LM3886 Heatsink size?

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

Thanks to all the contributions in this forum, I finaly have enough courage to start my first GC.

I've just built an LM3886 amp and it sounds amazing! Just one problem though.

The heatsinks heat up really fast and the amp cuts off. Roughly about playing 3-4 songs only.

The various parameters involved:

Using 24V 4 ampere transformer to power the LM3886, used with 4ohm speakers. Current heatsink size is 6" x 2.5"

I intend to increase heatsink size to 8" x 3.5" as this is the current maximun allowable without relocating the amp's layout within casing.

My question is, would 8' x 3.5" enough to keep the LM3886 cool?

If not, then can you advise on heatsink size you think suitable from experience?

Heatsink dimensions are in inches.

Thanks.
 
Hi,
6inch by 2.5inch of what?
Aluminium cooking foil or a 10mm thick chunk of aluminium with 50mm deep fins at 8mm pitch with the fins aligned vertically so that cooling air can pass over them?

What happens if you disconnect the load? Goes the heatsink cool down or remain the same?

Going to 8inch by 3.5inch is not a particularly big increase.

What voltage are you supplying to the chipamp? 24Vac could be about +-36Vdc, which is at top limit for 8ohm speakers.
Did you follow the instructions in the data sheet for both voltage and heatsink?
 
Sorry for not being detailed enough.

The heatsinks currently used are made of aluminium with 1" fins which I bought off the shelf from local electronic stores.

The proposed 8" x 3.5" heatsinks are made from same aluminiun material and also have 1" fins.

Going thru the GC Gallery, I notice most pics have rather small heat sinks.

The heatsinks cool down within about 20 minutes or so when I switch amp off. When I restart the amp, it works fine.
 
I'm using a heatsink that looks like a CPU heatsink, probably no bigger than 3" x 3". The heatsink is inside my chassis, so there is NO airflow around them. I have one per LM3886 (it is a stereo amp). I picked them up for $1 at a local surplus store.

I think the load is roughly 6 ohms, and while I don't remember exactly what the DC voltage was, it was an 18V - 0 - 18V transformer.

It never overheated, and it has been run for at least a few hours straight, so by that point, it should've reached it's peak temp.

You don't need much heatsinking, just make sure it is a good design. Anything that looks like it could be used for a CPU should work. Since your average processor will dissipate magnitudes more power than the LM3886, most of those heatsinks WITHOUT the fan should still work for you.
 
BrianDonegan said:
I think Andrew is trying to determine if your amp is oscillating, or maybe shorting, which woud account for your heat problems. If your amp is on, but not connected to a load, and it still gets hot, heatsink size is not your problem.

Yeah, I think you may have an oscillation because many people have used heat sinks that size or smaller (including me) without having heat issues.
 
Hi,
if it stays cool when no load is applied and quickly rises in temp to shutdown, then it seems like the speaker impedance is too low for the supply rail voltage.

Like I said earlier, 24Vac will give about +-36Vdc.
Have you measured that yet?
The manufacturer recommends maximum rail voltages of +-35Vdc for 8ohm and +-28Vdc for 4ohm. For these you must follow through the heatsink design process. It seems like you have omitted this design stage.
 
You need to take a few things into account.

Andrew has mentioned the first thing, which is that your rail voltages are too high for a 4 ohm load. You should reference the chip manufacturer spec sheet which will give you all the info you need regarding voltages, load, heatsink requirements.

Also, it will depend on your speakers, as a nominal 4 ohm load will actually dip lower than 4 ohms to varying degree depending on the speakers.

Another factor is how loud you're listening to them.

V
 
Hi guys,

I have solved the heat problem with my LM3886, by changing to larger heatsinks, measuring 8 x 3 1/2 inches.

The voltage rail is still high at 36.5V, but the amp is not shutting down now. The heatsinks do run hot but still touchable, without the "burn" sensation.

The LM3886 sound really wonderfull now, and after listening to them for 2 long nights, I can now see why my friend, almost gave up his high end audiophile for the LM3886.

Once again, thanks to all who assisted, especially Andrew T.
 
LM3886 hot even w/o load/speaker

Hello,

I've read through your comments. It looks like I have the problem that Andrew mentioned where if you don't have your speaker connected and the heatsink still gets hot, then your amp could be oscillating or spiking.

I'm sure it isn't spiking. I'm using DC voltage to bias the op-amp. +/-12 V powersupply & 1.3A. I have the amp set up as a voltage follower. The input is my ipod which as a max of 1.5 Vpeak (I measured with an oscope, but had no luck finding speaks on web for ipod output).

Since the amp is set up as a "voltage follower"/"current buffer", the output voltage is at most 1.5Vpeak at any one time. I do not have an external heat sink applied, but only after 1.5 or 2 minutes, the lm3886, starts to get too hot and you can hear the poor performance on the output.

My speaker is 8 ohms. It has been over 5 years since I took a detailed class on op-amp design. Although oscillation sounds familar, I don't remember exactly what to look for. Could this really be the problem though or is it something else. I have read the datasheet over and over, so I apologize if I missed something that is obvious to others.

~mel
 
minimum gain

Thanks for the comment Andrew..

The datasheet shows examples of various applications and how to calculate their gain, but it doesn't limit the gain ... as far as I saw (in the general spec data).

I rewired the circuit. It is now set up as an inverting op-amp with a gain of -6.5 (about) |A| = Rf/Rin = (93.2k/14.4k). Rin is in series with Vin and then to the inverting input of the op-amp. Rf is from the inverting input to the output of the op-amp.

Really Rf is a pot so the gain can go as low as 0. The noninverting side is grounded. Anyway, with the Gain at 6.5 and no speaker/load, it no longer gets extrememly hot. only a tiny bit warm.

If I do set the gain to be -1 (Rf=Rin= 14.4k), then w/o the speaker the op-amp does get really hot. Why would they design an op-amp to limit the minimal gain for your design? I'd like to be able to turn the volume all the way down to low some times. I could do this using the IPOD, but it would be nice to have one knob do everything.

If I have to, then I guess I can just use the mute button and limit my minimal gain and/or use ipod controls, but I"ll order some heat sinks and see if that helps as well. I'll read over the specs in more detail later, but on page 20. It is just showing basic equations for example circuit designs they have. I've set it up (with having a max gain of 6.5) to make sure the power supply can supply enough voltage and current at the output.

Also, I'm not wanting to make any power supplies right now. but if I were to take two old laptop power supplies (both are about 19V & 4 A) & hook them up as my dc bias power for the op-amps(instead of the two dc power suplpies I have now 12V & 1.3A), will the op-amp (741 & the LM3886 only draw enough current as what they need?).

Ohms law has to apply across the device right? So if it were 2 resistors in series across the supply, then it would be straight forward. I'd assume the voltage is set at 19V and the current is 19V/(R1+R2). But does it work a similar way for op-amp in general or do they always sink whatever current is not being used from the other components (that is still left in the power supply)?

Perhaps this is a simple question but we never went over power supplies much at Chico State in my undergrad anyway. Thanks bunches! :)
 
Member
Joined 2002
Paid Member
Hi melbsp,

Quote from datasheet:

"LAYOUT, GROUND LOOPS AND STABILITY
The LM3886 is designed to be stable when operated at a
closed-loop gain of 10 or greater, but as with any other
high-current amplifier, the LM3886 can be made to oscillate
under certain conditions. These usually involve printed circuit
board layout or output/input coupling."

regards
 
thanks Greg

Thanks Greg,

I was hoping that wasn't the case, but on the other hand, it will sound great when it is at the higher levels, so I'll just have to redesign and hear what it sounds like. Do you have any tips for power supplies that produce enough current and voltage to run LM3886 to get the most out of a 60W 8 ohm speaker.

V = sqrt(PR) = 22 volts
I = sqrt(P/R) 2.74A (actually about 6A needed from power supply for everythign)

I'm assuming if I need 2.74A to go through my resistor, then the power supply has to supply at least that much. You can't get something for nothing, right ... or is there some magic going on with the op-amp?

Also I'll be biasing 2 LM3886 & at least four 741 op-amps, so I'll need probably a total of about 6A. Assuming no "magic" happens.


Smiles ~ Melody
 
Gain of 18

Hi Greg (and Andrew),

so I rewired the circuit with a gain of 18. It now is running much longer with only a little heat sink on it. I guess I'll just use my ipod to control the volume, so that way I don't have to have a huge hink sink. I was trying to design it to use the knob on the device, but that is only because I'm a TA for a lab and that is what the student are required to do. I'll let them know that they can increase the volume using the knob of a LM3886 but only if the make the gain increase from 10.

Thanks for spotting that in the datasheet Greg. I wish they would make the important data like that in the table (quick reference), but teaches me that you really have to read all the information even if it is just on the applications.

Anyway, thanks bunches. It sounds loud enough for my "ballet studio" in my garage (currently only one LM3886 connected though). I'll try it with two later and if I need more power from power supplies, I'll message you all for advice.

Thanks again!

Melody
 
Re: thanks Greg

melbsp said:
Thanks Greg,

I was hoping that wasn't the case, but on the other hand, it will sound great when it is at the higher levels, so I'll just have to redesign and hear what it sounds like. Do you have any tips for power supplies that produce enough current and voltage to run LM3886 to get the most out of a 60W 8 ohm speaker.

V = sqrt(PR) = 22 volts
I = sqrt(P/R) 2.74A (actually about 6A needed from power supply for everythign)

I'm assuming if I need 2.74A to go through my resistor, then the power supply has to supply at least that much. You can't get something for nothing, right ... or is there some magic going on with the op-amp?

Also I'll be biasing 2 LM3886 & at least four 741 op-amps, so I'll need probably a total of about 6A. Assuming no "magic" happens.


Smiles ~ Melody
Hi,
transformer + rectifier + smoothing = PSU for a chipamp.
For reliable maximum into an 8ohm speaker use a 25Vac to 28Vac transformer. For two channels try to find a 4 secondary toroid. One pair of secondaries for each amplifier.
100VA to 150VA is adequate for one channel.

P=V^2 / R = I^2 * R.
which is the required answer? I. Then use I= SQRT[P/R]. But this is Iac.
Ipk= SQRT[2*P/R].

Minimum resistance of an 8ohm speaker is about 6r0 but can be less. When feeding a crossover to multiple speakers the minimum impedance can be even lower. If it is a 4 to 8ohm speaker then Re and minimum impedance can be as low as 3r0. These low impedance bass/mid drivers have become very popular.

Now calculate Ipk for 6r0.
Ipk=sqrt[2*60/6]=4.47Apk and for a 4 to 8ohm speaker about double this i.e. 9A.
Oh!!
That's more than the guaranteed maximum for most of the chipamps on the market.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.