classic Kay attenuator

Very nice, so long as your desired attenuation is one of the 64 available values! It should be straightforward enough to come up with an equation that gives what switches to throw to minimize the error bewteen your desired attenuation and what attenuations are available; something using Lagrange multipliers. That is, if you couldn't just guess it by inspection.

We've got some of those. But I think the switches need cleaning.. or something. I never took one apart.

I did open up an old HP 600 ohm attenuator. In a wooden case, but with metal cans inside. It was all resistors. I'd have to check it's step response.

George H.

as far as I can see it covers every 1dB step from 1dB to max atten.

NT

Yes, of course you're right, I'm not sure what I was thinking! 1 - 41 dB plus the 10 dB insertion loss. There are 64 possible switch settings, but some are redundant. The redundancy could be eliminated if the switches operated in a binary fashion, like the first switch divided by 32...etc. but I don't know off the top of my head how to do it with passives. It would certainly be more complicated.

Nevermind, the binary system wouldn't work as the ratios wouldn't be right to give nice 1 dB increments. I guess if there were a better solution they would have thought of it, right?

--
Low-inductance connections between switches.

h t

ith

t 1

if

d

I presumed they didnt do that simply because they didnt design the attentuator to give that much attenuation.

It would work fine.

NT

--
Right, and they did:

news/cdufc71g31eguvf1dubjumerp02s0t06k8@4ax.com

Or: news: snipped-for-privacy@4ax.com

;)

The parts values in the first two stages are wrong.

John

--
Didn't you say that already?

And so what?

The post wasn't as much about accurate parts values as it was about
responding to the inaccuracy of the statement, above:

"Nevermind, the binary system wouldn't work as the ratios wouldn't be
right to give nice 1 dB increments.  I guess if there were a better
solution they would have thought of it, right?" above.

As a matter of fact I could just have shown tees with attenuation
values and no R values at all.

In any case, I'm trying to ferret out where the error came from and as
soon as I do I'll post what I find.

Weren't you going to fix it?

Well, your bud Thompson apparently prefers wrong part values to missing part values.

You hardly needed to design an attenuator to demonstrate that the binary number system works. Of course, you didn't bother to do it right, or to consider why Kay didn't elect to go over 20 dB per stage.

*That* is the interesting issue here. Perhaps a binary-weighted attenuator is not actually a "better solution."

But then Thompson would, no doubt, insult you, as he insults me when I post circuits without values. But then, he hasn't made a lot of sense lately.

OK.

John

I shouldn't post when I'm tired. For some reason I was thinking about the voltage division ratios in the attenuator being binary, like 1/32nd,

1/16th, not giving 1dB increment. If you designed an attenuator that had a 32 dB switch, and a 16dB switch with the proper voltage division ratio to give those values, it would of course give you 2^n values in 1 dB increments. I'm sorry for the confusion.

--
Still am, but since you're not paying for it I'll do the work on my
schedule, not yours.

--
Decent. :-)

The biggest step there is 16 dB. Kay stopped at 20 dB. 32 isn't impossible - people sell 60 dB fixed attenuators - but it is probably not advantageous to go over 20 in one step.

This attenuator seems to have provisions for dual stages

ftp://jjlarkin.lmi.net/VAT-20.zip

which they might implement at the 40 or 60 dB level. It would be interesting to open up some high-dB attenuators and see how they do it.

Cite?

I often correct mistakes. I only demean people who deserve it.

Where did you get the numbers from?

John