[sdiy] 4069 vco gives v+ noise?

[Magnus Danielson]

From: Karl Ekdahl <_nial_ at yahoo.com>
Subject: Re: [sdiy] 4069 vco gives v+ noise?
Date: Wed, 14 May 2003 00:16:55 +0200 (CEST)

> Ok, the hum is the 50hz non-rectified voltage. I use a
> 15v AC without center tap, two 4700uF 40v filter caps
> (not 6800 as i thought) and a half wave rectifier. 

Eh, uh... half wave rectifier sounds a bit worrying really. You more than
double the time that the caps needs to hold the voltage. If you load it this
effectively means very large caps...

If we just considers rought estimates, then you have this relation:

C * /|U = /|Q = I * t

where t is the uncharge period. Now, the /|U is the peak voltage to lower limit
where the regulator gives up keeping the target voltage. So..

      ^
/|U = u - U
           reg,lim

Now... Just solving for the cap we need, we get:

    I*t
C = ---
    /|U

or... if we want to know the voltage droop:

      I*t
/|U = ---
       C

for the case we have:

^          _
u = 15 * \/2 = 21,213 V

U        = 15 + 2 V = 17 V (2,0 V droopout for a 7815 according to databook)
 reg,lim

t = 200 ms

/|U = 21.213 - 17 = 4,213 V

       /|U * C   4,123 * 4700 u   4,123 * 4,7 m   4,123 * 4,7
I    = ------- = -------------- = ------------- = ----------- = 96,89 mA
 max      t           200 m            200 m          200

So, there you have it. If you load it down (IN TOTAL for all regulators!!)
with more than about 100 mA then your regulators can't take the heat.

Proposed cures:

1) Full-wave rectifying. If you can do it, do it. You effectively half the t
   value and doubles the current you can handle without dropping too low.
   Choose transformer to allow you do full-wave.

2) Make sure your caps are large enought to handle worst-case load. Unless you
   divert to primary-switched converters, the normal linear regulators will not
   cope with the situation when you go too low on the voltage at the maximum
   intended load.

Even if these calculations where done in a very over-simplified manor, they
give you a good hint on where to go, and design by margin from that. You need
that margin since when the caps go old, their internal resistance rises and the
will not charge to the top and they will discharge to lower voltages even if
they sustain the capacitance... so design margins helps you to make the box
live longer.

Now, if you have three-phase you can do even better, since then you could do
three-phase rectifying and getting the t downto just 33 ms which means that the
caps can be a 6th of the size the half-wave rectifier, which is a considerable
gain if you also is going to toss obscene currents at the same time.
> 
> Could anyone help me with the math here? How do i
> calculate how big filter caps i would need? The load i
> will drive would be
> "the-average-load-of-a-beginners-first-homemade-small-modular"
> :)

OK... so you need a 1kA 400V three-phase line to your labbench and a good
cooling system to keep the fried flesh-steams in control, right? ;O)

I hope you've got a good clue about the caps from my formula toss (also, Don
should be happy now ;O)

Cheers,
Magnus - seems to find the same problem popping up all over the place