I find a simple 2 relay approach with resistors is best because a NTC needs time to cool off.
Suppose someone toggles the on off switch or the power falls off and comes back on again with a big toroid having only fraction of an ohm dc-resistance?
I used the "relay" approach for up to 250kVA transformers that got switched on and off every 30 seconds year in and year out.
No other delay than the switching of 3 "relays" (a few tens of cycles perhaps) was enough to tame the inrush current of those transformers.
A bank of caps with almost 1F cannot be any good when fed from such a big toroid.
Charging time will be almost nothing and current spikes go through the roof, making all sorts of troubles.
Who wants the noise induced from huge current spikes?
Suppose someone toggles the on off switch or the power falls off and comes back on again with a big toroid having only fraction of an ohm dc-resistance?
I used the "relay" approach for up to 250kVA transformers that got switched on and off every 30 seconds year in and year out.
No other delay than the switching of 3 "relays" (a few tens of cycles perhaps) was enough to tame the inrush current of those transformers.
A bank of caps with almost 1F cannot be any good when fed from such a big toroid.
Charging time will be almost nothing and current spikes go through the roof, making all sorts of troubles.
Who wants the noise induced from huge current spikes?
Last edited:
The NTC variable resistors are only in the circuit for the first few seconds of startup, ten seconds at the most, if you have the right size and ohms then they don't get hot enough to change resistance much. You use NTCs so if the bypass relay fails and the NTCs stay in the circuit then they can get hot and decrease in resistance so the transformer will get full current. NTCs handle the momentary power off and on quite well.
A 250kVA transformer? What are you powering, a tank? The largest Toroidal type transformer I have seen in a power amp was a 4kVA unit that was 8 inches in diameter and weighed 30 pounds.
A bank of caps in the range of 1F in a CLC arrangement is good. You can use a regulator circuit on the front end to eliminate ripple but for the output stage of a big power amp a large bank of capacitance to draw on reduces the supply ripple to very low percentage.
Limiting the current in to the transformer, I use 12 amps max as the current limit, with the NTCs limits the current out so the surge/spike is minimized, it is still a big pull but doesn't cause any trouble or noise.
A 250kVA transformer? What are you powering, a tank? The largest Toroidal type transformer I have seen in a power amp was a 4kVA unit that was 8 inches in diameter and weighed 30 pounds.
A bank of caps in the range of 1F in a CLC arrangement is good. You can use a regulator circuit on the front end to eliminate ripple but for the output stage of a big power amp a large bank of capacitance to draw on reduces the supply ripple to very low percentage.
Limiting the current in to the transformer, I use 12 amps max as the current limit, with the NTCs limits the current out so the surge/spike is minimized, it is still a big pull but doesn't cause any trouble or noise.
12A peak will work with almost no load attached, but with full load you should see something much higher, otherwise something does not add up.
At full load into 8ohm 155W out I have over 25A charging peaks, with 2 ohm 600W out I have about 60A peaks into my 50mF caps,
charging and de-charging is about symmetrical, toroid is a decent 2kVA 3% reg
At full load into 8ohm 155W out I have over 25A charging peaks, with 2 ohm 600W out I have about 60A peaks into my 50mF caps,
charging and de-charging is about symmetrical, toroid is a decent 2kVA 3% reg
Last edited:
Sounds like a 'direct-on-line' pump connection - they usually glitch everyone elses mains voltage on the same LV distribution, and a good reason to use a motor drive.
But at least a motor, once going, doesn't peak clip the mains voltage waveform like a simple rectified power supply with humongous first filter cap.
But at least a motor, once going, doesn't peak clip the mains voltage waveform like a simple rectified power supply with humongous first filter cap.
Exactly when large toroids with tight regulation and almost 1Farad of capacitors we need to deal with massive inrush.
Personally I favor the method used by bryston on most of their amps.
- Sensing zero crossing through an optocoupler.
- Power on the toroids by Triacs
All this managed by a PIC.
Using a DC filter and ground lifter as well (both).
The key is properly doing the programming in the PIC, to trigger the Triac at the optimal time, so almost no inrush happens.
No electromechanical relays, no resistors or whatever to dissipate a bunch of power at first before bypassing..
Very sensible way to do this.
- Sensing zero crossing through an optocoupler.
- Power on the toroids by Triacs
All this managed by a PIC.
Using a DC filter and ground lifter as well (both).
The key is properly doing the programming in the PIC, to trigger the Triac at the optimal time, so almost no inrush happens.
No electromechanical relays, no resistors or whatever to dissipate a bunch of power at first before bypassing..
Very sensible way to do this.
It would be nice if some cracker jack PIC programmers were pitching in to help with that stuff.
I played a tiny bit with an arduino before, and it's probably the easiest way to develop something, platform agnostic, quick and simple, with software available freely, and just connect the arduino directly to a computer via USB.
Now the PIC isn't quite as practical, but once programmed, there is no need for any extra stuff around the PIC besides what it really needs to fulfill its functions.
So PIC looks better for a final implementation, although a bit less easy to program than arduino.
I have yet to do any PIC programming, so that requires some learning curve. Plus I wonder what programming languages could be used on PICs, beside C and all the complexities involved. There must be some BASICs for PIC available, but how to find and use this?
I played a tiny bit with an arduino before, and it's probably the easiest way to develop something, platform agnostic, quick and simple, with software available freely, and just connect the arduino directly to a computer via USB.
Now the PIC isn't quite as practical, but once programmed, there is no need for any extra stuff around the PIC besides what it really needs to fulfill its functions.
So PIC looks better for a final implementation, although a bit less easy to program than arduino.
I have yet to do any PIC programming, so that requires some learning curve. Plus I wonder what programming languages could be used on PICs, beside C and all the complexities involved. There must be some BASICs for PIC available, but how to find and use this?
With the zero crossing detection done with an optocoupler, and that signal made available to a PIC, the program can simply apply some calibrated delay to turn on the Triac.
The key being finding the best delay for turn on from zero crossing, for the lowest impact from the inrush current.
Still, adding a series resistor between the bridges and cap banks can also reduce the demand, and those can be bypassed by relays after things settle enough..
The key being finding the best delay for turn on from zero crossing, for the lowest impact from the inrush current.
Still, adding a series resistor between the bridges and cap banks can also reduce the demand, and those can be bypassed by relays after things settle enough..
Hi,
Maybe you missed my post 26. You can built one using the schematic and the Arduino sketch attached to the post. Also instead using the Arduino UNO you can use the attiny85 that it is 8 pin micro processor and can be program using the Arduino IDE. It is a simple circuit that any one can built it. I have been using it to start/stop my amplifier for more the 5 years without any failure. It will ramp up / ramp down the AC at the start and at the stop. No inrush current.
Maybe you missed my post 26. You can built one using the schematic and the Arduino sketch attached to the post. Also instead using the Arduino UNO you can use the attiny85 that it is 8 pin micro processor and can be program using the Arduino IDE. It is a simple circuit that any one can built it. I have been using it to start/stop my amplifier for more the 5 years without any failure. It will ramp up / ramp down the AC at the start and at the stop. No inrush current.
I saw that with the arduino, and this can be the kind of thing to use as a test bed.
But for a final implementation, not very practical.
I have a few of those attiny85, and they're definitely not for a final use as is, but could be good for development.
The best thing about the PIC is they're just a regular chip that can simply be put in a socket, and the only stuff needed around it once programmed is what it needs to function and nothing else.
What I need to develop is something scalable and fully integrated. I'm designing a 4 way multi-amp system and everything will be in a single "rack" with all the housekeeping centralized for all amps and everything else, with proper sequential power up of all toroids, muting and all..
So in the end it needs to be programmed into a single chip to put in a socket, with enough I/Os for all the stuff it needs to supervise and control. There are PICs with many I/Os, but development can be done with smaller types to test each function, then later all be fully integrated into a larger system.
But for a final implementation, not very practical.
I have a few of those attiny85, and they're definitely not for a final use as is, but could be good for development.
The best thing about the PIC is they're just a regular chip that can simply be put in a socket, and the only stuff needed around it once programmed is what it needs to function and nothing else.
What I need to develop is something scalable and fully integrated. I'm designing a 4 way multi-amp system and everything will be in a single "rack" with all the housekeeping centralized for all amps and everything else, with proper sequential power up of all toroids, muting and all..
So in the end it needs to be programmed into a single chip to put in a socket, with enough I/Os for all the stuff it needs to supervise and control. There are PICs with many I/Os, but development can be done with smaller types to test each function, then later all be fully integrated into a larger system.
Im using a 2kva toroidal transformer with about 60-0-60 secondary VAC resulting in about 83V per rail and using about 0.1F / 100V capacitor esr 8mohm large one jumbo cap per rail and I have used a large NTC from ametherm which is the size of a biscuit and thick leads but when turned on initially without any triac in here the inrush is still there lights dim in house for a moment. The Resistance value of NTC is about 0.5ohm I guess since I had large capacitors in series. The total aggregated Joules for the capacitors to charge was quiet high so had to choose the NTC that way. This NTC becomes Hot as well of course and its bypassed after 3 secs. Im not happy to use it this way though.
If the first post circuit is implemented then will that circuit triggers at zero crossing?
As Harry said that he is happily powering up 4kva with almost a Farad Capacitors with ease.
If that can help the problem that would be great.
If the first post circuit is implemented then will that circuit triggers at zero crossing?
As Harry said that he is happily powering up 4kva with almost a Farad Capacitors with ease.
If that can help the problem that would be great.
You're not looking at the whole picture.
Doing soft start isn't the main and only duty for the PIC, but just one among many.
The PIC would be handling just about everything, all the housekeeping.
Muting input/outputs, sequencing power ups, sensing for speaker line shorts before unmuting amps, getting the info from over current and DC sensors to handle the situation and reacting appropriately, by muting, disengaging power rails, then sensing if the fault is still there to prevent going back to "normal" operation, and if the fault has gone, then go back to normal and unmute, with proper timing.
The PIC would need quite a few I/Os for all this, so of course it wouldn't work with the smallest model.
Switching on exactly at zero crossing isn't optimal.
The load is complex and not just capacitive or inductive. It's a difficult combo of all of this.
The transformers themselves have their inrush current, they need to magnetize their core and ramp up, but they also have rectifiers and cap banks on their secondaries, with totally different behaviors there.
So the detection of the zero crossing isn't just to turn on right on that spot, but to use that zero crossing reference point and do the turn on after a proper delay for best compromise for lower impact.
And using Triacs, the transformer isn't just turned on and left that way until it settles, it's turned on for a short time, then off, and then on again on the next cycle, etc, until the transformer core is fully magnetized and the cap banks are charged. Only then the Triacs can leave the transformer fully turned on. Until power off must be done...
The load is complex and not just capacitive or inductive. It's a difficult combo of all of this.
The transformers themselves have their inrush current, they need to magnetize their core and ramp up, but they also have rectifiers and cap banks on their secondaries, with totally different behaviors there.
So the detection of the zero crossing isn't just to turn on right on that spot, but to use that zero crossing reference point and do the turn on after a proper delay for best compromise for lower impact.
And using Triacs, the transformer isn't just turned on and left that way until it settles, it's turned on for a short time, then off, and then on again on the next cycle, etc, until the transformer core is fully magnetized and the cap banks are charged. Only then the Triacs can leave the transformer fully turned on. Until power off must be done...
i did mine the same way.
8 pin pic and a opto plus triac.
ramp up phase slowly and bobs your uncle.
power supply to pic was resistor dropper off the mains