[DESIGN] A general-purpose 600-ohm balanced attenuator

[PRR]

> Is the Daven or Shallco (or equivalent) better than building your own?

If you'd ever opened a Daven, you wouldn't ask that.

It is the difference between a 1939 John Deere farm tractor, and a $99 PepBoys lawnmower. Both cut grass, but everything about the Deere (or Daven) is 10 times bigger and will last 100 times as long, without buzzes and rattles and pieces falling off.

However if you are not harvesting hay every year, or potting up/down every 2.5 minutes all day and night live on the air, you may not wish to pay for such solidness.

 

[NewYorkDave]

Of course, the Daven is better-built--much better-built--than what a DIYer could put together with a garden-variety rotary switch. But I took the question as one of circuit topology. Circuit-wise, the Daven really is nothing but a very nicely-made rotary switch with a bunch of resistors. This is coming from a guy who loves the ol' things and collects them whenever they turn up at a reasonable price.

Daven doesn't make 'em anymore, but price out a new balanced attenuator from their successor, Shallco--or even second-hand, if you can find one--and my junky little box starts to look pretty usable :wink:

Besides, the ckt. I posted is more akin to a lab-type "decade attenuator" than a Daven/Shallco mixer control... in other words, it won't be subject to constant adjustment.

 

[alexidoia]

Hi,
I am sorry to post to this thread years after its first edition.
I need to find balanced audio attenuators of 600 ohms input output but they are way too expansive for me. That is why this thread comes to me as a rescuer as I am considering building them myself.

Will there be audio clicks/stops when switching from position to position or will it goes smoothly as a Daven pot ?
Has anyone here built one of this and have photos of it ?
And finally, am I understanding something wrong or are there in that schematics two different pots to be operated to attenuate the signal ?

Thanks for your input.
Alex

 

[Mlewis]

it should be two (three pole, 6 position, make-before-break) rotary switches. the left half of the circuit is all on one switch and is a fine level control (2db steps) whereas the right half of the circuit is for the second switch and is for coarse level adjustment (10db steps)

i've not built one yet and i'm no expert but that said, the audio might click a tiny bit during level change (most likely when using the course adjust switch) but only because the level changes so quickly when the switch makes contact not due to anything nasty going on. it might not though - i haven't tried it. if you want the switch to rotate smoothly maybe you could remove the spring stepper from the switch or something.

by the way, glad you did post, this had slipped through the net for me and was just what i'm looking for. cheers NYD!

 

[alexidoia]

Thanks,
Well if you do ome of them, let me know.
I will do one with only one RS and will post here photos and comments as I go along.

 

[PRR]

Someone asked me this in PM.

> 351 Ampex modification ... If I use a 1kpot, I need a double pole one as I am attenuating a balanced signal, could this do?

NYDave will argue with me. I say, in the studio, for transformer outputs, you can just use a single-gang pot. The transformer is not just "balanced", it is floating, no ground reference at all. You can put an unbalanced attenuator in a perfectly floating line without upsetting the balance.

The line is not "perfectly floating" due to stray capacitance. But this capacitance is small. The leads from the transformer to the pot are short, maybe a foot. The leads to the load are not long, maybe 30 feet or 10 meters. Typically the long line will be nearly perfectly balanced due to equal capacitance on both hot leads. The short link will run about 300pFd out of balance, which at 1K ohms gives bad balance above 1MHz, but good balance across the audio band.

Try it and see if I am wrong.

 

[rafafredd]

PRR, HOW exactly would you wire a single 1kpot at a 600Z balanced output?

 

[PRR]

> HOW exactly would you wire a single 1kpot at a 600Z balanced output?

 

[NewYorkDave]

NYDave will argue with me.

Well, not necessarily. The objection to an unbalanced attenuator in a "balanced" circuit is theoretical and in many practical cases it doesn't matter much. It would matter in some cases where the asymmetical source impedances would degrade CMRR at the receiving end--but again, it may not be enough to notice. Lots of old gear used bridged-Ts (unbalanced and floating) on inputs and outputs. And that's what I used on the output of Soundguy's Altec 436.. That double-deck open-frame rotary switch visible in the inside shot is a homemade 11-step (0 to 27dB plus "off") 600/600 ohm bridged-T.

For those who wanted to use a 1K-A pot in a 600-ohm circuit, I would caution that you should make sure the pot can handle the power. You wouldn't want to use a regular .25W carbon pot on the output of, say, a hefty tube comp or line amplifier where there may be over a watt of power involved. In a case like that, as long as your load impedance is going to be over 5kohms, a 500-ohm wirewound pot might be a better choice--but good luck finding one in audio taper!

 

[rafafredd]

Thanks a lot PRR. There wasn´t really no need for all that fancy drawing, but it´s a really nice drawing.

I just wanted to make sure I was thinking right. In fact I did this mod already and yes, it sure does works fine, at least in a transformer coupled in/out reality.

 

[alexidoia]

[quote author="NewYorkDave"]For those who wanted to use a 1K-A pot in a 600-ohm circuit, I would caution that you should make sure the pot can handle the power. You wouldn't want to use a regular .25W carbon pot on the output of, say, a hefty tube comp or line amplifier where there may be over a watt of power involved. [/quote]
point made.

In a case like that, as long as your load impedance is going to be over 5kohms, a 500-ohm wirewound pot might be a better choice--but good luck finding one in audio taper!

That brings me back to case one: building a pot myself. argggg patience is a key element in restauring old gear.

 

[Martin_T]

NYDave:

These are constant impedance, right?Could these atenuators be used to control line level in a passive summing box?

Martín

 

[alexidoia]

Okay,
I am in the process of building a 600 ohms constant impedance blanced output attenuator with only one switch 3 poles 12 positions.

NYDave gave me this link to calculate the resistors value:
http://www.microwaves101.com/encyclopedia/calcattenuator.cfm

According the NYDave initial schematics of this post, both of Z0 are 300ohms which in total makes 600ohms.

I don't really know yet how to use the calculator to calculate my resistor values. I found this link to explain me the design of bridge T attenuator:
http://www.fmsystems-inc.com/eng_tee.htm

In there they say that both of Z0 should be 600ohms. So I am a bit confused. Can someone take time to explain me what value I have to use for Z0 nd how to use the calculator.
thanks

Alex

 

[alexidoia]

Ok, this is what I came up with the resistors calculator, can anyone have a look to it and tell me if/where I am wrong.

[wrong schematics]

Thanks.
PS: here is a better quality (pdf) file if you have difficulties reading the values.
http://idoia.org/electronics/att.pdf
Alex

 

[NewYorkDave]

Well, one thing doesn't look correct right away: those 600-ohm resistors should be 300 ohm. Remember what i told you: in a balanced attenuator, the series elements must be one-half the values of the unbalanced version, while the shunt element is the same.

I'm afraid I can't look it over more carefully than that because I have a lot of work to do today...

 

[alexidoia]

[quote author="NewYorkDave"]Well, one thing doesn't look correct right away: those 600-ohm resistors should be 300 ohm.[/quote]

This is because of the link I gave above that says that it must be 600 and 600. I guess that only applies to a unbalanced att.

Remember what i told you: in a balanced attenuator, the series elements must be one-half the values of the unbalanced version, while the shunt element is the same.

Ok I will redo it. Thanks.

 

[alexidoia]

here we go:

Indeed, it makes more sense. I have also the same values than yours for the first switch of your schematics. The only thing is a resistor of the shunt (-10db), yours is 280, while according to my calculations, it should be 76.22. Is this an error ?
When you have time....
Thanks a lot.
Alex

 

[alexidoia]

Alright,
Although nobody replied to my previous schematics, I think it is ok and I am going to order resistors.
I will order metal film ones 1%. I have done a list of approximative values and I found them all @ 0.5W or 1W which I think is ok for my circuit. I have a couple of them which are 0.25W though and I wonder if it can be a problem.
Tell me if I am wrong but resistors in series share power, so if I have 1/4W resistor with one 1W that should be ok ?! right ?
If some of you are willing to buy some rotary switches or resistors, let me know we could organise a group order.
Thanks
Alex
PS: if someone could have a rough look at my schematics though, it would reassure me

 

[alexidoia]

[/quote]

Just to be sure, but the pot is not only connected to the "hot" it is also connected to the "cold", I would have put the pot only on the "hot" as a rehostat, why this ?

Alex

 

[NewYorkDave]

Alex,

The way you drew the wiring of the pot is correct.

As for the power rating of the resistors in your attenuator, you need to figure out how much power will be dissipated in each resistor for various steps of attenuation. A simple circuit like an attenuator can be simulated in SPICE with good results; you can check the power dissipation of each resistor and also check each step for correct attenuation while you're at it.

One hint: the SPICE programs I've used require you to enter peak rather than RMS voltages. Bear that in mind when checking the power dissipation of the resistors. For instance, suppose you want to make sure the attenuator can stand up to a +30dBM input. That's 24.5 volts RMS, but peak is 34.6V. Also, be sure your Bode plot display is set to read out in decibels rather than "real" values.