TO264 and TO3P packages are becoming orphaned as mainly for BJTs. Audio power is actually a small field compared to general power control, which is now the domain of Mosfets and little real development now goes into linear audio power. Even up-market audio is largely Mosfet, and a steady creep to class D is in progress as it improves.
What this means is that smaller package parts can be used because there is no second breakdown to be cautious of and if more power is needed, jam in more devices. For BJTs, we still need larger packages for the lower temperature safety margins but with commercial demand falling, who's making new parts and support products like thermal pads for low demand BJTs and LMosfets? I have a hard time getting even cut sheet silpads and mica is as scarce as hens teeth for the large pads!
I suspect we will be without TO264 case parts in the not too distant future. Smile and don't fret over capacitance straw men - enjoy Exicons, Alfets, On-semi and Toshiba's biggest as we have for decades, whilst you still can. 🙂
What this means is that smaller package parts can be used because there is no second breakdown to be cautious of and if more power is needed, jam in more devices. For BJTs, we still need larger packages for the lower temperature safety margins but with commercial demand falling, who's making new parts and support products like thermal pads for low demand BJTs and LMosfets? I have a hard time getting even cut sheet silpads and mica is as scarce as hens teeth for the large pads!
I suspect we will be without TO264 case parts in the not too distant future. Smile and don't fret over capacitance straw men - enjoy Exicons, Alfets, On-semi and Toshiba's biggest as we have for decades, whilst you still can. 🙂
I have just searched Farnell, Digikey and Mouser and I can't even find anything for TO264 or similar sizes, not to say ceramic insulators. 
I'm not surprised. I've never seen them in in the major distributors catalogues in the past either. You may see them advertised as TO3P(l) as opposed to TO3P👎, or even confused as simply T03P, normally recognized as smaller. Google those and see what you get for package descriptions.
On-semi have a penchant for re-categorizing their packages every so often in quickly forgotten types that mean little to us. Who remembers T03 as TO205-3?
They call their TO264 package TO3PBL, but who uses that?
In a standardised industry, manufacturers still resist by distinguishing their products with their own unique codes, stylistic touches and house standards - in other words, no standards. 😀
On-semi have a penchant for re-categorizing their packages every so often in quickly forgotten types that mean little to us. Who remembers T03 as TO205-3?
They call their TO264 package TO3PBL, but who uses that?
In a standardised industry, manufacturers still resist by distinguishing their products with their own unique codes, stylistic touches and house standards - in other words, no standards. 😀
Hi Can these be used: TO-3PL / TO-264 Silicone Insulating Sheets 29mm x 22mm | eBay
I Will use Them for mjl4302 if okay with you guys🙂 ?
I Will use Them for mjl4302 if okay with you guys🙂 ?
It is now easier to isolate the heat sink from the chassis than it is to insulate the power devices from the heat sink. Heat sink at +140V! If you stick your fingers through the fan you'll get TWO surprises! Have to separate the + and -, but what the heck, I've been separately compensating the thermals on each output and driver for the last 10 years anyway.
War chest full of MJL2119x, and original Toshiba types, as well as older TO-3's, and plenty of MOSFETs as well. Enough to say supplied for 50 years (no joke).
what about these? (my project is 2pair of MJL4302, PSU +-50V)
50 Pieces Silicone Insulation Pad Sheet 26mmx20mm for Package TO-3P Transistor | eBay
50 Pieces Silicone Insulation Pad Sheet 26mmx20mm for Package TO-3P Transistor | eBay
had that same thought more than a few times!
buy these stickers instead of kapton sticky tape
The thing I don't like about the sil pads is a limited shelf life > a vendor gave a me sample roll of the stuff (1st generation grey Berquist ) and a few years later it was ruined, stuck to itself and turned into delaminated mess.
Attachments
Last edited:
Indeed for EF OPS it is simpler to mount Bjt directly on heat sink , at the expense of 2 heatsinks, and insulation of heatsinks, but saves about
0.8 K°/watt thermal resistance per Bjt.
With lateral mosfets where there are no source resistors all the output transistors can be mounted on one heat sink without insulators.
Saves you the trouble of locating the !@$#$%^&* insulators. I have several thousand TO-3 micas, and the last time I used TO-247's I just cut the darned TO-3's up to use them. Yeah, the commutators still had to be insulated from the TO-264 bank.
Are lateral mosfets backside source? All the IR and IXYS stuff I used in the past were backside drain.
edit: Interesting, looked up the part number, backside source...neat.
Going to have to look up the die topology, very interesting..
jn
Last edited:
As the other Nigel says that is attractive when FET's. Exicon sell match gate voltage sets to get the best possible matchiing.
On a related issue. I wish we could buy all power devices in T03. IRF530/9530 for example. It is drain to mounting tab as with bipolar collector. If it could be source to case then so much the better.
Don't forget black heat-sinks help. If anodising is of good quality no washer required ( Arcam A60 ). FET's would be best aluminium to T03 case ( + heat grease ) as the anodising is a small heat barrier. One suspects etch primmer might work. My brother found even white paint works. He suspected it wasn't white at infra-red. He discovered that a fault on TV's could be cured by some matt black paint on a mini heat-sink. It was just enough to cure a problem that happened over time. He made an interesting point, a device at the limit gradually weakens.
On a related issue. I wish we could buy all power devices in T03. IRF530/9530 for example. It is drain to mounting tab as with bipolar collector. If it could be source to case then so much the better.
Don't forget black heat-sinks help. If anodising is of good quality no washer required ( Arcam A60 ). FET's would be best aluminium to T03 case ( + heat grease ) as the anodising is a small heat barrier. One suspects etch primmer might work. My brother found even white paint works. He suspected it wasn't white at infra-red. He discovered that a fault on TV's could be cured by some matt black paint on a mini heat-sink. It was just enough to cure a problem that happened over time. He made an interesting point, a device at the limit gradually weakens.
If the outputs are common emitter/drain, then the same trick works there as well. And this time you can install emitter/source resistors for multiple pairs.
I use the alf16x16w parts and they are middle pin is source GSD as opposed to vertical mosfet GDS.
As said before the source resistor can be left out if pre matched devices are bought. Exicon sold is 25 off devices. That would be 50 total. As they test them after prduction they put them in groups. Even if you don't use matched sets they tend to self compensate. I dare say if so power output is less than optimum. Better than using one set even so. If looking at the thermal simplicity and the ability to go direct to heat sink thius is no small thing. The problem with source resitors is they would need to be large in value to force good behaviour. The other FET's I mention IRF530/9530 need source resistors and a Vbe bias adjuster to thermally trachk the heatsink temperature , some designs take a risk on that. Exicon 10N/P20 and BUZ 900/905 ( and 2SK/SJ devices ) are source to case and only need a variable resistor as bias.
One thing if I were a rich man I would like to do is build more ideal devices using existing device silicon. BCV 61 and 62 as DIL 4 device, better still 2N5401 or 2N5551 as ready made current mirrors. Everything possible in TO3. Also a longtail pair with 2 x 100 R already fitted. This would be in DIL 8 and have access to the resistors. NPN and PNP versions. A ready made feedback pair in the style of a Darlington could be interesting. It would be sold under the understanding it needs it's stability to be checked. The idea of T03 out of T0220 came from the fakes. I suspect I have some. They work better than T0220. These are 2N3055. My FET's would be TO3530/9530
My post was ambiguous.
I'll try again:
Using vertical mosFETs and BJT where the middle pin/backplate is the collector/Drain, then in a Common emiiter/Drain output stage, the collectors/Drains are at the same output voltage and can be mounted directly on a common heatsink.
The emitters/drains connect to the supply rails either directly or via source/emitter resistors.
The common emitter/drain topology for BJTs and V mosFETs allows one to use resistors to aid stability and to use direct mounting to a common heatsink.
I'll try again:
Using vertical mosFETs and BJT where the middle pin/backplate is the collector/Drain, then in a Common emiiter/Drain output stage, the collectors/Drains are at the same output voltage and can be mounted directly on a common heatsink.
The emitters/drains connect to the supply rails either directly or via source/emitter resistors.
The common emitter/drain topology for BJTs and V mosFETs allows one to use resistors to aid stability and to use direct mounting to a common heatsink.
I think without being rude we all knew that. It is the ability of the special FET's that is so attractive. Case to metal and no source resistors. The Ron of most devices is a predicatable 1 to 1.5 ohms depending on exact use. This would need a very large RS to be of any advantage. Where you might fall down is when 1 amp required 1 V on one device and 1.5 V on another. Here one device might end up doing 70% of the work. What happens is the Ron shots up as it heats. thus the badly matched higher voltage device takes over until it heats up. This is not ideal but better than one device. If bought as matched sets thhen the ideal is more closely found.
As stated previously IRF530/9530 is Drain to centre pin. This suggests a T03 version would be Drain to case. Still usable in a complinerty pair if so. Same goes for 2n30555/2955 ( with BD 140/139? ). Actually a BD139 IRF9530 and BD140 IRF950 look to be an interesting pair. The FET could be direct to metal. I would make the resistor that allows opperation ( 100 R is bipolar in Self ) variable to get the best distortion.
As stated previously IRF530/9530 is Drain to centre pin. This suggests a T03 version would be Drain to case. Still usable in a complinerty pair if so. Same goes for 2n30555/2955 ( with BD 140/139? ). Actually a BD139 IRF9530 and BD140 IRF950 look to be an interesting pair. The FET could be direct to metal. I would make the resistor that allows opperation ( 100 R is bipolar in Self ) variable to get the best distortion.
The Thread title is about thermal insulators and the topic moved to the advantage of directly mounting high power devices to a common heatsink for better thermal conduction, by avoiding the insulator.
I have about 20 mica TO3 insulators, mostly from industrial surplus equipment from
the 1960s - 80s. They're mostly 2.5 thou, a few 3, 3.5 and 2.2s none are 1 thou, any
data on how much worse 2.5 thou is over 1?
I've read that most these days are all over the place on thickness, seems they don't make
them like they used to.
I'm very tempted to run the heat sinks hot whenever I can - simple solution.