MJ15003 & MJ15004 > The next step up ? bipolar T-03 ?

Hi there,
Going back in time, I remember the MJ15003 & MJ15004 being the easy go-to output pairs for 250W amps.
They were rugged with a strong Safe Operating Area.
What I would like to know is, are there any newer Higher Spec. T-03 bipolar pair devices available?
Perhaps higher Vce / Ptot or bigger S.O.A. ? (consumer available)
MJ21193/4/5/6 are the “newest” and highest rated TO-3’s. They are also available in several plastic packages. They are still old too. Above 80V they have a bit of advantage over the 15024/5. Below that, the 15024/5 is still more rugged.

Even their most recent plastic-only types are over 10 years old now.
MJ21195 and MJ21196

Has good gain
and depending on what your doing for power.
They Hold up well with Ft since Cop is around 500p compared to around 1000p for most.

From 1.5 to 2 amps Ft is actually pretty well matched around 5 MHz
very good pair for T03.
Depending on power of the amp design. Where I like to keep
them as far as how many output devices to use.

Ironically some TO3 have PNP which are faster.
Was thinking of doing a strange PNP Quasi with MJ15002
Since Ft seems to hold at 6 to 7 MHz with up to 2 or 3 amps current
Thank you very much - that's great info!
I'll wait and see if any posts come in re. 2SA, B, C etc.
No chance. The only original manufacturer that hasn’t completely dropped the TO-3 is OnSemi. No more Toshiba/Sanken/Hitachi. They all dropped it years ago.

If you wave a fistful of dollars at them you might get On to do a package run in TO-3 with the *J*3281 dies. It has been done, but it will COST. Entire lot quantities, too. Last manufacturer to do that was Krell, and that was over 20 years ago. They’ll want even more money to do house numbered parts now.

Second source makers like Central and Mospec do a respectable job with even older obsolete 2N’s, but they are no harder to make than a TIP41 (just bigger). I have no idea how their 15003’s stack up against original On’s - I’ve never had to try. There is no real NEED for the older types for new design or routine audio service work, just use MJ15024 for all the old types and be done with it. I’d be leery of ISC’s 2SD424 without some data - we know their MT200 clones don’t cut the mustard.
Thank you ...
I wondering about gain. The HF. performance sounds good as well.
I am thinking of DC rails of +/- 80 V.
Using .33 ohm emitter resistors, how many output devices (per rail) would you recommend
to drive 2 ohm load?
Six or seven. One could do it with five (ie PV CS800, and it would do 2 ohms with 15024’s). But the big boys all used six. My “CS800” clone used seven - the heat sinks had holes for banks of 8, so I filled it.
If you have a TO-3 based amp, the upgrade path is to apply modern circuit features to get the most out of those old parts. You may want to make a new PCB that fits in the existing space. Or replace the entire amplifier modules and reuse the powder supply, chassis and maybe heatsinks. Depending on your needs, class-D amps will give twice as much audio power from the same power supply. While older transistors were slow and week SOA, the biggest problem with older amps is usually the IPS and VAS.
What is the amplifier topology / schematic ?

Depends how much power is actually needed.
Usually something straight forward is with large but easy to obtain power transformer.

So about 800 VA or 800 watts
is a good range to think, far as already wound off the shelf.
So voltage rail could be good enough for 400 / 450 watts

depending if you using forced air cooling or not.
usually over 250 watts forced air is used.

your hoping for 1.5 to 2 amps per transistor.
So rough guess is 4 to 6 devices per rail

6 is overkill maybe....4 devices per rail or 8 devices total
sounds like typical heatsinks that are easy to find.

forced air be fine. I like to think of a amp as " 4 ohm"
but has enough to take 2 ohm punches in the face.
and VI limiters would be set to not trigger above 1.5 ohms.

obviously soft start and DC protect would be mandatory.
fans are on at 1/2 speed for noise and then a thermal trigger is
set for full speed.
I think that is exaggerated. I used Motorola custom parts a lot in the past. I doubt the batch quantity was anything like that for a TO-3. You don't get many big die per wafer
They probably wanted to discourage the practice, and to ensure that they only had to do this ONCE instead of repeatedly. And a custom package deal is more of a headache than custom screening and binning of otherwise standard products (Which is all Peavey and Crown ever did). And that was the PAST - in todays world nobody even wants to do that anymore. Mention “binning” at a design review and the test team will come after you with torches, pitchforks, and a NOOSE. Doesn’t matter how much money the customer waves at you - if it slows up the process and messes up the metrics….
Looking at transformers and filter caps - if I proceed with this "upgrade" for 2 ohm loads,
I have decided to go with only +/- 74 volt rails. I don't want to push the SOA into 2 ohms.
Figure on two and a half amps per transistor of reactive current. You could push three in 10 ms bursts even at temp, but the short circuit condition could be longer. Time-dependent SOA protection is a bit more of a PIA but could be worth it here.

At the woofer’s lowest impedance, it’s pure real. At maximum current draw it IS a resistive load. Don’t go thinking you need to drive 1.5 ohms at 90 or even 60 degrees because it just won’t happen. At least at low frequency. ESLs are an exception, but if you put out full power at 20 kHZ I pity your hearing. The SOA issue is mitigated by the amplitude and duration.