Challenge: 1 MHz FSK shock test on amplifier output – What would your scope show?

[tomsoon]

This doesn’t follow at all. Any proper amp design would filter out any frequencies it should not amplify. They are specifically designed to block and RFI at these frequencies.

That is a beautiful poetic question! But to look at it from another angle: a diamond doesn't change its structural properties or its value just because it's in the dark with no light source to observe it.
It is similar with an analog architecture. The native speed, stability, and phase linearity are internal properties of the circuit itself. Testing it at 1 MHz without filters is not about trying to "hear" the ultrasound.
The real question is: if a topology possesses the native capability to handle this high-frequency energy directly without falling apart, won't those exact same internal properties remain active when it operates back down within the standard 20 Hz - 20 kHz band? Can those hidden capabilities be the reason behind a deeper, blacker background and better spatial delivery in the audio range?
Thank you for bringing this fascinating perspective into the discussion!
Best regards,

 

[solderdude]

Hi,
Wow, that must have been quite a dramatic moment in the lab! Burning out one of those beasts during an EMC pulse test definitely proves how unforgiving and intense this kind of high-frequency energy can be [AS INDEX]. Thank you for sharing that experience!

Connected to the wrong side of the DUT... just a tick and switched off.

But perhaps—just perhaps—there is a beautiful bridge between the two perspectives. The goal of designing a topology that can survive an EMC-like 1 MHz FSK stress test into a 7.5-ohm load is not about trying to hear 1 MHz [AS INDEX].

to what end ?

Instead, if the architecture is fast and robust enough to remain stone-cold and perfectly stable under that kind of assault, it means that within the standard audio band, all micro-oscillations, transient smearing, and phase shifts completely vanish

Just high enough not to become problematic. For audio frequencies and well up to 100kHz is is pretty easy to get a robust design.

. And that absolute technical peace is exactly what allows us to sit back and listen to the music for 16 hours straight without any brain fatigue [AS INDEX].

I don't experience any of that.
It is usually caused by a transducer and not so much an amplifier being unstable or it must be a really crappy and 'colored' design.

 

[voodooless]

That is a beautiful poetic question! But to look at it from another angle: a diamond doesn't change its structural properties or its value just because it's in the dark with no light source to observe it.

That poetic bullshit…

The real question is: if a topology possesses the native capability to handle this high-frequency energy directly without falling apart, won't those exact same internal properties remain active when it operates back down within the standard 20 Hz - 20 kHz band?

No! This “logic” makes absolutely no sense at all.

 

[tomsoon]

I’d say, citation needed…

If you need phase response in de audio band, you measure it in de audio band…

Any music will be band limited anyway.

And please stop using AI beyond basic translations. It’s very clear you’re using it for more than just that. It is against the policy.

Official policy on use of AI

We have had recent issues come about regarding use of AI in member posts. I personally think this technology can be useful to us. And at any rate, hate having to police members left and right on what they post. That said, here are my thoughts: 1. If you are going to use AI, make it...

www.audiosciencereview.com

You are right about the policy, but let's look at the logic. A human mind asked these theoretical questions. The AI was only used as a precise tool to translate my thoughts into proper technical English so that we wouldn't misunderstand each other.
Should I write in broken, misleading English or write in Hungarian just to prove I am a human? This is a professional forum, and I wanted to give this community the respect of using correct terms so we can focus on the actual scenario.
I am a real person from Hungary, and my questions about 1 MHz FSK and voltage scaling are based on genuine curiosity. Let's focus on the transistors and the electronics, not the translation.
If this approach is not acceptable here due to the rules, then unfortunately, we must say goodbye. But despite all of this, I am genuinely glad I had the chance to be here.

 

[fpitas]

You are right about the policy, but let's look at the logic. A human mind asked these theoretical questions. The AI was only used as a precise tool to translate my thoughts into proper technical English so that we wouldn't misunderstand each other.
Should I write in broken, misleading English or write in Hungarian just to prove I am a human? This is a professional forum, and I wanted to give this community the respect of using correct terms so we can focus on the actual scenario.
I am a real person from Hungary, and my questions about 1 MHz FSK and voltage scaling are based on genuine curiosity. Let's focus on the transistors and the electronics, not the translation.
If this approach is not acceptable here due to the rules, then unfortunately, we must say goodbye. But despite all of this, I am genuinely glad I had the chance to be here.

It all starts with, none of us understand how your test will further the design of an amplifier. There is no use case for an audio amp where that test is representative.

 

[fpitas]

You raised the point about micro-phase variations earlier, and that is a real thing. But existing tests quantify the effects of that. See for example the 32-tone input signal test here: https://www.audiosciencereview.com/forum/index.php?threads/topping-mini-300-amplifier-review.68581/

 

[egellings]

Hi Everyone,
I am completely new here, and after looking around, I found this forum to be the most appropriate and capable place for this specific topic. It is truly an honor to be here, and it would be a privilege to get answers from the real pros on this board.
I am very curious to hear your expert opinions and see your simulations regarding a test measurement that pushes the boundaries between theory and practice.
We have a harsh test signal from a UNI-T UTG932E function generator, using a demanding FSK (Frequency Shift Keying) modulation:

• Carrier Frequency (CarrFreq): 1,000,000 kHz (that's 1 MHz!)
• Hop Frequency (HopFreq): 100,000 kHz
• Switching Rate (Rate): 10,000 kHz

Let's assume we feed this ultra-dense signal, packed with drastic frequency and phase jumps, into the input of an amplifier, while the output is loaded with a real, brutal 7.5-ohm dummy load.
My question to the experts:
How do you think a classic pure Class A, a Class AB hi-fi, or a modern high-end Class D (PWM) amplifier would react to this shock on the output?

That's a pretty severe signal, the likes of which an amplifier would never see while generating audio. I'm not sure what running that test on an amplifier would prove.

 

[DonH56]

Hi,
Thank you for catching that! Yes, it was absolutely a decimal vs. comma typo on my part—I meant 1.000000 MHz (1 MHz), not 1 GHz. I appreciate the correction! [Technical Articles]

Got it.

Also, I must say I love your quote by Aldous Huxley. It resonates deeply with anyone who truly appreciates the emotional depth of music. [Technical Articles]

Thank you. I am a musician (amateur) as well as a listener so it resonated.

To clarify the measurement setup: the 1 MHz FSK signal is applied directly to the input of the amplifier. The question is specifically focused on how the topology behaves under a real 7.5-ohm load, and what we would see on the output terminals. [Technical Articles]

Any decent amplifier will reject a 1 MHz input signal. We have AM radio stations in the USA that go from 530 kHz to 1.7 MHz and we do not want to amplify them for our speakers. To the extent that it handles the signal, I would expect that an audio amplifier will exhibit severe amplitude and phase changes in addition to high distortion (harmonic and otherwise) due to the signal being out of band and thus well beyond the ability of the amplifier's feedback loop. Similar to applying a 1 GHz signal to an RF amplifier rated for 10 MHz bandwidth; any response is likely to be "bad" and the only point of such a test is to see how well the amplifier rejects out of band noise.

Your point about poor RFI rejection at the output terminals in past designs is brilliant. That is exactly why this scenario is so fascinating to look at from a theoretical perspective.

Thanks, but it is a problem known for many decades, and one I and other designers and users have struggled with from time to time. Sneak paths from the output lead to demodulation/mixing and amplification of RFI on the speaker cables being retransmitted in the audio band. That is bad.

If an analog topology were fast and linear enough by nature, could it maintain a stable, clean output stage under this 1 MHz dynamic shock without relying on heavy input filtering or chokes? [Technical Articles]

There are many amplifiers that would handle the signal just fine, but audio amplifiers like any others are designed to suit a particular application. Making an audio amp wideband enough to handle a 1 MHz FSK signal means much more noise, requires much more power, and more sensitivity to undesired signals well out of the audio band.

The input filter is usually pretty simple, a first or second order filter created by a simple RC network. Inductors (chokes) are not usually used on the audio input.

Thank you again for your valuable insights, and I’m honored to have your expertise in this thread! [Technical Articles]
Best regards,

---

Thanks, but note my expertise is more in the RF/mW world rather than audio. The principles are the same, of course.

This may be an interesting thought experiment, but realistically the best response of an audio amplifier to a 1 MHz FSK signal is no response at all.

 

[pma]

Speaking about input of the amplifier, I measured about 17 years ago one of my preamps by signal generated from HP VHF generator up to 3 GHz and monitored the output by 3 GHz Spectrum analyzer. The preamp was able to work with 50 ohm load by design. The experiment did not bring any interesting outcome. I still have the screenshots in my archive. The GHz signal of course bypasses the normal audio path where it can, as an audio amp is not a GHz instrument.

and 100kHz spectrum. The audio preamp worked at 100kHz/50 ohm well (-84dB H2).

 

[fpitas]

I suggest you familiarize yourself with all the more or less standard tests Amir does here:

https://www.audiosciencereview.com/forum/index.php?threads/topping-mini-300-amplifier-review.68581/

If you have questions, I'm sure you'll get respectful replies.

 

[pma]

Non-standard tests are needed if you have troubles. And you have to be creative, then.
Like in the case below - preamp output stage oscillations at 253 MHz. No AP would disclose this. AP is a low frequency instrument and is not enough in the design stage.
This 253MHz oscillations raised audio range distortion and also hum components.

 

[tomsoon]

Speaking about input of the amplifier, I measured about 17 years ago one of my preamps by signal generated from HP VHF generator up to 3 GHz and monitored the output by 3 GHz Spectrum analyzer. The preamp was able to work with 50 ohm load by design. The experiment did not bring any interesting outcome. I still have the screenshots in my archive. The GHz signal of course bypasses the normal audio path where it can, as an audio amp is not a GHz instrument.

View attachment 552123

View attachment 552128

and 100kHz spectrum. The audio preamp worked at 100kHz/50 ohm well (-84dB H2).

View attachment 552130

Hi pma,
Thank you for sharing these highly educational measurements and text from your archives. Your Agilent and HP spectrum analyzer data perfectly illustrate why standard audio band measurements are completely insufficient during the development and design phase.
You are spot on—hidden parasitic oscillations in the hundreds of MHz range can easily fold back, increasing distortion and adding unwanted hum into the audible spectrum. Creativity and looking beyond standard textbook audio tools are exactly what circuit analysis requires.
I truly appreciate you bringing this kind of real-world RF perspective into the discussion.
Best regards,

 

[voodooless]

Hi pma,
Thank you for sharing these highly educational measurements and text from your archives. Your Agilent and HP spectrum analyzer data perfectly illustrate why standard audio band measurements are completely insufficient during the development and design phase.
You are spot on—hidden parasitic oscillations in the hundreds of MHz range can easily fold back, increasing distortion and adding unwanted hum into the audible spectrum. Creativity and looking beyond standard textbook audio tools are exactly what circuit analysis requires.
I truly appreciate you bringing this kind of real-world RF perspective into the discussion.
Best regards,

These kinds of measurements are obviously a useful tool during the design phase; this does, however, not show why an amp should accept crazy high-frequency signals. If anything, it shows why that should be avoided!

 

[tomsoon]

I do test amplifiers by pushing current into their outputs, to measure output impedance vs. frequency and output impedance non-linearity. But I will NOT participate in the topic as opened in this thread.

View attachment 552087 View attachment 552088

Hi everyone,
Thank you all for your engineering insights and for participating in this discussion. It has been an interesting topic.
Special thanks to amirm for bringing in the real-world measurements, and a big thank you for bringing up the Krell and Spectral philosophy—that example perfectly mirrors the core of high-speed analog design [June 1994 - Silicon Chip Online].
A very special thanks to pma for sharing the PicoScope current injection stress test from your archives [AS INDEX]. Your data regarding output impedance non-linearity during fast current transients perfectly illustrates the physical challenges of discrete analog topologies [AS INDEX].
Since the topic has been thoroughly explored, let's wrap up this little theoretical game.
Thank you again for your time and participation!
Best regards,

 

[antcollinet]

I wanted to see how the output stages, feedback loops, and architectures designed for high-end audio would react if they were pushed to handle this high-frequency energy directly under a real 7.5-ohm load.

Why?

What meaning does it have for audio frequency amplifiers? Does an amp that fails to operate in the presence of this somewhat extreme (in audio terms) test signal become less useful to me as a music listener? Does it simulate any real world environmental situation I might want my amp to withstand?

It is not a test (or even a question) that makes a lot of sense to me.

 

[voodooless]

thoroughly explored

Really?

 

[SSS]

That is a beautiful poetic question! But to look at it from another angle: a diamond doesn't change its structural properties or its value just because it's in the dark with no light source to observe it.
It is similar with an analog architecture. The native speed, stability, and phase linearity are internal properties of the circuit itself. Testing it at 1 MHz without filters is not about trying to "hear" the ultrasound.
The real question is: if a topology possesses the native capability to handle this high-frequency energy directly without falling apart, won't those exact same internal properties remain active when it operates back down within the standard 20 Hz - 20 kHz band? Can those hidden capabilities be the reason behind a deeper, blacker background and better spatial delivery in the audio range?
Thank you for bringing this fascinating perspective into the discussion!
Best regards,

This is total nonsense at least in my opinion as technical expert in audio as well in radio frequencies. An audio engineer will not build an amplifier for 1 MHz nor will test with this frequency. It makes simply no sense.

 

[tomsoon]

This is total nonsense at least in my opinion as technical expert in audio as well in radio frequencies. An audio engineer will not build an amplifier for 1 MHz nor will test with this frequency. It makes simply no sense.

Just scroll up a bit, pma clearly explains what this is all about.

 

[voodooless]

Just scroll up a bit, pma clearly explains what this is all about.

No, he doesn’t!

 

[pma]

No, he doesn’t!

So, instead of endless debates, you might bring your data? Number of posters bringing something else than usual 20kHz set here is close to zero. And I can make you sure that as a designer you need to do more than a usual set. Of course in case you know what you are doing. I am saying results and work that has been done count, debates do not.