Pad or Paste?

[sloper]

Goodaye all

 

Whats the thinking on using pads or paste on output transistors?

Just finished the driver board on a Sansui AU2900 had a couple of pads missing as well as a lot of parts.

 

In the past l always used paste for CPU's

 

regards Bruce

[bob_m_54]

Pads or paste and insulating wafers? If you use paste, you will still need to insulate them from the heatsink, via the mica washers. But these days silicone heat transfer sheet/pads are used by most manufacturers, so I would think they would be OK. Probably should get them from a reputable source though, just in case.

[mwhouston]

For me sili pads every time. Or sticky sili pads. 

[andyr]

2 hours ago, bob_m_54 said:

But these days silicone heat transfer sheet/pads are used by most manufacturers, so I would think they would be OK. Probably should get them from a reputable source though, just in case.

 

10 minutes ago, mwhouston said:

For me sili pads every time. Or sticky sili pads. 

 

I've used silicone pads (instead of the old-style mica washers) ... but I prefer ceramic pads ... 2mm thick.

 

Andy

 

[Batty]

I have a selection.

[mwhouston]

2 hours ago, Batty said:

I have a selection.

They look like sticky pads. 

[Batty]

Not sticky.

[mwhouston]

The sticky ones are good but you have to be very careful when finaly aligning your power FETs up with the heatsink. I put the stickies on the FTEs with the mounting bolts in. Then as is, keeping the now sticky FETs well back, part screw the bolts in getting closer. With the final align as i push the pads onto the heat sink and systematically tighten the bolts down. Last time i did it this way it worked well, best i have ever done on a Holton build, eight FETs in all. 

[aussievintage]

 

I guess I am old fashioned.   I prefer paste - with mica insulators if needed.

 

The one time I used a silicon pad under a power chip I had troubles with heat.   I see them as good for experiments where you might want to change the devices or heatsinks.

[mwhouston]

3 minutes ago, aussievintage said:

 

I guess I am old fashioned.   I prefer paste - with mica insulators if needed.

 

The one time I used a silicon pad under a power chip I had troubles with heat.   I see them as good for experiments where you might want to change the devices or heatsinks.

I used mica a lot but now just the pads. Hard to know which has lower thermal resistance. Its probably stated somewhere. 

[Avoxon]

On the Holton Inviso class A amp, 1.5 mm Al2O3 ceramic pads are used with 2 types of thermal paste (metallic paste between the pads and the heatsink). Note that the heatsink has been remilled after elox for a better thermal dissipation. The amp (50 W, 8 Ohm) never exceeds 60°C at 25°C room temperature). Better results with 1.5 mm pads as with 2 mm pads.

Edited April 15, 2021 by Avoxon

[mwhouston]

4 hours ago, Avoxon said:

On the Holton Inviso class A amp, 1.5 mm Al2O3 ceramic pads are used with 2 types of thermal paste (metallic paste between the pads and the heatsink). Note that the heatsink has been remilled after elox for a better thermal dissipation. The amp (50 W, 8 Ohm) never exceeds 60°C at 25°C room temperature). Better results with 1.5 mm pads as with 2 mm pads.

I realise Class AB amps run cooler but my eight power FETpc 300W Holton runs at about 50C at the HS with sticky silicone pads.

[Avoxon]

2 hours ago, mwhouston said:

I realise Class AB amps run cooler but my eight power FETpc 300W Holton runs at about 50C at the HS with sticky silicone pads.

Do you have both modules on the same heatsink, as in the picture? it would be very hard to do much better, whichever pad you would use, as the limiting step in that case would be the dissipation area of the heatsink. I use the same modules in another amp and use (per channel) a 900 cm2 heatsink with a Rth of 0.1 1 °C/W. The bias is set quite high, as the Exicons love bias. Your heatsink should be somewhere around 0.23-0.25, for two modules. To which value have you set your bias ? On another hand, running is at 50°C is no problem at all...

[Avoxon]

2 hours ago, mwhouston said:

I realise Class AB amps run cooler but my eight power FETpc 300W Holton runs at about 50C at the HS with sticky silicone pads.

One 500 module inside

[DavidH71]

Silicon pads were created for convenience so to speed up assembly time but mica washer and paste still have the best thermal transfer properties , I personally try to always use the mica washers and paste.

[Steffen]

Yes, or stick to output stages where all collectors are strapped together, so you just need a very thin sliver of paste

 

[mwhouston]

8 hours ago, Avoxon said:

Do you have both modules on the same heatsink, as in the picture? it would be very hard to do much better, whichever pad you would use, as the limiting step in that case would be the dissipation area of the heatsink. I use the same modules in another amp and use (per channel) a 900 cm2 heatsink with a Rth of 0.1 1 °C/W. The bias is set quite high, as the Exicons love bias. Your heatsink should be somewhere around 0.23-0.25, for two modules. To which value have you set your bias ? On another hand, running is at 50°C is no problem at all...

In this build the HS was inside a 3U enclosure between ventilation slots. I do run two silent 12V fans (@9V) at the back across the PCBs. The fans are shock mounted and can only be heard if you are standing over the amp with music off. Bias was low here in an effort to control heat. 

[Monkeyboi]

Some might say I'm "old school" but IME from both decades of DIY and repairs I choose mica insulators and a good quality thermal compound over silicone washers (sil pads)  and "stickies" any day.  Sure, sil pads and stickies are inexpensive, less messy and convenient as well as having their niche applications, BUT IMHO a correctly implemented combination of mica and thermal compound has a lower thermal resistance than the more modern replacements.

 

Some may argue the point, but when correctly installed the mica / thermal compound combination works extremely well. Unfortunately many novices have never been shown how to apply it correctly.  With thermal compound the secret is less is better (well to a point).  Many inexperienced constructors or repairers slap on way too much thermal compound and to add insult to injury fail to correctly tighten down the device(s) to the heatsink.

 

In the attached picture the output device was mounted on the heatsink using mica washer and thermal compound.  The thermistor is electrically isolated in the clip so only thermal compound was necessary. As you can see, only a small amount of compound is required.  Just enough so that just oozes out evenly  at the edges of the device when correctly secured.  Note in the far right corner of the picture the discarded light blue sil pad.

 

The proof is in the heat transfer.  A good thermal camera will quickly tell you if the mounting method is adequate when the device is operating.  There should not be a significant temperature difference between the surface of the device and the heatsink.  If the device is significantly hotter than the heatsink it indicates a greater thermal resistance in the mounting method.

 

 

Just be aware that some thermal compounds might be electrically conductive and therefore are not suitable to be used if the device you're attaching to a heatsink that is electrically connected to ground or exposed to be touched has to be electrically isolated.  Know your device characteristics. 

 

Cheers,

Alan R.

 

 

  

Edited April 16, 2021 by Monkeyboi

[Steffen]

Agreed, mica and a quality non-conductive paste (Gelid or Arctic Ceramique) is what I used in the past.

 

While we’re on this topic, do people usually use an adjustable torque screwdriver for mounting transistors to heatsinks? The transistor spec sheets usually say something around 0.8Nm, but I’ve never owed a torque screwdriver and was therefore always left guessing...

[Monkeyboi]

6 minutes ago, Steffen said:

Agreed, mica and a quality non-conductive paste (Gelid or Arctic Ceramique) is what I used in the past.

 

While we’re on this topic, do people usually use an adjustable torque screwdriver for mounting transistors to heatsinks? The transistor spec sheets usually say something around 0.8Nm, but I’ve never owed a torque screwdriver and was therefore always left guessing...

 

You've raised IMO a very valid point.

 

Even though I've used torque measuring / fastening tools in the past I rarely reach for them now for attaching semiconductors to heatsinks.  Instead I do it by feel, using multiple smaller tension advancements on each fixing point to get as even a pressure distribution across the device's fixing points as possible.

 

Even though following the manufacturer's recommendations is advisable, it can lead to disaster if the degree of torque ends up stripping the tapped thread in an aluminium heatsink or case.    Mostly this problem arises if the thread depth percentage is low.  IME it doesn't take too much over zealous tightening to ruin your day.   

 

Generally my approach is to torque bolts to a firm and "not likely to come loose" status initially.  Run the device to close to maximum thermal dissipation if the design allows that, then recheck the tension is still good.  Other techniques can be to used like shake proof washers or applying a small dab of nail polish lacquer to minimise the possibility of loosening.  I never use "Loctite" or similar substances as these sometimes lead to servicing headaches later.

 

Cheers,

Alan R.

[Batty]

Loctite softens with heat too.