My tpa3118 board - comments/critiques?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
OT:

Just for curiosity, this is a XL6009 boost converter powering a switched load (at 20Hz) in a distance of 10cm (3x18650 11V -> 36V), with batteries between sight picked up up with a 15cm wire-antenna at the scopes input...

Anyone ever complained about smps radiation? :trapper:

attachment.php
 

Attachments

  • boost.png
    boost.png
    19.2 KB · Views: 413
yes would be a conducted test. (fixture required)
They always require at minimum a ferrite line clamp or in severe case a seperate pi filter.

better ps have on board filter. the ps could still fail radiated test.

Antennas for radiated emissions are always calibrated ,Usually a log periodic.

Im not sure the scope probe on wire is telling complete story.
 
Had a free weekend, so had time to play around with my amp. Had a bit of "scratchy" noise in certain parts of music. Kept controlling my wiring, and re-checked the tracks on other sources. I ended up finding out even though I have 1cm between power leads and input leads they were not far enough apart. Moved the shielded RCA cable 2-3mm further away from the 24V power leads and the noise dissapeared. Will completely disassemble my amplifier and redo all wiring both shorter and further away from each other. Also added a very black background DAC this weekend. Really lifted the quality of sound.

20161105_140705.jpg
 
Member
Joined 2008
Paid Member
More Cowbell!

This is an international forum, so not sure how well-known the More Cowbell reference is outside the USA...

But anyway, real or imagined, on the cheap Chinese tpa311x boards, I always felt upgrading power supply decoupling caps to OS-CONs made a difference. One of my favorites was actually the Sure tpa311x (can't remember which x) that had eight PSU decoupling caps. After I replaced the stock caps with OS-CON, I really enjoyed that amp! Of course, I never used it to benchmark other designs, as I quickly moved on to the next project. But that amp always stands out in my mind with particularly fond memories.

So, in my mind, I always think, emphatically, More OSCON! And that immediately makes me think of the "More Cowbell" SnL skit. So I created yet another variant of this tpa3118 board, lovingly named More Cowbell.

How effective this amount of decoupling is I can't say. But given the way these PWM amps operate, I think there is a real benefit to getting power as clean as possible. I've seen other members use significant amount of capitors to provide clean, low impedance power to circuits (Abraxalito, I'm looking at you). So I figured why not?

The other limitation of these amps, discussed recently in this thread, is the matching of output filter to speaker load. So here I present a minimal ferrite bead filter. This assumes a very short run between amplifier and speaker, of course.

I envision these being used as monoblocks in a relatively low-power design. Specifically, I'm thinking 12v supply. Per the datasheet, with a 4R load, max power will be 20W (at 10% distortion, so you'll never actually go this high). At 90% efficiency the power supply only has to be capable of about 22W... Call it 24W for a little headroom and the convenient fact that there are a million 12V/2A power supplies available. With these modest current requirements, a very low noise power supply shouldn't be too hard to come by (and hopefully gets a further boost from the onboard caps).
 

Attachments

  • more_cowbell_20161118.png
    more_cowbell_20161118.png
    150.6 KB · Views: 206
what is the estimated inductance of FB assy.

if low enough maybe use an embedded inductor?

lots of old art around to design with eagle.

the newer designs get complex but using a two sided board is very workable.

so then eliminate two parts.

-
 

Attachments

  • image.jpeg
    image.jpeg
    118.6 KB · Views: 176
I'd say, it's "useless" from what the simulation says.

attachment.php


The voltage drop is domintated by the supplies ESR/resistance as the capacitance, even with 8x330uF, is still not enough to hold up the rails. Abraxalito is using a lot more capacitance, so a benefit is seen. With 8x330uF, the holdup time is less than 2ms for 1A current. Voltage then drops by 0.2R*1A in this example. Even then the rails voltage dropped by only 200mV or 1%.

(Things will change for 8x330000uF)
 

Attachments

  • noneed.png
    noneed.png
    7.3 KB · Views: 262
Last edited:
Member
Joined 2008
Paid Member
Somehow I missed this until now...

I'd say, it's "useless" from what the simulation says.

attachment.php


The voltage drop is domintated by the supplies ESR/resistance as the capacitance, even with 8x330uF, is still not enough to hold up the rails. Abraxalito is using a lot more capacitance, so a benefit is seen. With 8x330uF, the holdup time is less than 2ms for 1A current. Voltage then drops by 0.2R*1A in this example. Even then the rails voltage dropped by only 200mV or 1%.

(Things will change for 8x330000uF)

Just now I was looking at Panasonic 16SEPF1000M. This should work for a 12V supply; the cap is rated 16V, and 1000uF. Also, ESR is 0.012R (versus what looks like 0.050R in your sim). Does that improve the sim at all?

...Or, how about this: United Chemi-Con APSG160ELL222MJ20S. 16V, 2200uF, 0.008R ESR.
 
Not much at all. Keep in mind that ESR is measured at 100kHz, it's supposed to be higher at "DC". Compared to 100kHz and up, frequencies under 500Hz is "DC" for the cap.

Things will change from 100000uF and up. So assembling some more caps to get 0.1-1F with low resistance/inductance would do.

Will you hear it? Probably not.

Regards.
 
Hi guys i bought a NFB 11.28 and i have a big problem.. it i try to connect the NFB via RCA to my SMSL SA36A Pro (my speaker Amp) the NFB totally disappear from windows10
is a sort of incompatibility or i'm doing something wrong?

it's possible that the power supply of the speaker amplifier (24v 5A) make a incompatibility with nfb 11.28?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.