Class AB amp clipping at 55v under no load, help

The transformer is to small for 2 ohm load.
1200w/(60+60v) = 10A You should not load it more than absolute max 7A DC.
7A DC into an output stage that is conducting halv time will be 14A average or 20A top to the speaker.
20A *4ohm = 80v. That means at 4 ohms you load the transformer max if you play at the amplifiers limit.
For 2 ohms it would be vise to have 1 transformer more. And with respect to the mains fuses it can be good to first switch on one transformer and then the other. And maybe 2x40 000uF.
I recommend to have separate rectifiers so it is no risk for out of phase connection.
 
45V across 4 ohm is over 10A of current. It doesn’t take much resistance to bleed off a few volts. 1 ohm of parasitic resistance somewhere, and you’ve bled off 10V already. Your PSU has internal resistance, so do the wires, the output transistors aren’t exactly 0ohm across even at full saturation and you lose a little bit in the degeneration resistors. Designing for those kinds of currents can be challenging!
 
Food for thought:

You have gotten lots of good help. Please keep working on your project.

The SOA @ 160 volts, 25 degrees, 100ms single pulse is only 100ma per device (800ma to load).

This maybe OK for a real restive load, however the reactive speaker typically are 45-60 degrees. Reactive loads store energy and the SOA will be exceeded, maybe smoke & flames will follow.

I’m not trying to scare you, but there are several things that are unknow or not yet reveled. Fan, flyback diodes, input clamping, square wave overshoot, bode plot, thermal tracking, current limiting, ripple current ratings, etc.
 
Fully reactive load is 90 degrees out of phase, not 180. Peak current would occur at zero crossing, or about 80V on the transistor.

A fully reactive load makes no sound. Real speakers get to about 60 degrees, but the impedance minimum does not occur there. It’s practical to consider 30 degrees AT the minimum. The minimum physically occurs at ZERO degrees, but it’s still only a few percent above the minimum at 30 degrees for a real woofer driver that hits 60 degrees. Many do, but more phase is exceptionally rare. Such a driver has very high (double digit) Qms meaning very low mechanical losses. Then go and put any damping material in the box, and it spreads out that impedance peak and reduces maximum phase shift. So considering really crazy phase shift doesn’t make any sense for the real world. 30 degrees at Zmin, and for practical purposes the amp is safe.
 
SOA at full voltage is relevant. If there was no need of current there it would not be any need of flyback diodes.

I calculate about 10% of max current at max voltage to be safe with margin. The time for that current is also very short. I count with max 10 ms. At lowest voltage over the transistor 100ms. The period time at bass resonance is about 20 to 40 ms so i have a good margin there. At the resonance frequency the woofer has phase shift of nearly +- 90 degrees. But the impedance is very high there 25 to 50 ohms so the current is far from its max.
 
When the fly back diodes are conducting the transistor isn’t.

At resonance, the impedance is pure real. It’s on either side where you see phase shift.

Actually, in this case I’d be more concerned about the drivers. I’ve seen amps with 8-10 2SC5200’s on +/-80V hold up ok. SOA is on the light side, but it’s better than what they had in a Phase Linear by a huge margin. I’d be using the OnSemi parts - MJL21194 or MLJ3281. In either case, you could be drawing 200 mA of driver current at 40V into a resistive load. That’s far outside the SOA of a single C4793. Their power handling falls steeply above 20V, as do ALL 100-200 MHz epitaxial planar types. They can and will blow out, taking the output bank with them. For voltages above +/-55V, use two pairs, or switch to MJE15032. High end amps at these voltages use two driver pairs, or use “output” devices as drivers in EF3.
 
I see you are marking words here. I should have written "near the resonance frequency".
A problem? we often have is that we are not playing sinus waves.
If we take an assymetric pulse with lest say 30v for 10 ms from the amp to the 1mH speaker and than go max - what do you think will happen? If you consider how switching power supplys work you will find out how much energy is loaded in that flyback pulse.
Even if that dont happen often (someone pushing in the input contact when the amp is on) The owner of the amp can get very unhappy. Extension cord with RCA plugs.
Or a usual tone from a guitar. They often are very unsymmetric.
 
What is going on here ?
In his first post (as long as it will not be edited) OP clearly stated that
his amp clips at 55 volts peak, as seen on his scope.
How is it possible that we have about 40 posts to clean this mess up ?
Waste of time.
 
I have seen it too many times. A long time ago electricians learned how to nail up telephone cables in school.
They used that knowledge to mount home cinema systems later on. The one i was asked to help had a speaker cable short circuit after about 18 months.
About 20 years ago.
I have been working with dance bands and with a new place each night they just tape the speaker cords on sensitive places.
Usually it was the bass player that was responsible to fix defective speaker and microphone cords.
I thought every musician was able to identify the sound if the speaker cable was shorted.
 
I would be less concerned by a short than by the thermal dissipation, this amp loaded with 2R has about 500W dissipation
per channel when the amp is supplying 500W RMS/channel, that s a 1000W total TDP, it need a robust air flow to keep it within
safe operating margins.

On the security front there s not even a basic current limiter, let alone a SOA circuit, it would cost two transistors and a few
components per channel to implement one, not counting that as stated by wg_ski doubling the drivers is more cautious.