• 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!

(Suggestion on) Testing power amplifiers according to IEC 60268-3 standard

OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
What does it look like when remaining just under overload conditions ?
I mean... the amp is clipping substantially here.

I have captured the same overload (+10dB) at 20Hz and 1kHz, for you. The amplifier is in fact overloaded with 3.16x maximum rated input voltage.

A250W_20Hz_clipping.png

20Hz overload plot clearly shows PSU ripple, discharging of capacitor bank.


A250W_1kHz_clipping.png


These are important data and the IEC overload restoring test is excellent. Small amplifiers with tiny power supplies will have troubles. I will also test Hypex NC252MP, though I know that its behaviour under difficult conditions is worse than that of A250W4R class AB amplifier.

IEC overload restoring test greatly zoomed:
A250W4R_10dB_overload_recovery_bigzoom.png


Slow return of restored 20Hz sine to zero line is due to amplifier LF limit (3Hz/-3dB), LF time constant.
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Overload recovery from +10dB overload at 1kHz

A250W4R_10dB_overload_recovery_zoom_1k.png



Full record:
A250W4R_10dB_overload_recovery_full_1k.png
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Characteristics of d.c. offset protection circuits (abstract)

NOTE: These characteristics do not apply to amplifiers with only capacitive coupling to the load terminals.

Characteristics to be specified
The following characteristics should be specified:
a) rated maximum offset voltage at the load terminals, stated by the manufacturer in the specification, that does not operate the protection circuit;
b) rated response time of the protection circuit, optionally stated by the manufacturer in the specification, to an offset voltage of 30 % of the highest d.c. supply voltage of the amplifier output stage;

Methods of measurement
Methods of measurement for amplifiers which continuously sense the d.c. offset condition

The procedure is as follows:
a) The amplifier is brought under standard measuring conditions, and the input signal frequency is then changed to 20 Hz. A d.c. oscilloscope and a d.c. voltmeter are connected to the load terminals.
b) The d.c. balance of the amplifier is then slowly changed, by a suitable method which depends on the detailed design of the circuit and cannot be standardized, and the maximum reading on the d.c. voltmeter, before the protection circuit operates, is noted. This is the maximum offset voltage that does not operate the protection circuit.
--------------------------------------------------------------------------------

Measurement of DC offset protection of Hypex NC252MP

This amplifier is DC coupled and it transfers amplified DC voltage from input to output.


Input test signal - 20Hz sine + DC offset and amplitude jump after 1s
DC_protection_testsignal.png


Input signal spectrum before the amplitude jump
DC_protection_testsignal_spectrum.png



NC252MP measured response to the input signal
DC_protectiontest_NC252MP.png


The input signal level in the 1st second is at "standard measuring conditions" level of the output. Then comes the 20dB jump. One can see that NC252MP transfers DC average level of 10.69V to the output, without any action of the module DC protection. This is dangerous and may lead to speaker voice coil damage in case of DC error signal at the amplifier input. After 1s, when the input signal DC jumps of 20dB, the protection circuit finally takes action. It shuts down the amplifier output in 7.8ms after heavy DC clipping.

DC_protectiontest_NC252MP_dt.png
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Measurement of "DC protection" circuit action of NC252MP when input DC is continuously increased, in steps.

NC252MP_outputDCoffset_test_sm.png


The protection circuit did nothing until the output DC voltage level has reached 26V.
This is unacceptable!
And will fry any speaker.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Yep, there definitely is too little focus on protection and amp behavior on overload and clipping.
Current protection kicking in can be quite different from clipping in certain ways.
Usual tests only show behavior in resistive loads and not, or just on the onset of clipping.

As you already know these additional tests (and a lot of different tests can be done) and reporting takes up a LOT of time and space.
 

Mnyb

Major Contributor
Forum Donor
Joined
Aug 14, 2019
Messages
2,640
Likes
3,603
Location
Sweden, Västerås
If we view the class-D module itself as a commodity component amp integrators can buy it shifts the responsibility to them the implement proper protection schemes, outside the module itself ?

The DC blocking question itself is also a game off blame shifting putting the filter in the source or preamp may make these measure slightly worse .. The last defense. is the power amps DC servo , it can apparently quickly detect DC ?

With digital sources , one has to wonder if this should be implemented rigth into the DAC chips themselves by ESS and AKM et al. I think there are examples where there is DC offset in the audio content itself direct from the studio to you ?

But preceding components can fail and put out DC anyway so the power amp needs DC protection in all cases .

I repeat my own beliefs here that a good product is a lot more than audio performance alone.

Remember decades ago when audiophile approved amps often omitted protection circuits ? It's a bad engineering shortcut , they can be designed to not interfere with the signal afaik ?
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,840
Likes
5,774
If we view the class-D module itself as a commodity component amp integrators can buy it shifts the responsibility to them the implement proper protection schemes, outside the module itself ?

The DC blocking question itself is also a game off blame shifting putting the filter in the source or preamp may make these measure slightly worse .. The last defense. is the power amps DC servo , it can apparently quickly detect DC ?

With digital sources , one has to wonder if this should be implemented rigth into the DAC chips themselves by ESS and AKM et al. I think there are examples where there is DC offset in the audio content itself direct from the studio to you ?

But preceding components can fail and put out DC anyway so the power amp needs DC protection in all cases .

I repeat my own beliefs here that a good product is a lot more than audio performance alone.

Remember decades ago when audiophile approved amps often omitted protection circuits ? It's a bad engineering shortcut , they can be designed to not interfere with the signal afaik ?
It seems that this is a Hypex only matter.
I had a look both in Purifi and Ice Power papers which they clearly state the DC detection and block conditions.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Yes, with DC protection board we get slightly worse damping factor. But we get reliability. Without DC protection board we get better damping factor number, which is both pointless and inaudible. However, marketing numbers work and everything is done to support this approach.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Response of A250W4R amplifier to sudden deep input DC overload (-4Vdc at the input).

A250W4R_deep_DCoverload.png


The amplifier has input coupling capacitor and also DC servo. Even then the output DC offset protection circuit is activated.

Because of the ac input coupling, there is no special reaction if input DC is slowly rising, as it was here.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Overload recovery test of AIYIMA A07. Supply voltage = 48V, load 4ohm. The test is described here.

AIYIMA_A07_overloadrectest_sm.png


AIYIMA A07 fails to pass the test and shuts down after 0.2s. Turn off-on sequence needed to make it running again.

-------------------------------------------

The measurement was possible with shorter overload of 0.2s, however it is not according to the standard.

AIYIMA_A07_shortoverloadrectest_sm.png



Maximum short-term power is 148.6W/4ohm for 1s with 48Vdc power supply (current capability 10A).
AIYIMA_A07_shorttermpower_sm.png
 
Last edited:

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
If we view the class-D module itself as a commodity component amp integrators can buy it shifts the responsibility to them the implement proper protection schemes, outside the module itself ?
Yes. Ditto things like turn-on-turn-off thumps or other noises.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
UcD180HG response to deep DC- input overload

UcD180_DCoffset_response.png


The amplifier does not transfer DC from input to output. For slow input DC rise, the output does not follow the input. For sudden deep input DC overload (-9V), the DC protection has worked after 125ms, which is fine.
 

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,515
Likes
7,026
Location
Stow, Ohio USA
It seems that this is a Hypex only matter.
I had a look both in Purifi and Ice Power papers which they clearly state the DC detection and block conditions.

As @pma recently illustrated, the Hypex UcD module passes. This issue appears to be specifically intrinsic to the dual channel Hypex MP modules. Would be harder to overlook (EDIT) their own spec in the single channel case.

As mentioned, with robust sources, DC is not likely presented to an amp. This is somewhat confirmed by the lack of aftermarket remedies. We know a low cost remedy is available, but try to find one. In a market where snake oil remedies are often prevalent, the lack of them implies that the problem is considered to be a solved one. Will see if Hypex takes any action. With new designs already queued, unless the fix is inexpensive, suspect they may simply wait to address this in newer tech.
 
Last edited:

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,840
Likes
5,774
As @pma recently illustrated, the Hypex UcD module passes. This issue appears to be specifically intrinsic to the dual channel Hypex MP modules. Would be harder to pass their own spec in the single channel case.

As mentioned, with robust sources, DC is not likely presented to an amp. This is somewhat confirmed by the lack of aftermarket remedies. We know a low cost remedy is available, but try to find one. In a market where snake oil remedies are often prevalent, the lack of them implies that the problem is considered to be a solved one. Will see if Hypex takes any action. With new designs already queued, unless the fix is inexpensive, suspect they may simply wait to address this in newer tech.
I have to correct it then (already did)
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,840
Likes
5,774
It seems that this is a Hypex Hypex MP modules only matter.
I had a look both in Purifi and Ice Power papers which they clearly state the DC detection and block conditions.
Corrected.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
10MHz bw, PA5 well into clipping into an 8 ohm load:

It is the switching frequency, see below:

UcD180HG_burst_clipping_sm.png


It gets worse with rising amplitude (loss of feedback control):
UcD180HG_burst_clipping2_sm.png

(1kHz bursts)

UcD180HG_burst_clipping3_sm.png


Your measurement is still too "polite" to the amplifier :).

More burst measurements here.
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
This is the fluctuation of switching frequency during the 10ms 1kHz burst, it is also interesting:

UcD180HG_burst_switchfluctuation_sm.png
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Is the switching frequency actually changing, or is the frequency spreading due to the pulse width modulation? I thought the frequency switching was pretty much constant for these designs? Do not know for sure...
 
Top Bottom