• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

DIY 250W/4ohm amplifier based on "blameless" topology, and measurements

sabristol

Member
Forum Donor
Joined
Feb 23, 2021
Messages
46
Likes
73
#21
A perfect end to a perfect project

I’ve got some time off soo hopefully I’ll do something on this!
 

tomchr

Addicted to Fun and Learning
Technical Expert
Manufacturer
Forum Donor
Joined
Nov 5, 2018
Messages
654
Likes
1,872
Location
Calgary, Canada
#22
Very cool project! Nicely done. I especially appreciate that you simulated and measured with a complex load. Do you have any sense of whether the LF distortion comes from the load or from the amp?

I see you used the ModuShop 400 mm deep Dissipante chassis. One thing with them is that the heat sink is split into two 200 mm wide chunks. I realize I'm a bit late with my suggestion, but with two pairs of output devices, I would put one pair on each heat sink. That will make life a bit easier for the thermal system. That said, there's a fair amount of thermal connection between the two heat sinks, so it's probably not a huge deal to have them on the same heat sink.

Tom
 
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
1,604
Likes
2,572
Location
Prague
Thread Starter #23
Very cool project! Nicely done. I especially appreciate that you simulated and measured with a complex load. Do you have any sense of whether the LF distortion comes from the load or from the amp?

I see you used the ModuShop 400 mm deep Dissipante chassis. One thing with them is that the heat sink is split into two 200 mm wide chunks. I realize I'm a bit late with my suggestion, but with two pairs of output devices, I would put one pair on each heat sink. That will make life a bit easier for the thermal system. That said, there's a fair amount of thermal connection between the two heat sinks, so it's probably not a huge deal to have them on the same heat sink.

Tom
Hi Tom,
thanks for your kind words.
The LF distortion comes from highly nonlinear ferrite choke 18mH. This is intentional and simulates the real speaker LF nonlinearity. Finite DF then makes rise of LF distortion, but it would be in any case reflected in the speaker current even if the voltage source was ideal with zero distortion, as ideal voltage divided by nonlinear impedance makes nonlinear current.

To your second point, I am very well aware of the issue of the divided heatsinks in the 4U Modushop case. However I have nothing else here in the moment. Regardless this problem, the idle current remains stable enough probably for the reason of the big enough heatsinks and case. I understand that in case of miniature case and small heatsink that we can see with some projects it would be much worse and even the better thermal coupling would not prevent such amplifiers from overheating during the 1/3 power 1 hour test or during testing at maximum power. This amp passes 1 hour test at both maximum power and 1/3 power and stays just warm, not hot to touch.
 
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
1,604
Likes
2,572
Location
Prague
Thread Starter #25
Dummy load EPDR

I got a hint to show EPDR impedance of my dummy load as of post
https://www.audiosciencereview.com/...s-topology-and-measurements.21542/post-718212

So here it is and it probably explains why some amps have issues at high frequencies with it. EPDR stands for equivalent peak dissipation resistance.
Edit: the plots re-measured, one of my soundcards really has had issues.
dummy_load_EPDR2.png



I also need to add one comment. Starting with post #18
https://www.audiosciencereview.com/...s-topology-and-measurements.21542/post-726606
I changed the soundcard of my measuring system to get more BW and more predictable intrinsic system distortion as in the article 1a. It seems that in previous distortion measurements the low order distortion components might have been interacting with the soundcard distortion components, depending on their phase and amplitude and thus eventually reducing the distortion measured. That's the problem of not having the AP, however I cannot afford it for a DIY hobby. Fixed on March 31, plots of concern re-measured and re-posted.
 
Last edited:

tomchr

Addicted to Fun and Learning
Technical Expert
Manufacturer
Forum Donor
Joined
Nov 5, 2018
Messages
654
Likes
1,872
Location
Calgary, Canada
#27
The LF distortion comes from highly nonlinear ferrite choke 18mH. This is intentional and simulates the real speaker LF nonlinearity.
Fair enough. Thanks for confirming.

To your second point, I am very well aware of the issue of the divided heatsinks in the 4U Modushop case. [...] This amp passes 1 hour test at both maximum power and 1/3 power and stays just warm, not hot to touch.
Fantastic! That's all that matters.

I realize now that I made a mistake when I made the comment about the heat sink. I was thinking 250 W into 8 Ω, which would have been 400+ W into 4 Ω. You built to 250 W into 4 Ω (so probably 130-135ish W into 8 Ω). That makes a big difference in the thermal math and neatly explains why you can run the amp at 1/3 the rated power for hours without it overheating. My bad... Carry on.

Tom
 
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
1,604
Likes
2,572
Location
Prague
Thread Starter #28
Fair enough. Thanks for confirming.


Fantastic! That's all that matters.

I realize now that I made a mistake when I made the comment about the heat sink. I was thinking 250 W into 8 Ω, which would have been 400+ W into 4 Ω. You built to 250 W into 4 Ω (so probably 130-135ish W into 8 Ω). That makes a big difference in the thermal math and neatly explains why you can run the amp at 1/3 the rated power for hours without it overheating. My bad... Carry on.

Tom
No problem, Tom. Thank you!
 
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
1,604
Likes
2,572
Location
Prague
Thread Starter #29
Last edited:
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
1,604
Likes
2,572
Location
Prague
Thread Starter #30
This measurement is very interesting, to me. It is the THD vs. amplitude measured at 5kHz and 4ohm load, with magnified Y-axis scale. Measuring BW is 40kHz so we can see harmonics up to 8th. All H2 - H8 are used to calculate THD. We can nicely see the output stage transition point from class A to class B (green vertical line).

The maximum power in class A for a push-pull output stage is
Pa = 2 * Iq * Iq * Rload
here Iq = 0.2A (total idle current of the output stage), Rload = 4ohm
Pa = 0.32W

This phenomenon is easier to disclose at higher frequency like 5kHz (rather than 1kHz) but still we need enough harmonics in the analysis

A250W_THD_4R_5kHz_BW40k_3.png



Further measurements are made with different set of input dividers - so the resolution is higher but the measurement is limited to 6W. Plese note that the "Noise floor" is measuring system noise floor, not amplifier's. This is unfortunately not the AP with gradual switching of input dividers to minimize system noise.

A250W_THD_60uV-6W_4R_1kHz_BW40k.png


A250W_THD_60uV-6W_4R_5kHz_BW40k.png
 
Last edited:
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
1,604
Likes
2,572
Location
Prague
Thread Starter #32
Actually, it's P = I_peak^2*R/2, so 80 mW in your case.

Tom
No. Your formula is for SE class A, not for push-pull.
P = I_peak^2*R*2 for push-pull
Twice the idle squared for push-pull, half the idle squared for SE. You may verify in a simulator e.g.
 
Last edited:
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
1,604
Likes
2,572
Location
Prague
Thread Starter #33
This is the last measurement of THD at 5kHz/4ohm vs. power 20uW - 100mW measured again with another divider setup to get lower system noise.
A250W_THD_20uV-100mW_4R_5kHz_BW40k.png



Now there is something I am proud of. Next plot is the simulation of THD (without N) vs. Power at 5kHz and 4ohm - blue line, simulated in MC11. The red squares are the measured values. We can see they fit almost perfectly, only the power in the real amp is slightly lower - because of Power Supply voltage drop at power > 200W. The nowadays simulators are incredibly accurate. Bandwidth 40kHz in both cases.

A250_5kHz_dist_sim_real_s1.png
 

tomchr

Addicted to Fun and Learning
Technical Expert
Manufacturer
Forum Donor
Joined
Nov 5, 2018
Messages
654
Likes
1,872
Location
Calgary, Canada
#34
Ah. You're right. I can tell I haven't designed an output stage for a while. :)

P = (2Iq)^2*R/2 = 4Iq^2*R/2 = Iq^2*2R

Tom
 

tomchr

Addicted to Fun and Learning
Technical Expert
Manufacturer
Forum Donor
Joined
Nov 5, 2018
Messages
654
Likes
1,872
Location
Calgary, Canada
#36
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
1,604
Likes
2,572
Location
Prague
Thread Starter #37
Though I do not do it often, I have just also made distortion measurements into 8ohm resistive load. The plots are as follows:

A250W4R_8ohm_1kHz_dB.png

A250W4R load 8ohm, at 1kHz, both channels driven, THD vs. power, measuring BW 40kHz


A250W4R_8ohm_50W_thdfreq.png

A250W4R load 8ohm, both channels driven, THD vs. frequency at 50W power, measuring BW 40kHz

So we have the amplifier that makes 2x250W/4ohm and 2x150W/8ohm continuous power. This is IMO the entry level into "adult" power amplifiers ;).
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
6,715
Likes
16,879
Location
Gold Coast, Queensland, Australia
#38
@pma FTC specs are from 250mW to rated power. As we both know, many amplifiers are loathe to publish their low power THD (+N) numbers as they are dominated by noise. Many Class Ds are terrible in this regard.

How about a low power sweep or a spot set of frequencies at 250mW?

:)
 
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
1,604
Likes
2,572
Location
Prague
Thread Starter #39
How about a low power sweep or a spot set of frequencies at 250mW?
The area you are mentioning is under the noise of the measuring system (so the plot below some 0.5W does not reflect amplifier noise). I have the measurements with different setting of input dividers (thus the range of power is limited), see them attached below. There is nothing interesting happening. Range description is in the plots.

A250W4R_4ohm_1kHz_20uW-200mW_THD.png



A250W4R_4ohm_5kHz_5uW-50mW_THD.png

This measurement has probably revealed the real noise of the amplifier and not only the one of the measurement system
 
Top Bottom