The tale of a single rail.. Is it possible?

[zerokelvin99]

Hi all,

After many hours of internet searching and increasingly complex solutions being dreamed up, how would you achieve:

600w+ @4R powered by a 42v li-ion battery

The TPA3255 would be the logical choice upto 50V through a boost converter, as its an efficient & single rail device. Whilst also being simple to implement and cheap.

But, what is the best way to go higher power? I am currently looking at a pair of isolated (£££) boost converters to a +/- 70v supply to a (relatively) high idling IRS2092 based board. Alternatively, I would buck down to 12v, then allow a car amplifier to boost to whatever it needs. Aside from being an ugly solution, this is also painfully lossy.

Please, ideas on a postcard below

 

[bucks bunny]

12V car amps include converters and provide poor overall efficiency. Most of them are crap imho. And a suitable 12V lead-acid-battery - how big and heavy do you like it?

The simplest and most efficient solution would be the TPA-3255 with a stepup-converter.
Besides - I would not go for this - because the difference between 42V and 51V is not worth the effort. But if this 600W number is your holy grail then go for it.

 

[russc]

Two TPA3255 amplifiers & speakers running off 42V.
How long will your battery last at full power?

 

[zerokelvin99]

@bucks bunny
I agree most car amps are cr@p. I'm sure there are good ones out there - but I'm not convinced I'd be lucky enough to find one.

The TPA3255 does 300W at 4R (50V rails). What I would be looking at is 600W at 4R (+3dB), approximately 70V at a single rail for BTL or +/- 70 for SE.

@russc
It is to drive a single driver, so multiple amps is unfortunately not going to help.

With a sensible crest factor, and assuming that the idle power is low - there should be more than enough hours out of the 350Wh battery.

 

[bucks bunny]

Btw, did you consider a 50V powered TPA3255 feeding the speaker through a matching transformer with nprim:nsec=1:1,41?

 

[dtaylo3]

Build your own battery to the size and voltage you need. After all, a Tesla is powered by 4,000-6,000 small batteries in parallel and series. You can buy boards to stack from this site.

DIY Powerwalls & Battery Modules by Jehu Garcia – Jag35

 

[jasonhanjk]

Consider this, per cell max current is 3A.

 

[zerokelvin99]

@ bucks bunny

Thanks, good idea! I will look into that and shall try to figure out the losses expected. I assume the VA rating of the transformer would also scale with the expected crest factor? I am trying to keep this efficient and light if possible.

@dtaylo3

I have a prebuilt battery, which is 10S5P. I don't have a spot welder to hand - ideally I would keep this as is rather than tearing it apart. Also, this would be an issue still. If I were to run with an IRS2092 board (which is the only option as I see it) I would have to reconfigure it as a 32S1P battery for +/- 69v. This would firstly waste 1/3 of the batteries and also require a massive balancer or an array of BMS boards

@ jasonhanjk

Hi, my current battery is able to put out 1kW continuous, so no problems with output there.


-

So, aside from the output transformer, it seems that shifting from +43v to the +/- 70v for the dual rail amps to provide 600w@4R still seems like an overly complex endeavour for +3dB!

 

[bucks bunny]

Crest factor does not help down sizing the transformer. Any peak signal must not drive the core into saturation. So I would start with the size of 600VA mains toroid for f-3dB=50Hz. For f-3dB=25Hz twice that size.

 

[zerokelvin99]

So, turns out the ICEpower modules can run off a single rail (great!)

https://icepower.dk/products/other/a-series/#comparison-chart

The 500A can provide the 600W at 4R with a reasonable distortion (below 1%).

Power efficiency appears good as well, slightly above TPA3255 level:

TPA3255
10w - 70%
100w - 82%

ICEpower 500A
10w - 72%
100w - 87%

Cost isn't too ridiculous at £100 per module. I'll need to factor in a +/- rail generator too (or at least work out how to do it)

Seems like a good solution!

 

[kipman725]

they still need a +/-12V analog supply though and it looks like they need bolting to a heatsink?

Also power sequencing might be needed (+/-12V come on before VCC)?

Efficiency at high power makes sense as TPA series devices have relatively high Rdson for their MOSFETS. However average power levels will be below peak levels so perhaps in the real world this won't be noticeable.

 

[zerokelvin99]

I figured the +/- 12v supply, being low power, could work with a dual rail ebay module of some kind? (TBworkedout)

Heatsink wise, they are happy for it to be bolted to almost anything (see attachment). So, good news there

For sequencing, I can either use the ADAU1701 to control the mute - or wire in one of my collection of timed relay PCBs I suppose.

And yes, I was just happy to see that there wasn't an efficiency penalty for the higher power modules!

 

[zerokelvin99]

Looks like there are a few isolated dual channel options for powering +/- 12v lines.

With low current draw of DSP on same +12v rail, should be able to get away with a 9W unit happily

NSD15-D-MEAN WELL Switching Power Supply Manufacturer
https://www.mouser.co.uk/ProductDetail/MEAN-WELL/DKE15C-12/?qs=l0g2inPJSHNXfW%2BTPY63rA==
https://uk.farnell.com/mean-well/dpb09c-12/dc-dc-converter-2-o-p-9w/dp/3253288
https://uk.farnell.com/recom-power/rec15e-2412dz/dc-dc-converter-2-o-p-15w/dp/2918473