I have made this mistake myself. Thought 37 volts would be fine for a regulator rated for 40 volts. It was not.
Tube maximum voltage rating are pretty fuzzy limits that can normally be exceeded from time to time without issue.
Solid state data sheet maximum voltage limits are generally very HARD limits. Exceed them by even by a bit and reliability often goes right out the window.
From my experience rectifiers on the power line should be de-rated to about 1/2 the data sheet PIV required if you want long term reliability.
PIV required is generally accepted for a full wave center tapped rectifier as 1.414 * the total secondary voltage value.
In your case with 360V RMS on one side of the CT ..... 360V * 2 * 1.414 = 1,018V
So a 2KV diode is about right.
Why so much de-rating?
First there is that hard limit for the data sheet PIV value.
Semi manufactures uses the margin in their manufacturing process for maximum yield (profit).
So on a good day a 1KV diode batch (1N4007) may come out of that process with a PIV of 1.2KV or more. They all ship.
On a less good day the diodes may come out with a PIV between 980V and 1,001V. All the diodes that test at 1KV or more ship. Everyone.
So you never know what the actual hard limit of the PIV of any one diode may be unless you test it yourself.
The manufacture's process margin is for the manufacture to use, not the designer.
Add into the fun PIV is temperature dependent so measuring it yourself has issues.
Next the power line voltage varies, often by +/- 10% or more short term. By short term I mean minutes to hours.
The power line is never very clean. There are lots of voltage spikes from industrial equipment on the power lines and the power company switching in and out equipment.
Power spike clamps and line filters do work however they have to let a certain amount of over voltage through as voltage transients on the power line are common and trying to clamp everything exactly to normal voltages can not realistically be done. The diodes need margin for this.
The power switch on lots of equipment has no subber so due to contact bounce there can be lots of arcing on turn on or off creating voltage spikes. Still more margin needed.
Diode damage for over PIV events is often cumulative. Each over PIV event causes a high energy event inside the diode when the diode breaks down in reverse blocking and so it dissipates a large power transient. The diode's die is hurt but maybe not fatally the first time. Hurt it enough times and it fails later apparently for "no reason".
So I use only 1/2 of the diodes margin in 50/60HZ line power supplies.
Solid state data sheet maximum voltage limits are generally very HARD limits. Exceed them by even by a bit and reliability often goes right out the window.
From my experience rectifiers on the power line should be de-rated to about 1/2 the data sheet PIV required if you want long term reliability.
PIV required is generally accepted for a full wave center tapped rectifier as 1.414 * the total secondary voltage value.
In your case with 360V RMS on one side of the CT ..... 360V * 2 * 1.414 = 1,018V
So a 2KV diode is about right.
Why so much de-rating?
First there is that hard limit for the data sheet PIV value.
Semi manufactures uses the margin in their manufacturing process for maximum yield (profit).
So on a good day a 1KV diode batch (1N4007) may come out of that process with a PIV of 1.2KV or more. They all ship.
On a less good day the diodes may come out with a PIV between 980V and 1,001V. All the diodes that test at 1KV or more ship. Everyone.
So you never know what the actual hard limit of the PIV of any one diode may be unless you test it yourself.
The manufacture's process margin is for the manufacture to use, not the designer.
Add into the fun PIV is temperature dependent so measuring it yourself has issues.
Next the power line voltage varies, often by +/- 10% or more short term. By short term I mean minutes to hours.
The power line is never very clean. There are lots of voltage spikes from industrial equipment on the power lines and the power company switching in and out equipment.
Power spike clamps and line filters do work however they have to let a certain amount of over voltage through as voltage transients on the power line are common and trying to clamp everything exactly to normal voltages can not realistically be done. The diodes need margin for this.
The power switch on lots of equipment has no subber so due to contact bounce there can be lots of arcing on turn on or off creating voltage spikes. Still more margin needed.
Diode damage for over PIV events is often cumulative. Each over PIV event causes a high energy event inside the diode when the diode breaks down in reverse blocking and so it dissipates a large power transient. The diode's die is hurt but maybe not fatally the first time. Hurt it enough times and it fails later apparently for "no reason".
So I use only 1/2 of the diodes margin in 50/60HZ line power supplies.