# Is this even possible? Filter/impedance question

Discussion in 'Electronic Basics' started by billcalley, Aug 9, 2005.

1. ### billcalleyGuest

Hi All,

A company I'm contracting for told me to design a circuit I did not
even think was possible. But is it? They have an amplifier that needs
to see 1300 ohms at its output, even though the amplifier itself only
has an output impedance of 75 ohms. So far, so good. But they want a
*passive* highpass filter to be at the output of this 75 ohm amplifier!
In other words, they want the input to the highpass filter to present
the 75 ohm amplifier with 1300 ohms! (And then the HPF must also have
an output impedance of 75 ohms after all this). I would have thought I
needed an active buffer to accomplish this. How could I even begin to
design such a structure, much less maintain the HP filter shape? Has
anyone ever heard of this before?

Thank You,

-Bill

2. ### John LarkinGuest

You'd need a transformer to do this, and it would result in a
best-case voltage gain through the filter (ignoring losses) of 0.24.

Your customer sounds a little confused.

John

3. ### BanGuest

Did you really understand rightly? It is not that difficult to make a
highpass with 1300R input impedance, in fact the needed transformer itself
will act as a passive highpass if you design it right. The turns relation
must be 4.17:1and the attenuation will be 12.5dB depending on the losses in
the transformer. But since the amp is loaded with only 1.3k, its level is
almost +6dB higher. You will probably need a Zobel network on the output of
the amp to prevent gain peaking. I wonder why they want the high impedance
on the amp output, if a simple 1:1 transformer does the same thing without
any level loss.

4. ### Joe McElvenneyGuest

Hi,
Passive filters with asymmetrical in/out impedances are quite
common. Design an odd order Butterworth (or Tchebycheff) filter
using one of the impedances, figuratively cut it in half through
the middle reactance, scale one half to the second impedance and
then glue back together.

Look up p.18 of -

"Simplified Modern Filter Design" by Philip R. Geffe.

Cheers - Joe

5. ### jgreimerGuest

You left out most of the requirements of the filter. What is - the corner
frequency? - the slope (db/decade)? - the insertion loss?

My first thought is a pi or T filter but I need the above details first.

6. ### billcalleyGuest

Hi All,

Thanks for the replies! I know about asymmetrical filters, but
they are used when the filter's load and source impedances are not
equal. But in this special case both the source and load that the
filter sees *are* 75 ohms -- it's just that the filter's picky 75 ohm
source (the amplifier) wants to see 1300 ohms. How can I get a HPF
response out of that, considering that there is now a huge (purposeful)
mismatch created at the filter's input? How can any passive filter act
as a high impedance "buffer" like that?
The filter is a 600kHz (to 12MHz) HPF, with an insertion loss of
<2dB, 3 poles, >15dB RL, and <0.5dB ripple.

Thanks!

-Bill

7. ### qrkGuest

You can build passive LC filters with asymmetric input/output
impedance. Silly example:

GENERATED BY QUICKIE LC FILTER MAKER
Vin 1 0 AC 18.333V
Rin 1 2 1.3000k
C1 2 3 247.04p
L1 3 0 263.43u
C2 3 4 149.28p
Rout 4 0 75.000
..AC DEC 60 59.528kHz 6.0000MegHz
..PROBE
..END

However, the actual input and output impedances will change
significantly with frequency. Probably not what your looking for. You
could pad the input which would calm the input impedance variations
that the source sees. Of course, the pad will attenuate the signal
further. I think you need to chat with these guys and see what they
are really after.

8. ### Jim ThompsonGuest

They're probably clueless... 1300 ohms load comes out to 0.5dB
insertion loss.

Maybe they're French ?

...Jim Thompson

9. ### billcalleyGuest

That is what I thought guys -- thanks!

-Bill