hp-pn89400-14a_10 steps to a perfect digital demodulation measurement

8/14/2019 HP-PN89400-14A_10 Steps to a Perfect Digital Demodulation Measurement

1/16

Stepsto a

PerfectDigitalDemodulation

MeasurementProdu ct Note 89400-14A


2/16

2

Set the an alyzer 's center frequency and spa n for the s ignal under

test (vector mode) .

Hint: The cent er frequencies of the a nalyzer and inpu t signal mus t

ma tch t o within 3% of the s ymbol ra te (1% for 64QAM and 0.2% for

256QAM).

Check: The signal's spectr um should occupy 50-90% of the s pan , and not

extend beyond t he edges. It should appear as a distinct, elevat ed region

of noise 20 dB or more a bove th e an alyzer n oise floor. Broad, glitching

sidebands indicate bursts or other transients, which can be handled with

tr iggering or the P ulse Sear ch fun ction. These procedures ar e described

in steps thr ee and eight.

Set the input range as low as poss ible wi thout over load.

Hint: The green inpu t LED m ay or ma y not be on; the yellow overra nge

LED mus t not be on.

Check: Downra nge by one st ep; overload indicators sh ould tur n on.

Set u p Triggering , i f required.

Hint: Triggering is un necessary for most s ignals (see notes below

regar ding use of the HP 89400's interna l arbitr ar y waveform source).

For TDMA an d oth er bur sted signals, use Pulse Sear ch inst ead of

tr iggering (activated in step 8, r ight). Otherwise, choose trigger mode,

level, delay, holdoff, an d ar min g just a s you would with an oscilloscope,

in a ccordance with th e char acterist ics of your specific signal.

Hint: With t he Trigger men u selected, a display of Main Time:

Magnitude sh ows the curr ent t rigger level as a horizonta l cur sor,

super imposed on th e input waveform .

Check: With t riggering in use, the t ime domain display should show a

sta ble signa l tha t occupies >50% of the t ime span .

Select Digi tal Demodulat ion mode.

Hint: Be aware th at this mode retains th e settings from its previous

digital dem odulation m easur ement . If, for example, the pr evious s ymbol

ra te sett ing is incompa tible with your curren t frequency span , the

an alyzer may au tomat ically increase or decrease your frequen cy span .

Check: Make certa in tha t t he frequency span is still the same a s you set

in step 1.

The following general procedure is recommended as guidance for configuring the HP 89400 to demodulate and

characterize a wide var iety of digitally-modulated signals. Follow it st ep-by-step t o insure t hat no import ant setup

para meters a re missed or incorr ectly programm ed.


3/16

3

Enter the correct modulat ion format under Demodulat ion Setup.

Hint: This is simple unless you do not kn ow the modulation form at.

Ident ifying an un known signal form at is a significant challenge

requir ing specialized expertise.

Enter the correct symbol rate .

Hint: The symbol rat e needs t o be entered t o an a ccur acy of 10-100 ppm.

Hint: Ident ifying an un known symbol ra te is difficult, but it sometimes

helps to AM demodulate th e signal and look at th e high end of th edemod spectr um for a discrete signal at t he symbol ra te (it's easier to

see with avera ging). Also remem ber th at t he 3 dB bandwidth of the

signal is appr oximat ely equa l to the sym bol rat e. (Neither t echnique will

yield the accura cy required, but th e results m ay help you rem ember th e

actual numbers).

Select the resu lt length and points/symbol for the de sired display.

Hint: Result length is importa nt for burst ed signals, as you don't want

to include da ta symbols ta ken before or a fter t he bur st (i.e. you don't

want to demodulate noise).Hint: More points /symbol reveal more detail of the inter symbol

waveforms (for viewing eye diagra ms, er ror p eaks, et c.). Fewer

points /symbol allow you t o demodulat e more symbols per m easur ement

(remember: resu lt length * points/symbol must always be less than

max tim e points).

Turn on P ulse Search and s et the Sea rch Length (burst

s ignals only) .

Hint: Maximum search length can be ma ny times longer tha n a timerecord; set it to the burst "off" time plus twice the bur st "on" time t o

insure th at there's always a complete burst in every search.

Hint: This is a lso a good tim e to set up t he "Sync Search" function if

needed . Check online Help for a good descript ion of its capa bilities.


4/16

4

Select the f i l ter shapes an d alpha.

Hint: To choose th e right filters, r emember : the "measu red" filter should

simulat e th e inten ded receiver, while th e "reference" filter sh ould match

the product of the transmitter and receiver filters.Hint: If the a lpha is u nkn own, estima te it from (60 dB BW/3 dB BW)

1. EVM peaks between sym bols mean an incorr ect a lpha, but won't

usua lly affect t he E VM% readout (except in GSM). In t he worst case,

sta rt with 0.3 and ma nua lly adjust u p or down for lowest int ersymbol

peaks.

Final Checks .

Look for the following:

Overall shape of eye and/or constellation diagrams is regular and sta ble

(even if individual symbols a re very n oisy, th e char acteristic sha pe

should be evident). Eye opening should line u p pr ecisely with th e

display center line if not, twea k "clock a djust" for best center ing

(lowest EVM).

On the data t able display, the EVM% readout is fairly consistent from

measurement to measurement.

EVM vs. symbol trace is noise-like, but fairly uniform (short peaks are

OK). Abru pt chan ges within t he EVM tra ce signify problems, as do

sloping or V-shaped dist ribut ions. If pr oblems exist, consult the

ana lyzer's on-line H elp under "digital demodulat ion tr oubleshooting".

Spec ial procedures for using the HP 89400 arbitrary wa veform source

Demodulating signals created by the HP 89400's intern al ar bitrar y waveform source involves constra ints n ot

foun d with "real-world" signa ls. This is due to the fact tha t t he ar bitra ry signal is not continu ous, but subject t o

discontinuities in am plitude, phase, and/or timing each time th e waveform wra ps ar ound from end to beginning.

Therefore, in addition to th e above steps, always per form the following:

Use t r igger type = Internal Source .

Set a positive trigger delay to avoid the first 5-10 symbols of the waveform.

Choose a result length shorter th an th e arbitrary waveform dur ation (you can determine the length of thear bitra ry waveform empirically by increas ing result length unt il demod abru ptly loses lock).


5/16

5

Error vector m agnitude (EVM )

measurements can provide a great

deal of insight into the performance

of digitally-modulated signals. With

proper use, EVM and related m eas-

urements can pin point exactly the

type of degradat ions present in a

signal and can even h elp identify

their sources. T his n ote reviews the

basics of EVM m easurem ents on th eHP 89400 vector signal a nalyz ers,

and outlines a general procedure

that m ay be used to m ethodically

track down even the most obscure

signal problems.

The diverse t echnologies th at

compr ise today's digita l RF commu-

nications systems sha re in comm on

one main goal: placing digital bit-

stream s onto RF carriers and then

recovering them with accuracy,

reliability, and efficiency. Achieving

th is goal dema nds engineering timean d expertise, coupled with keen

insights into RF system performa nce.

Vector signal an alyzers such as the

HP 89400 perform th e time, fre-

quency, an d m odulation domain

ana lyses tha t pr ovide these insights.

Becau se th ey process signals in full

vector (magnitu de and ph ase) form ,

they easily accommodate the complex

modulation form ats used for digital

RF comm unications. Perha ps most

importan tly, these ana lyzerscont ribute a r elatively new type of

measu rem ent called "err or vector

ma gnitude" or E VM.

Pr imarily a m easur e of signal quality,

EVM provides both a simple, qua n-

tita tive figur e-of-mer it for a digitally

modulated signal, along with a far -

reaching methodology for uncovering

and a tta cking the under lying causes

of signal impairmen ts a nd distortion.

EVM measurements are growing

ra pidly in a cceptan ce, having

alrea dy been written into suchimporta nt system standards as

GSM1, NADC2, and PHS3, and they

ar e poised to appear in several

upcoming standards, including

th ose for digital video tr an smission.

This n ote defines er ror vector

magnitude and related measure-

ment s, discusses how they ar e

implement ed, and explains how

th ey are pr actically applied in

digita l RF commun icat ions design.

1 Global S ystem for MobileCommunications

2 North American Digital Cellular3 Personal Handyphone System

Usin g Error Vec tor Magn itud eMeas urem en ts to Ana lyze andTrou bles h oo t Ve cto r-Mod u late d Sign als


6/16

6

Unders tanding e rror

vector magni tude

EVM defined

Recall firs t t he ba sics of vector

modulation: digital bits ar e

tra nsferred onto an RF carr ier by

varying the carrier's magnitude

and phase such that , at each data

clock transition, the carrier occupies

an y one of several s pecific locations

on th eIversus Q plane. Each

locat ion encodes a sp ecific data

symbol, which consists of one or

more da ta bits. A constellation

diagr am sh ows the valid locations

(for example, the magnitude an d

pha se relat ive to the car rier) forall permitted symbols, of which

there must be 2n , given n data bits

tra nsmitted per symbol. Thus, to

demodulate th e incoming data, one

must accura tely determine the

exact m agnitude and pha se of the

received signa l for each clock

transition.

The layout of the constellation

diagram an d its ideal symbol

locations is deter mined generically

by the m odulation format chosen

(BPSK, 16QAM,

/4 DQP SK, et c.).The tr ajectory taken by the signal

from one symbol location to another

is a fun ction of the sp ecific system

implementation, but is readily

calculat ed nonetheless.

At any moment in t ime, the signal's

magnitude and pha se can be

measured. These values define the

actua l or "measu red" phasor. At the

sam e time, a corr esponding ideal or

"referen ce" pha sor can be calculated,

given kn owledge of the t ra nsm itted

dat a st ream , the symbol clocktiming, baseband filtering parameters,

an d so fort h. The differences

between these two phasors form t he

basis for t he EVM measurements

discussed in t his note.

Figure 1. Error vector magnitu de (EVM) and related quan tities.

MeasuredSignal

Q

I

Ideal (Reference) Signal

Phase Error (IQ error phase)

Magnitude Error (IQ error mag)

Error Vector

Figure 1 defines EVM and severa l

relat ed term s. As shown, EVM is

the scalar distan ce between t he two

phasor end points (the magnitudeof the difference vector). Expressed

anoth er way, it is the r esidual noise

and distortion r emaining after an

ideal version of the signal h as been

stripped away. By convention, EVM

is reported as a percentage of th e

peak signal level, usu ally defined

by the const ellation's corn er st at es.

While the err or vector ha s a pha se

value associated with it, th is angle

generally turns out to be random,

becaus e it is a function of both the

err or itself (which m ay or ma y not

be ran dom) an d th e position of the

dat a sym bol on t he const ellation(which, for all practical purposes, is

ra ndom). A more useful an gle is

measur ed between the actual and

ideal phasors (I-Q phase error),

which will be shown lat er t o contain

information useful in troubleshooting

signal pr oblems. Likewise,I-Q

magnitude error shows the ma gnitude

difference between th e actual an d

ideal signals.


7/16

7

Making EVM

measurements

The sequence of steps t hat compr ise

an EVM measurement are illustrat ed

in figur e 2. While th e HP 89400

Vector Signal Analyzers perform

th ese steps a utoma tically, it is still

useful to understa nd th e basic

process, which will aid in sett ing up

and optimizing the measur ements.

Step 2. Regenerating the

reference w aveform

The recovered data bits ar e next

used to create th e ideal referenceversion of the input signal. This is

again accomplished digitally with

powerful DSP calculat ing a

waveform t hat is both completely

noise-free and highly accur ate.

Step 3. Complex comparison

Taking the calculated r eference

waveform an d the a ctual incoming

waveform (both now existing as

blocks of digital samples), the two

need only be subtra cted to obtainthe err or vector values. This is only

slightly complicat ed by th e fact th at

both waveforms are complex,

cons istin g ofIand Q waveforms.

Fortu nat ely, the HP 89400's DSP

engine ha s sufficient power to

han dle this vector subt ra ction

and pr ovide the desired

measurement data.

Figure 2. Block diagram of the EVM measurement process.

Step 1. Prec is ion demodu lat ion

Following a na log-to-digital

conversion of the incoming signal,

a DSP 1 based demodulator recovers

the transmitted bitstream. This task

includes everyth ing from car rierand data symbol clock locking to

baseban d filtering. The H P 89400's

flexible demodulator can dem odu-

late signal form at s ra nging from

BPSK t o 256QAM, at sym bol rat es

from hundreds to several megahertz,

yet can be configur ed for the most

comm on signal types with a single

menu selection.

A final no te

The ana lyzer's ADC samples th e

incoming signal asynchronously, so

it will generally not provide actualmeasur ed data points at the exact

symbol times. However, a special

resa mpling algorith m, applied to

th e incoming ADC samples, creates

an entir ely new, accur at e set of

"virtu al sam ples", whose rate a nd

timing a re pr ecisely in sync with

the received symbols. (This is easily

seen on th e HP 89400 by observing

th e time-sample spa cings of an

input waveform, first in standar d

vector mode, and th en in digital

demodulation mode).

1 Digital Signal Processing


8/16

8

EVM Troubleshooting Tree

Measurement 1

Measurement 2

waveshapes asymmetric

symmetrical

noise tilted

Measurement 3

Measurement 4

Phase vs. Mag Error

IQ Error Phase vs.Time

Residual PM

Phase Noise

I-Q Imbalance

QuadratureError

Constellation

EVM vs. Time

error peaks

uniform noise

(setup problem clues)

Measurement 5

AmplitudeNon-Linearity

Setup Problems

Error Spectrum

discrete signals

distorted shape

flat

flat noise

sloping noise

Measurement 6

Spurious

Adj. Chan.Interference

Freq Response

Filter Distortion

SNR Problems

phase error >> mag error phase error mag error

Figure 3. Flow chart for analyzing vector modulated signals w ith EVM measurements.


9/16

9

Troubleshoot ing

wi th e rror vector

measurements

Measur ement s of err or vector

magnitude and related quant ities

can, when properly applied, provide

insight into the quality of a digitally

modulated signa l. They can also

pinpoint the causes of any problems

uncovered dur ing the testing pr ocess.

This section pr oposes a general

sequence for examining a signa l

with E VM techniques, a nd for

interpreting the results obtained.

Note: th e following sections a re n ot

intended as step-by-step procedures,but r at her a s general guidelines for

th ose who are already familiar with

basic operat ion of the HP 89400.

For a dditiona l inform at ion, consu lt

the instrum ent's on-screen H elp

facility or t he r eferences in t he

bibliography.

Measureme nt 1Magnitude vs . phase error

Descr ipt ion: Different er ror mech-

anisms will affect a signal in differentways, perha ps in ma gnitude only,

phase only, or both simultaneously.

Knowing the r elative amount s of

each t ype of err or can qu ickly confirm

or rule out certa in types of problems.

Thus, th e first diagnostic step is to

resolve EVM into its magnitu de and

phase error components (see figure 1)

and compa re th eir relative sizes.

Setup: from digita l demodulat ion

mode, select

MEAS DATA E rro r Ve ct or: Tim e

DATA FORMAT Data Table

Observe: When the average phase

error (in degrees) is larger than the

average magnitude error (in percent)

by a factor of about five or m ore,th is indicates t ha t some sort of

unwant ed phase modulation is th e

dominant err or mode. Pr oceed to

meas ur ement 2 to look for noise,

spur s, or cross-coupling problems in

th e frequency reference, phase-

locked loops, or oth er fr equen cy-

generat ing stages. Residual AM is

evidenced by magnitude errors that

are significant ly larger t han the

phase an gle errors.

In ma ny cases, the magnitude and

pha se err ors will be roughly equal.This indicates a broad category of

other potential pr oblems, which

will be fur ther isolated in meas ur e-

ments 3 th rough 6.

Measurement t ip

1. The error values given in t he dat a

table summar y are the RMS

avera ges of the error at ea ch

displa yed symbol point (except

GSM or MSK type I, which also

include th e inters ymbol errors).

-1 Sym 1 Sym

I - Eye

1.5

-1.5

EVM = 248.7475 m%rms 732.2379 m% pk at symbol 73

Mag Error = 166.8398 m%rms -729.4476 m% pk at symbol 73

Phase Error = 251.9865 mdeg 1.043872 deg pk at symbol 168

Freq Error = -384.55 HzIQ Offset = -67.543 dB SNR = 40.58 dB

0 1110011010 0110011100 0110011010 0100101001

40 0010100110 1000010101 0010010001 0110011110

80 1001101101 0110011001 1010101011 0110111010

120 1000101111 1101011001 1001011010 1000011001

16QAM Meas Time 1

Figure 4. Data table ( lower display) show ing roughly similaramounts of magnitude and ph ase error. Phase errors much largerthan magnitude e rrors wou ld indicate possible phase noise orincidental PM problems.


10/16

10

Measuremen t 2IQ phas e error vs. time

Descr ipt ion: Pha se error is the

instantaneous angle difference

between the m easured signal and

th e ideal reference signal. When

viewed as a function of time (or

symbol), it sh ows th e modulating

waveform of an y residua l or

inter fering PM signal.

Setup: from digital dem odulation

mode, select

MEAS D ATA IQ Error: Pha se

DATA FORMAT Ph ase

Observe: Sinewaves or other regular

waveform s indicat e an inter fering

signal. Uniform noise is a sign of

some form of pha se noise (random

jitter, residua l PM/FM, and so

forth).

Examples:

Measurement t ips

1.Be careful n ot to confuse IQ

Pha se Err or with E rr or Vector

Phase, which is on the sam e menu.

2.The X-axis is scaled in symbols.

To calculate absolute time, divide

by the symbol rate .

3.For m ore detail, expand the wa ve-

form by reducing resul t l ength

or by using th e X-scale marke rs.

4.The pra ctical limit for waveform

displays is from dc to approxi-

mat ely (symbol rat e)/2.

5.To precisely determ ine th e

frequency of a phas e jitt er spu r,

create an d display a user-defined

math function FFT(PHASEERROR).

For best frequency resolution in

the resulting spectrum, reduce

points/symbol or increas e

resul t l ength.

Figure 5. Incidental ( inband) PM sinewave is c learlyvisible - even a t only 3 degrees pk-pk.

Figure 6. Phase n oise appears random in the t ime domain.


11/16

11

Measuremen t 3Conste l lat ion diagram

Descr ipt ion: This is a comm on

graphical analysis technique

ut ilizing a polar plot t o display a

vector-modulated signal's magnitude

and ph ase relative to the carr ier, as

a function of time or symbol. The

pha sor values a t t he symbol clock

times are par ticularly important,

an d are highlighted with a dot. In

order t o accomplish this, a const el-

lation a nalyzer mu st kn ow the

precise carrier and symbol clock

frequencies and ph ases, either

thr ough an external input (tradi-

tional constellation displays) or

th rough au tomat ic locking

(HP 89400).


mode, select

MEAS D ATA IQ Measured Time

DATA FORMAT Polar: Constellation

(dots only)

or

Po lar: Vecto r

(dots plus int ersymbol

paths)

Observe: A perfect signal willha ve a uniform constellation that

is perfectly symmetr ic about th e

origin.I-Q imbalan ce is indicated

when t he const ellation is n ot

"square", tha t is when t he Q-axis

height does not equal the I-axis

width. Quadra ture er ror is seen in

an y "tilt" to t he constellation.

Examples:

Figure 7. Vector display show s signal path ( including

peaks) between symbols.

Figure 8. Conste l lat ion display s hows symbol points only,reveal ing problems such as compression (shown here) .

Measurement t ips

1. Resul t l ength (number of

symbols) determines h ow ma nydots will appear on the constel-

lation. Increase it to populate

the constellation states more

completely.

2. Points / symbol determines h ow

much det ail is shown between

symbols. To see peaks a nd over-

shoot, use four or more

points/symbol.

To allow a longer r esult length ,

use fewer points (in eith er case,

resul t l ength x points /symbolmust be less than max t ime

points) . One point per symbol

creates t rivial eye an d const el-

lation diagram s, because no

inter symbol data is collected a nd

all symbols ar e connected by

straight, direct lines.

3.To view the sp rea ding of symbol

dots m ore closely, move th e

mar ker to any desired state,press mkr -> ref lvl and then

decrease Y/div.

4.With norma lize ON , the

outerm ost stat es will always have

a va lue of 1.000. With normalize

OF F , the values are absolute

volta ge levels.


12/16

12

Measuremen t 4Error vector magn i tude

vs. time

Description: EVM is the difference

between the input signal and the

internally-generated ideal reference.

When viewed as a fun ction of symbol

or time, errors m ay be correlated t o

specific points on the inpu t wa veform,

such as peaks or zero crossings.

EVM is a scalar (magnitude-only)

value.


mode, select

DISP LAY tw o grids

ME AS D ATA ( A) e rr or v e ct o r: t im e(B ) IQ measured t ime

DATA FORMAT(A) linea r magn itu de(B) l inear magnitude

MARKER cou ple markers: ON

Observe: With mar kers on the two

tr aces coupled t ogether, position the

mar ker for the u pper (EVM) trace at

an er ror peak. Observe the lower

tr ace to see the signal ma gnitude for

the sam e moment t ime. Err or

peaks occurrin g with s ignal peaks

indicate compression or clipping.

Err or peaks t hat correlate to signalminima suggest zero-crossing non-

linearities.

The view of EVM vs. time is a lso

invaluable for spott ing setup

problems.

Measurement t ips

1.EVM is expressed a s a percenta ge

of th e out ermost (peak) sta te on

the constellation diagram.

2.The X-axis is scaled in symbols.

To calculate absolute time, divideby the symbol rate .

3.For more detail, expand the

waveform by r educing result

length or by using th e X-scale

markers .

4.With points/symbol >1, the

err or between symbol points can be

seen. An EVM peak between each

symbol usu ally indicates a problem

with baseban d filtering -- either a

misadjust ed filter or a n incorr ect

value for alpha entered into the

ana lyzer during setup.

5.The E VM waveform can also

highlight t he following setu p

problems:

Example:

Figure 9. EVM peaks o n this sign al (upper trace) occurevery t ime the signal magnitude ( lower trace) approacheszero. This is probably a zero-crossing e rror in anampli f icat ion stage .

Figure 10. V-shaped EVM plot du e to incorrectsymbol c lock rate .

Figure 11. EVM becomes noise at en d of TDMAburst; use sh orter result lengt h .


13/16

13

Measuremen t 5Error spectrum (EVM

vs. frequency )

Descr ipt ion: The error spectru m

is calculated from the FF T1 of th e

EVM waveform , and resu lts in a

frequency domain display tha t can

show details not visible in the t ime

domain. Note the following

relat ionsh ips which apply to the

error spectrum display:

frequency span = points/symbol x symbol rate1.28

frequency resolution ~ symbol rat eresult length

Setup: from the digital demodu-lat ion mode, select

MEAS DATA error vector: spectrumDATA FORMAT log magni tude

Figure 12. Interference from adjacent (lower) chann elcauses uneven EVM spectral distribution.

Example:

Observe: The display shows the

err or-noise spectr um of the signal,

concentr ated within the bandpass

an d th en r olling off ra pidly on

either side. In most digita l system s,

non-uniform noise distribution or

discret e signal peaks indicate th e

presence of externally-coupled

interference.

Measurement t ips

1.To change th e span or resolution

of th e EVM spectr um, a djust only

the points/symbol or th e result

length according to the equat ions

given a bove. Do not adjust the

ana lyzer's center frequency or

span , or t he signal will be lost!

2.With a linear m agnitu de display,

spectrum calibrat ion is EVM

percent. With log magnit ude, it is

in decibels relat ive to 100% EVM.

3.Averaging in th e an alyzer isapplied prior to demodulat ion;

thus, the er ror spectru m cannot

be smoothed beyond wha t is sh own.

4.Fr equency calibrat ion is absolute,

with the carr ier frequency in the

cent er. Use th e offset marker to

determ ine specific interference

frequencies relat ive to baseban d.

Figure 13. Switching pow er supply interference appearsas EVM spur, offset from the carrier by 10 kHz.

1 Fast F ourier Tran sform


14/16

14

Figure 14. In traditional netw ork analysis, the stimulu s

signa l does not n eed to be p erfect (flat, noise -free, etc.).It i s careful ly measured at the DUT input and th enratioed out of the measurement results .

Figure 15. With an HP 89400, the D UT inpu t sign al doesnot need to be measured or even provided, because i t i scalculated (regenerated) from the me asured signal , andis already in ideal form.

Examples:Measuremen t 6Channel frequency respon se

Description: This powerful, unique

HP 89400 measur ement calculates

the r atio of the measur ed signal to

th e reference signal. Becaus e the

latter is internally-generated a nd

ideal, it allows a frequ ency response

measur ement to be made across an

entire modulated system without

physically accessing t he modulator

input. Having this virtual baseband

access point is importa nt because

in most cases, such a stimu lus point

is usua lly una vailable, either

because it is a) inaccessible, b)

digitally implement ed, or c) the

aggregate of separ ate Iand Q inputs.

Setup: from the digital demodu-

lat ion mode, select

MATH Define Function: F1 =MEASSPEC/REFSPEC

MEAS DATA F1DATA FORMAT log ma gnitu deor phaseor group delay

Observe: The measurement results

show the aggregate, complex

tr ans fer function of the system

from the baseband Iand Q inputs

of th e modulator t o the point of

measurement. View the results

as a magnitude ra tio (frequency

response), a ph ase r esponse, or even

as gr oup delay. In h igh perform an ce

modulators, even the smallest

deviat ion from flat r esponse/linear

phas e can cause ser ious

perform ance problems.

Figure 16. The flatnes s of this digital TV transmitter isabout 0.5 dB from the modulator input to th e pow erampli f ier output, w ith the transmit equal izer on. Thismeasurement required no interruption of the videotransmission.

ref meas


15/16

15

Measurement t ips

1.See measurement t ips under

Measurement 5 "Error spectru m"

regar ding how to adjustfrequency span and flatness.

2.Use this t echnique primarily to

measure passban d flatness.

Dynamic ran ge is generally

insufficient for stopband rejection

measurements.

3.With data format:phase, the

deviat ion from linear -phase

response is easy to read, because

aut omatic car rier locking has

alrea dy removed any pha se slope.

4.This is typically a rat her n oisy

measur ement, because the

distribut ion of energy across the

passband is uneven an dconst ant ly varying. Thus, the

SN R1 of any individual frequency

point can vary dr am atically from

one measur ement to the next.

Averaging is a pplied prior to

demodulation an d will not h elp.

For group de lay measurements,

choose a wide aperture to

smooth out t he data.

5.Traces may be averaged manually

using tr ace mat h a s follows: save

a measu rement into D1; then

change to a display of F2 =(F1 + D1) / K1, with F 1 as defined

above, and th e num ber of

measurement s to be taken stored

in K1. As each meas ur ement is

made, save the resulting tra ce

into D1, repeat ing unt il all K1

measurement s have been taken.

Bibl iography

Blue, Kenneth J . et al. "Vector Signal Analyzers for Difficult Measu rem ents on Time-Var ying and Complex

Modulat ed Signals." Hewlett-Pa ckard J ourna l, December 1993, pp 6-59.

Hewlett-Packar d Compan y. "Using Vector S ignal Analysis in t he I ntegra tion, Troubleshooting a nd Design ofDigital RF Comm unications System s," Product Note HP 89400-8, Publication Nu mber 5091-8687E,

Palo Alto, CA. 1994.

Voelker, Kenn eth M. "Apply Er ror Vector Measu rem ent s in Comm un ications Design." Microwaves & RF,

December 1995, pp 143-152.

1 Signal-to-Noise Rat io


16/16

For more informat ion aboutHewlett-Packard test and measure-ment products , appl i cat ions ,serv i ce s , and for a current sa l e s

of f i ce l i s t ing , v i s i t our web s i t e ,http://www.hp.com/go/tmdir. Youcan also contact one of the followingc e n t e r s a n d a s k f o r a t e s t a n dmeasurement sales representative .

United States:Hewlett-Packard CompanyTest and Measu rement Call CenterP.O. Box 4026En glewood, CO 80155-40261 800 452 4844

Canada:Hewlett-Packard Canada Ltd.5150 Spectru m WayMississauga, OntarioL4W 5G1(905) 206 4725

Europe:Hewlett-PackardEuropean Marketing CentreP.O. Box 9991180 AZ AmstelveenThe Netherlands(31 20) 547 9900

Japan:Hewlett-Packard J apan Ltd.Measurement Assistance Center9-1, Takakura-Cho, Hachioji-Shi,Tokyo 192, J apa nTel: (81-426) 56-7832Fax: (81-426) 56-7840

Latin America:Hewlett-PackardLatin American Region Headquar ters5200 Blue La goon Dr ive, 9th F loorMiam i, Florida 33126, U.S.A.(305) 267 4245/4220

Australia/New Zealand:Hewlett-Packard Australia Ltd.31-41 Joseph Str eetBlackburn, Victoria 3130, Australia1 800 629 485

Asia Pacific:Hewlett-Packard Asia Pa cific Ltd.17-21/F Sh ell Tower, Times Squ ar e,1 Matheson Str eet, Causeway Bay,

Hong KongTel: (852) 2599 7777Fa x: (852) 2506 9285

Data Subject to ChangeCopyrigh t 1997Hewlett-Packard CompanyPrinted in U.S.A. 7/975966-0444E

hp-pn89400-14a_10 steps to a perfect digital demodulation measurement

Documents