PEP? Are Those Power Meters Showing Us PEP Peak Envelope Power, or just Peak Power?

Discussion in 'Amateur Radio Amplifiers' started by KQ4X, Dec 31, 2012.

Thread Status:
Not open for further replies.
ad: L-HROutlet
ad: l-rl
ad: Left-2
ad: L-MFJ
ad: Left-3
ad: abrind-2
ad: Subscribe
  1. W2VW

    W2VW XML Subscriber QRZ Page

    You're thinking of W3PHL.
  2. W5BIB

    W5BIB Premium Subscriber QRZ Page

    WHEW !!... is it over ?? MILLER TIME......:)
  3. K4FMX

    K4FMX Ham Member QRZ Page

    Thank you for your advice of reading up on various forms of power sensors.
    I do not profess to be any sort of expert in power measurement but I do think that I have more than a basic understanding.
    But as to "diversion" I think I have to give you the award on that one. :>)

    The problem that I see here is your terms of what is and is not contained in PEP and what I am describing.
    I realize that voltage peak reading meters are going to read the peak sum of all signals present.

    What I started off saying was that IMD will add to the PEP of a signal and give an error in true PEP reading on a peak reading meter.

    By true PEP I am considering only the wanted signal and not the distortion products. PEP readings will differ with and without distortion products.

    If an amplifier is tuned for maximum output the PEP is going to be the same whether IM products are contained in the output or not, as read on a peak reading meter.
    But if the amplifier is operated below the maximum output level then adding or deleting IM products will be seen on the meter. (PEP will be greater with IM products)

    The whole reason of my last post was to try and show that the author of the paper was saying that when using the indirect method that PEP could be computed accurately by 2 times the average power reading.

    *(indirect method, measuring average power of the signal with a thermocouple type meter)

    Quote from the author's paper:
    "This is not technically correct because the PEP is always less than twice the average power in the presence of harmonic and intermodulation distortion."

    My point was that if the PEP is LESS than 2 times the average power in this situation then PEP can not consist of anything but the two times sum of the two original tones of the signal.

    Harmonics or IM products can not be part of true PEP or the PEP would be at least 2 times the average power as the average power meter is going to give the sum of both tones plus the IM products and the harmonics when reading average power.

    No peak reading power meter involved here.

    I am not trying to "cross over" between finding PEP by the average power method and the peak voltage reading method. Only trying to define what the author is considering to be contained in PEP.
    What do you think? Would PEP be LESS than 2 times average power?

    Please, let's stick to this one point and not shotgun the subject.

    Gary K4FMX

  4. G0HZU

    G0HZU QRZ Member

    Well first of all I apologise if you may have taken my recommendation about reading up on power sensors as a bit patronising Gary. It was meant in good faith and wasn't meant to be patronising :)

    My knowledge about power sensors is reasonably good too and it came from reading up about them and also from using them regularly here and at work. I thought you were mixing up the characteristics of each sensor type in your thought process and I'm sorry if I got this wrong and caused any offence.

    Now I've read your post above and it does contain some bits that I may be misinterpreting because of your interpretation of PEP. So I'm not going to comment on your theory until we both agree about what PEP really means.

    Note that I'm not criticising the pdf article. I just don't see it as contributing very much info about Vpk type power sensors apart from the place in the article where it states the extreme sensitivity to harmonics? But we both know and agree about that extreme sensitivity anyway.

    Not sure where this average power meter analysis is headed but here's my answer below although I'm not sure it is of any use to you or anyone else as I don't see my answer as particularly enlightening...

    Note that you are asking me to comment on the large signal behaviour of a hypothetical test amplifier of which I know nothing about in terms of configuration, complexity or bias class or filtering etc.

    However, I'll give my answer for a 'typical' amplifier test at highish drive level :) Assuming two clean RF tones set a couple of kHz apart are fed to something like a single stage RF amplifier then I'd expect the PEP to become measurably less than 2 times average power as the amplifier starts to generate significant IMD. I think that's the answer you were expecting from me anyway.

    However, I suppose if you used a multi stage amplifier which generated significant IMD in the driver(s) then the relationship about PEP always being <= 2* average power could get microscopically challenged at certain drive levels if the driver distortion terms affect the final amplifier. But we are talking tiny differences here at acceptable IMD levels and I'm falling asleep thinking about the relevance of this.

    Still a bit confused about this anyway... In what way is this answer going to be useful? What would you say if I simply gave an answer of yes and what would you say if I said no?

    IMO the definitive way to measure PEP is to look in the time domain and cherry pick the RF cycle with largest Vrms. Then compute across to average power as (Vrms * Vrms)/50. Some of the Vrms could be a tiny contribution from harmonic terms but it is likely to be vanishingly small even with -30dBc harmonics.

    If the harmonics are filtered to well below -50dBc then the PEP can also be computed from the Vpk of the RF cycle using a Vpk detector as Vpk*Vpk/100 as there will be minimal uncertainty caused by the phase of the harmonic(s) if the harmonics are this small.

    If the harmonics are significant then the measurement uncertainty with the Vpk detector starts to grow dramatically hitting +/-6% on a -30dBc harmonic.

    That's my view on it?
  5. K4FMX

    K4FMX Ham Member QRZ Page

    A few days ago I reread the HP app notes on power sensors but they don't have much on measuring PEP. Lots of info on various peak power and average power etc.

    Again, getting back to what the author of the paper in question said,
    "This is not technically correct because the PEP is always less than twice the average power in the presence of harmonic and intermodulation distortion."
    Why would you think that would be the case unless he is only considering the power in the two tones to be part of PEP?
    I could see that happening if the amplifier was driven into compression where the average power could go up while the PEP does not. But below any compression?

    Not trying to put you on the spot, just trying to establish why this would always be true.

    Gary K4FMX

  6. G0HZU

    G0HZU QRZ Member

    I'm still unsure as to why you are taking us down this path Gary?

    However... you are asking again for a one size fits all 'true or false' answer for all cases and I can't give it because the large signal characteristics of an amplifier will depend on its topology, biasing and also the device type. eg A FET device will have an inherently non linear characteristic for large signals as it has a square law response.

    I think the author of the article was just referring to a basic model of an amplifier or passive device here. i.e. a basic large signal model where the device follows
    the classic input/output graph with a graceful trend into compression. With such a simple model the IMD are generated due to the subtle onset of compression
    of the peaks of the signal where the two signals are phased together. In this case the PEP will always be less than 2 * average power because the model will
    show flat topping of the RF cycle.

    I could probably design/simulate something to break the rule when measured but it would be a horrible and dirty design full of distortion as it would have to have an undesirable non linear gain characteristic. i.e. It would have to show notably higher gain for higher level drive signals compared to low level drive signals. But then this is not my rule to defend anyway? I'm not that bothered to try and break it either?

    More to the point I don't see any link here with respect to the Vpk detector issue?
  7. G0HZU

    G0HZU QRZ Member

    Ah yes, but with a Vpk detector (rather than Vrms) you would get a +/- 4% measurement uncertainty window with 2 harmonics each at -40dBc. eg 2nd and 3rd harmonic each at -40dBc.

    There are some ham power meters that claim high accuracy on cw signals but what the maker won't tell you is that they will lose their spec if you get a couple of harmonics popping up.

    Eg some ham meters meters with Vpk detectors claim +/- 5% accuracy on a cw signal into a 50R load but if you get two harmonics at -40dBc this will degrade to a measurement uncertainty window of +/-9% depending on the phases of the harmonics.

    Not such a good spec anymore?

    The whole point of this thread is to highlight that nearly all ham power meters are simply Vpk detectors on the FWD and REV ports of the coupler. In order to read across to average power they rely on the signal being a pure sine wave.
  8. K4FMX

    K4FMX Ham Member QRZ Page

    Well ok, here is where I am at on this.
    An amplifier with a two tone input signal that has an output of 250 watts average power in each tone has a total average power output of 500 watts. It is not driven into compression. There is headroom available. The PEP will be 1000 watts, two times the total average power. Let's say that harmonics and IM products are down some 50 dB and we ignore them in power measurements.

    Now we introduce a 3rd order IM product that is -30dB below one tone. That would be .25 watts for each 3rd order IM signal.
    (There are two of those present. An upper and lower, sum and difference)
    Still operating the amplifier below compression.
    If we measured the average power again we would have 500.5 watts average power out.
    Calculating PEP output would give us:
    250 watts = 111.8 volts
    250 watts = 111.8 volts
    .25 watts = 3.54 volts
    .25 watts = 3.54 volts
    Total volts = 230.68 = E^2/R = 1064.26 watts PEP with 2 IM signals.

    Any time there are more than two tones in the signal we can not just double the average power to find PEP. In this case we have 4 tones present with unequal values so we can add all the voltages together to calculate PEP.
    We should measure the same PEP on a peak reading diode type watt meter as it is going to read the highest peak voltage (sum of all signal voltages).
    Average power can still be measured accurately with a thermal type power meter. It will read 500.5 watts average power.

    I don't see how PEP could be less than 2 times average power when IM products or harmonics are present unless the amplifier is driven into compression.

    Gary K4FMX

  9. G0HZU

    G0HZU QRZ Member

    Look Gary, I'm getting seriously confused here :) You are the person who introduced the above quote from a pdf article. It's not my pdf and it's not my quote to defend. If you think it's flawed then why did you bring it to the forum in the first place? It isn't relevant anyway?

    You now seem to be clutching a 'long term average' power meter in one hand and in the other hand you have somebody else's rule about 'long term average power' vs PEP that YOU introduced that YOU don't fully agree with. Where is the sense in that???

    I've already told you it's possible to design a type of amplifier to break that rule. What on earth has this rule got to do with typical Vpk reading meters sold to hams anyway?

    No, it won't give an 'error'

    Any harmonic or IMD contribution to the largest RF cycle at the crest is VALID if you take the literal meaning of PEP as average power in 1 RF cycle at the modulation crest. PEP is not a measurement of signal quality. You don't get to decide which bits that contribute to the cycle are wanted and which are not. It is just a measurement of average RF power contained in that single RF cycle.

    The problem with a typical ham meter is that the VALID harmonic energy will be misrespresented by a Vpk reading meter because you cannot read across from the Vpk to the Vrms of that (harmonically) distorted RF cycle at the peak using Vrms = Vpk*0.707. You can only do this with a sinewave. A Vpk meter will suffer significant measurement uncertainty depending on the phase of the harmonic. However, a Vrms meter will measure the harmonic contribution correctly and get the PEP correct.

    IMD will not affect the purity of the RF sinewave at the modulation crest. Obviously, this is based on a typical ham SSB radio operating up in the MHz/GHz region modulated with audio. i.e. a huge difference in modulation frequency vs RF frequency. Also I assume there are no instability induced distortion terms anywhere or any other types of spurious eg from the radio's synthesiser.

    So can you stop blowing smoke and PROVE your above quote about an IMD induced 6% error in PEP reading with a Vpk meter is true or retract it?

    (please? :) )
  10. K4FMX

    K4FMX Ham Member QRZ Page

    The purpose of my introducing that pdf article was to question/clarify what signals PEP actually consisted of. Earlier I stated that only wanted signals should maybe be considered in true Peak envelope power as that was what we would be interested in. This of course could only be measured with a selective/tuned power meter.
    I didn't say that I agreed or disagreed with the article nor am I asking you to defend it. I was looking for some input as to why it would be stated that PEP is always less than 2 times average power when harmonics and IM products are present. The article stated that 2 times average power was the most accurate way of measuring PEP without harmonics and IM products of any significance.

    I probably should have left the last sentence off my last post as I was not necessarily trying to connect to the pdf article in question.
    "I don't see how PEP could be less than 2 times average power when IM products or harmonics are present unless the amplifier is driven into

    My last post was to show that with the IM products present there is only a small increase in average power but a rather significant increase in PEP.
    Do you agree with what I said in my last post?

    The same amplifier with and without IM products present will show approximately 6% more PEP with the IM products?

    I previously called it an "error" in meter reading but that may not be the correct term. However I am still saying that there will be a 6% difference in PEP reading in the presence of the -30 dB IM product verses no significant IM product level.
    Would you agree with that?

    I understand what you are saying with the harmonics causing the error.

    Gary K4FMX

Thread Status:
Not open for further replies.

Share This Page