Thermistor for an 120VA toroid transformer

those typical peak currents are probably achieved in the real world. But the peak is very short term.
A fuse takes time to heat up and rupture.

To prevent nuisance blowing when starting up a transformer or motor, it is usual to fit a fuse that is rated to about three times what the wattage would predict.
for a 160VA transformer fed from 230Vac the rated current is ~700mAac.
The fuse that would normally be fitted is the nearest to 2.1Aac. Use a T2A fuse for a 230Vac 160VA transformer.
However that fuse will not blow at exactly 2A. It is rated to pass 2A almost for ever.
At double it's rated current it will last for a long time. Consult the manufacturer's datasheet to see if it is 1minute or 10minutes or 30minutes. Yes it varies a lot and could allow a lot of incidental damage as a result of how long it takes to rupture.

A safer way to fuse your transformer is to close rate the fuse. Use T800mA and expect it to fail repeatedly due to nuisance blowing.
The way to avoid nuisance blowing of a close rated fuse is to fit a soft start circuit.
You may find that a fuse even smaller will allow your transformer to power up and for the equipment to function perfectly well for years with this smaller fuse.
Try it.

You decide whether to try T600mA, or T800mA, or T1A, or T1.2A, or T1.6A or T2A fuses.
I know which end of the range I would choose.


2009-02-28 9:40 am
I have just been doing a bit of experimenting on this very subject. I am using a couple of the old Maplin GA28F MOSFET amps with a 300VA toroidal transformer and 10,000uF of smoothing caps. I found that when I plugged the thing in it blew an ordinary 3A fuse in the plug, but seemed OK with a 5A one. There was an audible buzz from the transformer at power-on that died away after a second or two. Obviously, the current draw should be insignificant unless playing music very loudly (unless using a Class A amp).

Reading around the subject, it is apparent that the size of the power-on surge varies depending on the point in the mains cycle when the power is applied and also at what point it was last switched off! (The core of the transformer holds the magnetic flux even when switched off, apparently).

I decided I didn't like the idea of an uncontrolled short circuit every time I switched the amp on - I would forever be bracing myself for the fuse to blow, and I like the idea of low-ish current fast fuses in the plug if I can get away with them (plus a lower-rated 'T' delay fuse in the amp itself).

A common approach seems to be to use a NTC (negative temperature coefficient) thermistor, as you say, but I also read that this is really horrible for several reasons:

(a) It needs to run hot to work.
(b) If your amp doesn't draw much current for a while the thermistor cools down and its resistance goes up again. This sounds as though it would cause the supply voltage to dip at the next loud transient..?
(c) It needs time to cool after power-down (up to a minute), so if you turn the amp on straight away, it won't have the desired effect.
(d) It isn't very 'pure' in that it's a permanent, not-insignificant resistance in your supply. Kind of makes the idea of solid gold mains plugs etc. seem a bit silly! (if they're not already)

Other types of electronic surge limiters include clever circuits with TRIACs to switch the circuit off and on at the 'correct' phase in the mains cycle, and apparently someone has patented a method of pulsing the transformer with unipolar pulses at power-up to pre-magnetise the core to a known state before fading in the full cycle. I would love to have a play with that, but I don't like the idea of the TRIAC's 'crossover distortion' even when fully on. It could be bypassed with a conventional relay, however.

Which brings me to my experiments with a simple series resistor and bypass relay which I just thought I'd report on, in case it is of use to anyone. The idea is to power-on the amp with a resistor in series with 'Live' and, after a short delay, to bypass it by activating normally open relay contacts across it. There are quite a few commercial circuits that do it, but I wasn't quite sure what I might expect using the technique.

I made a circuit using a small 9V 1VA transformer/rectifier/smoothing cap to provide isolated 12V-ish DC power for a power-on delay circuit based on an RC and LM311 comparator - the LM311 output transistor can pull a relay directly. I used a 12V relay with 16A contacts and also included a (mains rated) 120R/100n snubber across the relay contacts (a good idea or not?).

The question was, what size resistor to use. And the worries were, what if the relay contacts failed for some reason? How hot would the resistor get? Would the initial surge pop a wire-wound resistor anyway? And does the inclusion of a series resistor prevent the amp from powering up correctly? That is, might I get a speaker 'thump' that I never get otherwise?

I fitted a 22R 25W aluminium-clad resistor, figuring that the amp would have to be playing pretty loud to exceed 25W-ish anyway. And to be doubly certain, I also fitted a thermal switch with bi-metallic contacts, rated at 10A and 70 deg C in series with the resistor and glued to it with thermal adhesive. The idea being that if the resistor went above 70 deg C it would switch off temporarily, and no smoke would issue forth. 22R limits the absolute current draw to a lot less than the 50, 60, 70A surge that people talk about, but I figured probably not enough to upset the operation of the amp. I set the delay to 2s.

Well, the result is the amp now works happily with a 3A fast fuse in the plug - no blown fuses yet, and the audible buzz seems to have gone. No speaker thump either. The resistor doesn't get at all warm and I think the bi-metallic switch is overkill. Should I increase the resistor to 47R?
I have just been doing a bit of experimenting on this very subject. .................................
Well, the result is the amp now works happily with a 3A fast fuse in the plug - no blown fuses yet, and the audible buzz seems to have gone. No speaker thump either. The resistor doesn't get at all warm and I think the bi-metallic switch is overkill. Should I increase the resistor to 47R?
yes, I suggest you increase the resistor to well over 22r.
I would also suggest that you try fitting a T1.6A fuse and if that blows too frequently change it to T2A.

If you go to >=60r, then you might be able to start up reliably with a T1.2A fuse.

T3A supplies ~700W forever without out rupturing. How hot could 700W get your flammable support?
I'm not certain that "pre-magnetising" the core will make much difference. Bringing up the AC gradually will work, by spreading the problem over a greater period.

The basic problem is that randomly applying AC to an inductor can result in a DC current equal to the peak AC current. The DC current then decays with a time constant of L/R. The mains has a low resistance. A toroidal transformer has high L and low R, so the time constant is high. Adding initial resistance by some method both reduces the initial AC voltage (so reducing the initial AC and DC currents) and shortens the time constant so the DC decays more quickly.


2009-02-28 9:40 am
Hi DF96

Thanks for the info. This stuff falls into a major gap in my knowledge!

This is the company with the "pre-magnetising" patent, supposedly:
Function of TSRL and TSRLF.: EMEKO, Michael Konstanzer

"Premagnetising of the Transformer with unipolar Voltage Time impulses until the remanence reaches the max. point. Then full switch on, then close the bypass relay contact."

@Andrew T

I've increased the R to 47R, and everything seems OK (so far) with a T1A fuse.

(and I meant to say "2 x 10000uF" of smoothing caps, yesterday)
If you are worried about a failure mode causing excessive heating of the series resistor, use a thermal fuse in series with it, in thermal (but not electrical) contact with the resistor. If it overheats, the thermal fuse protects your amp and you.

I have a bench power supply with a soft-start in it. The series resistor is not a standard power resistor but a PTC thermistor (PTC, not NTC). Normally, the amount of energy dissipated by the PTC on a given start-up is not sufficient to heat it up. But in a failure where it stays in the circuit, the current draw would cause it to heat up, increasing its resistance until it won't heat up any more. It is therefore self-protecting. I wouldn't recommend this for an amp though, as it would result in low rail voltages which could result in amp malfunction, which could result in speaker damage, etc.
Thanks for the replies.
Since I am very new to this subject, I chose the safer way. The usage of a single NTC sounds very easy, but I will have a part with very high temperatures in my amp. I got myself a ready built soft starter from ebay in Germany. It is probably overpriced and oversized. Have a look: Einschaltstrombegrenzung - Softstart - Ringkern bei Verstärker (endet 09.12.10 00:08:07 MEZ). I have a toroid with 120VA, so an T 0.7A will fit?

Thanks for your help

regards eladmi
Last edited:
I have a toroid with 120VA, so an T 0.7A will fit?
I don't know if T700mA is available, but I know you can buy T600mA, T800mA and T1A fuses.

220Vac and T700mA allows long term 154W to be supplied. This should run indefinitely when the amplifier is working properly into the correct load.
However the start-up condition even with the soft start fitted may blow the fuse. Try various fuse rating until you know what kind of life you get from your fuse. Blowing a dozen times a year when "I have done nothing wrong" is too often.
Blowing once every two years may be acceptable to you, you decide.
since you have 700mA fuses, then I suggest you add a few more ohms to your soft start. You can add more NTC and/or more resistors. I consider two CL60 as the minimum for effective soft starting of a biggish toroid transformer.

Smaller transformers can manage with much more resistance, but when it is switched out (bypassed) the mains gets another peak demand that could still blow the fuse. This is where NTCs come into their own. They adjust to suit the demand. But 3 or 4 CL60 for a single 120VA transformer is just too many £.
Particularly when I have 15p 20r and 10r 5W resistors.
I have used unbypassed NTC, but they can give a PSU which has a slightly negative output impedance in the region below about 1Hz. This could tip a circuit with marginal LF stability over the edge into motorboating. Most circuits probably won't notice.

As AndrewT says, you can always bypass after a time delay if it proves to be a problem. This has the added advantage of allowing the NTC to cool down so you can switch off and on again quickly, provided the bypass drops out immediately (as it should anyway).
Here are some pics of the toroid and the softstart:



The connectors on the softsstart module are labeled (from left to right):

- LOAD 10A
- AC 220V IN

I am intended to connect them this was:

- LOAD 10A <- live to to torodial transformer
- NEUTRAL <- neutral to tororidal transformer
- NEUTRAL <- neutral AC
- AC 220V IN <- live AC

Please correct me.