Prev: constant power (APC) current driver design
Next: High side current monitor with latch
From: John Fields on 19 Mar 2010 12:55
On Fri, 19 Mar 2010 16:23:44 GMT, Jan Panteltje
<pNaonStpealmtje(a)yahoo.com> wrote:

>On a sunny day (Fri, 19 Mar 2010 16:20:08 +0100) it happened oopere
><me(a)somewhere.net> wrote in <ho04n9$d9l$2(a)defalla.upc.es>:
>
>>
>>
>>Jan Panteltje wrote:
>>> On a sunny day (Fri, 19 Mar 2010 09:21:13 +0100) it happened oopere
>>> <me(a)somewhere.net> wrote in <hnvc5q$a5u$1(a)defalla.upc.es>:
>>>
>>>> Candide Voltaire wrote:
>>>>> I'm looking for a passive rc-network which produces a delay of 200ns
>>>>> for a 5MHz sinusoidal voltage.
>>>>> Which is the minimum number of RCs to achieve this? Are there some
>>>>> schematics optimized for this purpose?
>>>>>
>>>>>
>>>>> Candide
>>>> Since the required delay is a whole period, the answer is yes, there is
>>>> an optimum schematic, although it is a degenerate case of RC, with R=0,
>>>> C=0 (inexpensive)
>>>>
>>>> Vin----R--�--Vout
>>>> |
>>>> C
>>>> |
>>>> gnd
>>>>
>>>> Pere
>>>
>>> Na, C=0 and R=0 is extremely expensive.
>>> Impossible to make even.
>>> Better just use some meters of coax cable to delay one period :-)
>>
>>But that's not a passive rc circuit ;)
>
>But it is: a cable has a capacitance and resistance.
>The inductance comes for free, and makes it possible :-)
>And the cable method WORKS.

---
Yeah, but since it's got L in it, it's not what the OP _asked_ for.

JF
From: whit3rd on 19 Mar 2010 16:02
On Mar 19, 4:25 am, Candide Voltaire <candideguev...(a)gmail.com> wrote:
> On Mar 19, 9:21 am, oopere <m...(a)somewhere.net> wrote:

> > > I'm looking for a passive rc-network which produces a delay of 200ns
> > > for a 5MHz sinusoidal voltage.

> > an optimum schematic,... is a degenerate case of RC, with R=0,
> > C=0 (inexpensive)

> OK that's correct, do you also have an answer in case we want 180ns?

Depends; do you accept a phase-lead network, or does it have to
be phase-lag? A single RC stage can have zero to 90 degrees lag,
it'd take four or more to get to the 180 ns mark, and there is quite
a lot of attenuation. The 36 degree phase lead is easier.
From: Candide Voltaire on 19 Mar 2010 16:26
On Mar 19, 4:19 pm, oopere <m...(a)somewhere.net> wrote:
> Candide Voltaire wrote:
> > On Mar 19, 9:21 am, oopere <m...(a)somewhere.net> wrote:
> >> Candide Voltaire wrote:
> >>> I'm looking for a passive rc-network which produces a delay of 200ns
> >>> for a 5MHz sinusoidal voltage.
> >>> Which is the minimum number of RCs to achieve this? Are there some
> >>> schematics optimized for this purpose?
> >>> Candide
> >> Since the required delay is a whole period, the answer is yes, there is
> >> an optimum schematic, although it is a degenerate case of RC, with R=0,
> >> C=0 (inexpensive)
>
> >> Vin----R--·--Vout
> >>            |
> >>            C
> >>            |
> >>           gnd
>
> >> Pere
>
> > OK that's correct, do you also have an answer in case we want 180ns?
>
> > Candide
>
> A passive RC network will not exhibit constant amplification over
> frequency. This means that you can not say that it is a pure delay.
> Having said that, a single RC section is able to produce -90 deg of
> phase shift. Two cascaded RC sections will give you an all pole second
> order response able to produce up to -180 deg, and so on. If you require
> 180ns (-324 deg), you would need 4 sections.
>
> But you need to tell more about your problem. What is the intended
> application? Do you really need a pure delay? Does it really need to be
> RC-only? What signal levels are involved?
>
> Pere

Ok here's more details, I want the attenuation of the RC-network
=<20%. Is it still possible achieve this by using only 4 RCs? if not
how many do we need?
As it should be a passive linear network solution, the signal levels
involved are not that important but to make things a bit more
practical lets say we have an amplitude of 1Volt at the input.
Considering the intended application I was and still am just wondering
whether the use of a transmissionline could/can be avoided by using a
passive RC-network
I'm also curious to know whether there is a better solution than just
cascading the passive RC-sections (in any case I really want to use
only Rs and Cs, no active components, no transmission lines, no
transformers...)

thanks to all who are willing to share their ideas with me

Candide
From: Tim Williams on 19 Mar 2010 19:20
"Candide Voltaire" <candideguevara(a)gmail.com> wrote in message
news:6bf6ab3a-bb8b-4f88-8eed-dcb1f9c0c1aa(a)z11g2000yqz.googlegroups.com...
> Ok here's more details, I want the attenuation of the RC-network
> =<20%. Is it still possible achieve this by using only 4 RCs?

No.

> if not how many do we need?

I don't think there is any point on the curve where attenuation is that much
less than phase shift (in some arbitrary units). For instance, for
attenuation > 20dB, you get a phase shift of nearly 90 degrees from one RC.
To get more 360 degrees phase shift requires slightly more than four RC
stages, and each one must have an impedance about 10 times higher than the
previous, otherwise the response gets screwed up more. Accordingly, the
attenuation is closer to 80dB, i.e., your signal is gone altogether.

> I'm also curious to know whether there is a better solution than just
> cascading the passive RC-sections (in any case I really want to use
> only Rs and Cs, no active components, no transmission lines, no
> transformers...)

a. Use inductors.
b. Use op-amps (all-pass filter).

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms


From: Glenn Gundlach on 20 Mar 2010 01:06
On Mar 19, 4:25 am, Jan Panteltje <pNaonStpealm...(a)yahoo.com> wrote:
> On a sunny day (Fri, 19 Mar 2010 09:21:13 +0100) it happened oopere
> <m...(a)somewhere.net> wrote in <hnvc5q$a5...(a)defalla.upc.es>:
>
>
>
>
>
> >Candide Voltaire wrote:
> >> I'm looking for a passive rc-network which produces a delay of 200ns
> >> for a 5MHz sinusoidal voltage.
> >> Which is the minimum number of RCs to achieve this? Are there some
> >> schematics optimized for this purpose?
>
> >> Candide

>
> >Since the required delay is a whole period, the answer is yes, there is
> >an optimum schematic, although it is a degenerate case of RC, with R=0,
> >C=0 (inexpensive)
>
> >Vin----R--·--Vout
> >           |
> >           C
> >           |
> >          gnd
>
> >Pere
>
> Na, C=0 and R=0 is extremely expensive.
> Impossible to make even.
> Better just use some meters of coax cable to delay one period :-)

At 8nS / foot that's 25 ft which might cause space problems though it
would work well. A BAL, Allen Avionics ot Mathey delay line would take
a lot less space but cost more and not be as easily adjustable. It
still needs an amp to make up the 6+dB insertion loss - or maybe
that's acceptable.

G²
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4
Prev: constant power (APC) current driver design
Next: High side current monitor with latch