bdr Receiver

Mar/Apr 2006 © ARRL

Jaders Prastgard63505 EskilstunaSweden

Blocking DynamicRange in Receivers

By Leif Åsbrink, SM5BSZ

An explanation of the different procedures anddefinitions that are commonly used for blocking

dynamic range (BDR) measurements.

IntroductionBlocking dynamic range (BDR)

may actually mean quite differentthings at different laboratories. Cir-cumstances and what definition ofBDR is being used affect performancecomparisons among receivers in thepresence of a strong, off-channel inter-fering signal.

Dynamic range in general is theratio (or difference in decibels) be-tween the weakest signal a system canhandle and the strongest signal thesame system can handle simulta-neously — without an operator switch-ing attenuators or turning volume

potentiometers. The concept is quitegeneral and by no means limited toradio receivers. Human sensors likethe ears and the eyes have very largedynamic ranges, for example. The un-damaged ear can detect a 1 kHz soundwave at a level of 10–12 W/m2 while theupper limit is about 1 W/m2, where westart to feel pain. The dynamic rangeof our ears is thus about 120 dB. Oureyes can detect the light from a starin the dark sky when about ten pho-tons per second reach the retina,which converts to something like10–13 W/m2. The Sun, with its 300 W/m2, does not damage our eyes unlesswe look straight into it.

Another example of dynamic rangeis the dynamic range of a vinyl musicrecord. It may be on the order of 60 to80 dB only, much less than the dy-namic range of our ears.

The above examples show the dy-namic range for a single signal. Thecorresponding dynamic range for areceiver is not particularly interesting.It relates the strongest on-channel sig-nal the radio can handle, without se-rious distortion (or damage), to theweakest signal it can receive. We mayobserve that a local very strong SSBstation sounds severely distortedwhen we tune to it even though thesame station does not cause any in-terference when we listen to other sta-tions on the same band. This is notlikely to be any problem to a radioamateur. The example is given just toshow the wide meaning of the conceptof dynamic range.

The rest of this article will deal withthe dynamic range pertinent to situa-tions where a single strong signal iscausing interference to the desired sig-


nal. This kind of dynamic range —blocking dynamic range (BDR) — maybe the most important figure of meritfor some radio amateurs while othersmay find that the interference createdby the simultaneous presence of twoor more strong signals is the limitingfactor. In such cases, the intermodu-lation-free dynamic range is the num-ber to look for when selecting whatreceiver to use.

The intermodulation-free dynamicrange is discussed frequently in ama-teur publications. It may be techni-cally challenging to measure properlybut there is no controversy about whatit means.

The Weakest SignalThe weakest signal a receiver can

handle is usually taken as the totalnoise power that reaches the outputas measured with a true RMS voltme-ter. This means that the low end of thedynamic range is taken as the signalat the antenna input that doubles theoutput power from the receiver. Themeasurement has to be done withoutAGC, and the doubled power outputmeans that the weak signal powerequals the noise power at the output.Any room-temperature resistor pro-duces a noise voltage that would trans-fer –174 dBm/Hz to a matched coldresistor. With the RF preamplifier dis-abled, a typical HF receiver may pro-duce 20 dB more noise with a room-temperature dummy load at the inputthan would an ideal receiver thatwould not add any noise of its own(only amplifying the noise from thedummy load). A receiver adding 20 dBof noise is said to have a noise figureof 20 dB. If the bandwidth were500 Hz, the noise floor referenced tothe antenna input would be –174 + 20+ 27 dBm = –127 dBm. (Note that 10log 500 ≈ 27.) This signal level is some-times improperly called MDS (mini-mum discernible signal) for such atypical receiver, even though a CWoperator would easily copy a signalthat is 10 dB weaker.

Picking the noise floor as the lowend of the dynamic range is typical forall dynamic ranges, not only in radioreceivers. The noise floor power is pro-portional to the bandwidth and there-fore a receiver will have 10 dB moredynamic range when measured at abandwidth of 200 Hz compared towhen it is measured at a bandwidthof 2 kHz. It is the same receiver,though, and the dynamic range differ-ences that depend on bandwidthshould not be included when differentreceivers are compared.

For that reason, receivers should 1Notes appear on page 39.

Figure 2 — The loudspeaker output from an FT-1000D with a dummy load at the antennainput. All settings are identical to those of Figure 1.

Figure 1 — The loudspeaker output from a Yaesu FT-1000D transceiver with a very leakytransformer close to it causing a fairly strong, but still absolutely inaudible, 50-Hz humsignal.

be measured at a standardized band-width. It could be 1 Hz, 500 Hz,2.4 kHz, or something else. Since thereis no de-facto standard, each labora-tory has to specify what bandwidththey use to allow correct comparisonsto results from other labs.

The Strongest SignalThere are two vastly different BDR

definitions that may give completelydifferent, although perfectly reproduc-ible, results. Michael Tracy, KC1SX,described them in QST.1 He writes:“BDR as a lab measurement normallyrefers to the point at which the weak(presumed desired) signal is reducedby 1.0 dB (‘blocked’) by the presenceof a strong (presumed undesired) sig-nal.” The procedure followed at the

ARRL Lab for BDR measurements isset up to find the point of 1 dB gainloss.2 A quite different definition de-fines the strong signal as the signalthat causes a predetermined degrada-tion of signal-to-noise ratio (SNR) fora weak signal (1 dB or 3 dB). In somecases, the two different definitionsmay give the same result; but in othercases, the resulting BDR may differby 40 dB or more.

The two different phenomena thatlie behind the two definitions are calledsaturation and reciprocal mixing, al-though the physical phenomena in thereceiver may be something else. A re-ceiver that uses front-end AGC to avoidsaturation, and that compensates forthe gain loss after the bandwidth-defining filters, will lose sensitivitywhen the off-channel signal turns thefront-end AGC on. As a result, the noise


floor will increase to cause a reducedSNR for a weak signal even though thetotal gain is unchanged. An RF ampli-fier that is supplied with inadequatelydecoupled dc voltage or that has a highimpedance for audio frequencies on thebase and very low impedance for au-dio frequencies on the emitter willamplitude-modulate the strong off-channel signal. Both these exampleswould be classified as reciprocal mix-ing even though the mechanism isquite different from what the classifi-cation implies.

The ARRL Lab definition of BDRthrough the procedure in Note 2 trans-lates to this definition in plain En-glish: “The BDR is the amount abovethe noise floor for an off-channel sig-nal that degrades the receiver forstrong signals.” It may be appropriatefor crowded HF bands where the de-sired signal typically is high above theinternal noise floor of the receiver. Areceiver with a very high number ac-cording to this definition is convenientto use because the operator does nothave to use an attenuator to shift thedynamic range up and down as condi-tions change. Radios that reach150 dB in the ARRL test reports are

Figure 3 — The loudspeaker output from an FT-1000D with a noise source at the antennainput. All settings are identical to those of Figure 1.

Figure 5 — The loudspeaker output of an FT-1000D with a14.178 MHz on-channel signal at a level of –77 dBm (10 dB belowthe on-channel CP1) as a function of the off-channel signal levelat 14.158 MHz. The dots are the on-channel signal level measuredat 5Hz bandwidth, the small crosses are the power (10 Hz to24 kHz) and the big crosses are the power measured without theon-channel signal.

Figure 4 — The loudspeaker output of an FT-1000D with a14.138-MHz on-channel signal at a level of –77 dBm (10 dB belowthe on-channel CP1) as a function of the off-channel signal levelat 14.158 MHz. The dots are the on-channel signal level measuredat 5 Hz bandwidth, the small crosses are the power (10 Hz to24 kHz) and the big crosses are the power measured without theon-channel signal.

not uncommon. The other definitiontranslates to this in plain English:“The BDR is the amount above thenoise floor for an off-channel signalthat degrades SNR for a very weaksignal.” This BDR is typically about100 dB.

The first definition is similar to thedynamic range for ears and eyes asdefined above. It is the total useful sig-

nal range but it does not contain anyinformation about the ability to detecta weak signal in the presence of astrong undesired signal. The seconddefinition is the true dynamic rangein the case of a single dominating off-channel signal. It defines the strongestsignal that may be present simulta-neously while the weakest signal isdetected.


Table 1

MDS measured with a wideband RMS voltmeter and noise power in 1 Hz bandwidth asmeasured with a spectrum analyzer in the loudspeaker output.

Transformer

ON OFFMDS(dBm) –110.3 –122.5NPD(dBm/Hz) –151.8 –151.8

Measurement Accuracy Issueswith the Weakest Signal

Finding the noise floor by connect-ing an RMS voltmeter to the audiooutput may give inaccurate results.The human ear does not respond wellto hum from the ac mains frequencyof 50 or 60 Hz. Figure 1 shows theloudspeaker output from a Yaesu FT-1000D HF transceiver that had a veryleaky transformer placed close to it.This pure 50-Hz hum is absolutely in-audible despite the fact that it isabout 30 dB above the noise floor inthe spectrum, which has a bin widthof 12 Hz. This inaudible 50-Hz humdominates the loudspeaker signal. Itactually lifts the reading of a typicalRMS voltmeter by 12 dB and its pres-ence degrades the MDS value by thesame amount. The FT-1000D performsabsolutely as well with the leakytransformer next to it; there is no au-dible difference whatsoever and thehuge sensitivity difference is an arti-fact of the measurement procedure.

Even without the leaky trans-former, there are some low-frequencyspurs in the FT-1000D, as can be seenin Figure 2. This signal is 27 dBweaker than the 50-Hz hum causedby the leaky transformer, so it is onlyabout 2.5% of the total power thatreaches the loudspeaker and the er-ror it introduces is only about 0.1 dB.Figures 1 and 2 were recorded in CWmode with the FT-1000D bandwidthset to 500 Hz. It is pretty obvious fromFigure 2 that noise outside the desiredbandwidth gives a noticeable contri-bution to the total power seen by anRMS voltmeter. The passband is wellvisible in Figure 3, where a noisesource is connected to the antennainput. From Figure 3 we can estimatethe noise floor in a 500-Hz windowabove the passband to be about 28 dBon the relative dB scale. From Fig-ure 2, we can read the noise floorwithin the passband as 35 dB. Thatmeans that the noise power in therange 900 to 1400 Hz is only 7 dB be-low the noise power within the desiredpassband, and therefore causes a read-ing that is about 20% higher than itwould be if a selective voltmeter wereused. The noise above the passband iswell audible, but it is not really dis-turbing and it does certainly not de-grade the effective sensitivity of theFT-1000D by as much as 20%, as theMDS value would indicate.

By use of exactly the same settingsas for Figures 1 to 3: Preamp off (IPO),AGC off, max RF gain, 50% AF gain,500 Hz bandwidth in CW mode, MDSwas measured at 14.120 MHz as de-scribed in Note 2 using an RMS volt-

meter that was flat from 10 Hz to24 kHz. The measurement was madewith and without the very leaky trans-former that makes the difference be-tween Figures 1 and 2. The results areshown in Table 1, which shows the noisepower density (NPD) referenced to theantenna input. These settings formkind of standard settings for a trans-ceiver test and they are used in all mea-surements presented in this article.

The noise power density is 22.2 dBabove –174 dBm/Hz, which meansthat the noise figure of this particularreceiver is 22.2 ±1.2 dB with the con-trols set as in this test. The noisepower density can be used to computethe noise power that would pass anideal filter with a bandwidth of500 Hz. The result is –124.8 dBmwhich is 2.3 dB better than the resultobtained in the MDS measurementwith the RMS voltmeter. The noisepower density is a selective measure-ment and any 50-Hz hum does notaffect it at all.

Since the actual detection is donein a much narrower bandwidth than500 Hz either by machine or by a hu-man being, I argue that the noisepower density in the vicinity of thesignal is the true noise floor for anMDS measurement. Any widebandnoise, hum or high-frequency audiosignal that may add several dB to anRMS voltmeter measurement is insig-nificant — and in those cases whereit is audible, there should be a notesaying, “This radio suffers from humor high frequency noise.” It may beappropriate to mention here that themains hum originating in poorly fil-tered power supplies typically is apulse train at twice the mains fre-quency. Contrary to the magneticallycoupled pure 50/60-Hz sinewave, 100/120-Hz pulse trains are well audible.

The noise power density can be mea-sured easily with a spectrum analyzerthat will give the noise surrounding acarrier in dBc/Hz. With a carrier ofknown power at the antenna input, thenoise power density comes out directly.

SNR loss for a weak signal is chosen,there are no accuracy problems be-cause the definition requires SNR fora weak signal to be monitored. Thedefinition in Note 2, where the read-ing of an RMS voltmeter is monitoredwhile the off-channel signal is in-creased, that may lead to unexpectedresults sometimes.

To illustrate the problems, the BDRof the FT-1000D was measured withthe procedure in Note 2. The loud-speaker output was also monitored ona spectrum analyzer on which thelevel of the on-channel signal wasmeasured. The on-channel response ofthe FT-1000D is linear from the noisefloor at –122.5 dBm up to –70 dBm.The point of 1-dB compression is–67 dBm. Following the procedure inNote 2, the on-channel signal level wasset to –77 dBm. Figures 4 and 5 showthe result of the measurement at a fre-quency separation of 20 kHz aboveand below the desired frequency.

An inspection of Figures 4 and 5shows that the reciprocal mixing noiseis equal to the noise floor power (MDS)for an off-channel signal level of–26 dBm regardless of whether the on-channel signal is above or below theoff-channel signal. The dynamic rangeis therefore –26 dBm – (–122.5 dBm)= 96.5 dB. If the noise floor (MDS)were measured by means of the noisepower density, the same dynamicrange would result because then thereciprocal mixing noise power densitywould equal the FT-1000D noisepower density at a signal level of–28.3 dBm.

The FT-1000D has lost sensitivityby 3 dB at a signal level of –28.3 dBmalthough the gain is unaffected up to–22 dBm (20 kHz below) and –14 dBm(20 kHz above) an off-channel signal.The first nonlinear phenomenon thathappens is that the gain increases. Thepoint of 1-dB compression as measuredwith a selective voltmeter is +8 and+10 dBm, respectively. When measur-ing the loudspeaker output with anRMS voltmeter, 1 dB compression oc-curs at +12 dBm when the on-channelsignal is 20 kHz above the off-channelsignal; but when the on-channel sig-nal is 20 kHz below the interference,

Measurement Accuracy Issueswith the Strongest Signal

If the BDR definition in terms of


the reading of the RMS voltmeter re-mains constant within 0.5 dB all theway up to an interference power levelof +20 dBm (100 mW). It would prob-ably remain constant at much higherinput levels until the front end wasdamaged.

It should be quite clear from Fig-ures 4 and 5 that the procedure usedup to now at the ARRL Lab may givevery inaccurate results because theRMS voltmeter does not distinguishbetween the desired signal and noise.Using the average from 20 kHz aboveand 20 kHz below, BDR as definedfrom the point of 1-dB saturation is131.5 ±2.2 dB above MDS using theARRL definition for MDS, or133.8 ±2.2 dB when MDS is definedin terms of the noise floor power den-sity. The much higher values reportedin QST is caused by the use of an RMSvoltmeter.3

ConclusionsWhen using the concept of dynamic

range in Amateur Radio, we shouldrefer to signals present simulta-neously at the antenna input. Thismeans that BDR — implying thatblocking means that the ability to copythe desired signal as blocked by astrong off-channel signal — for theFT-1000D is 96.5 dB. When the de-sired signal is placed at –77 dBm (seeNote 2), the point of saturation, whichwas +20 dBm in QST (see Note 3) hasto be compared to –77 dBm for a dy-namic range of 97 dB, not to the MDSvalue measured under quite differentcircumstances. The value of 150 dBreported in QST is not the dynamicrange for two simultaneously presentsignals. It is the dynamic range for asingle signal and is not of much inter-est to a radio amateur. The point of1-dB saturation is 133.8 dB above thenoise floor, which means that the op-erator can place the system noise floorup to 37 dB above the internal noise

floor of the FT-1000D without seriousproblems because of gain compressionon strong signals. This leaves ad-equate margin for a VHF enthusiastwho uses the FT-1000D together witha transverter that might lift the noisefloor by 10 dB and a mast mountedpreamplifier that has to lift the noisefloor by another 16 dB to get a systemnoise figure that is within 0.1 dB ofthe preamplifier itself.

Both BDR definitions convey im-portant information on the usefulnessof a receiver. ARRL is considering theuse of an AF spectrum analyzer andthe reporting of both measurementsin future product reviews.

Editor’s note: Leif ’s observationsabout human vision form an apt anal-ogy to receiver dynamic range. Thebaseball outfielder who drops a fly ballwhen the Sun shines in his eyes knowsabout blocking effects. He employs anattenuator (his sunglasses) to reducethe interference but the attenuationmay, in fact, put the image of the de-sired signal (the baseball) below hisnoise floor. That’s the same as what afront-end AGC does to your receiver: Itraises the noise figure and may actu-ally make it harder to recognize thedesired signal.

The point is that you must measurethe small-signal end of dynamic rangeunder the same interference conditionsused to measure the large-signal end.Only then do you truly measure dy-namic range. In addition, the band-width in which noise-floor power ismeasured must be accurately known. It’snot enough to dial in a nominal 500-Hzfilter and assume its effective band-width is exactly 500 Hz. An uncertaintyof ±150 Hz in bandwidth would resultin additional uncertainty of at least2.6 dB in noise powers between receiv-ers. Passband and stopband shapes af-fect effective bandwidth. You cannot goby the 3-dB points alone.

As early as 1998, we noted the dif-ference between blocking dynamicrange and reciprocal-mixing dynamicrange on these pages. Now it is a mat-ter of how previously measured dataare corrected or models recharac-terized, and procedures updated, in-cluding measurement uncertainties. Tothe best of our abilities, we must adoptprocedures that accurately measurereal effects in receivers.

A 3rd-order IMD dynamic range mea-surement on an analog-to-digital con-verter, for example, may be superfluouswhen performance is dominated by over-load instead. Likewise, another mea-surement, such as BDR, should not becharacterized as “noise-limited” whenthe noise contribution can be discernedseparately from the modeled effect.

Notes1M. Tracy, KC1SX, “Product Reviews —

In Depth, In English,” QST, Aug 2004,pp 32-36.

2Ed Hare, KA1CV (now W1RFI), “SweptReceiver Dynamic Range Testing in theARRL Laboratory,” QEX, Jun 1996, p 16.

3Product Review, QST, Mar 1991, pp 31-36.

Leif was born in 1944 and licensed in1961. He holds a PhD in physics andworked with research on molecularphysics for 15 years at the Royal Insti-tute of Technology. Since 1981, he hasbeen running his own company devel-oping various electronic products. He isthe inventor of the magnetic intermodu-lation EAS system now owned by Check-point. Leif is essentially retired andworks mainly with Amateur Radio re-lated technical projects.

Documents

bdr Receiver