On Class D Amplifiers Measurements

[DonH56]

My father used to have "radio breakthrough" on various systems through the 1970s and when we first moved to Queensland, there was an AM radio station on 1197 with its main transmitter around 7km away. You'd could pick it up with a piece of wet string. But as a little boy, a single germanium diode, a capacitor and a slug tuned coil, salt crystal earpiece would give me great reception with 20cm of wire. But only that station and in several places across my 'dial'. Even pocket transistor radios got overwhelmed...

Most of the demod radio breakthrough he suffered, came via the speaker wires (NFB I guess) and/or the phono stages in his various 1970s pieces (they were brand new then). I've had zero trouble with it, ever, but as you point out Don, the AM stations are mostly gone and if they remain are running tiny fractions of the ERP they once were.

Exactly my experience (natch). The argument that speaker outputs are low-Z does not account for their steeply rising impedance with frequency and the feedback path was (probably is) a very common "sneak" path for AM. As for phono stages, a fair part of my tech time was spent adding bypass caps to phono front ends as we had some big AM radio stations around the area.

 

[restorer-john]

Promoting your business, March Audio?

No. He has great faith in the products, stands behind them and fiercely defends them. An admirable quality, even if sometimes, the wood can be difficult to see for the trees.

 

[restorer-john]

as we had some big AM radio stations around the area

And now we have them built in our amplifiers... ba dum tss.. Just had to say it!

 

[Xulonn]

Lucille Ball hearing strong radio signals via her temporary lead dental fillings has never been confirmed or disproven, but one thing that she did is appreciated by me and millions of other science fiction fans - and she had no idea what the show was about. A very "strange but true" tale.

Lucille Ball is the reason we have 'Star Trek' — here's what happened
Wikimedia Commons - Lucille Ball kept the Starship Enterprise in flight.
The comedy icon's company, Desilu Productions, was responsible for Gene Roddenberry's original "Star Trek" series.
Desilu was one of the largest independent production companies in Hollywood and of course was the driving force behind Ball's star-making vehicle "I Love Lucy," which ran from 1951 to 1957.

But it was also responsible for "The Andy Griffith Show," "The Untouchables," "The Dick Van Dyke Show," and more.
Ball and then-husband and eventual "I Love Lucy" costar Desi Arnaz formed Desilu in 1950. Ball made most of all of the creative choices while Arnaz handled the business. The two worked as partners for years until they divorced in 1960, and Ball purchased Arnaz's share of the company in 1962.

Ball was the head of a major studio, and thus one of the most powerful women in Hollywood at the time.
When the landmark "The Untouchables" ended its run in 1963, Desilu desperately needed another big hit. Herbert Solow, who was hired to find projects for the studio, brought Ball two proposals: one for Roddenberry's "Star Trek" and another for "Mission: Impossible."

It was clear that the "Star Trek" pilot would be expensive to film, but Ball — who actually believed the series was about traveling USO* performers — overruled her board of directors and got the pilot produced.

The pilot, titled "The Cage," famously flopped. However, NBC pulled an unlikely move and ordered a second pilot, which came to be called "Where No Man Has Gone Before," only retained Leonard Nimoy's Mr. Spock from the first pilot, and became the show it is known as today. Ball agreed to finance this reshoot, again over the preferences of her board of directors.

So Ball is the one who let "Star Trek" live long and prosper. Thanks, Lucy.[/QUOTE]

*The United Service Organizations Inc. (USO) is an American nonprofit-charitable corporation that provides live entertainment, such as comedians, actors and musicians, social facilities, and other programs to members of the United States Armed Forces and their families, even for military personnel in war zones.

 

[amirm]

Had no idea of connection between Lucille Ball and Star Trek. Thanks for posting that! Amazing how many fond memories we would have been without, if she had not funded the series.

 

[restorer-john]

Had no idea of connection between Lucille Ball and Star Trek.

Everything ends up with Star Trek at some point on a geek forum.

Class D to out of band signals to radio breakthrough to Lucille Ball's dental fillings to Star Trek. Pretty straightforward association I'd say.

 

[March Audio]

Promoting your business, March Audio?

I have a vested interest in class D, I make no bones about that.

The technical issues are there to be debated. They have nothing to do with the business. However the same old class D myths just keep coming around. They are of course tweeter frying radio transmitters that cant drive difficult loads and shut down at the drop of a hat.

First thing I did was acknowledge your point - this is a real thing that can happen. All I have done is put it into perspective about what level of RF might be present on an amp input in the real world and the simple mitigations for it. As Amir and Don have mentioned, ingress of RF is not just a problem for class D.

You just dont see these issues manifest themselves with in band measurements. However I have suggested that Amir perform some measurements up to 1 MHz of the output of DACs so we can see whats there.

 

[maty]

In traditional class A and class AB the working bandwidth of the faster amplifier is usually limited over 350 kHz to avoid problems with RF.

The higher the limit, the more care must be taken in the design and in its correct implementation.

Also the circuit and PCB design matters.

 

[RayDunzl]

Class D to out of band signals to radio breakthrough to Lucille Ball's dental fillings to Star Trek. Pretty straightforward association I'd say.

The Moon today is in a Waning Gibbous Phase.

That explains it.

Or maybe not:

You should be feeling the benefits of your hard work for the last two weeks. Your metaphorical crops are in abundance and you should be seeing some (albeit small) outcome from your previously set goals and intentions. Now's the time you'll be feeling full of love. You'll want to give back to those around you — maybe you treat your spouse to dinner or buy your best friend a gift because you saw something and thought of them.

 

[solderdude]

Recap.

OP measured some things that are to be expected of class D amps when the input is not properly filtered.
Measuring equipment sweeping during an audiotest that doesn't hit the freq. the class D operates on won't ever show the problem.
The reason being such expensive measurement equipment doesn't have 'garbage' up there.
So measures perfect.

March Audio recognizes the issue and says he ADDS a filter to ensure it doesn't happen in his amps.
Problem solved in his amps.
Problem possibly not solved in other amps using the same modules.

Potential problem: garbage with freq. close to the class-D amps oscillator (or multiples) entering the amp and creating audible aliasing.
Again, with a good filter won't happen.
Not so good filter (demonstrated by OP) a real possibility.

Audio equipment using DACs (which one isn't) that has UNINTENDED but non-the-less present crap in the specifiied range (unlike measurement equipment).
This seems to be a real thread. Radio could be if very close to the device with lost of power and using a source with a rather high output R (no uncommon in hifi) ... maybe.

Solution is simple. Add a sufficient filter on the input of class-D amps there where the modules aren't having sufficient filtering.
OP can easily make one and check the efficiency.
March Audio can easily show the effect his filter has by sweeping to 1MHz with and without his filter and looking in the audible band.
I mean every tech probably has a tone generator reaching 1MHz.

Not the fault of the measurement crew.
Not an audiophile mystery of why some folks don't like class-D.
An omission of the amps designer, perhaps thinking what he did was enough.
May not be a potential issue with all class-D amps.
Depends on what audio source is used and its post filtering.

 

[audimus]

Which brings it back to the point of the OP that the measuring review of a class D amp should include something additional over the classic test to ensure that they have sufficient out of design bandwidth filtering at the input because of how they behave. This is not a test of Class D as a whole but of individual implementations, just as individual implementations and design/care differentiate one from the other in these reviews. A Class D amp without input filtering that shows audible spikes if one exists might be grounds for a decapitated pink panther.

 

[solderdude]

I would agree class-D amps may need an additional test that has to be standardised.
Most likely that specific test should show no influence on wide bandwidth analog amps (or very low level) but could expose input filtering issues on some amps.
When using class-D amps one could probably also pick their source to have little to no crap in it above 100kHz measured all the way up to at least 1MHz.

 

[boXem]

I would agree class-D amps may need an additional test that has to be standardised.
Most likely that specific test should show no influence on wide bandwidth analog amps (or very low level) but could expose input filtering issues on some amps.
When using class-D amps one could probably also pick their source to have little to no crap in it above 100kHz measured all the way up to at least 1MHz.

Why limiting this test to class D? Demodulating out of band noise in audible band can be done in many ways.
Moreover, I believe the issue highlighted here is not general to class D, but associated to self oscillating amplifiers.
For me testing should be agnostic.

 

[March Audio]

Which brings it back to the point of the OP that the measuring review of a class D amp should include something additional over the classic test to ensure that they have sufficient out of design bandwidth filtering at the input because of how they behave. This is not a test of Class D as a whole but of individual implementations, just as individual implementations and design/care differentiate one from the other in these reviews. A Class D amp without input filtering that shows audible spikes if one exists might be grounds for a decapitated pink panther.

Only if you can show that many dacs output a sufficient level of RF garbage to create a problem. I would suggest that is the first place to start. Take a look at the output of a bunch of dacs to see whats coming out and if there is something to be concerned about or if you are chasing ghosts.

Unless you have a large AM transmitter in your back garden at the right (wrong) frequency then external RF its extremely unlikely to create a problem. The source to the amp, as was pointed out by someone else, is low impedance. You have screened cables.

 

[restorer-john]

For me testing should be agnostic.

I agree. Let's just test all amplifiers regardless of class or topology out to 100KHz and be done with it. So-called Hi-Res needs that bandwidth anyway.

*runs and hides*

 

[RayDunzl]

Unless you have a large AM transmitter in your back garden

There's a neighborhood a few miles from me that has like five transmitter sites around it...

Looking south, then north, from the one location on another street:

 

[March Audio]

There's a neighborhood a few miles from me that has like five transmitter sites around it...

View attachment 39203View attachment 39204

In which case you are in deep doo with any amp, class D or A/B. Youll be hanging ferrites off of any and all wires Mind you I dont think they are AM transmitters.

My current RF environment. Stations at about -50dBm

 

[BDWoody]

Promoting your business, March Audio?

Sounds more like two heavyweights, who respect each other, going at it a bit...
I actually appreciate these interactions as there is always something to be gleaned.

 

[boXem]

So-called Hi-Res needs that bandwidth anyway.

It needs 48 kHz

 

[restorer-john]

It needs 48 kHz

48kHz Sampling rate? (~22kHz BW) What happened to 192kHz?