A forum member just bought a Zero Zone TPA3255 amplifier board and supplied it with just below the recommended 2x28Vac. A nice looking board. Quickly, the LM2575 chip (buck) for the auxiliary 12V supply burned off to an extent where a part of the chip-housing left the chip.
Photos from "before" showed it was an LM2575S Buck regulator with an absolute maximum for the input voltage of 40V. The recommended supply voltage (2x28Vac), when rectified, can pass 40Vdc in idle mode of the amplifier. We believe(d) this to be a reason for the LM2575S to break down and an LM2575HVS should have been used instead.
Studying the LM2575 datasheet, 330uH is shown as the Buck choke. On a photo of the board I notice a "100" marked choke of the type used for cheap class D amplifier output filters. "100" should mean only 10uH. I made an estimate of the ripple current in the choke and I arrive at 16A peak!
The switching frequency of the LM2575 is 52KHz which should correspond to a switching period of 19.2us. The Buck ON-period is found to be (12V/40V)x19.2us=5.8us. The current rise in the 10uH choke during 5.8us with 28V across the choke (40V-12V), I find to be (28V/10uH)x5.8us=16.2A!
The 1A rated LM2575 is trying to handle a current ripple that should be 16A. If I'm right, it will never succeed and the Buck choke is far too small. Very likely the reason for the LM2575 to burn.
Can anybody see if I make a mistake in my calculations?
Photos from "before" showed it was an LM2575S Buck regulator with an absolute maximum for the input voltage of 40V. The recommended supply voltage (2x28Vac), when rectified, can pass 40Vdc in idle mode of the amplifier. We believe(d) this to be a reason for the LM2575S to break down and an LM2575HVS should have been used instead.
Studying the LM2575 datasheet, 330uH is shown as the Buck choke. On a photo of the board I notice a "100" marked choke of the type used for cheap class D amplifier output filters. "100" should mean only 10uH. I made an estimate of the ripple current in the choke and I arrive at 16A peak!
The switching frequency of the LM2575 is 52KHz which should correspond to a switching period of 19.2us. The Buck ON-period is found to be (12V/40V)x19.2us=5.8us. The current rise in the 10uH choke during 5.8us with 28V across the choke (40V-12V), I find to be (28V/10uH)x5.8us=16.2A!
The 1A rated LM2575 is trying to handle a current ripple that should be 16A. If I'm right, it will never succeed and the Buck choke is far too small. Very likely the reason for the LM2575 to burn.
Can anybody see if I make a mistake in my calculations?
Last edited:
May be no error or several of them, First, check that really 100µH coil is really 100 and no 10.
Second, LM IC's are known to be faulty. For buck regulator, my experience that L4970's from ST are the bests, with MOSFET output, easy to sync and has soft start circuitry.
Also, larger choke inductance means less current ripple, but slower compensation network, so, more sag under audio peaks.
I made a current shared out of phase voltage regulator using two L4974's running at 100KHz and controlled from a HCF4047 oscillator and divider, whit high degree of confidence: 10 years doing it job.
Second, LM IC's are known to be faulty. For buck regulator, my experience that L4970's from ST are the bests, with MOSFET output, easy to sync and has soft start circuitry.
Also, larger choke inductance means less current ripple, but slower compensation network, so, more sag under audio peaks.
I made a current shared out of phase voltage regulator using two L4974's running at 100KHz and controlled from a HCF4047 oscillator and divider, whit high degree of confidence: 10 years doing it job.
A forum member just bought a Zero Zone TPA3255 amplifier board and supplied it with just below the recommended 2x28Vac. A nice looking board. Quickly, the LM2575 chip (buck) for the auxiliary 12V supply burned off to an extent where a part of the chip-housing left the chip.
Photos from "before" showed it was an LM2575S Buck regulator with an absolute maximum for the input voltage of 40V. The recommended supply voltage (2x28Vac), when rectified, can pass 40Vdc in idle mode of the amplifier. We believe(d) this to be a reason for the LM2575S to break down and an LM2575HVS should have been used instead.
Studying the LM2575 datasheet, 330uH is shown as the Buck choke. On a photo of the board I notice a "100" marked choke of the type used for cheap class D amplifier output filters. "100" should mean only 10uH. I made an estimate of the ripple current in the choke and I arrive at 16A peak!
The switching frequency of the LM2575 is 52KHz which should correspond to a switching period of 19.2us. The Buck ON-period is found to be (12V/40V)x19.2us=5.8us. The current rise in the 10uH choke during 5.8us with 28V across the choke (40V-12V), I find to be (28V/10uH)x5.8us=16.2A!
The 1A rated LM2575 is trying to handle a current ripple that should be 16A. If I'm right, it will never succeed and the Buck choke is far too small. Very likely the reason for the LM2575 to burn.
Can anybody see if I make a mistake in my calculations?
Many thanks for the investigation.
Hummm let me correct one thing :
recommended PSU is 1X28V @ 300W (not 2X28V)
It means that mine delivers about 10.7A on the single 28V AC Output.... (4 wires, input 220V , single output 28V)
Here the exact reference of my Toroidal PSU :
TST 300/013 INDEL - Transformateur: toroidal | 300VA; 230VAC; 28V; 10,71A; 2,9kg; H: 58mm; TST300W/28V | TME - Composants electroniques
And here the link to the recommended one :
300W Toroid Transformer 0-115V 0-115V to 0-28V for TPA3255 amplifier board | eBay
Both seem to be different... Maybe the error comes from my PSU choice ?
I see on their pictures that it seems to have more wires on it....
Last edited:
Thanks Dani, 1x28Vac. I have been dealing with too many symmetrical power supplies lately.
NB: I believe I have more 330uH in my cupboard so I can send you one if you like.
Many thanks, do not worry ==) I already orderd a LM2575HVS + LM2575S ...
So I will let you know before soldering it to avoid any mistake =)
if I should buy a new PSU, I would like a link to the good one (Amazon or Ebay)
Last edited:
Hi Dani,
Let's wait a bit more for comments if I am wrong in my assumption. Else, the 10uH choke should be replaced with a choke having a much higher value. TI suggests 330uH which should leave a current ripple of some 0.5A. With the present 10uH choke, you just risk blowing your replacement LM2575 as well. The impedance of 10uH is far too low.
I hope that there is no more engineering faults on the board.
Your PSU (the transformer) is fine.
Do you know someone who has a good soldering iron and is used to do work on electronics boards?
You/he needs to have a "cutter" for electronics, something like this: Platon americain Original. PLATO 170 Wishful pince bricolage pinces pinces diagonales electroniques pinces de coupe laterale coupe fil-in Accessoires pour outils electriques from Outils on AliExpress
Let's wait a bit more for comments if I am wrong in my assumption. Else, the 10uH choke should be replaced with a choke having a much higher value. TI suggests 330uH which should leave a current ripple of some 0.5A. With the present 10uH choke, you just risk blowing your replacement LM2575 as well. The impedance of 10uH is far too low.
I hope that there is no more engineering faults on the board.
Your PSU (the transformer) is fine.
Do you know someone who has a good soldering iron and is used to do work on electronics boards?
You/he needs to have a "cutter" for electronics, something like this: Platon americain Original. PLATO 170 Wishful pince bricolage pinces pinces diagonales electroniques pinces de coupe laterale coupe fil-in Accessoires pour outils electriques from Outils on AliExpress
OK, I will wait, in the worst of the case I will put back the LM2575S and buy a 20VAC PSU instead....
I will bring the card to my local Electronic Shop so that they can make the soldering action : I am very very bad in Soldering...
Of course. It is the choke marked with "100" closest to the LM2575 IC. It corresponds to the 330uH choke on the top of page 2 of the LM2575 datasheet: https://www.onsemi.com/pub/Collateral/LM2575-D.PDF
On the board they have put 10uH while 330uH is advised. Quite a difference. 330uH has the marking "331".
Attachments
Hi Dani,
I understand that my doubt about the Buck choke value is confusing. How come that a board that should just work is stated to have more important flaws?
ON Semi's datasheet ( https://www.onsemi.com/pub/Collateral/LM2575-D.PDF ) has a good explanation about continuous or discontinuous operation on page 17. A 10uH Buck choke should give a rather extreme discontinuous operation. ON Semi has, on page 13, choke selection graphs. For the LM2575-12, with 40V input voltage and a maximum output current near 0.4A, a Buck choke of 1500uH ("H1500") is recommended. Even above the 330uH I mentioned before.
This indicates support to my worry about the 10uH.
I understand that my doubt about the Buck choke value is confusing. How come that a board that should just work is stated to have more important flaws?
ON Semi's datasheet ( https://www.onsemi.com/pub/Collateral/LM2575-D.PDF ) has a good explanation about continuous or discontinuous operation on page 17. A 10uH Buck choke should give a rather extreme discontinuous operation. ON Semi has, on page 13, choke selection graphs. For the LM2575-12, with 40V input voltage and a maximum output current near 0.4A, a Buck choke of 1500uH ("H1500") is recommended. Even above the 330uH I mentioned before.
This indicates support to my worry about the 10uH.
As soon as I get the LM2575S + LM2575HVS I will let you know here.
As that moment, I wish you could tell me which one I need to provide to my electronical shop so that they can replace it.
Last edited: