LM4780 in bridged or parallel to keep Pdmax down?

[jaste]

From the datasheat I understand that it should run cooler in prallell then in bridged config, right?

My problem is that I´ve made a enclosure that i really like but it seems to have to small heatsink. At the moment I´m running each chip in parallell configuration but I might aswell run them bridged if that would make them cool down a bit, but that seems to not be the case...

[artquake]

check your situation with THIS ...

or tell us more about it ...

Any schematic of what you assembled ?

artQuake

[jaste]

I´ve tried to get something out of that aplication but it just tells me that my supply voltage (+-37Vdc) is to high to keep it cool. That might be the case but i´m to blind to see... I want it to run cooler, but i try to tweak the case aswell.

[pacificblue]

Depends on the load you drive. Look for AN-1192, where both operation modes and thermal considerations are explained.

You should also be aware that an LM4780 has a different temperature perception from a human being. What you may find too hot might be just cosy for the IC.

[artquake]

it seems to have to small heatsink

Why to small ? how hot does it run then ? Have you measured it ? What load are you driving ? Have you measured DC-Offset before paralleling them ?
If the application tells it's too high to run it cool, why not using a suitable supply voltage ?

Look for AN-1192, where both operation modes and thermal considerations are explained.

These considerations are very optimistic due to heat, in fact you'll need mostly a bigger heatsink as they recommend ... i've somewhere calculated it here somewhere as i had a smoking BPA300 ...

A servo is a good way into low heat dissipation ... or even e good matched pair of those chips ... and high speaker impedance ... and above all; high efficiency speakers

[jaste]

Well I´m driving a pair of 8 ohm speakers and intend on doing that for some time.
I can´t really lower the supply voltage if I don´t buy another tranny and that´s not an option for my economy at the time.

I thought that the chips didn´t need as serious matching in bridge as they do in parallell so therfore they would run cooler. the DC offset I can´t remember but it was low enough, around 70mv I think.

[pacificblue]

How do you match chips?

In parallel operation you should match the gain setting resistors Ri and Rf or use a DC servo.

These considerations are very optimistic due to heat, in fact you'll need mostly a bigger heatsink as they recommend

A fan was recommended to improve heat dissipation, because each ICs power dissipation capability is limited to 30W for the TF type and 40W for the T type. With fan that could be increased to 45W for TF and 60W for T.

A servo is a good way into low heat dissipation ... or even e good matched pair of those chips ... and high speaker impedance ... and above all; high efficiency speakers

...and reasonable listening volumes?

Even low efficiency speakers give you somewhere around 80 dB/W/m. Here in Germany you are obliged to wear hearing protection at workplaces with that amount of noise. And that is only 1W average. Make it 10W, for greater listening distance. And now share that among 3 LM4780 or 6 LM3886 that you would need for BPA-300. Could it be that your BPA-300 started smoking for different reasons? And that your bigger heatsink is curing the symptoms instead of the disease?

[artquake]

Originally posted by pacificblue
How do you match chips?

Having a prototype circuit and using a "Prüfsockel" donno the exact word in english, and so mesuring all chips under +/- the same conditions ... Maybee there are easyer ways to check the chips for which i don't know ?

Originally posted by pacificblue
In parallel operation you should match the gain setting resistors Ri and Rf or use a DC servo.

... and still then not forget to add 0R5 ballast resistors ...

Originally posted by pacificblue
Could it be that your BPA-300 started smoking for different reasons? And that your bigger heatsink is curing the symptoms instead of the disease? [/B]

The disease was in fact the unproper layout of the pcb and also in not using a servo, and the curing was that i had trimmers on each LM3886 for the offset adjustment, but oscillations where the real reason of: learning by burning ...

@jaste:
Have you measured how hot the heatsink runs ? Can you touch the heatsink by hand ?

artQuake

[AndrewT]

4780 is two channels,
two of these on a single heatsink will dissipate upto 25W from +-37Vdc when in quiescent condition.

Do you need 4channels of amplification?

Remove one chip and see if the other runs any cooler.

I usually recommend that one should double the cooling capacity that National show in their literature. They test the chip to it's extreme (Tj=150degC) for their heatsink recommendation, but don't tell you that this will trigger the protection systems MUCH earlier than running the chip cool.