Design problem

wer

to

The 2N3055 has a lot more silicon area the the 2N3904. Are you sure that your 6k8 at R1 isn't being dragged down the collector leakage current? The first data sheet I could download for the 2N3055 had Iceo at up to 5mA at 150C, which is enough to drag Vc down to 15V.

The Motorola 2N3055 was notorious for being a smaller (and faster) device than RCA's origininal 2N3055, and migh have less leakage current.

-- Bill Sloman, Nijmegen

"Bill Sloman"

The Motorola 2N3055 was notorious for being a smaller (and faster) device than RCA's origininal 2N3055,

But Motorola are very naughty and like to substitute chips in legacy devices at whim - for example, the chips used in Motorola or " ON " 2N3773s and MJ802s have long ( since the mid 1980s) been identical to the ones in their MJ15003s.

God knows what chip is actually inside a current stock " ON " 2N3055G.

Whatever is most convenient and cheapest for them to use, no doubt.

... Phil

wer

to

That's a really odd looking circuit, what's it supposed to do?

This circuit assumes the Vbe's of the transistors will cancel. That may not work very well in real life. Self-heating of Q2 looks like a thermal runaway hazard to me. Cascoding would help that, but the Vbe match is still a hazard.

Can Spice account for self-heating of transistors, and the subsequent change in Vbe? It would have to know the heatsinking situation if it did.

What's the intent/application? Do you need DC coupling?

John

Are you looking for som,ething that works when a semiconductor's junction temperature hits 204 degrees C? Or in an ambient temperature of

Furthermore: Once or twice back around 1980 or in the early 1980's I saw figures for life expectancy or MTBF for such devices as a function of temperature. I remember the numbers not looking good (MTBF under around

Can you add more Q2's (with emitter R's) in parallel?

Don't know if it'll work at 204C, but at least it will spread the dissipation across multiple transistors.

Ed

Well, i used 2N3055 because it was the only power NPN that Spice had..there are "equivalents" like the MJD50, the TIP50, the STLD128DN etc.

It is for what the oil patch calls a Casing Collar Locater (CCL). Made with 2 very strong Alnico magnets opposing and a coil in between. When assembly passes a pipe collar, the magnetic field gets "bent" some and nominally generates a 1mV pulse; (say) first positive then negative; roughly 1mSec at nominal tool speed. The idea here is to get more gain than that obtained by the SIE circuit, as well as have a more uniform gain over temperature.

Please see my response to Fred Bloggs on application. Yes, DC coupling is a virtual necessity - but the sensing coil could be AC coupled to the amplifier; presumedly there is a CCL amplifier (Titan Industries,i believe) using a FET and the coil had to be capacitively coupled so as to not ruin the FET biasing. Now i started with Q2 being a 2N3055 equivalent (TIP-50) and R4 about

Plastic / epoxy package; SMT and 204C ambient (note: all of the "wrong" answers). If one does a design on that basis, it usually works well to 185C. I have a few isolated designs that are reliable all the way up, but the dizzy-pay-shun is under 1mW.

That is a good thought, but (say) if 10 were used, that would mean 10 times the base current, making R2 crudely speaking 10 times lower and killing the overall gain by that factor. Cascode Q2 or zener-in-series would seem a better choice. See attached.

wer

to

Leakage across reverse biased Si junctions doubles for every 10oC rise in temperature- so that is about a factor of 10^6 increased leakage over room temperature spec, or 1mA at 200oC for every 1nA of room temp ICBO, also the HFE should be about double the room temp gain.

That doesn't sound like it needs DC coupling.

I've experimented with mosfets over temperature. They work up to maybe

I'd guess that some commercial opamps might work pretty hot. Maybe ask various applications engineers at the semi companies. I have a connection to someone who designs high-temp downhole electronics... I'll ask her.

Uniform gain implies lots of open-loop gain and feedback.

John

llar

(Q3)

lower

ng to

That doesn't absolve you from thinking about the way leakage currents increase exponentially with temperature. Your reply to John Larkin suggests that the reverse leakage current in your zener diode might have been the real problem - get the zener hot enough and you don't need to activate any kind of breakdown mechanism to sink the current from the resistor in series, so the voltage across the zener diode can sink to close to zero.

-- Bill Sloman, Nijmegen

collar

nt (Q3)

ly lower

ming to

tc.

There is such a thing as high temp linear

llar

(Q3)

lower

ng to

een.

ed text -

I'd worry about the zeners at that temperature. Mind you I say "worry" because I do not know. But I suspect they will bring you grief.

Ed

Along these lines, Motorola did a die shrink on us in the 1990's on some power FETs... We found out the hard way they weren't as good as they had been in the past in our inverters. They quickly lost a pretty good customer at that time.

I am curious is the ON Semi MJ15024 TO-3 power NPNs are as good as they used to be. Those have been the replacement NPN of choice for most audio power amplifiers I have worked on.

boB

That last change I am aware of was in the mid 1990s after plastic pack versions of the same device were released ( ie MJL21194). Because lead frames are used, plastic power devices cannot normally use the exact same chip design as a TO3 version. The TO3 chips had used two termination areas for the emitter structures as opposed to one for plastic versions.

AFAIK, all the changes are in fact performance upgrades of the earlier versions - particularly in terms of SOA.

The only major hassle is for repairers of older amplifiers - as the various versions have somewhat different Vbe characteristics and may not share current well installed in parallel groups.

..... Phil