IcePower module

[Phil Allison]

Hi to all my fans,

just completed repairs on one of these:

http://www.sound-power.ru/files/doc/ICEpower125ASX2_datasheet.pdf

Found it fitted inside a mini-size bass instrument amplifier, which the owner had just purchased over the net and failed on him at first power up.

Smelling a rat, I checked the AC switch on the rear panel first - blow me down, it was set to 120VAC and we live in a 240V country.

Reset the switch, then it drew zero AC current - no surprise.

Replaced the blown 5AT fuse in the IcePower module and then the PSU began hiccupping - drawing an amp or two very briefly then shutting off every few seconds.

IME feeding double AC supply voltage to a unit with a SMPS is almost always fatal, sure the AC fuse blows but semis in the HV part of the circuit fail in a fraction of a second. But these must be OK this time because the PSU was hiccupping - ie working but sensing an overload so shutting down.

Soon enough, multimeter testing showed two small power mosfets on the output side of the SMPS tranny were shorted while two others seemed OK. But what the heck are mosfets doing there ?

Well, it's the fist time I have come across seen a *synchronous rectifier* in such an unit. IcePower had certainly gone all out to reduce losses and heat in this module by using a synchronous bridge.

The particular TO-252 fets here are made by ON, number FDD86110 rated at 100V and 8mohms on resistance.

Removing the duds cleared all shorts and after fitting some new ones ( delivered to me by Element14, from their Singapore warehouse ) I gingerly Variaced the unit up from zero.

No hiccupping this time and it began to run normally - after which it passed all my usual bench testing, no problems.

FYI:

AC current draw was 120mA at idle, 2.4A at full power (340W rms/4ohms)

No PFC and inrush surges were about 25A for a few milliseconds.

There was a residual sine wave signal on the audio output of about 1V at 500kHz - no biggie except it makes THD testing a right PITA.

I find it odd the way the SMPS failed, just two mosfets in the secondary rectifier, nothing else. For a brief time, the DC supply to the switching fets must have been nearly double voltage.

Seems the two rectifier fets failed SHORT instantly and protected the rest of the circuitry - remember the power supply is NOT regulated, just a square wave inverter running at 100kHz.

I also feel that supplying an expensive amplifier to a buyer living in a 240V country with the AC inlet set to 120V and giving NO warning is criminal. BTW the amp was fitted with a regular IEC inlet which most of the world uses for 240VAC.

Any comments?



.... Phil

 

So you didn't have to really get into the drive circuitry, right ? I don't wish that on anyone, even you.

I saw one in a bass amp years ago. It had carbon tracks near the input. (on the board) It was fried, I looked into getting a replacement but for that unit it would be $100 for the part so the job didn't get done. Except for the big boys here, musicians are cheap, and by necessity because they are broke.

It had arced over from hot to cold, you could see it. And the same guy had another amp or something in it with the same type of fault, well not really but the same cause I would guess. Too much voltage between hot and cold, both of them.

One unit I fixed, but the one with the Icepower no. I quoted it for the whole module which I found for a hundred bucks. I might do like you and fix the choppers and rectifiers and all that but I am not getting really into it. Too small, probably a four sided board. I would probably have to charge a hundred bucks to not change a hundred buck module.

I don't want to work for three bucks an hour but I will sometimes for my business. It makes up for the times when I make a hundred bucks an hour. Otha, worked for the olman in his days. Got rid of Vette bodies, scrapped the steel and somehow dealt with the fiberglass n shit. He told the olman "I takes the good with the bad". That whole episode is a (his)story in and of itself.

Anyway, nice that you go it working. But you know if you want to get into the intricacies of it that is a pain in the ass and for a hundred buck thing is not worth it. But this was, kinda.

I would guess you charged fro diodes, transistors, caps and whatever, how much did you actually make ?

 

[Phil Allison]

jurb...@gmail.com wrote:

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

So you didn't have to really get into the drive circuitry, right ?
I don't wish that on anyone, even you.

** Well tanks pal - it's got enough SMD to choke a brown dog.

It had arced over from hot to cold, you could see it.
And the same guy had another amp or something in it with the
same type of fault, well not really but the same cause I
would guess. Too much voltage between hot and cold, both of them.

** Maybe hit by a lightning spike ?

Anyway, nice that you go it working. But you know if you want to
get into the intricacies of it that is a pain in the ass and for
a hundred buck thing is not worth it. But this was, kinda.

** No full schem available anywhere - a new Ice Power 125A module would cost me about A#300 plus postage from the USA or Europe. If that were needed, I would get the owner to buy it for me.

I would guess you charged fro diodes, transistors, caps and whatever,
how much did you actually make ?

** Only the two mosfets were gone, plus a 5AT fuse. I billed the owner A$150 for my time, so it came just inside my estimate/quote of A$200.

But one thing still worries me, the amp output operates as bridged pair and may fail if the ever shorted. It only has a single Speakon on the back, but the owner may well try to use a Speakon to jack lead. Bad news with a 350W amp.

So I'm gonna include a Speakon to Speakon, made with a metre of fancy round cable, plus a matching socket sitting on one end for him to use - for an extra consideration of course !!

Also I will advise NOT sitting it on top of his bass cab, cos doing that shakes stuff to bits.


..... Phil

 

[Winfield Hill]

Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp

My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!


--
Thanks,
- Win

 

[Phil Allison]

Winfield Hill wrote:
------------------

Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp


My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!

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

** FYI to Win:

The majority of SMPS integrated with a high powered audio amplifier on the same PCB ( or otherwise) are simple, square wave inverters.

In the wide, wonderful world of power audio engineering, there is simply no need for a regulated or supply frequency free PSU - in order to feed DC power to such an amplifier.

Linear amps ( class B or AB or whatever ) are easy to design with almost any desired degree of PSRR - while class D amps can also have that same ability with little difficulty. Just include a sample of the ripple voltage in the feedback loop.

However, professional power amps ( 2kW per ch and above ) need their PSUs to incorporate PFC - so that the RMS current draw does not become some insane number. Otherwise, it's a PITA to supply enough AC current power to run dozens of the damn things as needed for a major concert.

With PCF, the DC supply for the following *square wave inverter* is virtually ripple free and *regulated* - so the amp has the same power spec almost regardless of supply variation. Consider that the AC supply in use may well be a diesel generator, on the end of a long length of three phase cable.

The various IcePower modules ( engineered in Europe, by the B&O company of Denmark ) are decent designs mostly intended for the domestic, hi-fi market - but complete bastards if you are asked or fix one that has failed for any reason.

Same comment goes for most SMD boards of any complexity.




.... Phil

 

[John S]

On 9/19/2019 8:27 PM, Phil Allison wrote:

Hi to all my fans,

just completed repairs on one of these:

http://www.sound-power.ru/files/doc/ICEpower125ASX2_datasheet.pdf

Found it fitted inside a mini-size bass instrument amplifier, which the owner had just purchased over the net and failed on him at first power up.

Smelling a rat, I checked the AC switch on the rear panel first - blow me down, it was set to 120VAC and we live in a 240V country.

Reset the switch, then it drew zero AC current - no surprise.

Replaced the blown 5AT fuse in the IcePower module and then the PSU began hiccupping - drawing an amp or two very briefly then shutting off every few seconds.

IME feeding double AC supply voltage to a unit with a SMPS is almost always fatal, sure the AC fuse blows but semis in the HV part of the circuit fail in a fraction of a second. But these must be OK this time because the PSU was hiccupping - ie working but sensing an overload so shutting down.

Soon enough, multimeter testing showed two small power mosfets on the output side of the SMPS tranny were shorted while two others seemed OK. But what the heck are mosfets doing there ?

Well, it's the fist time I have come across seen a *synchronous rectifier* in such an unit. IcePower had certainly gone all out to reduce losses and heat in this module by using a synchronous bridge.

The particular TO-252 fets here are made by ON, number FDD86110 rated at 100V and 8mohms on resistance.

Removing the duds cleared all shorts and after fitting some new ones ( delivered to me by Element14, from their Singapore warehouse ) I gingerly Variaced the unit up from zero.

No hiccupping this time and it began to run normally - after which it passed all my usual bench testing, no problems.

FYI:

AC current draw was 120mA at idle, 2.4A at full power (340W rms/4ohms)

No PFC and inrush surges were about 25A for a few milliseconds.

There was a residual sine wave signal on the audio output of about 1V at 500kHz - no biggie except it makes THD testing a right PITA.

I find it odd the way the SMPS failed, just two mosfets in the secondary rectifier, nothing else. For a brief time, the DC supply to the switching fets must have been nearly double voltage.

Seems the two rectifier fets failed SHORT instantly and protected the rest of the circuitry - remember the power supply is NOT regulated, just a square wave inverter running at 100kHz.

I also feel that supplying an expensive amplifier to a buyer living in a 240V country with the AC inlet set to 120V and giving NO warning is criminal. BTW the amp was fitted with a regular IEC inlet which most of the world uses for 240VAC.

Any comments?



... Phil

Yes. Well done, Phil. I enjoy reading about troubleshooting methods. Thanks.

 

[Lasse Langwadt Christense]

fredag den 20. september 2019 kl. 11.23.11 UTC+2 skrev Winfield Hill:

Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp

My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!

not quite china prices but they have a few with psu

http://www.cadaudio.dk/cadpwmmodules.htm

 

[bitrex]

On 9/20/19 12:16 PM, jlarkin@highlandsniptechnology.com wrote:

On Fri, 20 Sep 2019 12:07:58 -0400, bitrex <user@example.net> wrote:

On 9/20/19 11:54 AM, jlarkin@highlandsniptechnology.com wrote:
On Fri, 20 Sep 2019 08:28:01 -0700 (PDT), Lasse Langwadt Christensen
langwadt@fonz.dk> wrote:

fredag den 20. september 2019 kl. 11.23.11 UTC+2 skrev Winfield Hill:
Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp

My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!


not quite china prices but they have a few with psu

http://www.cadaudio.dk/cadpwmmodules.htm

Half-bridge class-D amps are interesting. As they pull power out of
one supply rail, they pump power into the other one.

All those huge-power specs must be some sort of intermittent peak
music thing. There's no way those tiny boards can get rid of much
heat.



You don't think this board can do 100 watts "RMS" into 8 ohms?

http://www.cadaudio.dk/d2050s.jpg

I bet it can just fine

With that big heat sink, maybe so. But they have 500 watt units, and
kilowatt boards, that look scary.

We're designing a 120 watt full-bridge class D amp, and I'm sure going
to have a big heat sink and forced air flow.

The problem with naive class D for consumer audio applications is that
the efficiency is quite good, way better than a linear amp, when running
flat out but nobody uses an amp that way, at normal volume levels the
power efficiency stinks and is not that much better than a class B.

In cheap audio gear they put the cheap kind in there cuz they're cheap
and you have to heat sink a linear amp for the worst case. If you want
great efficiency at low power outputs on a class D you have to resort to
many of the same tricks as a linear amp like rail tracking/switching for
a fixed supply or dynamically boosting the supply voltage of SMPS supply
for the peaks

 

On Fri, 20 Sep 2019 12:07:58 -0400, bitrex <user@example.net> wrote:

On 9/20/19 11:54 AM, jlarkin@highlandsniptechnology.com wrote:
On Fri, 20 Sep 2019 08:28:01 -0700 (PDT), Lasse Langwadt Christensen
langwadt@fonz.dk> wrote:

fredag den 20. september 2019 kl. 11.23.11 UTC+2 skrev Winfield Hill:
Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp

My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!


not quite china prices but they have a few with psu

http://www.cadaudio.dk/cadpwmmodules.htm

Half-bridge class-D amps are interesting. As they pull power out of
one supply rail, they pump power into the other one.

All those huge-power specs must be some sort of intermittent peak
music thing. There's no way those tiny boards can get rid of much
heat.



You don't think this board can do 100 watts "RMS" into 8 ohms?

http://www.cadaudio.dk/d2050s.jpg

I bet it can just fine

With that big heat sink, maybe so. But they have 500 watt units, and
kilowatt boards, that look scary.

We're designing a 120 watt full-bridge class D amp, and I'm sure going
to have a big heat sink and forced air flow.

 

On Fri, 20 Sep 2019 08:28:01 -0700 (PDT), Lasse Langwadt Christensen
<langwadt@fonz.dk> wrote:

fredag den 20. september 2019 kl. 11.23.11 UTC+2 skrev Winfield Hill:
Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp

My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!


not quite china prices but they have a few with psu

http://www.cadaudio.dk/cadpwmmodules.htm

Half-bridge class-D amps are interesting. As they pull power out of
one supply rail, they pump power into the other one.

All those huge-power specs must be some sort of intermittent peak
music thing. There's no way those tiny boards can get rid of much
heat.

 

[bitrex]

On 9/20/19 11:54 AM, jlarkin@highlandsniptechnology.com wrote:

On Fri, 20 Sep 2019 08:28:01 -0700 (PDT), Lasse Langwadt Christensen
langwadt@fonz.dk> wrote:

fredag den 20. september 2019 kl. 11.23.11 UTC+2 skrev Winfield Hill:
Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp

My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!


not quite china prices but they have a few with psu

http://www.cadaudio.dk/cadpwmmodules.htm

Half-bridge class-D amps are interesting. As they pull power out of
one supply rail, they pump power into the other one.

All those huge-power specs must be some sort of intermittent peak
music thing. There's no way those tiny boards can get rid of much
heat.

You don't think this board can do 100 watts "RMS" into 8 ohms?

<http://www.cadaudio.dk/d2050s.jpg>

I bet it can just fine

 

[bitrex]

On 9/20/19 11:54 AM, jlarkin@highlandsniptechnology.com wrote:

On Fri, 20 Sep 2019 08:28:01 -0700 (PDT), Lasse Langwadt Christensen
langwadt@fonz.dk> wrote:

fredag den 20. september 2019 kl. 11.23.11 UTC+2 skrev Winfield Hill:
Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp

My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!


not quite china prices but they have a few with psu

http://www.cadaudio.dk/cadpwmmodules.htm

Half-bridge class-D amps are interesting. As they pull power out of
one supply rail, they pump power into the other one.

All those huge-power specs must be some sort of intermittent peak
music thing. There's no way those tiny boards can get rid of much
heat.

What heat? the high quality ones are like 95% efficient just like a SMPS.

 

On Fri, 20 Sep 2019 12:07:58 -0400, bitrex <user@example.net> wrote:

On 9/20/19 11:54 AM, jlarkin@highlandsniptechnology.com wrote:
On Fri, 20 Sep 2019 08:28:01 -0700 (PDT), Lasse Langwadt Christensen
langwadt@fonz.dk> wrote:

fredag den 20. september 2019 kl. 11.23.11 UTC+2 skrev Winfield Hill:
Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp

My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!


not quite china prices but they have a few with psu

http://www.cadaudio.dk/cadpwmmodules.htm

Half-bridge class-D amps are interesting. As they pull power out of
one supply rail, they pump power into the other one.

All those huge-power specs must be some sort of intermittent peak
music thing. There's no way those tiny boards can get rid of much
heat.



You don't think this board can do 100 watts "RMS" into 8 ohms?

http://www.cadaudio.dk/d2050s.jpg

I bet it can just fine

Do they spec that it can?

The audio business is dominated by various techniques of creative
lying. The aerospace business isn't.

 

[John Larkin]

On Fri, 20 Sep 2019 12:13:44 -0400, bitrex <user@example.net> wrote:

On 9/20/19 11:54 AM, jlarkin@highlandsniptechnology.com wrote:
On Fri, 20 Sep 2019 08:28:01 -0700 (PDT), Lasse Langwadt Christensen
langwadt@fonz.dk> wrote:

fredag den 20. september 2019 kl. 11.23.11 UTC+2 skrev Winfield Hill:
Phil Allison wrote...

just completed repairs on one of these:
http://www.sound-power.ru/files/doc/ICEp

My comment would be, it's nice to have
AC power integrated with the amplifier.

I've been getting class-D amplifier PCBs
from China, very cheap, yet many of them
seem well engineered and well built. But
none of them include a power supply, nor
do they offer a suitable power supply.
But, damn, these high-power supplies can
be more complicated than the amplifier!


not quite china prices but they have a few with psu

http://www.cadaudio.dk/cadpwmmodules.htm

Half-bridge class-D amps are interesting. As they pull power out of
one supply rail, they pump power into the other one.

All those huge-power specs must be some sort of intermittent peak
music thing. There's no way those tiny boards can get rid of much
heat.



What heat? the high quality ones are like 95% efficient just like a SMPS.

1KW x 5% is 50 watts. If indeed it is 95% efficient.

Some of the wire on those inductors looks absurd for the specified
currents.

 

[whit3rd]

On Friday, September 20, 2019 at 10:20:11 AM UTC-7, John Larkin wrote:

1KW x 5% is 50 watts. If indeed it is 95% efficient.

Some of the wire on those inductors looks absurd for the specified
currents.

Yeah, but... the usual situation in audio is that you allow 5x headroom (current
over 20% of the peak value is always going to make distortion). It's part of
the creative lying process to rely on statistical low averages for dissipation, while
quoting the peaks...

I tried a class D amplifier design around 1980; it just wasn't feasible to handle
the current spikes with low-cost hardware (in fact, for some reason, my best
choice for power transistor was PNP silicon). That high current makes a strong
impetus to use short wires, so the excessively-compact design
of the IcePower module looks... efficient.

 

[John Larkin]

On Fri, 20 Sep 2019 11:14:56 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

On Friday, September 20, 2019 at 10:20:11 AM UTC-7, John Larkin wrote:

1KW x 5% is 50 watts. If indeed it is 95% efficient.

Some of the wire on those inductors looks absurd for the specified
currents.

Yeah, but... the usual situation in audio is that you allow 5x headroom (current
over 20% of the peak value is always going to make distortion). It's part of
the creative lying process to rely on statistical low averages for dissipation, while
quoting the peaks...

You can buy a 300 watt Chinese amp that comes with a tiny wall wart.

I tried a class D amplifier design around 1980; it just wasn't feasible to handle
the current spikes with low-cost hardware (in fact, for some reason, my best
choice for power transistor was PNP silicon). That high current makes a strong
impetus to use short wires, so the excessively-compact design
of the IcePower module looks... efficient.

I'm planning to use the TI TPA3255. We have the eval board and it
seems to be unbreakable. But it will have a heat sink and some planned
air flow.

 

[Winfield Hill]

John Larkin wrote...

I'm planning to use the TI TPA3255. We have the eval
board and it seems to be unbreakable. But it will
have a heat sink and some planned air flow.

That chip has on-board MOSFETs, with high Ron, so it
can't be optimized for high efficiency. I prefer to
separate the two functions. For example, an IRS2092
clocks to 0.8MHz, and it has optimized MOSFET pairs
to go with it, like the IRFI4020H. Or select other
MOSFETs with much lower Ron, and proper thermal tabs,
like the 330-watt IRFB4227, 20-milliohms (that's 5x
lower), used in some 1kW-rated amplifiers.


--
Thanks,
- Win

 

[bitrex]

On 9/20/19 3:53 PM, John Larkin wrote:

On Fri, 20 Sep 2019 11:14:56 -0700 (PDT), whit3rd <whit3rd@gmail.com
wrote:

On Friday, September 20, 2019 at 10:20:11 AM UTC-7, John Larkin wrote:

1KW x 5% is 50 watts. If indeed it is 95% efficient.

Some of the wire on those inductors looks absurd for the specified
currents.

Yeah, but... the usual situation in audio is that you allow 5x headroom (current
over 20% of the peak value is always going to make distortion). It's part of
the creative lying process to rely on statistical low averages for dissipation, while
quoting the peaks...

You can buy a 300 watt Chinese amp that comes with a tiny wall wart.


I tried a class D amplifier design around 1980; it just wasn't feasible to handle
the current spikes with low-cost hardware (in fact, for some reason, my best
choice for power transistor was PNP silicon). That high current makes a strong
impetus to use short wires, so the excessively-compact design
of the IcePower module looks... efficient.

I'm planning to use the TI TPA3255. We have the eval board and it
seems to be unbreakable. But it will have a heat sink and some planned
air flow.

For low powers of a watt or two at audio frequency from a uP I just run
PWM into an LM386 pulling double duty as a 2 pole Sallen-Key filter, a
Bessel/Butterworth hybrid from the TI Filterpro designer. 2.2k, 4.7k,
0.47u x2. inverting input biased through a divider from the output with
a diode drop to compensate for the internal offset. the old '386 can
drive inductive loads and small motors too and take a lot of punishment

 

[bitrex]

On 9/20/19 5:11 PM, bitrex wrote:

On 9/20/19 3:53 PM, John Larkin wrote:
On Fri, 20 Sep 2019 11:14:56 -0700 (PDT), whit3rd <whit3rd@gmail.com
wrote:

On Friday, September 20, 2019 at 10:20:11 AM UTC-7, John Larkin wrote:

1KW x 5% is 50 watts. If indeed it is 95% efficient.

Some of the wire on those inductors looks absurd for the specified
currents.

Yeah, but... the usual situation in audio is that you allow 5x
headroom (current
over 20% of the peak value is always going to make distortion).
It's part of
the creative lying process to rely on statistical low averages for
dissipation, while
quoting the peaks...

You can buy a 300 watt Chinese amp that comes with a tiny wall wart.


I tried a class D amplifier design around 1980; it just wasn't
feasible to  handle
the current spikes with low-cost hardware (in fact, for some reason,
my best
choice for power transistor was PNP silicon).   That high current
makes a strong
impetus to use short wires, so the excessively-compact design
of the IcePower module  looks... efficient.

I'm planning to use the TI TPA3255. We have the eval board and it
seems to be unbreakable. But it will have a heat sink and some planned
air flow.


For low powers of a watt or two at audio frequency from a uP I just run
PWM into an LM386 pulling double duty as a 2 pole Sallen-Key filter, a
Bessel/Butterworth hybrid from the TI Filterpro designer. 2.2k, 4.7k,
0.47u x2. inverting input biased through a divider from the output with
a diode drop to compensate for the internal offset. the old '386 can
drive inductive loads and small motors too and take a lot of punishment

oh, and cap to bypass the internal feedback so it has high open loop gain

 

[John Larkin]

On 20 Sep 2019 14:22:45 -0700, Winfield Hill <winfieldhill@yahoo.com>
wrote:

John Larkin wrote...

I'm planning to use the TI TPA3255. We have the eval
board and it seems to be unbreakable. But it will
have a heat sink and some planned air flow.

That chip has on-board MOSFETs, with high Ron, so it
can't be optimized for high efficiency. I prefer to
separate the two functions. For example, an IRS2092
clocks to 0.8MHz, and it has optimized MOSFET pairs
to go with it, like the IRFI4020H. Or select other
MOSFETs with much lower Ron, and proper thermal tabs,
like the 330-watt IRFB4227, 20-milliohms (that's 5x
lower), used in some 1kW-rated amplifiers.

The fets in the TPA3255 are 85 mohms, and in mono mode I'll parallel
the two half-bridges. I only need 120 watts!

The advantage of a single amp chip is all the protections that it
includes. And the small size and low parts count. And price, although
that's not a big deal on this project... one output transformer costs
10x the TI chip. One heat sink will cost a bunch more than an amp
chip.