot do form an attenuator which drops the level by about 6dB. Changing the pot to 2K5 will drop the level by another 12dB.

The problem with using pots is the off resistance. Most regular pots will will drop the level by about 60dB in the fully off position. If you want a true mute then it would probably better to use a separate mute switch for this to both short the outputs and disconnect them from the inputs.

Cheers

Ian

I think the original design included a mute, incorporated into the input selector, but I omitted it.

 

[Kingston]

I've made a few balanced passive monitor controllers using rotary switches and ran into problems I'd like to point out.

Do not skimp on the quality of the rotary switches or you are in for regret. They all get scratchy and roughly speaking you are paying for the amount of years you want them usable.

Don't try anything less than 24-steps. You will find out you wanted more resolution no matter what.

The biggest problem is scaling the resistance along those steps. Besides the fact good rotaries are super expensive and mostly non-existent these days it may take several tries to scale the steps to something usable. It's most difficult to get right on the quiet end right next to the last mute step.

 

[ruffrecords]

The biggest problem is scaling the resistance along those steps. Besides the fact good rotaries are super expensive and mostly non-existent these days it may take several tries to scale the steps to something usable. It's most difficult to get right on the quiet end right next to the last mute step.

Can you explain this a little more? This is not a fader, you just use it to set a comfortable and repeatable listening level.

Cheers

Ian

 

[nielsk]

My habit has always been to use the CUT switch when I want the monitors cut, so the fact that attenuators don't really turn ALL the way down to cut was never an issue for me...
Separate L & R cut switches are very useful, too
For me, wiring the CUT switches so they disconnect the source and connect +&- of the destination to each other works best, any insight on this aspect?

BTW, so an LO2567 is 10K:600 ? I have not been able to find the sheet on that one, seems like it could make a nice tube O/T!

 

[CurtZHP]

If your source is capable of driving a 600 ohm load then you could divide all the values by 10 and get exactly the same performance plus the ability to drive longer cables.

This got my attention, because the computer that houses the sound card and the amp that powers the speakers both live in a closet just outside the control room, so I'd be looking at running cables a little longer than a few feet.  Not ridiculously long, but maybe a few meters.


Interesting tidbit of information:  According to the docs, the output impedance of the sound card is listed as <100Ohms.  Input impedance of the amp is 20K.

 

[abbey road d enfer]

Interesting tidbit of information:  According to the docs, the output impedance of the sound card is listed as <100Ohms.

That is quite typical of soundcards; it doesn't mean it is capable of driving 600 ohms load.

Input impedance of the amp is 20K.

That is also quite typical; most modern connections are bridging, based on quite low source impedance and rather high load impedance, allowing distributing one signal over to several receivers, and also inserting some passive controls without too much losses.

 

[CurtZHP]

Interesting tidbit of information:  According to the docs, the output impedance of the sound card is listed as <100Ohms.

That is quite typical of soundcards; it doesn't mean it is capable of driving 600 ohms load.

Input impedance of the amp is 20K.

That is also quite typical; most modern connections are bridging, based on quite low source impedance and rather high load impedance, allowing distributing one signal over to several receivers, and also inserting some passive controls without too much losses.

As, ruffrecords said earlier, when the pot is all the way down, it effectively shorts the two input resistors, so at minimum, it would be about a 10K load, right?  Trying to figure out what the load is with it all the way up.  Obviously, with no attenuator in place, it would be whatever the input impedance is of the amp, right?

 

[Kingston]

The biggest problem is scaling the resistance along those steps. Besides the fact good rotaries are super expensive and mostly non-existent these days it may take several tries to scale the steps to something usable. It's most difficult to get right on the quiet end right next to the last mute step.

Can you explain this a little more? This is not a fader, you just use it to set a comfortable and repeatable listening level.

Cheers

Ian

in that case you can just make a barrel connector with single attenuation setting soldered in.

I needed my monitor controllers to provide flexible fader-like use.

 

[CurtZHP]

I'm just trying to determine the best topology for this thing.  Is it OK to go with a larger or smaller pot for a different range?  I guess I could try that.

I'm also wondering about just doing a variable L-pad type control.  For balanced stereo, I'd need a four-gang pot, right?  Bleh!

I had a web page bookmarked a while back that explained pretty concisely the differences between a few types of attenuators in terms of impedance matching, etc.  Darned if I can find it now!

 

[abbey road d enfer]

Trying to figure out what the load is with it all the way up. 

With Rp=nominal pot value, Rin=input impedance of amp or powered monitor and Rs=value of series resistor
2.Rs+Rp//Rin
Plugging your values
2x4.7k+(10kx20k)/(10k+20k)=>16.06k
You could drop the values by a factor 5, for about 2k min load on the soundcard.

Obviously, with no attenuator in place, it would be whatever the input impedance is of the amp, right?

Correct.
[/quote]

 

[abbey road d enfer]

I'm just trying to determine the best topology for this thing.  Is it OK to go with a larger or smaller pot for a different range?  I guess I could try that.

The range would not be different, since one end is supposed to be minus infinity. But the taper may be different; the higher Rs vs Rp, the closer the taper to the pot's intrinsic taper, but it may not be what you want, and the cost is higher base attenuation. You may run out of juice. In fact the Rs=1/2Rp is a sensible choice, since it spreads a little the taper at the upper end, giving more control of high volume. It is essential that you adjust adequately the sensitivity of your amp or powered monitor.

I'm also wondering about just doing a variable L-pad type control.  For balanced stereo, I'd need a four-gang pot, right?  Bleh!

The only advantage would be the beter matching of stepped attenuator.

I had a web page bookmarked a while back that explained pretty concisely the differences between a few types of attenuators in terms of impedance matching, etc.  Darned if I can find it now!

Constant Z attenuators have no advantage here because the impedance of the source does not match that of the load (100r->20k).
Balanced attenuators have better CMRR, but it's moot for short cable lengths in a typical "home" studio.

 

[CurtZHP]

Balanced attenuators have better CMRR, but it's moot for short cable lengths in a typical "home" studio.

Short is a relative term.  And I'm a short guy, so a ten foot cable is long to me!

 

[ruffrecords]

The biggest problem is scaling the resistance along those steps. Besides the fact good rotaries are super expensive and mostly non-existent these days it may take several tries to scale the steps to something usable. It's most difficult to get right on the quiet end right next to the last mute step.

Can you explain this a little more? This is not a fader, you just use it to set a comfortable and repeatable listening level.

Cheers

Ian

in that case you can just make a barrel connector with single attenuation setting soldered in.

I think there is a little more to it than that. It is essentially a volume control for your monitor speakers or phones.

I needed my monitor controllers to provide flexible fader-like use.

So it sounds like you use your monitor controller differently.  I am fascinated by how people use their gear. Can you elaborate?

Cheers

Ian