Thermal Track Transistors

[Ipanema]

Hi,

Has anyone tried the thermal track transistors from ONSEMI NJL1302D and NJL3281D? How well does it work compared to typical diode/transistor thermal tracking?

Onsemi even publish a schematic based on the new transistors.

http://www.onsemi.com/site/products/summary/0,4450,NJL1302D,00.html?tax=796

 

[Arius]

Short answer - NO, I haven't had the chance to try them out yet.

I did open a thread in the Solid-State forum about the possibility of collaborating on a Leach clone variant using these devices. No takers so I am proceeding ahead with a couple of friends. I just started the layout. Sims and schematics done. Will let you know how it turns out especially the Iq tracking.

 

[Jan Dupont]

These are supposed to be MJL1302/3281's with build in diode.
I have compared the thermal track NJL's with the MJL's and can't hear or meassure any difference........

 

[Arius]

According to OnSemi, using the ThermalTrak will allow you to up the bias a bit more (and yet remain thermally stable). This supposedly yields better performance when driving low impedance loads. Might be just marketing but they've an app note with before/after diagrams.

Bleh - just noticed Digikey upped the prices on the ThermalTraks by at least $1.00. At today's prices - they cost more than Lateral MOSFETs.

I'll be trying them out with 6 internal diodes in series to replace the VBE Multiplier and as part of the VBE multiplier (on the B-C leg). Open to suggestions as to other ways to use them.

 

[sam9]

they cost more than Lateral MOSFETs

Usually they come down in a few months.

Biggest question (to me) with the Thermal Traks is do you get suboptimal bias because you are forced to bias in discrete steps of ~0.6V-to-0.7V or whatever the drop is. Can you "shim" the bias with a seties pot and is it worth the trouble.

Second question on using these is: suppose you have parrallel so many pairs that you have to leave some of the diode disconnected -- what is the best strategy for insuring the that the devices with the unused diodes get tracked OK? Presumably they just "go along for the ride".

 

[Ipanema]

The main advantage is supposely the inbuilt diode will track faster to reach thermal equilibrium in no time. As for lazy ppl like me, it take away the hassle to mount external thermal track diode/transistor especially TO92 package on the heasink/output transistor.

Since a reference design is also given, why not try it out. BTW, any comment on the reference design?

Regards.

 

[forr]

These transitors are a very old demand from many enginneers involved in the design of classe B amps. At last manufcturers provide them. This is the second time in a few weeks I heard about devices incorporating thermal compensation. The fisrt one mentionned Sanken.
Well, why all at the same time after having so long waited ?

~~~~~ Forr

§§§

 

[sam9]

You can find the Sanken datasheets at the bottom of this page:

http://www.sanken-ele.co.jp/en/

They are darlingtons, appearently with a 5 diode string parallel to the base and with a built in 0.22 ohm emitter resistor. No hint of application info.

 

[Arius]

Since a reference design is also given, why not try it out. BTW, any comment on the reference design?

The OnSemi design is a lousy design. It is meant to showcase the transistors (note the words reasonable performance). Anyways, there's plenty of good BJT designs around that we can adapt to use the ThermalTraks.

Biggest question (to me) with the Thermal Traks is do you get suboptimal bias because you are forced to bias in discrete steps of ~0.6V-to-0.7V or whatever the drop is. Can you "shim" the bias with a series pot and is it worth the trouble.

Yeah - that's my biggest question too. In the end, I'm thinking of going with a standard VBE multiplier and have the diode string between base and collector (ala Leach). This way, I can still use a preset to set Iq and have the diode string do its job.

what is the best strategy for insuring the that the devices with the unused diodes get tracked OK? Presumably they just "go along for the ride".

That's another of my questions too. I'll be using 4 pairs. Not too sure of myself, I am thinking of using 3 diode pairs and maybe have one diode as a temp sensor for overtemp and the last diode unconnected. I wonder about designs that have say, 10 pairs.

 

[sam9]

Nice to know I'm only one with such quibbles. It ocurrs to me it may be these sort of quibbles that discouraged anyone from making devices like this.

If someone can take a look at those Sankens and figure out how they can be used, that would be interesting as well. The datasheets left me scratching my head.