| |
G8MNY > TECH 22.07.05 10:20l 99 Lines 4019 Bytes #999 (0) @ WW
BID : 55204_GB7CIP
Read: GUEST DG4NAU OE7FMI
Subj: Reducing Electronic RF QRM 1/2
Path: DB0FHN<DB0FOR<DB0SIF<DB0FSG<DB0RGB<OK0PPL<DB0RES<ON0AR<GB7CIP
Sent: 050722/0907Z @:GB7CIP.#32.GBR.EU #:55204 [Caterham] $:55204_GB7CIP
From: G8MNY@GB7CIP.#32.GBR.EU
To : TECH@WW
By G8MNY (New Mar 05)
There are 3 principles to removing QRM & 1 not.
1/ Stop the RF being generated at the source.
2/ Stop Source RF from radiating.
3/ Stop Rx system seeing the RF QRM.
4/ Living with it.
1/ In electronic systems the use of radio frequencies is common, but the need
for wideband signals is less common.
Lets look at a logic gate that runs at 1MHz but uses a fast 40MHz totem pole
output stage. This produces a high current power rail pulse @ > 120MHz on
the switching edges.
Gate ³ ³ High bandwidth
Current ÄÄÄÁÄÄÄÄÄÄÁÄÄÄ High QRM
+5V ÄÄÂÄÄÄÄ
³_
______ ³ `\ ______
/ 1MHz \_____ ÄÄ´GateÃÄÄ ³ ³____
Low Bandwidth ³__./ 40MHz edges
low level QRM ³ Medium Bandwidth
0V ÄÄÁÄÄÄ Medium level QRM
The output fast edge shape of often wanted, but power rail noise is not.
Filtering is normally proved on the power rails to provide am on PCB path
for these high pulse currents..
PCB ANT
\³/ \³/
+5V ÄÄÂÄÄÄÄÄÂÄÄÄ 5V ÄÄÂÄÁÄÄÄÂÄÄÁÄÄ
³ ³_ But due to ) ³_
=== ³ `\ PCB track 1uH ) ³ `\
10n ³ ÄÄ´GateÃÄÄ lengths this ) Ä´GateÃÄÄ
³ ³__./ often looks 10n ³ ³__./
³ ³ like this... === ³ \³/
0vÄÄÁÄÄÄÄÄÁÄÄÄ 0v ÄÄÁÄÄÄÄÄÁÄÄÄÄÄÄÁÄÄ
One cure is to put a C directly across any noisy ICs, these can be found by
scopeing the power rails on a PCB looking for the worst RF noise sources.
Then putting a Cs (surface mount?) across the power pins without loads of
track en-route.... Cs of 10n-10u can make difference to power rail noise
radiation.
PCB ANT
\³/ \³/ Much less
5V ÄÄÂÄÁÄÄÄÂÄÄÁÄÄ RF Noise
) ÚÄÄ´_ on PCB
1uH ) ===³ `\
) ³ ´GateÃÄÄ
10n ³ ³ ³__./
=== ÀÄÄ´ \³/
0v ÄÄÁÄÄÄÄÄÁÄÄÄÄÄÄÁÄÄ
It is not usually possible to add L to the power rails to help the
decoupling.
Data bus radiation is more difficult to cure as there are more lines & the
RF power level is normally quite low from any particular line.
But filtering of a R or LC is sometimes effective (eg on a PC printer port),
but both type must consider the timing delay & loss & any undamped resonance
on the data line.
__ QRM __ Much less __ Very little
³ `\ \³/ ³ `\ \³/ ³ `\ \³/
ÄÄ´GateÃÄÄÁÄ Ä´GateÃÄ100ÄÂÁÄ Ä´GateÃÄ((()ÄÂÁÄ > Z load
³__./ ___ 1MHz ³__./ | stray ³__./ L ³
³ ³ ³___³ ³ === C ³ 1uH === 1n
ÄÄÄÁÄÄÄÄÄÄ ÄÄÁÄÄÄÄÄÄÄÄÄÁÄÄ ÄÁÄÄÄÄÄÄÄÄÄÄÁÄ
Noise Source With R C the RF With L & C the
dB -6dB/Octave for Harmonics drop Harmonics drop
³ ³ | odd harmonics @ 12dB/O after @ 18dB/O after
³ ³ ³ ³ | -3dB point the damped
³ ³ ³ ³ ³ ³ | where Xc=R resonant Freq.
ÀÄÁÄÁÄÁÄÁÄÁÄÁÄÁÄÁÄÁÄÁÄÁÄÁ
0 1 2 3 4 5 6 7 8 9 11 MHz
The actual radiation from PCB track is proportional to the frequency in
addition to any resonant track. So a 1MHz square wave may be very weak at
11MHz as the PCB track length is very small compared to a 1/4 wave, but
radiate quite loudly at 50 to 144MHz where the track becomes a good aerial.
These principles also apply to SMPSUs as well as TV/Monitor Line stages,
just that it is very difficult to apply in those high power cases.
In part 2... Filtering @ source, Reducing QRM seen @ Rx, Operate when no QRM/
remove anoyance DSP etc.
Why don't U send an interest bul?
73 de John G8MNY @ GB7CIP
Read previous mail | Read next mail
| |
