1. Too bad you do not have enough space, sometimes tubes make things work well, and with simple ccircuits:
2. A 5AR4 rectifier takes some time to warm up, because it is indirectly heated.
But you can easily delay the warmup even more.
Take a separate 6.3V transformer winding (dedicate it to the 5AR4).
But the 5AR4 needs 5V at 1.9 Amps (the filament is 2.6 Ohms when warm).
6.3V - 5V = 1.3V
1.3V/1.9A = 0.68 Ohms
Consider the 5AR4 filament, it is less than 1 Ohm when cold.
It actually draws 5A or more at startup.
Now, connect 6.3V, a series 0.65 Ohm resistor, and the 5AR4 filament in series.
At startup, the 6.3 Voltage will be divided by the series 0.65 Ohm, and the cold 5AR4 filament.
This is an automatic soft start of the 5AR4 filament, and even more of a soft start of the cathode.
3. You seem to have room for a heatsink and solid state device, or a relay and circuitry to run it . . .
so don't you have room for a rectifier tube?
2. A 5AR4 rectifier takes some time to warm up, because it is indirectly heated.
But you can easily delay the warmup even more.
Take a separate 6.3V transformer winding (dedicate it to the 5AR4).
But the 5AR4 needs 5V at 1.9 Amps (the filament is 2.6 Ohms when warm).
6.3V - 5V = 1.3V
1.3V/1.9A = 0.68 Ohms
Consider the 5AR4 filament, it is less than 1 Ohm when cold.
It actually draws 5A or more at startup.
Now, connect 6.3V, a series 0.65 Ohm resistor, and the 5AR4 filament in series.
At startup, the 6.3 Voltage will be divided by the series 0.65 Ohm, and the cold 5AR4 filament.
This is an automatic soft start of the 5AR4 filament, and even more of a soft start of the cathode.
3. You seem to have room for a heatsink and solid state device, or a relay and circuitry to run it . . .
so don't you have room for a rectifier tube?
Last edited:
Success with capacitor multiplier delay circuit
Thankfully I decided to investigate further why reality wasn’t matching the simulation.
Turned out I’d been doing all my measurements using a blown MOSFET. I blew a few more before I realized that I should put the delay circuit after the choke not before it.
The first image shows where I now have the delay circuit. There are a lot of filter capacitors in the power supply left over from when I was asked to convert the amp from dual triode 6550 SE to dual 6S4S SE.
The second shows the final schematic for the delay. The first timer switches 400V to the capacitor multiplier at 12s after switch-on.
The second timer switches the HT directly through to the output and isolates the capacitor multiplier. One of the requirements was no MOSFETs in the HT supply.
The 1uF capacitor in the multiplier is an MKP low leakage type.
R11 and C9 ensure that there is 8ms or so overlap between the direct HT and the capacitor multiplier HT to ensure that there are no switchover spikes in the HT.
R8 discharges the capacitor multiplier capacitor through the 6.2V Zener.
The first oscilloscope screenshot shows the original transient caused by a relay switching on the HT after the heaters had been on for 30s. This is what caused the alarming cone excursion.
The remaining oscilloscope screenshots show the HT (blue) and the speaker output.
After 12s the HT rises over about 20s. There’s still a small transient in the output when the HT starts ramping up but it’s only about 1/20th of the original 4Vpp transient and only just audible. I could reduce it further by increasing the size of the resistor that charges the cap multiplier capacitor and extending the delay before the full HT is switched through.
At t=44s when the HT is switched through directly there’s a 100mV negative spike in the output caused by the ~4V jump in the HT. This causes an audible but not loud click.
The second last screenshot shows the output transient at 500ms/division. It’s about 0.5Hz. I’m having some PCBs made up; last image.
The resistors are all off-centre because I moved the footprint origin to the centre and didn’t change the 3D models. There’s room for a heat sink on the regulator and MOSFET but they aren’t really required.
There’s just enough room to mount the assembled PCB on the inside side of the amp chassis.
Thankfully I decided to investigate further why reality wasn’t matching the simulation.
Turned out I’d been doing all my measurements using a blown MOSFET. I blew a few more before I realized that I should put the delay circuit after the choke not before it.
The first image shows where I now have the delay circuit. There are a lot of filter capacitors in the power supply left over from when I was asked to convert the amp from dual triode 6550 SE to dual 6S4S SE.
The second shows the final schematic for the delay. The first timer switches 400V to the capacitor multiplier at 12s after switch-on.
The second timer switches the HT directly through to the output and isolates the capacitor multiplier. One of the requirements was no MOSFETs in the HT supply.
The 1uF capacitor in the multiplier is an MKP low leakage type.
R11 and C9 ensure that there is 8ms or so overlap between the direct HT and the capacitor multiplier HT to ensure that there are no switchover spikes in the HT.
R8 discharges the capacitor multiplier capacitor through the 6.2V Zener.
The first oscilloscope screenshot shows the original transient caused by a relay switching on the HT after the heaters had been on for 30s. This is what caused the alarming cone excursion.
The remaining oscilloscope screenshots show the HT (blue) and the speaker output.
After 12s the HT rises over about 20s. There’s still a small transient in the output when the HT starts ramping up but it’s only about 1/20th of the original 4Vpp transient and only just audible. I could reduce it further by increasing the size of the resistor that charges the cap multiplier capacitor and extending the delay before the full HT is switched through.
At t=44s when the HT is switched through directly there’s a 100mV negative spike in the output caused by the ~4V jump in the HT. This causes an audible but not loud click.
The second last screenshot shows the output transient at 500ms/division. It’s about 0.5Hz. I’m having some PCBs made up; last image.
The resistors are all off-centre because I moved the footprint origin to the centre and didn’t change the 3D models. There’s room for a heat sink on the regulator and MOSFET but they aren’t really required.
There’s just enough room to mount the assembled PCB on the inside side of the amp chassis.
