This subject caught me off guard because I have never used toroids, personally. But a little research clears the fog:
https://www.electrical4u.com/magnetizing-inrush-current-in-power-transformer/
The problem is that zero crossing is exactly the wrong time to power up a transformer. Normally, after a transformer is running properly, the current at the voltage zero crossing is the opposite peak current, from which the current will build in the other direction until the zero crossing in the opposite direction. If you start at zero crossing, the current builds from zero for the half cycle instead of from the opposite peak, so it potentially saturates the core. This is only a problem because transformers are wound for a voltage that pushes their limits, but manufacturers are not about to spec their product below the absolute maximum. If you used a "240V" transformer at 120V, the problem would not exist, and I'm sure some vendors do sell a more conservative product.
So theoretically, for the transformer's sake, if you switched at the voltage peak when the current is normally zero, there would be no danger of core saturation, but no power supply is just a transformer, and switching at the voltage peak is not what a typical capacitor input DC power supply likes. However, such supplies normally only draw current on the sine wave peaks anyway, so maybe it's not a problem.
So using a relay bypassed NTC is the standard solution, but some sort of phase sync'd switch would be better. Before I retired, I refurbished an adjustable phase synced relay that was used for inrush testing, so I know it can be done, and the NTC would be unnecessary. Relay closure time is not something we usually worry about, but it is critical in this application.
https://www.electrical4u.com/magnetizing-inrush-current-in-power-transformer/
The problem is that zero crossing is exactly the wrong time to power up a transformer. Normally, after a transformer is running properly, the current at the voltage zero crossing is the opposite peak current, from which the current will build in the other direction until the zero crossing in the opposite direction. If you start at zero crossing, the current builds from zero for the half cycle instead of from the opposite peak, so it potentially saturates the core. This is only a problem because transformers are wound for a voltage that pushes their limits, but manufacturers are not about to spec their product below the absolute maximum. If you used a "240V" transformer at 120V, the problem would not exist, and I'm sure some vendors do sell a more conservative product.
So theoretically, for the transformer's sake, if you switched at the voltage peak when the current is normally zero, there would be no danger of core saturation, but no power supply is just a transformer, and switching at the voltage peak is not what a typical capacitor input DC power supply likes. However, such supplies normally only draw current on the sine wave peaks anyway, so maybe it's not a problem.
So using a relay bypassed NTC is the standard solution, but some sort of phase sync'd switch would be better. Before I retired, I refurbished an adjustable phase synced relay that was used for inrush testing, so I know it can be done, and the NTC would be unnecessary. Relay closure time is not something we usually worry about, but it is critical in this application.
The problem is not the bandwidth. The problem is that toroidal transformers are often wound with too few turns; too little inductance to avoid saturation in a half cycle at the rated voltage. Normally, each half cycle is preceded by a half cycle of the opposite polarity, which precharges the current in the opposite direction. But if the power is applied at voltage zero crossing, there is no current precharge. See the graphs in the link above.
Chris see if you can find a 500VA say 100Vct laminated core transformer for sale these days?
I use a very simple method for a soft start borrowed from pioneer, a 3.3ohm/25w resistor and a thermal fuse underneath in series and a relay across them to short out. The relay coil is connected to the unregulated supply with a v drop resistor
I use a very simple method for a soft start borrowed from pioneer, a 3.3ohm/25w resistor and a thermal fuse underneath in series and a relay across them to short out. The relay coil is connected to the unregulated supply with a v drop resistor
I like to have an on/off switch mounted on the front panel, and I happen to prefer those round, illuminated switches called "anti vandal" (examples below). Unfortunately most of them are not rated for mains voltage and definitely not rated for mains inrush current. So I put together a board called H9KPXG which includes both soft start, and also on/off control using a low voltage, low current switch. Schematics and PCB Gerbers are here on the Forum, just search for the board name.
Hi Rick,
Easy, you order one ... Any custom voltage and current will be an orderable item, and if you find something stock (catalogue), so much the better. But you have a good point. Most people opt for the easier, cheaper way out. But, if you take the time and trouble to build an amp, how much extra effort is getting the exact power transformer that is the best for your application? Extra windings for the voltage amp section are now possible - cheaply. That and you can now increase the raw voltage to regulate those for the voltage amp stage.
My point is simply that an E-I core has many advantages and we aren't trying to market a low height case. If you can choose between the two, get one that actually works for you instead of having to install other measures to look after excesssive inrush current. Now, a large, high power amplifier needs inrush current limiting regardless of the transformer type. I would still go with an E-I.
My tube amp design is made from all custom transformers from Hammond. The output transformers make all the diffrence in the world, and may as well have the exact supply voltages I wanted. I will and do order exactly what I need if there isn't a catalogue item available. Mind you, I tend to design to available transformers and will use two to make up special windings if needed.
For solid state amps, I have often done as you suggested. Easy, cheap and automatically pulls in when the caps have charged enough. Still, having a timed pull-in reduces the current during closing so maybe a long term advantage there.
Easy, you order one ... Any custom voltage and current will be an orderable item, and if you find something stock (catalogue), so much the better. But you have a good point. Most people opt for the easier, cheaper way out. But, if you take the time and trouble to build an amp, how much extra effort is getting the exact power transformer that is the best for your application? Extra windings for the voltage amp section are now possible - cheaply. That and you can now increase the raw voltage to regulate those for the voltage amp stage.
My point is simply that an E-I core has many advantages and we aren't trying to market a low height case. If you can choose between the two, get one that actually works for you instead of having to install other measures to look after excesssive inrush current. Now, a large, high power amplifier needs inrush current limiting regardless of the transformer type. I would still go with an E-I.
My tube amp design is made from all custom transformers from Hammond. The output transformers make all the diffrence in the world, and may as well have the exact supply voltages I wanted. I will and do order exactly what I need if there isn't a catalogue item available. Mind you, I tend to design to available transformers and will use two to make up special windings if needed.
For solid state amps, I have often done as you suggested. Easy, cheap and automatically pulls in when the caps have charged enough. Still, having a timed pull-in reduces the current during closing so maybe a long term advantage there.
Hi Mark,
Yes, and then you can also use remote control to power stuff up and down.
I've used those vadal-proof switches and prefer momentary (for the above reasons), I also use normal every day power switches. Whatever suits the application. Especially if I am retrofitting a soft start, I'll use the existing switch.
Everything lasts longer when you control inrush current, and also the turn-off spikes that can happen. I use an MOV for that, right across the primary while also affording lightning protection to a small degree. I use 150 vrms MOVs these days, but found they greatly extended the life of Marantz power switches. Just the MOV, no soft start.
Can you share the H9KPXG? Or send me a link via PM?
-Chris
Yes, and then you can also use remote control to power stuff up and down.
I've used those vadal-proof switches and prefer momentary (for the above reasons), I also use normal every day power switches. Whatever suits the application. Especially if I am retrofitting a soft start, I'll use the existing switch.
Everything lasts longer when you control inrush current, and also the turn-off spikes that can happen. I use an MOV for that, right across the primary while also affording lightning protection to a small degree. I use 150 vrms MOVs these days, but found they greatly extended the life of Marantz power switches. Just the MOV, no soft start.
Can you share the H9KPXG? Or send me a link via PM?
-Chris
Couple of years back I've made a simple circuit board as soft starter that also shorts out a big resistor. Found this idea somewhere on the internet. It sits in a big isolation transformer thingy that I use to make my lab a little safer. Probably not even needed (its 500VA I think and that doesn't trip the breakers anyway). But still nice 
Its not the prettiest but a very simple solution. It works with a capacitive dropper taken directly from mains and builds up a small charge across a couple of big caps. That prevents the relay from being activated straight away. When voltage reaches about 10VDC, the relay latches and switches on. I guess it has a delay of about 1 second or so.
I'm not too worried about arching of the contacts (because of the parallel big resistor) but I did use a DPDT relay and combined the contacts as a precaution.
Its not the prettiest but a very simple solution. It works with a capacitive dropper taken directly from mains and builds up a small charge across a couple of big caps. That prevents the relay from being activated straight away. When voltage reaches about 10VDC, the relay latches and switches on. I guess it has a delay of about 1 second or so.
I'm not too worried about arching of the contacts (because of the parallel big resistor) but I did use a DPDT relay and combined the contacts as a precaution.
- Relay is a 24V omron G5RL-1-E relay.
- electrolytics are both 470 uF
- MF cap is 0.33 uF, 275V. Be sure to use X2 and have a fuse in line (always good practice ti fuse these sort of things anyway)
- I used an oversized power resistor of 47 ohm (it can handle some power in case the relay fails).
- Don't mind the rest of the diagram, that's simply how i did the wiring inside this thing.
- There's some space to add a resistor in parallel over the coil of the relay so you have some control over the voltage across it. I forgot to put that in the circuit diagram.
Last edited:
That part.
Thermistors need to get hot to have low resistance. In a class AB amp that doesn't get the thermistor hot enough, you'll have this series resistance that varies with how loud you crank the amp. Like a very long time-constant dynamic range expander, as the amp will be able to deliver slightly higher power as the more you crank it.
In my experience, NTC thermistors are less reliable than wirewound power resistors used in soft-starts.
Many soft start PCBs short out the inrush current limiter when the inrush event is complete. Series resistance drops to zero ohms whether the amp is drawing big currents from its power supply, or small currents. Here's how Texas Instruments describes it in their application note AN-1849 (image below). Both Douglas Self's power amp textbook and Bob Cordell's power amp textbook, recommend this short-it-out procedure. The H9KPXG design includes the idea, giving it the name "bypassing" the inrush current limiting resistor.
_
_
Attachments
E
I am very happy with the triad standard Toroid I chose for the BC-1 reference design, quiet as a mouse, where as the transformer in the Dh-200 is noticeably noisier but it’s still acceptable and not in need of change.
Imo it’s a lot of effort, easier said than done, plus the cost differential might be uneconomical, time involved etc. for a one of. Making the chassis larger/higher just to accommodate the transformer also leads to extra unnecessary expense, everything adds up.
I am very happy with the triad standard Toroid I chose for the BC-1 reference design, quiet as a mouse, where as the transformer in the Dh-200 is noticeably noisier but it’s still acceptable and not in need of change.
check out this thread:
https://www.diyaudio.com/community/...ution-for-large-toroidal-transformers.161419/
https://www.diyaudio.com/community/...ution-for-large-toroidal-transformers.161419/
Hi tiefbassuebertr,
Instead of forever providing links, why not actually put the relavent information in as long as you aren't repeating previous information. Following a link is a pain in the rear!
Hi Nick,
Too complicated. The existing solutions are really simple, and that means more reliable. Don't forget to, you are hitting the rectifiers and filter caps with this waveform. It is repetitive and high current spikes (capacitor charging). We want to reduce current spikes. Overall your suggestion is a soft start, but not when you look at each individual pulse.
Simple, straight forward designs are often the best and most reliable (within reason).
-Chris
Instead of forever providing links, why not actually put the relavent information in as long as you aren't repeating previous information. Following a link is a pain in the rear!
Hi Nick,
Too complicated. The existing solutions are really simple, and that means more reliable. Don't forget to, you are hitting the rectifiers and filter caps with this waveform. It is repetitive and high current spikes (capacitor charging). We want to reduce current spikes. Overall your suggestion is a soft start, but not when you look at each individual pulse.
Simple, straight forward designs are often the best and most reliable (within reason).
-Chris