Talk:Crest factor

Table re-organization
1. waveforms in table should be normalized to RMS, not peak. this is the normative normalization method.


 * As per the grid lines, the waveforms in the table are normalized to a 'peak' level of one. I don't know exactly what you mean by 'normative normalization method' but normalizing to peak value of 1 makes the most sense to me visually. --Kvng (talk) 19:26, 8 March 2010 (UTC)

2. since this article is about crest factor (peak to average factor), the relevant columns should be put closer to the beginning. i suggest this order:

waveform type, waveform plot, peak value, crest factor (linear), crest factor (dB)

RMS value of 1 is implied for all waveforms so it doesn't need to be a column.

non-relevant columns should be pushed to the end, or removed. right now they are at the center of the table, even though this is not an article about mean-rectifed magnitude, and waveform factor. (and waveform factor is not even related to crest factor/PAPR..)

3. all these mathematical derivations (square roots and ) are obscuring the main data. consider hiding them / moving them.

217.132.27.200 (talk) 21:19, 4 February 2008 (UTC)

let me correct myself re #2 - when waveforms are normalized to 1 RMS, the "peak value" column is equal to the "crest factor (linear)" column. so both columns should be merged into 1 column. this makes the table even simpler and more understandable.

I'm ready to do the corrections myself, but i would like some OK first so that some whimsical admin won't revert my work.

217.132.27.200 (talk) 21:23, 4 February 2008 (UTC)

Table
The table of the article is wrong.

The mean values of sine, triangle and square wave signals are all 0.

The values given are for (perfectly) rectified signals.

KjellElec


 * Yes. Also, the meaning of waveform factor is not explained. Omegatron 04:33, 21 October 2006 (UTC)


 * It appears to be the RMS divided by the mean, though I don't know how it is used. — Omegatron 07:52, 9 March 2007 (UTC)


 * I've made some bold edits to the table. Waveform factor seemed only to complicate things it is gone. As a reference point, peak level in all cases is 1. So perhaps we don't need the Peak magnitude column. The stragglers at the end of the table might want to be moved to their own table. --Kvng (talk) 19:26, 8 March 2010 (UTC)


 * You are correct and the article is still wrong. Sine triangle, and square have PAR=1.Skeptonomicon (talk) 22:36, 12 April 2016 (UTC)

Also: QPSK (which has zero-crossings in the constellation diagram) is incorrectly shown to have lower PAPR than OQPSK (which has no zero crossings and would ideally be constant modulus, except for overshoot); this shows the table is incorrect. — Preceding unsigned comment added by 2610:148:610:3C0D:B464:91F3:2A41:BDEE (talk) 17:22, 3 January 2020 (UTC)

Reformat
Please break up the table into individual images and editable math formulas in an HTML table. I'll do it if no one else does, but I don't have time right now. — Omegatron 19:38, 20 October 2006 (UTC)

The table should also be expanded to include Gaussian white noise, pink noise, compressed mastered CD audio, raw audio from a microphone, etc. etc. — Omegatron 19:59, 20 October 2006 (UTC)

Starting:



Obviously we should create dedicated images. These are just placeholders. Omegatron 04:12, 21 October 2006 (UTC)


 * I created better images and they are now in the article and here. — Omegatron 05:05, 9 March 2007 (UTC)

http://www.dliengineering.com/downloads/crest%20factor.pdf

I think we should more clearly discuss the mean values are obtained by rectification. At a quick-glance, these values seem incorrect, though they are just rectified.


 * What do you mean? — Omegatron 05:05, 9 March 2007 (UTC)

Waveform factor is the ratio of DC average to RMS and is used to scale resistors for measurements with DC or AC meters. The waveform factor for the half wave rectified sine wave should be 2.22 as the DC average is VP/Pi. —Preceding unsigned comment added by 212.77.217.195 (talk) 11:32, 26 September 2007 (UTC)

Peak Value?
Suppose we have a waveform whose peak values are asymmetric. That is, the magnitude of the negative peak is different from the magnitude of the positive peak. Which peak is used when calculating crest factor: the higher or lower magnitude? Or should the average of the peak magnitudes be used?

65.161.52.184 21:22, 18 December 2006 (UTC) Scott L.


 * The higher value would be used. — Omegatron 05:05, 9 March 2007 (UTC)

Whole premise of article is inconsistent, Term is ill defined
The term peak to average power ratio or crest factor, as used in electronics, is the peak of the power envelope divided by the average of the power envelope, and is not typically applied to signals directly. PAPR and PAR are almost exclusively used to talk about envelopes, though I am less sure that crest factor is limited to envelopes.

As an example of the inconsistency, a QPSK signal with all zeros (no modulation) is a sine wave, yet the table has different values for sine wave and QPSK. QPSK is correctly shown as the ratio of the peak power envelope to the average power envelop, but the sine is ignoring the envelope and incorrectly listing a PAR greater than one. The PAR of a sine wave is one. Skeptonomicon (talk) 22:31, 12 April 2016 (UTC)

Also, in the table, QPSK (which has zero-crossings like a sine wave) is shown to have lower PAPR than OQPSK (which has no zero crossings) and would ideally be constant modulus (except for overshoot); this shows the table is incorrect. — Preceding unsigned comment added by 2610:148:610:3C0D:B464:91F3:2A41:BDEE (talk) 17:15, 3 January 2020 (UTC)

Crest factor vs PAR
The source I am reading says that PAR is "identical to the traditional crest factor", implying that crest factor is an older term. — Omegatron 04:19, 26 July 2007 (UTC)

I still hear the term crest factor used all the time. Much more, in fact, than PAR


 * Me too, in my field it's much more common. It probably depends what field you're in --mcld (talk) 13:01, 29 January 2009 (UTC)


 * "Crest factor" is in IEE Std. 100. "Peak to average ratio" is not. A quick look on Google initially shows PAR meters are tropical fish tank light meters measuring Photosynthetically active radiation. We need a reference for PAR meter being a recognized term in some industry, else I'd be inclined to cut it out. --Wtshymanski (talk) 16:38, 29 January 2009 (UTC)
 * Yes, "crest factor" is defined in IEEE Std. 100 as the ratio of a peak quantity to the RMS value of the quantity. Examples provided are: (1) & (2) voltage, (3) any periodic function, etc.

I'm sorry, but "Average" or "Mean"?. And the crest factor is the "peak to rms" or "peak to average (or mean)"? It's very confused for me.--Vmsa (talk) 01:57, 31 May 2013 (UTC)

crest factor for stochastic signals
The crest factor as given here does not to seem very practical for stochastic signals. The reason is that it can always happen that there occurs an rare, but extremely high value. This high value would determine the crest factor for the whole measurement. I think that practically, some kind of "decaying maximum" is used in real measuring devices. In other words, the maximum is "forgotten" after a while. The article should describe how this is typically done in practice. --62.159.14.3 (talk) 12:32, 12 June 2009 (UTC)
 * For real-time measurements there's also a practical problem with using a true average. In practice, there are time constants associated with both peak and average readings. These are commonly first-order low-pass filters. --Kvng (talk) 19:07, 8 March 2010 (UTC)

Isn't MS an indication of average power - not RMS? —Preceding unsigned comment added by 116.193.184.242 (talk) 08:59, 21 October 2010 (UTC)


 * No: Root_mean_square 71.167.72.23 (talk) 14:58, 14 April 2014 (UTC)

PAR vs PAPR
I'm confused. PAR interpreted literally cannot be a synonym of crest factor, since average voltage != RMS voltage. For any waveform without DC, the average voltage is 0, so PAR is always infinite? I thought PAR was implicitly always a power ratio, since average power = RMS voltage and they then produce the same number. Now the article says PAR and PAPR are separate things? — Omegatron (talk) 16:31, 18 September 2013 (UTC)

PAPR of a sine wave
I've got a CCDF measurement up in my Keysight VSA software. I'm inputting a 2GHz sine wave, and the CCDF tells me the PAPR is roughly 0.2dB. Wikipedia told me 3dB, and now I'm spiraling into a world of confusion.

Your measurement for 2GHz sine wave aka CW is correct. PAPR for such a signal is 0dB. Wikipedia and also some authors have fooled themselves to believe that sine wave has PAPR of 3dB. It is true if the sinewave is an envelope of a RF signal.

In case a RF signal has an envelope of sinewave, in other words is an AM-modulated RF signal, it will have PAPR of 3dB if AM modulation index is 100%. In case a RF signal has an envelope of DC, you could call it a sine wave if you want, it will have PAPR of 0dB.

External links modified
Hello fellow Wikipedians,

I have just modified 7 external links on Crest factor. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20090821020320/http://broadcastengineering.com/mag/broadcasting_transitioning_transmitters_cofdm/ to http://broadcastengineering.com/mag/broadcasting_transitioning_transmitters_cofdm/
 * Added archive https://web.archive.org/web/20141130224524/http://www.ti.com/lit/ml/sloa082/sloa082.pdf to http://www.ti.com/lit/ml/sloa082/sloa082.pdf
 * Added tag to http://zeidler.ucsd.edu/students/thesis_sthompson.pdf
 * Added archive https://web.archive.org/web/20130526003113/http://www.atsc.org/cms/pdf/pt2/Wells_ATSC_paper_on_T2.pdf to http://www.atsc.org/cms/pdf/pt2/Wells_ATSC_paper_on_T2.pdf
 * Added archive https://web.archive.org/web/20060427062603/http://rfdesign.com/mag/radio_crest_factor_analysis/ to http://rfdesign.com/mag/radio_crest_factor_analysis/
 * Added archive https://web.archive.org/web/20100302151718/http://www.rane.com/par-c.html to http://www.rane.com/par-c.html
 * Added archive https://web.archive.org/web/20080628071722/http://www.atis.org/glossary/definition.aspx?id=2773 to http://www.atis.org/glossary/definition.aspx?id=2773
 * Added archive https://web.archive.org/web/20120318202105/http://www.atis.org/glossary/definition.aspx?id=6766 to http://www.atis.org/glossary/definition.aspx?id=6766

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 10:54, 14 August 2017 (UTC)

Don't write 'Effective value'; write 'RMS'
In first paragraph, don't write 'Effective value'; write 'RMS'. If you write 'effective value' that phrase should be defined and/or hyperlinked. It's just another layer for the reader to try to figure out. Just use 'RMS' then everyone knows what the crest (peak) is being compared to. 12.33.223.210 (talk) 18:57, 4 October 2018 (UTC)
 * I just asked my wife and she doesn't know what "RMS" means. "Effective value" is at least English and not acronymese. --Wtshymanski (talk) 16:23, 9 October 2018 (UTC)

Modulation types in the table
The introduction or the heading of the table should have a clear statement how the crest factor and PABR are calculated for modulated signals. One can't take the usual definition at face value since the modulated signal is a sine wave. Without any amplitude modulation (prefect phase or frequency modulation with a bandwidth much less than the carrier frequency) the usual definition would yield the crest factor of a sine wave, 1/√2, not the value of 1 usually given for such a signal. The German version mentions this. Unfortunately I am not knowledgeable enough to correct it in a good way - experts please jump in!

By the way, the values for QPSK are inconsistent; a crest factor of unity (if correct) should lead to PABR = 0 dB. - Anastasius zwerg (talk) 19:08, 2 March 2023 (UTC)


 * Agree that values for QPSK are inconsistent and should change. Furthermore, in the reference for QPSK, where the 1.76dB value came from, it was derived from a model with oversampling but without Nyquist filtering. In the words of the author, "Clearly this restriction is no realistic." In reality the PABR is a function of the Nyquist filter roll-off factor (PABR decreases as roll-off factor increases) and typical values are much higher than 1.76dB. Jarrielcook (talk) 21:36, 30 June 2023 (UTC)