FreeDV---peak power vs average power

[Rick, W4XA]

Howdy all, 

I have been doing a few tests to determine power levels and have made measurements  with 3 peak reading watt meters and 2 (UN-modified) NON-peak reading Bird 43 watt meters (1 with a 1000w slug and another with a 250W slug) 

Knowing that most peak reading watt meters are really optimized for voice envelope measurement it's not surprising that the 3 peak reading meters indicate slightly more than 3 times higher than the Bird 43.

For example, Just now, I was indicating 50W on the Birds  using both the 1000W and 250W slugs.

OTOH, my  Array Solutions Power Master II indicates about 150W, the Mercury IIIs amplifier built-in (peak-reading) watt meter indicates around 150-175w and the watt meter in the Palstar HFAUTO  tuner indicates around 130-150W (all peak reading) 

It's not surprising that those peak reading watt meters would read higher than the Bird,  but I if I remember correctly, there's supposed to be  a roughly 1dB difference between average and peak power with RADE.

I could understand if I only had 1 Peak reading and 1 average reading meter (one might be inaccurate)  but with 3 different peak reading meters and 2 Birds with different slugs(but agreeing with each other) ,  it would appear that Peak power is considerably more than "1dB" (even more than 3dB) 

None of this probably matters,  but I am just wondering what the "real deal" actually is.

73/Rick

W4XA

[Peter Reichelt]

Hi Rick 

See this article from David VK5DGR https://freedv.org/david-july-2025-waspaa-paper-papr-rade-v2-snr-bbfm-otc-demo/ where he measured approx 4.5  dB when he expected about 0.8dB.

73 Peter VK5APR

[Takehiko Tsutsumi]

Hi Rick,

Thank you for sharing your work results for us.

In fact, I did the similar measurement reading David's blog pointed by Peter. Here is the result done January 21, 10 days earlier than your post.

See the table below.

1. Notes to be able to read table written in Japanese.

Title: Transmission Filter Bandwidth and Measured Peak Power 

First line: Bandwith (kHZ)

Second line: Transmitter Fiter Bandwidth (Hz). Thetis (SDR) transmitter filter setting LOW and HIGH.

Third line: Average Power (W) Measured by DAIWA Analog Meter (W). The center value of needle movement.

Fourth line: Peak Power measured by ATU-100 (antenna tuner). The value is the maximum reading of LCD for more than 10 seconds.

Fifth FPAPR and PAPR(dB) By the reason that two meters are uncorrelated, multiplied 1.25 correlation factor to average power to get PAPR and PAPR dB. The correlation factor is obtained as Peak Power is 20W and Average Power is 16W to the same unmodulated CW signal of Thetis program.

2. Comments comparing your and my measurement works.

1)  It is fairly similar values obtained between two independent works, i.e. your "slightly more than 3 times higher" and my PAPR such as 2.7 at 2.5kHz and 3.7 at 2.0kHz. 

2) Have you or are you going to measure the effects of transmitter bandwidth and the PAPR like I did?

3) I tend to agree your conclusion i.e. "None of this probably matters" if it is based on the judgment from SSB nature of PAPR characteristic saying more than 6dB (?). However, I wish to remind that it is better to set your transmitter filter as much as wide to the legal limit i.e. 3.0kHz. I believe modern transceivers supplied by radio amateur venders can offer almost 3.0kHz bandwidth values. Otherwise, you may break the legal limit of power as it is defined by peak, not average.

4) I do not know exact PAPR design target of "digital voice systems", perhaps not RADE designers one, except "smaller is better if you manage harmful things" by my measurement done.  Agree?

Regards,

I am glad to receive your reply if you are interested in to continue your thread.

take

de JS1MAV/JA5AEA

 

2026年2月1日日曜日 16:13:17 UTC+9 Peter Reichelt:

[Rick, W4XA]

Hi Takehiko, 

I am sure what you mean by "Otherwise, you may break the legal limit of power as it is defined by peak, not average"

That might be a problem only if one is operating at or near the maximum PEP allowed in their particular country (1500W in the US)   I don't know of many operating near those power levels here.  I tend to use the lowest power for reliable operation and certainly no more than around 500W PEP measured by 3 different PEP indicating watt meters.

I tried using the 500-2500 limit offered by the ICOM IC705 because it appears that there are seemingly spurious components of the transmitted signal that exceed the supposed 1.5kHz RADE bandwidth.

I attempted to verify this (quite unscientifically) using 2 different type SDR receivers (KIWI and N. UTAH SDR's) 

Visually, it appears that there are components outside that 1.5kHz bandwidth that were "removed" or reduced to  2.0 kHz by using the IC705's own SSB-D TX bandwidth controls  (500-2500Hz is evidently the lower limit of the control in the IC705)

Then varying those limits while observing the SNR of the receiving station did not appear to produce a reduction  in SNR when adjusting from the widest setting to to the lowest of  500-2500Hz

Obviously the best way to measure would be using a calibrated spectrum analyzer but using the "poor-mans" analyzer (spectrum scope on today's transceivers and SDR's)  seems to produce some believable results!

It appears that those signal components outside of 1.5kHz are produced at the audio level (meaning the USB audio coming from the software/computer) and are "faithfully" and accurately  being transmitted by the radio.   

What I do not know is if those components of the signal are being produced by the encoder or by the computer itself.

Cheers, 

Rick

W4XA

[Rick, W4XA]

Correction:    I am NOT sure what you meant  by legal limit peak/average power etc......

[David Rowe]

Fine work Take,

As per the blog post I have been looking into the PAPR issue for RADE V1
and V2.  In simulation I get around 4.3dB PAPR when RADE V1 is passed
through a 2700 Hz SSB style filter.  This is a good match to your results.

The 0dB PAPR from RADE V1 is only possible by using the full 8 kHz
bandwidth of the sampled signal, which might be possible with a SDR Tx.

It also suggests that operators narrowing their Tx bandwidth with
filters will be delivering a poorer SNR to the receiver.  For a given
peak power Tx, the PAPR is higher and hence RMS power is lower with a
narrower filter.  Filtering does of course clean up the transmitted
spectrum.

RADE V2 has a much cleaner Tx spectrum: narrower 99% occupied power
bandwidth (800 Hz or so) and better PAPR (around 3 dB). Hopefully the V2
signal will pass without any PAPR changes through most radios, allowing
all operators to achieve a higher RMS Tx power.

Cheers,
David

On 9/2/26 16:32, Takehiko Tsutsumi wrote:
> Hi Rick,
>
> Thank you for sharing your work results for us.
>
> In fact, I did the similar measurement reading David's blog pointed by
> Peter. Here is the result done January 21, 10 days earlier than your post.
>
> See the table below.
>
> 1. Notes to be able to read table written in Japanese.
>
> Title: Transmission Filter Bandwidth and Measured Peak Power
>
> First line: Bandwith (kHZ)
>
> Second line: Transmitter Fiter Bandwidth (Hz). Thetis (SDR)
> transmitter filter setting LOW and HIGH.
>
> Third line: Average Power (W) Measured by DAIWA Analog Meter (W). The
> center value of needle movement.
>
> Fourth line: Peak Power measured by ATU-100 (antenna tuner). The value
> is the maximum reading of LCD for more than 10 seconds.
>
> Fifth FPAPR and PAPR(dB) By the reason that two meters are
> uncorrelated, multiplied 1.25 correlation factor to average power to
> get PAPR and PAPR dB. The correlation factor is obtained as Peak Power
> is 20W and Average Power is 16W to the same unmodulated CW signal of
> Thetis program.
>
>

> --
> You received this message because you are subscribed to the Google
> Groups "digitalvoice" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to digitalvoice...@googlegroups.com.
> To view this discussion visit
> https://groups.google.com/d/msgid/digitalvoice/24137114-c873-409a-994c-e6689fad8e94n%40googlegroups.com
> <https://groups.google.com/d/msgid/digitalvoice/24137114-c873-409a-994c-e6689fad8e94n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Rick, W4XA]

David, 

Thanks so much for clarifying the PAPR differences with V1 and V2!    I did only a few try's over the air and used the Reporter SNR's for real-time results.  

I didn't see a reduction in SNR when varying the TX bandwidth from 100-2900    to 500-2500 Hz    but I did notice the slight difference in actual bandwidth viewing the actual signal on both the N. Utah SDR and a couple of different local Kiwi's

If I read you correctly, we should always  use maximum bandwidth available in the actual radio settings?   ( for example 100-2900 Hz in the ICOM IC705 SSB-D tx setting? ) 

Regards, 

Rick/W4XA

[Richard Lamont]

FWIW I did some tests today with RADE V1 and an ANAN 100W rig with adaptive pre-distortion. The nominal PEP was 100W but I don't have a way of measuring PEP accurately.

With the SSB TX bandwidth set to 150-2850 Hz, a Bird 43 showed 48 Watts.
With the SSB TX bandwidth set to 675-2275 Hz, a Bird 43 showed 41 Watts.

This was with a 500W slug, so these figures should be treated with caution. On the face of it they imply PAPRs of 3.2dB and 3.9dB respectively.

In practice, operators typically do not want their 'whiskers' to overlap an adjacent signal. So in terms of channel spacing the 99% occupied bandwidth may be less relevant than, say, a -40dB emission bandwidth. If we could keep that under 2kHz then users might be happy to work at 2kHz spacing.

Regards,

Richard G4DYA

[David Rowe]

Hi Rick,

> If I read you correctly, we */should always  use maximum bandwidth
> available in the actual radio settings?/*  ( for example 100-2900 Hz

> in the ICOM IC705 SSB-D tx setting? )

For the current RADE V1 waveform yes.

Although in practice it might only matter for weak signals where every
dB counts.  RADE won't sound any better with 23dB of SNR if you already
have 20dB.  As pointed out by others on this thread you may wish to
limit the Tx bandwidth to be more friendly to adjacent users.

These differences will be hard to measure over the air at the Rx side as
real channels bounce around so much, you need to do it on the bench at
the Tx side, as reported by a few people on this thread.

Cheers,
David,

[Takehiko Tsutsumi]

David,

Greeting.

It has been long time after I put my very short comment to your last July blog about "Localized SC-FDMA". Unfortunately, there was no reply from you at that time.

Again, I revisited the article titled "Single Carrier FDMA for Uplink Wreless Transmisson" pubished by IEEE Vehicular Technology Magazine September 2006 and I suggest you try to expand  the gap of RADE V2 subcarrier frequency, three times for example, and measure the effect of PAPR reduction. See Figure 5 (a) LFDMA (=Localized) Blue Solid Line  vs. IFDMA (=Distiributed) Red Solid Line. This is 64 subcarriers QPSK example much larger than RADE V1 and V2 with single carrier OFDM case but I believe we are talking the same topic and 4G experience is telling us that we should spread the subcarriers as much as possible, in our case from current 800Hz to legal limit i.e. 3kHz if we absolutely want PAPR reduction and one third narrow channelization like 700D is not the subject. Keep in mind both narrower and distributed channelizaation techniques has the same SNR and frequency efficiancy improvement effects.

Regards,

take

de JS1MAV/JA5AEA

2026年2月11日水曜日 8:41:23 UTC+9 David Rowe:

[Takehiko Tsutsumi]

To whom, who can not find the article itsef,

I found the figures of the article mentioned in my previous mail from images through Google searh.

Avobe is the explanation of "Distibuted" and "localized". Localized Mode is current RADE V2 channel structure and Distributed Mode is my suggetion to try.

Avobe l (eft figure) is Figure 5 (a) mentioned in my previous mail. The article auther used I(nterleaved ) FDMA to explain Dstibuted Mode. Distributed Mode PAPR is "ZERO" and Localized Mode is "7dB at e-3 Pr"  in the case of 64QPSK, and 4 Interleaved case. I suggest "three interleaved case" to David exactlly pictured in Figure 1 to fit legal limitation of 3kHz.

Regards,

take

de JS1MAV/JA5AEA

2026年2月11日水曜日 10:46:21 UTC+9 Takehiko Tsutsumi:

[David Rowe]

Hi Take,

Thank you for suggestions.  My time is limited and I cannot explore
every suggestion or idea that is sent to me.  I encourage you or anyone
else who is interested to investigate this area.

Thanks,
David

> --
> You received this message because you are subscribed to the Google
> Groups "digitalvoice" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to digitalvoice...@googlegroups.com.
> To view this discussion visit

> https://groups.google.com/d/msgid/digitalvoice/e811cabd-dd51-428c-86de-753fe3ce4b85n%40googlegroups.com
> <https://groups.google.com/d/msgid/digitalvoice/e811cabd-dd51-428c-86de-753fe3ce4b85n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Takehiko Tsutsumi]

Rick,

Please refer FCC rules in your country. SSB measurement of power is "peak" and not "average". On the other hand,  other modes CW and FT8. is "average".   the bandwidth is 2,700Hz by -6dB bandpass filter which is equivlent to 3kHz with 99% energy method in Europe and Asia.  8KHz reqirement which David mentiond is our of questionand you may be jailed by the operation in Japan.

Am I answering your questionn properly?

Regards,

take

de JS1MAV/JA5AEA

2026年2月11日水曜日 3:12:07 UTC+9 Rick, W4XA:

[Takehiko Tsutsumi]

Rick,

I interprete he has been saying he prefers 8kHz bandwidth setting but it is ilegal and therefore we should always  use maximum bandwidth available in the actual radio settings. ICOM IC705 SSB-D tx setting has 3kHz setting with default setting as FIL1(3.0kHz). Chek the low and high to get 3.0kHz with your rig.

Regards,

take

de JS1MAV/JA5AEA

2026年2月11日水曜日 5:20:32 UTC+9 Rick, W4XA:

[Takehiko Tsutsumi]

Richard,

I am glad you are teting with pure signal enabled. 

I agree with your 48W nd 41 W readings by each bandwidth. It is the same trend with me.

On the other hand, I suggst you will find to measure PEP properly with each bandwidth for your self or you will find the best realiable number by ths thread. As David stated "For a given
peak power Tx, the PAPR is higher and hence RMS power is lower with a narrower filter" and my measured data support his statemenet.

I hope the PEP would not change by pure signal enabled or disbled.

Regards,

take

de JS1MAV/JA5AEA

2026年2月11日水曜日 7:34:57 UTC+9 Richard Lamont:

[Takehiko Tsutsumi]

David,

No, SDR venders would never agree to "full 8kHz bandwidth modification" unless otherwise each regulatory bodies in the world change their regulation clause. Evenn I know it is a piece of cake for SDR and I am temptating to wewrite my SDR program code to be able to see <0.8dB PAPR with full 8kHz  in my shack with dummy load. As my time is limited as well, I have benn believing you for a while.

Regards,

take

de JS1MAV/JA5AEA

2026年2月11日水曜日 5:01:48 UTC+9 David Rowe:

[Takehiko Tsutsumi]

Richard,

I am a little bit confused. If you have ANAN100 and Thetis combination, you can measure with Thetis SeTup>DSP>AGC/ALC checking "Use peak meter readings for TXCOMP and ALC" . I hope the value is changed by bandwidth.

Regards,

take

de JS1MAV/JA5AEA

2026年2月12日木曜日 13:04:43 UTC+9 Takehiko Tsutsumi:

[Rick, W4XA]

Take, 

I am not sure we are talking about the same thing.

In the US, maximum allowable power output is 1500W Peak Envelope Power (or less in certain circumstances) .  This would be referring to AM and SSB and measured by either a calibrated scope or a peak reading wattmeter.

With other modes like FM, PM,  CW,  RTTY,  FT8 or other "continuous wave" like modes, average and peak power are virtually  the same.

Peak reading wattmeters appear to be optimized for analog AM or SSB , but are they accurate with all digital modes?

In all practicality, I don't think it matters much since it seems few people would operate using power levels near the legal maximum.

I think David was telling me that it's not necessary to limit the bandwidth in the transmitter and doing so would make SNR suffer at the receiving end. 

The audio signal bandwidth is controlled by the software before it ever gets to the transmitter making any filtering at the transmitter unnecessary. 

If I visit Japan  (which I have many times in the past)   I will be sure not to use any radio devices!  I certainly do not want to be jailed there!

Cheers, 

Rick/W4XA

[Takehiko Tsutsumi]

David,

Thank you for your quick response.

I thought my suggestion is valuable for your RADE V2 project, but I understand your situation.

By the way, I found out the article I introduce is received following award.

  2021 BEST MAGAZINE PAPER AWARD Recognizes the best magazine paper published in the IEEE Vehicular Technology Magazine Hyung G. Myung, Junsung Lim, David J. Goodman for the paper entitled “Single Carrier FDMA for Uplink Wireless Transmission,

Anyway, I am glad I could introduce a valuable article for IEEE society after 15 years publication to the audience of this forum.

Regards,

take

de JS1MAV/JA5AEA

2026年2月12日木曜日 7:53:12 UTC+9 David Rowe:

[jdow]

Rather than boatloads of conjecture perhaps injecting some science or even engineering into the picture might help. Mooneer has developed a multiple frequency modulation scheme. Normally that suggests the peak voltage may be as high as N times the average power with the average power being N times the power in a single frequency. Plot out the sine waves to show what I mean.

WAY back in the 70s a fellow wrote up an idea in one of the RF design trade magazines of the time. He noted that a direct digital frequency synthesizer is equivalent to a Yamaha sampler of the era. Well, he didn't mention that connection but it is obvious upon reading his article if you know both synthesizers and music samplers. He found an application for a many many channel signal generator, one signal for each AM broadcast channel. Bury a lossy cable in the road in areas with potential problem choke points. Light it up with warnings strong enough to override the broadcast channels when there is a mass pile-up or the like. So he simply programmed all the frequencies into his DDS look up ROM. It worked well but had a very nasty peak to average level issue. So he got imaginative. He did not start all the sinewaves at the same instant. Each channel got a phase shift relative to the others. He dramatically reduced his peak to average level. I believe I passed this on to Mooneer a long time ago. If he followed this the peak to average level will be less.

With that in mind, and SDR technology on the table, setup a 500 kHz wide give or take spectrum display with an SDR and a paper clip for a super-low efficiency antenna. The number you are going after is how much power can you transmit without dramatic increases in the frequency band you occupy. So start low (amplifier rating / 2N) and run your power up until you see the splatter become too much, I'd go for 50dB down but ham amplifiers are probably not that good. That gives you the power not to exceed. And it MAY be astonishingly low compared to your expectations.

The other approach is to simply simulate generating N channels of sound at some 1 unit level and at each sample in your simulation of the modulating waveform measure the power. Calculate PEP however you want from that data.

Guessing and "it seems real" is usually trumped by actual testing when you visit the real world. (And passing along your results might be useful.)

{^_-}    Joanne/W6MKU

--
You received this message because you are subscribed to the Google Groups "digitalvoice" group.
To unsubscribe from this group and stop receiving emails from it, send an email to digitalvoice...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/digitalvoice/2f341d52-d2ce-4232-8e0c-c7441e331825n%40googlegroups.com.

[Richard Lamont]

Take,

I'm using piHPSDR not Thetis.

Regards,

Richard

> >> https://freedv.org/david-july-2025-waspaa-paper-papr-rade-v2-snr-bbfm-otc-demo/ <https://freedv.org/david-july-2025-waspaa-paper-papr-rade-v2-snr-bbfm-otc-demo/>

> >> To view this discussion visit https://groups.google.com/d/msgid/digitalvoice/24137114-c873-409a-994c-e6689fad8e94n%40googlegroups.com <https://groups.google.com/d/msgid/digitalvoice/24137114-c873-409a-994c-e6689fad8e94n%40googlegroups.com> <https://groups.google.com/d/msgid/digitalvoice/24137114-c873-409a-994c-e6689fad8e94n%40googlegroups.com?utm_medium=email&utm_source=footer <https://groups.google.com/d/msgid/digitalvoice/24137114-c873-409a-994c-e6689fad8e94n%40googlegroups.com?utm_medium=email&utm_source=footer>>.

> >
>
> --
> You received this message because you are subscribed to the Google Groups "digitalvoice" group.

> To unsubscribe from this group and stop receiving emails from it, send an email to digitalvoice...@googlegroups.com <mailto:digitalvoice...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/digitalvoice/13a9594f-26df-4458-9736-98a01e514663n%40googlegroups.com <https://groups.google.com/d/msgid/digitalvoice/13a9594f-26df-4458-9736-98a01e514663n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Takehiko Tsutsumi]

Richard,

I checked PiHPSDR manual and 3.8 Meter section says

 During TX, the output power is displayed, provided that the radio actually reports this power. The output power meter can be calibrated (see the PA menu). If the SWR exceeds a threshold for SWR warnings (the default is 1:3, but this can be changed in the TX menu), the SWR indicator turns red. If, in addition, SWR protection is enabled in the TX menu, the output drived will be reduced to zero if the SWR exceed that threshold. Furthermore, the ALC (automatic level control) value of the transmitter is shown. Negative ALC values (at least in peak mode) indicate that the volume of the TX input audio could be increased to get full output power. Further info on the meters (e.g. switching between ,,peak” and ,,average” reporting) is described in the Meter menu.

I also checked real PiHPSDR on Kubuntu and clicked menu on the right top of the window.  Then, PiHPSDR - Meter widget says TX ALC Reading has three option "Peak", "Average", "Gain". 

My Redpitaya does not have power senser and I can not further check "Average" and "Power" are properly working or not.

 

You said "The nominal PEP was 100W but I don't have a way of measuring PEP accurately.".

In this measurement, you should be able to measure "Average" power without Bird 43 and  you can calibrate the value with Bird 43 if PiHPSDR properly works.

Please let's me know the result.

Regards,

take

de JS1MAV/JA5AEA

2026年2月12日木曜日 20:41:20 UTC+9 Richard Lamont:

[Takehiko Tsutsumi]

Rick,

I do not say all gigital modes are peak reading. FT8 is a digital but itoday t is measured by either "average" or "peak" because everbody knows PAPR of FT8 is  "x1" or "0dB" as Joe Tayler declared it is a popular pure Continuos Wave in his papers and we confirmed his paper is correc by measurement . So, we are doing the same thing. But it is not easy as the modulation is a new RADE OFDM and  we are informed it depends on the transmitter bandwidth,

Yes, it is quetionable whether analog AM or SSB optimized Peak reading is acuurate for RADEs or not. It should solved either David or we have to discuss what Peak measurement is appropriate or not. I agree with you that is does not matter as long as RADE stays within 3kHz audio spectrum and measure enough time period to be able to have confidence it is the peak of measurement.

Well,  I think PAPR is either zero or 3dB is matter for Dxing but is it not a matter for NVIS i .e above 10dB SNR condition provided when the bad is opened. I have been monitoring JA stations whether they found any concers about RADE V1 performance about path loss for a year but I can say they are generally positive about it.  I guess they can enjoy Dxing with push buttoned QSO by FT8 without language concerns.  How about US about the RADE QSO performance between West cost and East for example ? Are they care about 3dB difference?

I have been talking about clauses of Radio Wave laws.  You know they are very kind as long as you behave as gentleman.

Regards,

take

de JS1MAV/JA5AEA

2026年2月12日木曜日 15:54:46 UTC+9 Rick, W4XA: