Power On Surge Protection Using NTC Thermistors

MikeC · « **on:** 3 Mar 2006, 07:26 pm »

Ginger posted about using these here: http://www.audiocircle.com/index.php?topic=2322.msg19431#19431&highlight=inrush+current+thermistor#19431

My question is regarding the power draw of the 100W amp at idle or low power. I am looking at incorporating an NTC thermistor to protect my power relay and rectifiers from the switch on surge of the 500VA transformer I have for the bass amp in my (nearing completion) Orion setup. I can see how these devices can be really useful if power drawn is relatively constant, eg soft starting a vacuum cleaner motor or for low voltage lighting, but I have a concern with using them in an audio amplifier. When Ginger posted about these, he only quoted the full load current draw in his example. Given the transformer efficiency (from the quoted losses) of about 92%, a 500VA 30-0-30 transformer will draw 2.5A at full load at a nominal 220V. I can put in a 3A 25-ohm (at 25°C) thermistor which would limit the switch-on current to 9.2A. This will have a hot resistance of 0.48-ohm at full current. As per Ginger, the effect of this will be negligible at 0.008 ohms effective at the secondary. My concern is at lower current draws, the resistance will increase considerably, so that if only 0.1A is drawn from the mains, the resistance will be about 18 ohm or 0.3 ohm on the secondary. At low power levels (around 0.01A drawn from the mains), the resistance will approach 25 ohms or 0.42 ohm at the secondary. This will undoubtedly limit instantaneous power, as I will instantaneously drop 150W across an almost cold thermistor if a large peak comes along. Thie resistance will obviously drop very quickly as the thermistor heats up.

Alternatively, I can put a 10-ohm thermistor in line. This will give a worst case scenario of 0.17 ohms across the secondary, and restrict instantaneous power by 60W when cold, but will only limit switch on current to 23A.

How significant are any of these effects? A soft start circuit is undoubtedly better but it will take some thinking to squeeze one into the chassis now. Am I really worrying for nothing about limiting current at low power levels or worrying about longevity of what is a relatively cheap relay? Can someone (Hugh?) help me with the current drawn by a 100N+ at idle?

All help will be appreciated.

Regards

Mike

PSP · « **Reply #1 on:** 3 Mar 2006, 10:00 pm »

Hi Mike,
I have a similar question about using thermistors for inrush in power amps. The analyses that I have seen are based on steady state arguments. Of course, the switch-on transient is considered, but then the analysis treats the case where the current draw is near maximum and the thermistor has reached its steady state temperature.

I wonder about transients: say that the music is soft, slow, sweet. Then Igor picks up the biggest stick in the orchestra, winds up, and smacks the biggest drum in the back row, and keeps beating that drum for 20-30 seconds while the trombones, trumpets and low bass go nuts too. Then the pipe organ comes in...

During the quiet section, the thermistor has been running cool (low power draw = low temperature = high resistance). The first few seconds of profound noise may well be handled by the power supply caps, and after a long noisy passage, the thermistor will eventually get hot (and resistance decrease), but I worry about how the music will sound when the thermistor is playing catch up. Thermistors have significant mass, and I'm skeptical that their thermal response can be quick enough to follow the dynamics that are so important in music.

I'm building a 100N+ with 43v rails to run Orion bass with a 625VA torroid and 20,000uf per channel. I had planned to use the LC Audio soft start circuit, but Hugh was concerned that the slow startup might drive the amp to produce large DC offsets and stress the drivers. He recommended starting the system up naked, and then leaving the amps on. Mine start up naked without dimming the room lights, but I suppose the diodes may be sweating a tad (I used Digi-Key p/n 568-1667-5-ND, 20A continuous, 125A peak) and it hasn't blown the first few times I've turned it on.

I'm interested in what others think or have tried.

Peter

AKSA · « **Reply #2 on:** 4 Mar 2006, 03:51 am »

Hi Mike, Peter,

Two issues here; switch on surge, and on-resistance. Any resistance in the primary circuit is not well tolerated, largely because the charge pulses in the filter caps of the power supply have very high current, around 50A for very short periods, and this load reflects into the primary, rapidly dropping voltage.

Second issue is switch on surge. Transformers operate on inductance which limits current only after a few half cycles have passed through the windings. The situation is exacerbated with capacitor banks after the secondary because they must be charged at switch on. In truth, the mains will withstand high current surges because there are many fractional HP electric motors in homes and these too require huge surges, particularly appliances like AC compressors. Your mains should easily be able to withstand 50A surges; most fuses, mechanical and metallic, are slow blow devices which can tolerate this for short periods.

The transformer type is very important. Trafos long on wire and short on iron, like toroids, have very low inductance and the switch on surge is heavy, typically around ten times the max current. EI, or rectangular, lam types, have much more iron and their inductance is higher, so they exhibit a much lower switch on surge. If you are concerned about deep switch on surges, the EI transformer is the way to go.

In closing, I would try everything possible to avoid use of a thermistor. Your mains is 240V I believe, so your capacity for surges should be very good. I would try the zero protection route for the moment, and see how it goes. You might be quite surprised how well the house wiring tolerates the situation!

Peter, what is the model and engine type of your Olds? They appear to be very reliable cars, with only a few issues, such as faulty alternators. I have a 1986 250cid Ford Falcon fitted with LP gas, and it just keeps going..... Up to 250K miles now, and utterly reliable.

Cheers,

Hugh

PSP · « **Reply #3 on:** 4 Mar 2006, 03:00 pm »

It's an Olds Cutlass Ciera, 1989, four cylinder. It was my mom's car until she died a few years back. It's kind of a barge on the road, but very reliable and gets me to work in the morning.

I've got about 150K miles on it and everything is still apparently happy. There are a few rust spots (we use salt on the roads here to soften/loosen the ice enough so that the snow plows can chip it off), and I would prefer something that was a bit more fun to drive.... but I've got an audio habit

and my lovely wife is thinking about redecorating the house

so I'll probably be a barge captain for a few more years...

MikeC · « **Reply #4 on:** 4 Mar 2006, 04:22 pm »

Hi Peter

You have pretty much the same idea as me, I'm also running 42V rails with more capacitance than standard (14100uF per rail) and 2 x 20A Schottky Diodes for rectification. I may add another cap per rail later, but I need to get running first. Your concerns about using thermistors exactly mirror mine.

Hugh

Thanks for the advice. I am not too concerned about the house wiring, but am more concerned about welding the relay contacts. I would ideally like to keep the amps on all the time, but it looks like we will have a winter of major power outages as our total installed generating capacity for the country is exactly on the projected demand over winter, and the operational (i.e. not broken or mothballed for cost saving) capacity is about 5-6 GVA below this.

As for cars, that's a whole new discussion.

Cheers

Mike

ginger · « **Reply #5 on:** 6 Mar 2006, 03:30 am »

Guys,
You have identified the major problem with inrush current limiting NTC thermistors. You have to select them such that they have negligable hot resistance at steady state current draw which is almost impossible for a Class AB Amp.

The alternative, used by a lot of tube amp designers, is to use a fixed resistance in the line which is subsequently shorted out by contact closures on a time delay relay. For tube amps the delay is typically 40 or 50 seconds to allow tube heaters to stabilise BUT for a SS Amp then a delay of 1 to 2 seconds is adequate. All this adds cost and complexity and necessarily involves you in additional mains side wiring and the attendant risks involved with that.
I eventually abandonned the NTC thermistors as impractical and just made sure I was using "ANTI-SURGE" fuses. These fuses do NOT have an "anti-surge" action as the name might suggest but have a surge current tolerance - that is they do not blow on quick transient surge currents, but rather have a "time delay" action.

For guidance of those wanting to add a time delay relay. You need (for example):
Relay Base/Socket - finder 9002SMA to suit 8 pin relay and timer (Farnell 426-9093) @A$13.20
Timer Module - finder 8600-0240-0000 (Farnell 426-9044) @ A$96.73
Relay - finder 60-12 Series (8 pin) 230 V AC Coil (Farnell 431-679) @ A$22.05
OR 110 V AC Coil version (Farnell 431-667) @ A$19.76

Note that the Timer Module is suitable for both 230V AC or 110V AC.

The relays are 2 pole change over. a different socket (11 pin) and 3 pole change over (finder 60-13 Series) relays are also available which adds only a few dollars and might be worthwhile if you want the extra pole for some other function or to switch 3 amplifiers etc.
So there is around a $A150 cost to SAFELY add this functionality. I haven't bothered. Relays specified above are 250V AC 10A rated.

Similar Relay bases, timers and 8 or 11 pin plug in relays are available from other manufacturers such as OMRON etc.

If you are going to play with these maybe a separate Amplifier Power distribution box incorporating the above would be the go so that you can protect multiple amplifiers. More Cost for plugs, sockets box etc.

You need ONE MORE THING - an appreciation of the fact that you are playing around with mains voltages and the SAFETY PRECAUTIONS REQUIRED to do this. If this worries you - GOOD, it damn well should. If you don't feel confident, then don't try it. If you still want it - ask an electrician to wire it for you.

Cheers,
Ginger

Geoff-AU · « **Reply #6 on:** 7 Mar 2006, 12:41 am »

Rod Elliott's soft-start circuit here would probably work out quite cheap, and follows the same idea.

I'd much rather go with a current-limited startup that switches off, allowing the mains to directly energise the transformer. Also it's only really important if you're using a bigger transformer (500VA and up, as MikeC is) - I'm using 2x160VA in my AKSA55 and don't bother with soft start. You can hear the transformers buzz/hum softly and briefly as they energise but that doesn't matter - as long as you avoid switching the amp on and off all day it won't affect longetivity.

AKSA · « **Reply #7 on:** 7 Mar 2006, 04:36 am »

Thanks Ginger!!

Geoff,

I strongly discourage people from using soft-start circuits. The reason is that during the switch-on of the AKSA a DC pulse of around 12V peak is applied to the speaker. It's of very short duration, less than 100mS on switch-on and about 500mS on switch-off, but any attempt to soften inrush current by lengthening the procedure will increase these durations and possibly threaten the voice coils of the bass drivers.

With normal, instantaneous switch-on these DC pulses are of no significance, not even with a Lowther.....

Cheers,

Hugh

PSP · « **Reply #8 on:** 7 Mar 2006, 03:53 pm »

I just completed my 100w AKSA N+ with 43v rails, 625VA torroids, and 20,000uf per channel. Following Hugh's advice, I did not use a soft-start circuit (I will soon be selling two LC Audio soft start circuits that I bought for this purpose, if someone is interested...).

Turn on/off with this amp (115VAC mains in the US) is entirely uneventfull. I've made the following concessions to the inrush issue:

Geoff-AU · « **Reply #9 on:** 8 Mar 2006, 05:59 am »

Hugh,

fair call, I don't notice any switch-on noise at all with my AKSA55 - assuming I follow the correct procedure of preamp first, then amp.

If there is cause to use a soft-start then perhaps a couple of muting relays on the amp outputs that cut over a couple of seconds after the soft-start has bypassed would be the way to go. Again, it would only be necessary with a hefty amp!