Radioberry output power to full blast

[Samuel Lourenço]

Hi,

After the last Raspi OS update I was forced to update the driver, since the output power got somewhat lower. However, right after the Radioberry driver update, the output power went full blast, as in pedal to the metal, pegged at 23 dBm. The tune drive and PA settings are ignored, and the issue persisted after updating piHPSDR as well.

I recall that the PA settings were glitchy, because the power would back down after certain points. Now it is pegged to the max, beyond it was able before. The Tx output is connected to my 10 W amplifier from 60 dBm, which accepts 0 dBm, and thus I'm using a 20 dB attenuator. I assume that some IAMP setting was unlocked, but it is not healthy for the Radioberry nor for the amplifier, and the latter can go to 25 W momentarily when tuning, which is not great.

By the way, on a slightly different topic, I'm in the process of building a 1.2 W amp for the Radioberry, and I'm designing it to accept 12 dBm, and gain is set to roughtly 9 V/V, so as to output 31 dBm. My rationale for such high gain is that it provides more flexibility, enabling compatibility with many versions of software and hardware, and excessive gain can always be removed by the use of attenuators or in software. However, it would be essential for this to be ironed out. The output power should be limited to 10 W, and the tuning power even less so.

I need help, and by the way, the current script for piHPSDR installation does not compile, so I'm still using the old script.

Kind regards, Samuel Lourenço

[pa3gsb]

Hi Samuel,

I have seen multiple complaints for the same issue.

It is not clear at least for me what your setup is .

https://github.com/pa3gsb/Radioberry-2.x/wiki/Radioberry-Registration   you can find a way to describe your setup in terms of installed software .

This makes it possible to help you better.

73 Johan

PA3GSB

Op maandag 15 december 2025 om 09:59:18 UTC+1 schreef samuel.f...@gmail.com:

[Samuel Lourenço]

Hi,

The main issue here is that it seems to me that the driver is not in a stable state. I would suggest creating a stable production driver for RPi 4 and another one for RPi 5, and let the development do itself on other repositories.

Anyway, I'm using RPi 4 and Aursinc Radioberry with CL25, and an external 10 W amplifier from 60 dBm, the last one controlled by a relay board of my creation.

Before the update, the PA gain controls per band worked somewhat. Now they don't. I ask what will get broken next? I have a suggestion: make it so that my board never goes to TX, or gets stuck in Tx, while at it. :)

Anyway, I can't have a radio that changes everytime. And I'm considering forking the driver and do my own edits, including ripping the spyware part completely off.

Kind regards, Samuel Lourenço

--
You received this message because you are subscribed to the Google Groups "Radioberry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to radioberry+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/radioberry/ce0b1543-e662-4dcb-9a18-1506c0191b86n%40googlegroups.com.

[pa3gsb]

Samuel,

Thanks for your positive feedback.

I have added some improvements and maybe some new issues but anyway enjoy.

73 Johan

PA3GSB

Op dinsdag 16 december 2025 om 09:20:19 UTC+1 schreef samuel.f...@gmail.com:

[Samuel Lourenço]

Hi,

Meanwhile, how can I apply the workaround? IIRC, the script does not keep the source code. It would be essential to set the output power to a constant value for all bands. My objective is to have 20 dBm to the amplifier, so it can output 10 W for all bands.

Kind regards, Samuel Lourenço

[Samuel Lourenço]

Hi,

I've decided to download and reinstall the driver and piHPSDR. The issue is now fixed and I can adjust the output power on each band. Now I can output 10 W and tune it to 5W, which is great! Thanks!

Kind regards, Samuel Lourenço

[Łukasz Radzikowski]

Hi everyone,

I’m experiencing a similar issue. In my case, full power is available only on the 80 m band, about half on 40 m, very little on 20 m, and almost nothing on 10 m – and this is not related to the amplifier’s linearity. Measurements taken directly at the Radioberry TX port show a clear voltage drop across a 50 Ω dummy load.

In my case, the drop in power occurred after an update and coincided with building the amplifier.

When I started working on the amplifier, I first measured the Radioberry output level and then built the amplifier. It turned out that it didn’t work properly across all bands. It took me some time to realise that the issue was actually with the Radioberry.

Samuel, could you let me know which firmware and gateware versions, which SDR application, and which version of Raspberry Pi OS you are using to achieve consistent output power from the Radioberry? I have tried various combinations, but I can’t seem to reproduce my original configuration.

Thank you and best regards,
Łukasz SQ5BAR

[Samuel Lourenço]

Hi Lukasz,

I'm using the following:
- Radioberry driver: v73.3 (which doesn't say much, because it is a moving target)
- piHPSDR: v2.6 (WDSP v1.29)
- Raspi OS: Bookwork
- Raspberry Pi 4 and Radioberry V2.0, accompanied with a cased 10 W amplifier from 60 dBm and Radioberry Relay (custom RX/TX board)

You will have to adjust the power for all those different bands. I've already maxed out the output capability on 10 meters, and still doesn't produce the full 10 W, but rather 8W. I've had to reduce the output for the lower bands a bit. I can confirm that this non-linearity is from the Radioberry itself. BTW, don't use the tune function from piHPSDR (the output is reduced by default on purpose), but an external program like WSJT-X to generate the tone, when doing the calibration. It will be much closer to the performance you want to get.

Also, mind that you will, apparently, output less power when doing voice, but it is all an illusion. If you whistle to the mic, you will see.

The Radioberry should output about 20 dBm, or something in the vicinity of that. Have that in mind when using an amplifier. In my case, I've had to use a 20 dB attenuation.

Kind regards, Samuel Lourenço

[Łukasz Radzikowski]

Hi Samuel,

thank you for your detailed reply — that helps a lot.

In my case the situation seems a bit more severe than what you describe. I understand and accept that Radioberry is not perfectly flat across bands and that some adjustment per band is needed, but what I’m seeing is much stronger:

• 80 m → full level
• 40 m → about half
• 20 m → very low
• 10 m → almost no signal

What is important:
this behavior is already visible directly at the Radioberry TX output on a 50 Ω dummy load, and even earlier — I can see the drop already on the AD9866 pins, so it’s definitely not caused by the amplifier.

I also verified that:

• the issue persists across different software combinations (including older builds),
• it appeared after an update, but now I cannot reproduce the original “good” configuration,
• probing the signal (even after the first stage) significantly affects the level, which suggests the source impedance may be higher than expected or the output stage is very sensitive to loading.

Compared to your case (≈20 dBm and only moderate drop on 10 m), mine looks more like a strong frequency-dependent attenuation rather than just normal non-linearity.

A few questions that might help narrow it down:

1. Do you also observe any noticeable drop already at the Radioberry output itself (before your amplifier), especially on 20 m / 10 m?
2. Did you ever compare behavior between different gateware versions or kernel versions?
3. Are you using the standard 1:1 output transformer or did you modify the output stage?

Regarding the tune function — thanks for pointing that out, I’m already using an external tone source for measurements.

At this point I’m trying to determine whether:

• this is just a “normal” limitation of Radioberry TX,
• a gateware/configuration issue,
• or possibly something degraded in the AD9866 output stage or its surrounding circuitry.

Any additional observations from your setup would be very helpful.

Johan, maybe you have more insight into this?

Best regards,
73 Łukasz
SQ5BAR

[Samuel Lourenço]

Hi Lukasz,

The attenuation you are seeing is not normal. I suspect a counterfeit
AD9866 chip, which is probably a rebranded equivalent, although I
didn't find inferior alternatives for the AD9866. Or the sampling rate
is somehow severely limited. Did you measure the actual output
frequency when producing a tone? I think it would be interesting.

Somehow, on 10 meters, the behavior you are having is as if you were
approaching the Nyquist frequency. In reality, you should be well
below that, even on the top band. For instance, I did experiments
using the AD9851 as a AM emitter, and I could see some attenuation at
10 meters, despite the frequency being set at 150 MHz (25 MHz x 6).

Other hypothesis, is that the output transformer (balun) on your
Radioberry is either defective or inadequate. It would be interesting
to observe the signals before and after the transformer.

To see if that is not a gateware issue, you will have to try to
equalize the output at different frequencies via piHPSDR. Mind that my
results are after equalization, but before equalization I would still
see 5 W being produced at 10 m. I've confirmed that the limitation is
not from the amplifier itself, but from the Radioberry. The amplifier
that I use is actually very linear. I might add, though, that I've
seen different behaviors between gateware and piHPSDR versions, but
nothing that dramatic.

Kind regards, Samuel Lourenço

> To view this discussion visit https://groups.google.com/d/msgid/radioberry/b4da2bbe-443c-452f-bc22-533a1bc27442n%40googlegroups.com.

[Łukasz Radzikowski]

Hi Samuel,

Thank you. We already did some measurements here.

Before the recent update, 10 meters was working correctly, but now I see only about 180 mVpp on that band.

What I observed is:

• about 720 mVpp at 3.5 MHz into 50 ohms,

• about 680 mVpp at 7 MHz,

• about 360 mVpp at 14 MHz,

• about 180 mVpp at 28 MHz.

I also checked the signal directly around the AD9866 output side (pins 50 and 46), and the attenuation already seems to be visible before the output transformer, so the transformer alone may not be the only cause.

I also replaced the output transformer with different ones (turn ratios 1.4:1 and 2:1), but this did not improve the situation.

Another thing I noticed is that a direct 50-ohm load appears to load the Radioberry output very heavily.

Additionally, even using a x10 oscilloscope probe significantly affects the measured signal level. Before the recent update, I had no issue determining output levels on each band with a 50-ohm load, but now the measurement itself seems to influence the result.

I am using a 3-stage amplifier, and initially the measurements indicated that a 28 dB attenuator (PI-Pad 56R-300R-56R) at the Radioberry output was required, and this worked correctly across all bands. Now this is no longer the case.

I also tried modifying the first stage of the amplifier to a high-impedance input (emitter follower), but even that seems to load the AD9866 too much.

So at this point, the most likely suspects seem to be:

• the AD9866 itself,

• a gateware / sampling-related issue,

• or the output stage/loading conditions.

Best regards,
Lukasz SQ5BAR

[Samuel Lourenço]

Hi Lukasz

If you had the Radioberry working correctly, then I must conclude that
the fault is due to a software issue. I've already stated many times
that there should be a development branch which doesn't affect the
production branch, but the developer of the project still has the
notion that the Radioberry is purely experimental, while people like
us use it for real purpose communications. Any of my insdights on how
development should be done were completely ignored.

Anyway, if you have some spot with a large storage space (say 11 GB),
I can provide you a known good image for the Radioberry 4. You could
also try to recompile the driver and piHPSDR, to see if it helps.
Sometimes, even OS updates can change things.

Kind regards, Samuel Lourenço

On Tue, 31 Mar 2026 at 06:30, Łukasz Radzikowski

> To view this discussion visit https://groups.google.com/d/msgid/radioberry/d118dbee-0e90-47be-aae6-302d45f212b8n%40googlegroups.com.

[Łukasz Radzikowski]

Hi Samuel,

Thank you for your reply.

I have already tried multiple combinations, including compiling different versions of the firmware, gateware, piHPSDR, kernel, and various Raspberry Pi OS releases. Unfortunately, the issue persists across all of them.

Of course, I would be happy to test a known good image.

I will send you the upload link in a private message.

Kind regards,
Lukasz SQ5BAR