All power supplies for high power audio amps have a current limiter resistor on the primary of their transformers for the obvious reasons.
However, why is this not a practice on the SECONDARY ?
I have never seen this done on any amplifier.
Here is the reason for my question:
I am building an amp (more like a welding machine LOL) with a 2.5kva toroidal which will have secondaries at 2 x 70 volt AC and 520,000 uF of filter capacitance. With such a beast, on the secondary, such high capacitance will present an instantaneous short circuit, as soon as the power is switched on.
Is a secondary current limiter not needed on account that if a lot of current is drawn at the secondary, this will also translate to a proportional current increase on the primary and hence collapse the secondary voltage so that it literally self limits ? Even though that sounds reasonable, I have my doubts.
Or would it be always safer to insert a 5 ohm thermistor (limits current to 70/5 = ~14A at switch on) in series to each secondary, just in case ?
As more power is required & current drawn, thermistor's resistance would reduce and hence effectively become more transparent in the secondary, but obviously at the expense of generating some heat.
As you can imagine, such capacitors are extremely expensive and have to be very heavily protected.
Would a secondary current limiter not immensely improve the long term reliability and the life of the rectifier and the capacitors ?
Thoughts please ?
Thanks
However, why is this not a practice on the SECONDARY ?
I have never seen this done on any amplifier.
Here is the reason for my question:
I am building an amp (more like a welding machine LOL) with a 2.5kva toroidal which will have secondaries at 2 x 70 volt AC and 520,000 uF of filter capacitance. With such a beast, on the secondary, such high capacitance will present an instantaneous short circuit, as soon as the power is switched on.
Is a secondary current limiter not needed on account that if a lot of current is drawn at the secondary, this will also translate to a proportional current increase on the primary and hence collapse the secondary voltage so that it literally self limits ? Even though that sounds reasonable, I have my doubts.
Or would it be always safer to insert a 5 ohm thermistor (limits current to 70/5 = ~14A at switch on) in series to each secondary, just in case ?
As more power is required & current drawn, thermistor's resistance would reduce and hence effectively become more transparent in the secondary, but obviously at the expense of generating some heat.
As you can imagine, such capacitors are extremely expensive and have to be very heavily protected.
Would a secondary current limiter not immensely improve the long term reliability and the life of the rectifier and the capacitors ?
Thoughts please ?
Thanks
Last edited:
If you want to put them there, so put them there. This is DIY after all. You could also use a relay shorting a resistor, that would engage when the capacitor charges up to a certain voltage. A bit more complex than a two terminal solution, but better.
Unsure if you'd ever "hot-replug" with the thermistors, but they could still be quite hot upon a new power on/off cycle. I understand these devices as a "one time" protection, not a repeat protection - which is entirely reasonable for consumer equipment.
Unsure if you'd ever "hot-replug" with the thermistors, but they could still be quite hot upon a new power on/off cycle. I understand these devices as a "one time" protection, not a repeat protection - which is entirely reasonable for consumer equipment.
Yes the available secondary current will be limited at mains turn on by a few mechanisms.
Firstly, the primary winding in-rush current is limited - and if as you infer you will be using a primary side NTC then that will likely be the dominant bottleneck for peak current for a few cycles. During that time, the primary magnetising current has to grow in order to provide secondary voltage to charge the caps. The secondary current will have its own current limit due to loop impedance. The available peak charging current will be way under whatever would or could ever stress the caps themselves.
As far as protecting the caps, I doubt there is anything you could do to damage them from an over-current event. If you were concerned about belts and braces protection, then perhaps look at a few thermal switches in the local ambient of the caps, and an overvoltage protection relay.
I would suggest caution about how you choose your NTC protection on the primary, and how you intend to include over-current protection on the mains primary and secondary. The link below goes some way to providing insight in to those processes.
https://www.dalmura.com.au/static/Valve%20amp%20fusing.pdf
Ciao, Tim
Firstly, the primary winding in-rush current is limited - and if as you infer you will be using a primary side NTC then that will likely be the dominant bottleneck for peak current for a few cycles. During that time, the primary magnetising current has to grow in order to provide secondary voltage to charge the caps. The secondary current will have its own current limit due to loop impedance. The available peak charging current will be way under whatever would or could ever stress the caps themselves.
As far as protecting the caps, I doubt there is anything you could do to damage them from an over-current event. If you were concerned about belts and braces protection, then perhaps look at a few thermal switches in the local ambient of the caps, and an overvoltage protection relay.
I would suggest caution about how you choose your NTC protection on the primary, and how you intend to include over-current protection on the mains primary and secondary. The link below goes some way to providing insight in to those processes.
https://www.dalmura.com.au/static/Valve%20amp%20fusing.pdf
Ciao, Tim
- You answered a question. Current limiter on a primary limits primary and secondary current, simultaneously. But if there is free space in a case and money then it is ok to put additional secondary current limiter. If you transformer has such a design that it has low own inrush current - you may possibly use only secondary charging current limiter.
Toroidal type of transformers typically have larger (the largest among others) its own inrush current, so they love to have primary inrush current limiters. Especially large ones (I use NTC with smaller ones). Otherwise all commutation elements (switches) and fuses will be stressed to much every turn on.
Last edited:
I keep the idea that - to keep the switches and relays (mechanical esp) as far on the outside of the signal as possible, and to put the relay, or switch, on the highest voltage part of the chain. The higher voltage of course means less losses across the contacts. I also run no rail fuses, but fuse the mains, individually for each transformer. If you run some kind of soft start circuit, you can also set the fuse closer to your max draw for each power supply.
Thanks for all your responses. I think I have settled on doing the following:
I will put a primary limiter with a resistor and a time delay shorting relay after about 4, 5 seconds. Even tho only a few cycles are needed to limit at the primary, extending the turn on time will also limit the excessive current on the capacitor charging until they reach a level which will no longer be considered a short circuit.
I will not put a limiter on the secondaries.
Thanks
I will put a primary limiter with a resistor and a time delay shorting relay after about 4, 5 seconds. Even tho only a few cycles are needed to limit at the primary, extending the turn on time will also limit the excessive current on the capacitor charging until they reach a level which will no longer be considered a short circuit.
I will not put a limiter on the secondaries.
Thanks
If big toroids are switched on in the mains zero crossing, they can saturate and easily draw 50A, welding switches and relay contacts shut. Hence a primary limiter is highly recommended.
What about splitting the cap bank, with an additional secondary NTC between a 22-47mF and the main cap on each rail? This will also reduce hum (more so in quiet passages), and I doubt the 22-47mF will get damaged. It costs a fragment anyway.
What about splitting the cap bank, with an additional secondary NTC between a 22-47mF and the main cap on each rail? This will also reduce hum (more so in quiet passages), and I doubt the 22-47mF will get damaged. It costs a fragment anyway.
I highly recommend an NTC
Resistors can start fire if the relay doesn't complete the circuit. Parallel some CL-90. Do the steady state computations if you must.
Resistors can start fire if the relay doesn't complete the circuit. Parallel some CL-90. Do the steady state computations if you must.
Of cause it is much reliable to use NTC termistors but not resistors. We used powerful wired resistors in old times when there wasn't NTC termistors at all.
Not if you have a soft starter on the primary side. The soft starter will limit the inrush due to saturated core but also charging currents if the delay time is long enough.
Well, you shouldn´t
You can do that better at the primary side.
Definitely not.
A properly designed primary one will do both jobs better.
You seem to believe primary and secondary currents are not relatd and should be tackled independently.
It is not so, they are *tightly* related.
Yes, I will use 20 amp thermistors at the primary with about 4 seconds of delay.
I think that should be enough to limit the secondary current also to a safe limit.
I think that should be enough to limit the secondary current also to a safe limit.
Why the need for such a massive amount of capacitance?
Is this due to the "more is better" syndrome permeating society?
This amp will be used at VERY hi volume levels for very long times (almost commercial use, fan cooled). Hence, to keep the ripple as low as possible at those current levels, it was appropriate to choose those caps.
I would not use a NTC for a device that uses more than 100W because a NTC continues to use significant amounts of power to stay hot. Perhaps use a NTC and a relay or TRIAC. A thyristor ~dimmer cct on iether side avoids the heat issues of a NTC but it's a lot more complicated, and, thyristors and transformers (inductors) can be a bad mix. Consider using a fusable resistor, assuming it will only open when a fault occurs. This may simply be a resistor with a known failure mode, ie open.
I recommend you not use an NTC without considering their basic use guidelines. You should calculate what throughput power is passing through an NTC at power on due to all the in-rush contributions, otherwise the NTC may be damaged and may explode. Imho you should choose an NTC that will be continuously operated at between about 20-30% and about 90% of its rated current, even if you are then bypassing the NTC with a relay contact, as a redundancy measure, in which case you need to know your continuous mains AC current requirement during idle, just as much as the current level during maximum practical signal conditions.
That appears to presume the use of a very large NTC device, as even a 20-23mm disk type NTC has a maximum allowable dissipation of circa 5W at highest allowable continuous current, so such a device would often be selected for use in an application where it may dissipate circa 2-3W continuously.
Last edited:
I see, but I'm curious....
But what sort of signal would this amp be dealing with?
Vocals, music, a constant tone of some sort?
And is this an indoor or outdoor venue?
Wattage of this amp?
It will be mostly indoor rock music, at about 70 - 80% of its max power.
The amp is a 500 Watts/channel RMS design.
The amp is a 500 Watts/channel RMS design.
The highly rated infamous Mcintosh MC-2300 belts out 600W bridged, 300/per.
And this is conservative ratings of course.
Yet it only needs 39kuf/39kuf caps per channel.
And this is conservative ratings of course.
Yet it only needs 39kuf/39kuf caps per channel.