• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Weird Behaviours I've Noticed Whilte Testing One of Topping's NFCA Based Amplifiers

mike7877

Addicted to Fun and Learning
Joined
Aug 5, 2021
Messages
718
Likes
144
I have four of Topping's products with their "NFCA" design (G5, LA90 Discrete, L70, and L30 II)

This evening I've been testing out my L30 II to find its limits for driving the highs of my project (project is turning high end 2-way passives into actives with a 4 channel sound card and two stereo amplifiers - active mode for lower level, critical listening, passive for louder, less exact)

With the L30 II, preliminarily I've found the design to respond... unconventionally. When Topping talked about NFCA in product descriptions and the like (never in too much detail! lol), I got the impression there are 3 or maybe 4 stages with cleverly applied feedback. Well, from the way the L30 behaves, I think there's a bit more to it...

How does it behave, though?
First, quickly, how would a normal ab amplifier respond? Like a simple car stereo chip TDA7377

PSX_20240227_195811.png


While the TDA7377 is designed to run at about 11.6 to 14.8V, its given operating range is 8V to 18V. Up until some time in the last 10 years (varies my mfr.), whenever a car was running, its 12V system was up at 14.2-14.7V the entire time. Charge voltage for the battery for hours at a time when it's full is bad for lead-acid batteries. Cathode corrosion is the technical term for what happened to the battery whenever the car was on for hours at a time. For ease of discussion, it's now 2012 (except for the existence of NFCA).

From datasheet, because it's much easier to understand and write a chart than a long written description of the chart:
PSX_20240227_201720.png


As supply voltage increases, power output at a given distortion increases. The line nearer the top of the chart is power at 10% distortion when the signal is a 1kHz sine wave, bottom is 1% distortion. Subtract another 10-20% to get the rating at which clipping and [potential] near-clipping behaviour isn't part of the signal, which in the TDA7377's case is 0.02% (single ended 4 ohm load and 14.4V supply - you can see in the chart above that 1% distortion with the same load and supply conditions is at ~5.2W, not 4)

Sorry if that wasn't necessary for you - it's over now (and we're all on the same page).

What does the NFCA class ab amplifier do?


Initially I thought its output could be described by a chart similar to the one above (even in power output, but at 8 ohm nominal, not 4!)
Why?
Because when I had music on and its level reached clipping, if I increased supply voltage, there was clipping, no more.

I wanted to get the L30 II's real continuous (RMS) power rating, and see how its supplied voltage affected power output. Only approximately though, because I'm using speakers as the load, not a resistor with a known value which doesn't appreciably change at different frequencies - I only know the approximate impedance of my speakers at various frequencies.

The L30 II is supplied its power by an ugly wall wart with a traditional transformer tucked inside - 1000mA / 15VAC out. I have two no-name variacs, so can turn 120V into 0-180V. Actually I've got a (claimed 70 watt, really 18 watt) 120 to 240V travel transformer (18 watts because the chassis is 65C after temperature stabilization is done with 18 watts drawn lol). If I plug that transformer into the end of the two wannabe variacs, I've measured 370something volts. With 18 watts drawn it'd probably drop to like 350 thoug haha. Anywayyyy

So music is on - say it's rock. The kick drum has the most energy. Say the drummer's foot was particularly heavy with one beat and we were really close to clipping already - his foot pushes the situation critical: clipping. Ok, not critical, but it's interesting what happens after - the next kick, quieter than the one before the one which clipped, also clips. Whaaaat? This isn't because the supply drooped and hasn't recovered yet, either. Something about reaching clipping puts the amplifier in a state where the output signal has to, for some length of time, remain at least some distance away from the rails (or a voltage near the rails) for the amp to recover. Recover? Recover: amplifier able to reach the same levels it did before the drummer's lead foot broke our schitt.

OK weird behaviour number TWO

60Hz sine wave on the output, 6VAC (16.8V peak-peak)
1 second goes by and it's fine. 2 seconds, still fine. 3 seconds and it's not clipping, but there's periodic clicking occurring. It sounds like a fart ripping inconsistently. Maybe 8Hz. On the scope (sorry, no picture) the sine wave is affected in what appears to be a very odd way... like you'd expect clipping to look, except it happens before the peak voltage. Imagine a normal clipped waveform with no artifacts (hitting the rail and that's it) - if the rail was 15% higher, the clipping wouldn't exist. OK, now we're looking at the negative side of the waveform, at 270 degrees (the negative peak). 180 degrees is the middle 0. Make a square centered over the waveform between 255 to 265 degrees. Rotate it clockwise 90 degrees. That's what it looks like. it doesn't happen every cycle - maybe 8 times per 60 (yes it's a perfect sine wave, not changing amplitude). What does it look like? It looks like ~15% before where clipping should occur the waveform reaches for the sky (actually reaches for the dirt cause we're looking at the negative side).

It gets weirder! 4 seconds go by and it's getting worse! 5 seconds go by and worse! It stabilizes around 15 seconds after starting.

What would you do to make the amplifier not do this? I'd give it more voltage.
What makes it so the amplifier does not do this? Give it less voltage!

Say the peaks of music reach +-12V with no distortion. You can also play a sine wave at +-12V, but only for 2 seconds. It's somewhere around half that voltage you'll have to attenuate to, to be distortion free after 15 seconds.

Say you start at 8V. It'll go for longer before distorting, but you'll still end up distorting in time and needing to reduce to 6V. BUT NOT IF...

Not if you reduce the supply voltage! That's right! Reduce the supply voltage and continuous power (the amount of power able to be supplied after approximately 15 seconds...) goes up! But if you reduce the supply voltage and it's music, you can't reach the 12V peaks!

The good thing about this is the following: Billy Idol, for example, a 50ms clip of a song off of Vital Idol with one kick drum kick, bass holding one note, and guitar playing a chord (not the most dynamic is what i'm getting at) is 12V peak, 2.5V RMS. So the L30 II can make 6V RMS and 12V peak, which seems to allow for the amp to continually reach the 12V kicks without the average power being too high, pulling down the peak possible voltage...


Anyway, those are the weird behaviours I noticed with the NFCA which I haven't noticed with any of my other amps. I don't know if the L70, LA90 Discrete, or G5 behave exactly like this, but it wouldn't surprise me.
 

Attachments

  • Screenshot_20240227_201658_Drive.png
    Screenshot_20240227_201658_Drive.png
    139.3 KB · Views: 39
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,963
Likes
244,777
Location
Seattle Area
So music is on - say it's rock. The kick drum has the most energy. Say the drummer's foot was particularly heavy with one beat and we were really close to clipping already - his foot pushes the situation critical: clipping. Ok, not critical, but it's interesting what happens after - the next kick, quieter than the one before the one which clipped, also clips. Whaaaat? This isn't because the supply drooped and hasn't recovered yet, either. Something about reaching clipping puts the amplifier in a state where the output signal has to, for some length of time, remain at least some distance away from the rails (or a voltage near the rails) for the amp to recover. Recover? Recover: amplifier able to reach the same levels it did before the drummer's lead foot broke our schitt.
I have noticed something similar to this in attempting to measure peak and max power. I have to play a lot in setting the low and high watermarks for the analyzer to find the max power point. In the last amp I tested, setting the lower bound to higher voltage allowed the measurement to work which is reverse of what one would think.
 

jasonhanjk

Active Member
Joined
Mar 24, 2019
Messages
253
Likes
182
Composite amplifier do not work that well in clipping condition. The op amp will try to correct the clipping but we all know that's impossible to correct. :p

In my country when the AC mains is 50Hz, I will stress test at 60Hz.
 

Anton D

Major Contributor
Joined
Mar 17, 2021
Messages
1,034
Likes
1,233
Mike, your post is so brilliant! Thanks for making me a little smarter.

Such great stuff.
 
Last edited:
OP
M

mike7877

Addicted to Fun and Learning
Joined
Aug 5, 2021
Messages
718
Likes
144
I have noticed something similar to this in attempting to measure peak and max power. I have to play a lot in setting the low and high watermarks for the analyzer to find the max power point. In the last amp I tested, setting the lower bound to higher voltage allowed the measurement to work which is reverse of what one would think.

Strange.

Oh! I've been wondering: That test you do with headphone amps-- the output voltage vs. distortion at different loads test (usually 12 or 20 ohms to 600 ohms with 5 or so steps between) - for how long do you hold each measurement at its highest level (like where distortion is ~-30dB THD+n) before switching to the next resistor?

-
The 12-600 ohm testing..., would you consider adding 6 ohms to the repertoire? With a 6 ohm test done (and passed) the words "though not its purpose, this headphone amplifier is capable of driving speakers at moderate levels" could be put the review. Anyone seeing this with doubts about their tricky headphones being able to be driven by whatever this is, would wind up with complete confidence in it!

Of course, some amps will be better suited than others.
Spec-wise, output impedance should be no higher than a quarter ohm, circuit protection related to current (ie. over or short) should not have engaged during the distortion/power test, and if it did, only with very audible distortion present. And there should be no "extra" distortion on the signal below clipping level (compared to any of thee easier-to-drive loads).
And other-wise, the unit should be able to dissipate the extra heat generated!.

When a headphone amp does not meet the standard for the previous quotation to apply (except maybe output impedance - up to half an ohm is tolerable if other requirements are met), then the words "should the need arise, this headphone amplifier should be able to drive a pair of speakers quietly" can be said in its place.
If the unit can't handle the heat, something like "this headphone amplifier gets hot when playing loud with low impedance loads. Keep an eye on temperatures during loud lengthy listening sessions if your cans are between 12-30 ohms!

Pretty involved idea... but you get what I'm getting at right? Forget all the minutia - do you like the idea of including a 6 or 8 ohm test with the headphone amps?

1709111255756.png

(visual example of properly behaving HPA at low impedance, high impedance, and all impedances in between lol)
 

Anton D

Major Contributor
Joined
Mar 17, 2021
Messages
1,034
Likes
1,233
I've been searching around, is this a common measurement or finding?

If not, it should be. Potential huge impact on the sound between devices that may nominally be considered 'electronically transparent' or 'indistinguishable' from one another.
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,610
Likes
18,748
Location
Netherlands
I've been searching around, is this a common measurement or finding?
I’d say, nothing special here. The amp is driven past its design limits: except it to not behave linearly.

Clearly the amp cannot drive more that 6VRMS into 12R, so less into the lower impedance of a speaker. It’s bound to be unhappy. This has little to do with the NFCA topology.

TLDR: don’t overdrive the amp.
 

Anton D

Major Contributor
Joined
Mar 17, 2021
Messages
1,034
Likes
1,233
I’d say, nothing special here. The amp is driven past its design limits: except it to not behave linearly.

Clearly the amp cannot drive more that 6VRMS into 12R, so less into the lower impedance of a speaker. It’s bound to be unhappy. This has little to do with the NFCA topology.

TLDR: don’t overdrive the amp.
But that recovery time!

Is one type of amp more prone to this slow recovery than another?

A general discussion from our past...


 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,610
Likes
18,748
Location
Netherlands
But that recovery time!
That may be more of a power supply issue. Possibly it has some over current/voltage protections that recover slowly. Maybe someone with more knowledge of the lm317 knows how it behaves. Certainly it allows for various current limiting schemes.
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,881
Likes
6,451
Location
Berlin, Germany
The 12-600 ohm testing..., would you consider adding 6 ohms to the repertoire? With a 6 ohm test done (and passed) the words "though not its purpose, this headphone amplifier is capable of driving speakers at moderate levels" could be put the review. Anyone seeing this with doubts about their tricky headphones being able to be driven by whatever this is, would wind up with complete confidence in it!

Of course, some amps will be better suited than others.
Spec-wise, output impedance should be no higher than a quarter ohm, circuit protection related to current (ie. over or short) should not have engaged during the distortion/power test, and if it did, only with very audible distortion present. And there should be no "extra" distortion on the signal below clipping level (compared to any of thee easier-to-drive loads).
And other-wise, the unit should be able to dissipate the extra heat generated!.
A "one-fits-all" headphone amp is a true engineering challenge because the impedances and sensitivities vary so much, much more than for any speakers-only amp.

Some HP amps therefore have individual outputs, one for sensitive low impedance IEM's and another for regular headphones with moderate impedances and sensitivities. This also makes protection features more easy to design.
 
Top Bottom