Heatsinks on rectifiers?

[richie00boy]

I have built a prototype PSU for a (sub) amp using one bridge rectifier for each secondary. I have used discrete 20 amp TO220 pack (dual) diodes, but not used any heatsink. The transformer is 500VA and the amp is 400W class-d.

At normal use they are totally fine, but tonight I want to take the sub to a party and it will be cranked up more than normal. Has anybody any experience of running these kinds of diodes without heatsinks?

Please, no comments like 'it would be a good idea' as I know that. I'm after REAL experiences. I'm really pushed for time so it would be nice if I had a good probability of getting away with not using heatsinks for tonight.

[AndrewT]

Hi,
no experience here, but fit at least 20C/W flags to each.

It will increase power dissipation by a factor of about 3,6.

If you have room for 15C/W then fit them.

[jackinnj]

Originally posted by richie00boy
I have built a prototype PSU for a (sub) amp using one bridge rectifier for each secondary. I have used discrete 20 amp TO220 pack (dual) diodes, but not used any heatsink. The transformer is 500VA and the amp is 400W class-d.

At normal use they are totally fine, but tonight I want to take the sub to a party and it will be cranked up more than normal. Has anybody any experience of running these kinds of diodes without heatsinks?

Please, no comments like 'it would be a good idea' as I know that. I'm after REAL experiences. I'm really pushed for time so it would be nice if I had a good probability of getting away with not using heatsinks for tonight.

MBR20100 Schotkeys perhaps ? -- go to page 3 of the PDF where you will see the necessary derating between heatsinked and ambient.

[richie00boy]

Thanks Jack. Yes they are MBR20100 Schottky - excellent guess! I did look at the derating curve... It doesn't look that promising for a 4 ohm load. But I was hopeful of some real world experience saying otherwise.

One thing from the datasheet that puzzled me - how come the square wave current is less than the DC current rating?

[I_Forgot]

I am going to make some assumptions here, but you can adjust the numbers for any differences between my assumptions and your actual circuit...

At 400W rms out, and 90% efficiency from the amp, you're looking at 444 W that need to be supplied by the rectifiers (400W/0.90). That corresponds to about 2.6A rms through each diode (I=sqrt(444W/4Ohms)/4rectifiers). Let's call it 3A just be provide some margin. Approximately 1V will be dropped across each diode, meaning each diode will dissipate (or maybe I should say have to dissipate) about 3W.

There are 2 diodes in each of your packages, so each package will have to dissipate 6W. I'd guess they're going to get pretty warm without any heatsink. If you allow a 30 C temp rise above ambient (taking them up to 50C case temp- pretty hot) you need a heatsink that is a most 5C/watt (on each package) (30C/6W) or 2.5C/W for the two packages (30C/12W). That's a pretty small, inexpensive, heatsink.

I would say to be safe there's no reason not to use some sort of small heatsink. If you can just screw the tabs on the packages to the amplifier box you can probably dissipate that heat. Think about what will happen to your amp modules if the rectifiers burn up. A small heatsink is cheap insurance...

I_F

[richie00boy]

Thanks for your response but your calculations are flawed I think.

That corresponds to about 2.6A rms through each diode (I=sqrt(444W/4Ohms)/4rectifiers)

I disagree with that as the diodes are in series so current passes through all of them not divided by 4.

One saving grace though is that because I have used a dual bridge approach the load is shared between the bridges, with one doing the upper half of the sine and the other doing the lower, do duty cycle is halved.

I have seen a reputable 300W MOSFET amp module with built in PSU and they only use axial type diodes so I'm hoping that my beefy TO220 ones will be sufficient.

[peranders]

One single TO220 can take 1-1.5 watts => the bridge can deliver in average 3-4 A

[AndrewT]

Hi,
surely the Shottky half bride will drop about 1v when passing current. A bit less through the two diodes on low current and a bit more on higher current.
So it follows that on full duty it can pass about 1A to 1.5A.
If run as part of a dual bridge it will be on a duty cycle of slightly over 50% so the maximum current without a heatsink is less than 2A to 3A.
Adding a heatsink will allow this half bridge to pass a lot more than this.
Guessing that the Vrails are about +-70V then 140V * 2A = 280W and @ 3A = 420W. The 500VA and 400W ClassD run at full power will just exceed these powers indicating that a heatsink is nescessary for PA/party work. Into clipping as seems the norm at discos then significant overheating can be expected.
Fit some flags (15 C/w to 20C/W) and hope. Or big sinks (4C/W to 8C/W) to be sure.
As I said earlier no experience of these but a little logic tells you that a heatsinkable package is designed to be used with a heatsink.

[richie00boy]

Thanks for everybodys responses. I get what you are saying AndrewT but I needed a different approach

As it happened I thought 'sod it' and used the sub at the party. I'm sure it wasn't anywhere near full chat but it was kicking out some reasonable SPL, and after 5 solid hours it was still playing nice as ever And DJ moi went down a storm again

I will be fitting 18C/W heatsinks to the final version, but it just goes to show...

[AndrewT]

Hi,
glad to hear it worked.

Any chance you could do that again and tell us the Vdrop across the half bridges? And how hot they get? if at all!