managing inrush current advice

[black_labb]

hello again everyone
i've been working on the bass amp and am wondering if inrush current should be managed, or if it is not an issue. the amp is running 6 6550's/kt88's (decided 8 is unneccesary).
the power supply is 2 torroidial transformers ( 450vac for the plates, 250ac for the screens and the preamp) the plates should be around 630vdc with 350uf (2x470uf in series and 2x220uf in series, may add an extra pair of 220uf's if i feel neccesary as i have them around).
the screen/preamp supply is running ~350vdc with 220uf just after the rectification.

would it be beneficial to use some inrush current management? i've seen inrush current thermistors reccomended, but they seem to be only rated at 265v for the ones that are available. obviously it wouldnt be a good idea to run that on the 630vdc supply. are these meant to be run on the input supply to the power transformers? (ie connected to the wall voltage after the fuse).
are there any other methods that appear to be good? i was considering puting a resistor on the standby switch to the amp so that all current from the trannies would run through it, but this didnt seem like a great solution, and its hard finding switches 3pdt switches rated at the current (i'm sure they can be found in a more specialised sources.

thanks everyone

[tubeswell]

Another way might be hefty power resistors on either side of the HT winding before the rectifier. Merlin b has an article on limiters (which altho' it applies to tube rectifiers, would also apply to SS AFAICT):

The Valve Wizard

[Chuck H]

At 630V plates I can't be too sure. Voltage spikes and all. I've had some trouble with B+ spikes on less sensitive projects than this. But it wasn't the tubes that had a problem. Rectifier tubes seem to be especially sensitive to inrush current. I haven't had a problem with power tubes...Yet. Maybe put it on a scope and watch what happens switching off standby. That should tell much of the story.If it seems to be a problem you could use a Pi filter type arrangement AFTER your initial B+ smoothing with a smaller cap on the B+ side. I think the smaller cap would slow down the inrush current but you would still have the benefits of the large filters eliminating ripple and hum.

Chuck

P.S. Test the idea on Duncan PSU. It's a good tool, and very accurate IMHE.

[black_labb]

thanks guys
i think i may bypass the standby switch with a resistor. i'm not sure what the best method of applying a standby to the plates. 630v is alot of voltage for most switches available. would the resistor across the standby make it safe enough to use a 250vac switch for 450vac? seems to be pushing things...

i have always thought of just leaving the B+ on the plates like in my holden/wasps, but i'd like to keep it a bit safer inside for working on when standby is on.

[Steve Conner]

Hey Black labb

If the fuses don't blow at turn-on, no inrush current management is needed.

Yes, those NTC thermistor thingies are meant to be used in the mains line. But several in series might work for your HT.

The classic method is "step-start" where you have a big-ass resistor in the line that gets shorted out by a relay after a second or two.

No equipment is safe to work inside while it's on, standby or not. The only truly "safe" condition is with the plug pulled and the capacitors discharged.

[Merlinb]

Originally Posted by Steve Conner

If the fuses don't blow at turn-on, no inrush current management is needed.

The classic method is "step-start" where you have a big-ass resistor in the line that gets shorted out by a relay after a second or two.

+1

A step-start is dead easy, and can prevent speaker thump. Here's an age-old way of doing it: you just adjust the limiting resistor to give the desired warm-up time until the amp's own current pulls the relay in.

Also, an NTC thermistor on the primary also limits inrush current on the secondary of course- the transformer reflects impedances both ways after all.

[Steve Conner]

FWIW, here's the step-start circuit I used on a solid-state hi-fi amp I built years ago. It uses the "amp's own current" method as recommended by Merlin. The amp had a huge bank of smoothing capacitors, and would blow fuses at turn-on without the step-start.

http://scopeboy.com/psu.gif

Instead of a switch, it has start and stop buttons like a motor starter. The stop button is also hooked up to the protection circuitry, so the amp turns itself off on serious faults.

[black_labb]

thanks steve and merlin, seem like a good idea. i think i will explore it a bit more

[R.G.]

If you use step-start, be very, very sure that the resistor is inside some kind of flame-proof enclosure, or pick a resistor with a power rating that will let it stay engaged forever without overheating.

If the relay ever fails to pull in, the resistor stays engaged, and the value usually works out to be such that the resistor goes up in smoke and flames. The one sure thing I know about relays is that they will fail to operate someday. When that day comes, you might not be there to recognize the disaster as it happens and stop the fire from spreading.

I believe it was exactly this which caused the popularity of NTC inrush devices.

Well, OK, also that they're much smaller and cheaper than relays and power resistors...

[Steve Conner]

This is one advantage of my circuit: If the relay fails to pull in, the circuit won't latch on, and the resistor won't catch fire unless you hold the start button in while pondering where all the smoke might be coming from.

The second pole of the Off switch is to stop you from burning out the resistor by holding both buttons in at once.

Nevertheless, I blew one of those gold Dale resistors during testing, shooting smoke clouds out of the ends. It just couldn't take the dissipation of repeated starts.

Resistors with built-in thermal fuses are available, they were popular in old TVs.

[defaced]

What would 130 volts do? I've run my setup at 500v into 440uf and not had an issue (50 or so starts before I disassembled it to build the next prototype). I'll shortly know what 560v into 340uf does. Does anyone know if Duncan's PSU designer is accurate with it's simulation of inrush current (I think that's one of the options for observation)?

[Diablo]

While on the subject of inrush current limiters, can someone explain the meaning of the codes on them? I'm using one that reads "NTC 100D-11", and I'm curious how many amps it's rated for. I've googled until I have a headache. The room temperature resistance is 107 ohms, according to my measurement. I'm using this on the standby switch to protect the single rectifier tube (5U4GB) from the surge when charging up the filter capacitor. Good or bad idea? This is a Tweed Twin 5E8A.

[black_labb]

i'm not sure diablo, why not check the datasheets?

[Enzo]

100 is the cold resistance (room temperature) in ohms. D is the type, 11 is the diameter

[black_labb]

maybe i will use a thermistor on the plate transformer and have the B+ on the plates in standby, and have a resistor across the standby switch so that the grids and preamp have a slow start as well.

the slow start circuit looks really useful. dont think i'll use it though, doesnt seem completely neccesary and i'd rather not use a relay if i can avoid it. seems to be a complicated way of doing what the thermistor will do. if the plates werent on at startup then it would probably be a good alternative.

[black_labb]

thinking about it a bit more i'm leaning towards putting the plate supply onto the standby switch and putting a resistor across it as well. it will still be running 450VAC but with a half wave (does that make it better?) but i think having the resistor it will tend to draw the current through that instead of sparking. the resistor will charge the caps and there wont be much of a voltage across it after a couple seconds anyway. if that doesnt work then i'll go with the thermistors.

[mooreamps]

Originally Posted by black_labb

thanks steve and merlin, seem like a good idea. i think i will explore it a bit more

Unfortunately he fails to give you suggested values as a starting point. Maybe you can figure these yourself, but perhaps there are others on this forum who would have need for a little more "explanation" than just some vague circuit posted on a forum with nothing more than, "OK, here is what it looks like. Good Luck !! "


Therefore, for the benefit of the others, I'll type in a brief circuit analysis of how this thing works. For example, if the filter cap is the typical value of 22 uF, and if the cap will charge to full value in 5 time constants, it will take approx 1.1 seconds for the cap to charge to the full value of B+ ; or in this case we could say "+300 volts". But the zener will fire the relay at 12 volts.

Well, if in the first time constant the cap charges to 96 volts ; if the given B+ is 300 volts ; then the relay will fire even before the first time constant ; R*C ; which is 10K times 22uF ; or 10 e+3 times 22 e-6 ; which equals approx 0.22 seconds. Not much of a delay, is it ???


-g

[Merlinb]

Originally Posted by mooreamps

Well, if in the first time constant the cap charges to 96 volts ; if the given B+ is 300 volts ; then the relay will fire even before the first time constant ; R*C ; which is 10K times 22uF ; or 10 e+3 times 22 e-6 ; which equals approx 0.22 seconds. Not much of a delay, is it ???


-g

You misunderstand the circuit operation. The relay will only fire if the current through it is sufficient. The incoming voltage is not constant- it is pulsing.

When you switch on, the cap charges slowly, but the inrush current to the cap can be made small, so that it doesn't fire the relay. It is only once the valves have warmed up and start drawing load current that the relay will fire. Obviously the limiting resistor may need a little adjustment depending on circumstances. For example, if R is too small then the relay might pull-in during the initial cap inrush, then drop out again almost immediately, also known as chatter.

[mooreamps]

Originally Posted by Merlinb

You misunderstand the circuit operation. The relay will only fire if the current through it is sufficient. The incoming voltage is not constant- it is pulsing.

When you switch on, the cap charges slowly, but the inrush current to the cap can be made small, so that it doesn't fire the relay. It is only once the valves have warmed up and start drawing load current that the relay will fire. Obviously the limiting resistor may need a little adjustment depending on circumstances. For example, if R is too small then the relay might pull-in during the initial cap inrush, then drop out again almost immediately, also known as chatter.

I'm still not so sure I would have drawn it this way. You would need to find a relay that would have it's coil fully energized with the value of plate current, and I'm not so sure I would put a 12 volt coil on a 300 volt, or in some cases a 600 volt DC power line. Instead, I would use a 12 volt dc power supply ; and my 12 volt DC regulator 9 second voltage ramp circuit to do the time delay for the relay coil, and then use a pair of relay contacts themselves to do the switching.


-g

[Merlinb]

Originally Posted by mooreamps

I'm not so sure I would put a 12 volt coil on a 300 volt, or in some cases a 600 volt DC power line.

The relay never sees more than 12V across it thanks to the zener, no matter what the load current. In this sense its task is actually less stressful than if the coil were fed from a low voltage PSU while the contacts worked in high voltage areas, because in this case the coil and the contacts are all much closer in potential.

[mooreamps]

I don't know....... I'm just not seeing the time delay in this...


-g

[Enzo]

Ask yourself this then: When the amp is first flipped on and that B+ first filter charges in an RC instant, but the tube heaters have not warmed up yet and are not conducting, how much voltage is dropped across the relay coil?

yes, there is B+ ON the relay coil, but not across it. If the tubes are cold, they are not conducting, so there is no current drawn through the relay coil. Once they start conducting, current starts to flow through the relay coil, and when sufficient current flows through it, it pulls in and shunts the limiting resistor.

At least that is what I see there.


Contrast a separate delay circuit that fires up after X seconds no matter what, consider that if the heater fuse is open, your delay still applies B+ to cold tubes, while the self actuating circuit leaves in the resistor.

or perhaps I am missing something.

[mooreamps]

Originally Posted by Enzo

Ask yourself this then: When the amp is first flipped on and that B+ first filter charges in an RC instant, but the tube heaters have not warmed up yet and are not conducting, how much voltage is dropped across the relay coil?

yes, there is B+ ON the relay coil, but not across it. If the tubes are cold, they are not conducting, so there is no current drawn through the relay coil. Once they start conducting, current starts to flow through the relay coil, and when sufficient current flows through it, it pulls in and shunts the limiting resistor.

At least that is what I see there.


Contrast a separate delay circuit that fires up after X seconds no matter what, consider that if the heater fuse is open, your delay still applies B+ to cold tubes, while the self actuating circuit leaves in the resistor.

or perhaps I am missing something.

You, sir ; are not missing anything . . . . .

In fact, I pulled a few data sheets of typical 12 relays. I found the nominal coil resistance at 1000 ohms. So, you would need a plate current of 12 mills to pull in the relay contacts. But, as Enzo so assutly points out, a cold power tube does not start conducting until what ; 14 or 16 seconds after initial power on, in which case "no" current is flowing and the full potential of B+ is still felt upon the plate of the power tube. In essesence, the circuit achives "nothing"...

Secondly, a separate PS for the relay coil can be tied to filament power, especially since if you run your filaments at "12 volts" . . . . .

-g

[Merlinb]

Originally Posted by mooreamps

in which case "no" current is flowing and the full potential of B+ is still felt upon the plate of the power tube. In essesence, the circuit achives "nothing"...

The purpose of the standby is not to stop the voltage being on the plates (the plates don't care a jot, they're just bits of metal). It's purpose is to limit the inrush current to the PSU caps, which is does very well indeed.

[mooreamps]

I would not really care about in-rush on such a "small" value of capacitance. Further, I would still not be wanting to put a 12 volt relay coil on a high voltage power line, along with the noise maker zener. I'd do it the old fashion way by using tube rectifier, and allowing the 8 to 10 sec warm up time to ramp up the high voltage "gracefully".

-g

[defaced]

So... a hand full of diodes later and I most definitely had inrush current problems with the attached schematic. Right now I'm using a series resistor which is bypassed by a relay. I'm ok with that solution because I plan to use a relay for standby anyway, but I would also like to learn a bit, which brings me to my question: how does one choose a thermistor for an application? Meaning, should a 5 amp thermistor be used in a 3 amp application? Should it be rated for maximum current or typical idle current? You know, all that fun stuff. If there's an idiots guide website or something I'd be interested, I was unsucessful in finding one.

Note: on the schematic I'm missing a 1N4007 diode between the relay contact and the first filter cap, 4th figure down The Valve Wizard.

[black_labb]

i ended up using a 47ohm resistor across the standby switch for the plate voltage (should run around 630v at idle, was at around 660v with 2 tubes instead of 6). the 47ohm (5w wirewound) resistor seems to have failed and gone open, but it could have been a cold solder joint (it worked at first though)

i had assumed that a 5w resistor would have enough thermal intertia to withstand the charging of 340uf with 670v. i may have been wrong. i'll have a closer look and change to a smaller resistance or a 10w resistor.

[defaced]

With the 100k resistor I'm using the B+ comes up to about 150v during standby, still low enough to mute the amp. By my math, that has to dissipate about 3 watts during initial start-up which falls to 1.6 watts once the B+ settles to 150v.

Why a smaller resistor? Seems to me you need to go the other way (larger resistor) to further limit current. Duncan's PSU simulator can accommodate for transformer winding resistance, which is what your series resistor effectively is.

Another question, black labb and I are both using toroidal power transformers. Because of their increased efficiency over EI transformers, is that alloying them to generate higher current spikes at turn on? As mentioned way earlier in the thread, 130 volts lower and 50uf higher with an EI transformer and I did not have this problem.

[black_labb]

i actually havent had any problems. the 47ohm resistor went open and the amp has run fine without it. i'm thinking of the longevity of the caps. and possibly blowing fuses on startup. hasnt happened yet though.

and i thought about doing it higher resistance. i either have to make it higher or lower. i might go 100k+. the grids and preamp supply is running the high resistance method. i thought i would try the low one for the plates as i didnt want much voltage on the switch for long as it is rated for 250vac and i am running 450vac rectified into it. the idea is that there is a low resistance across the switch so the switch has no need to spark over. then again, 100k resistor is better protection than a 47ohm resistor that has gone open.

on another note, what do people/manufacturers do for standby switches for amps running high voltages? surely they cant underrate a switch on something that has to be safety approved?

[Steve Conner]

Hi guys

Yes, toroidal transformers have worse inrush than EI types. This is for two reasons:

They're more efficient than EIs (lower leakage inductance and winding resistance) so they can dump more current from the mains into your caps at startup.

They have more inrush current in their own right. The core often saturates on startup if you turn on the mains switch at just the wrong instant. All transformers do this, but toroids do it worst: the tapewound core saturates hard, and the low winding resistance allows lots of current to flow. 50-100 amps wouldn't be surprising for a big toroid running off 120V (25-50 amps for 240v)

Combine these two effects and you can get quite a bang on startup.

Standby switches for high-powered amps are quite tricky. The safety approval ratings are a real issue, and even if they weren't, most switches are rated for AC and struggle to break high-voltage, high-current DC. I like to use a double-pole switch and put the two poles in series for better arc breaking, like Fender did in the 300PS. Putting the switch in the screen circuit would help, because the voltages and currents are smaller.

[R.G.]

Originally Posted by Steve Conner

Standby switches for high-powered amps are quite tricky. The safety approval ratings are a real issue, and even if they weren't, most switches are rated for AC and struggle to break high-voltage, high-current DC. I like to use a double-pole switch and put the two poles in series for better arc breaking, like Fender did in the 300PS. Putting the switch in the screen circuit would help, because the voltages and currents are smaller.

+1.

This is one reason I really like power MOSFETs as a standby switch. They are rugged enough to live in the high voltage environment, even with really big surges to fill filter caps, and do not arc at all. You get the high current stress off the standby switch entirely, and depending on the setup, may even be able to have the standby switch itself run with only low voltage DC on it.

One interesting setup that doesn't get much exploration is photovoltaic MOSFET drivers. With these widgies, you can float an N-channel MOSFET on top of B+ and drive the MOSFET gate from the current in an LED at the input to the MOSFET PV driver. These things naturally don't provide hard switching to the MOSFETs, so you get softer switching than you do with metal contacts, ramps and not cliffs.

[Steve Conner]

I remember seeing the MOSFET standby switch in some Yorkville amp. For my own MOSFET regulated amp, I just wired the standby switch up to disable the regulator, which causes the B+ to fade in and out over a few seconds.

However, many hobbyist amp builders are probably working with tubes because they couldn't get transistors to stay alive. High voltage MOSFET switches aren't necessarily a hassle-free addition to your amp. Voltage isn't necessarily a problem in itself though, 800V MOSFETs are fairly common and I've seen them up to 1200V.

The photovoltaic isolators are great if you can find them. You get ready-made "DC solid-state relay" modules that consist of a MOSFET driven by a PVI.

[R.G.]

Originally Posted by Steve Conner

I remember seeing the MOSFET standby switch in some Yorkville amp. For my own MOSFET regulated amp, I just wired the standby switch up to disable the regulator, which causes the B+ to fade in and out over a few seconds.

That works nicely.

High voltage MOSFET switches aren't necessarily a hassle-free addition to your amp.

I think "hassle free" and "high voltage" may be mutually exclusive.

I was trying to design switching power supplies back when only bipolars were available to do that. It was a real challenge to get bipolars to live in a 400V environment at all. Modern MOSFETs are a walk in the park by comparison. But you do have to understand that at high voltages, the inductance and capacitance of things like wiring makes a difference, and that inadequate layout and insulation can literally reach right out and bite you.

The photovoltaic isolators are great if you can find them. You get ready-made "DC solid-state relay" modules that consist of a MOSFET driven by a PVI.

Availability is getting better. It used to be only International Rectifier. Toshiba joined up, now Panasonic and Vishay are making them. Digikey stocks a variety of them. They're a little pricey ($2-$10 depending) but they do something which is otherwise very impractical.