I'm in the process of putting together a couple of power supplies for some upcoming guitar amp experiments. I've never really paid much attention to inrush currents before, but this time I made a quick estimate of just how big that initial inrush current can be, and that made me wonder about the necessity for inrush current protection.
So: Are there any guidelines for when you need to add inrush current protection? What sort of inrush current protection is appropriate for a high-voltage power supply (approx 330 - 340 V DC) for a class AB tube guitar amp, where power supply current will vary with output power? Will an NTC thermistor work in this application?
In particular, I'm thinking about a power supply that will use a 50 VA Triad N68x to generate roughly +340 V and +170 V DC rails. The plan is to use 1N4007s and 100 uF filter caps to make up a positive and a negative half-wave rectifier, each spitting out roughly 170 V DC. Grounding the (-170 V) rail leaves you with 0V, +170 V, and +340 V.
The part that gives me pause is the realization that initial turn-on conditions are essentially a dead short across the 120 RMS transformer secondary until those 100 uF filter caps charge up. That is a scary-big initial current.
There will be an additional RC filter stage to reduce ripple to reasonable levels, but I think that has negligible effect on inrush current, so we can forget about it.
-Gnobuddy
So: Are there any guidelines for when you need to add inrush current protection? What sort of inrush current protection is appropriate for a high-voltage power supply (approx 330 - 340 V DC) for a class AB tube guitar amp, where power supply current will vary with output power? Will an NTC thermistor work in this application?
In particular, I'm thinking about a power supply that will use a 50 VA Triad N68x to generate roughly +340 V and +170 V DC rails. The plan is to use 1N4007s and 100 uF filter caps to make up a positive and a negative half-wave rectifier, each spitting out roughly 170 V DC. Grounding the (-170 V) rail leaves you with 0V, +170 V, and +340 V.
The part that gives me pause is the realization that initial turn-on conditions are essentially a dead short across the 120 RMS transformer secondary until those 100 uF filter caps charge up. That is a scary-big initial current.
There will be an additional RC filter stage to reduce ripple to reasonable levels, but I think that has negligible effect on inrush current, so we can forget about it.
-Gnobuddy
I suggest you are just scaring yourself. If you are trying to do something quite different from thousands of other amps do, then well worth technically assessing your specific situation - but your amp seems typical.
Yes you can make measurements, and do simulation design, but that is more commonly related to choosing an appropriate mains fuse rating and type.
Hi gnobuddy .
Yes, you are right, you have a large inrush current peak at the beginning, but do the Math and you´ll see values are high but not as much as you think, and, more important, fit accepted Engineering practice (or else
)
I´ll write below the basic Math.
Scary big? .... did you actually calculate it?.
Fact is there is resistance, lots of it, in series with those diodes AND in series with that primary. (built in resistance).
1) measure transformer HV secondary resistance.
I would not be surprised at values of the magnitude of, say, 10/20 ohms, or thereabouts.
2) now measure mains/primary resistance ..... same thing.
Multiply it by square of voltage ratio to get impedance ratio so you get value of "reflected" primary impedance; in this case since it´s a 1:1 transformer, ratio is 1 so simply add primary to secondary resistance to get btotal series bresistance.
In this case I cheated and calculated total resistance (based on transformer rated regulation) and found it to be almost 40 ohms.
Yes, it´s a cheaply made transformer.
So your peak current surge will be around 120V*1.4142/40r=4.25A
Will in fact be somewhat higher seen from the Mains side, because transformer even if unloaded will also take some surge current to magnetize its core, but nothing to write home about.
But *please* measure and post results, mine are just estimations based on a datasheet (and having designed and wound a few thousand transformers )
2) as of: The plan is to use 1N4007s and 100 uF filter caps to make up a positive and a negative half-wave rectifier, each spitting out roughly 170 V DC. Grounding the (-170 V) rail leaves you with 0V, +170 V, and +340 V.
please don't.
I hate goofy voltage doublers for many reasons, and good news is you don´t actually need them here .
I think you have 120V mains.
In that case use secondary as 120V primary (transformers are *fully* reversible) , connect both 120V "primaries" in series, and feed a full wave bridge with 4 x 1N4007.
Put both filter caps in series and send transformer center tap you just created to cap center tap.
Instant cleaner +340 and +170V at your disposal and diodes will be happier.
And for peace of mind read 1N4007 datasheet surge capabilities in the datasheet, you´ll be surprised
Yes, you are right, you have a large inrush current peak at the beginning, but do the Math and you´ll see values are high but not as much as you think, and, more important, fit accepted Engineering practice (or else
)have you?
I´ll write below the basic Math.
Yes.
A thermistor never hurts, of course, but probably you don't need one.
A short? .... definitely.
Scary big? .... did you actually calculate it?.
Fact is there is resistance, lots of it, in series with those diodes AND in series with that primary. (built in resistance).
1) measure transformer HV secondary resistance.
I would not be surprised at values of the magnitude of, say, 10/20 ohms, or thereabouts.
2) now measure mains/primary resistance ..... same thing.
Multiply it by square of voltage ratio to get impedance ratio so you get value of "reflected" primary impedance; in this case since it´s a 1:1 transformer, ratio is 1 so simply add primary to secondary resistance to get btotal series bresistance.
In this case I cheated
and calculated total resistance (based on transformer rated regulation) and found it to be almost 40 ohms. Yes, it´s a cheaply made transformer.
So your peak current surge will be around 120V*1.4142/40r=4.25A
Will in fact be somewhat higher seen from the Mains side, because transformer even if unloaded will also take some surge current to magnetize its core, but nothing to write home about.
But *please* measure and post results, mine are just estimations based on a datasheet (and having designed and wound a few thousand transformers
)2) as of: The plan is to use 1N4007s and 100 uF filter caps to make up a positive and a negative half-wave rectifier, each spitting out roughly 170 V DC. Grounding the (-170 V) rail leaves you with 0V, +170 V, and +340 V.
please don't.
I hate goofy voltage doublers for many reasons, and good news is you don´t actually need them here .
I think you have 120V mains.
In that case use secondary as 120V primary (transformers are *fully* reversible) , connect both 120V "primaries" in series, and feed a full wave bridge with 4 x 1N4007.
Put both filter caps in series and send transformer center tap you just created to cap center tap.
Instant cleaner +340 and +170V at your disposal and diodes will be happier.
And for peace of mind read 1N4007 datasheet surge capabilities in the datasheet, you´ll be surprised
I use salvage NTC thermistors at 100 W output. The 2.5 ohm ones salvaged out of ATX (PC) power supplies work fine at that low level.
You pick a thermistor from the family datasheet by figuring the full running current, then pick a unit rated for that and not objectionable in cold resistance.
The old equipment I re-purpose never had thermistors since they weren't invented then, but they make the behaivoir of the unit nicer without that big thump from the transformer. Also I use $3 450 V rated B+ capacitors (nichicon, panasonic) instead of the $15 525 v surge rated ones, with no failures yet. Even though my ST70 came with a 525 v rated capacitor, desogmed back when AC voltage was 115-117 instead of 125-127 like now.
NTC thermistor in primary of a vacuum tube amp also removes fear of "cathode stripping" due to SS rectifiers instead of slow to turn on vacuum tube rectifier. I can't prove cathode stripping is true, but not hammering the cathodes with instant B+ voltage is cheap - about $2 if you buy the thermistor and the cinch terminal strip to solder it to. New vacuum tubes are not cheap.
As an aside, I think 1n4007 are too short for B+ voltage rectifiers. Give them a good coating of dust, locate near the coast on a foggy night, see what happens. I can't buy longer package rectifiers, but there are plenty of dead computer displays on the curb to salvage them out of. Or old CRT TV's.
You pick a thermistor from the family datasheet by figuring the full running current, then pick a unit rated for that and not objectionable in cold resistance.
The old equipment I re-purpose never had thermistors since they weren't invented then, but they make the behaivoir of the unit nicer without that big thump from the transformer. Also I use $3 450 V rated B+ capacitors (nichicon, panasonic) instead of the $15 525 v surge rated ones, with no failures yet. Even though my ST70 came with a 525 v rated capacitor, desogmed back when AC voltage was 115-117 instead of 125-127 like now.
NTC thermistor in primary of a vacuum tube amp also removes fear of "cathode stripping" due to SS rectifiers instead of slow to turn on vacuum tube rectifier. I can't prove cathode stripping is true, but not hammering the cathodes with instant B+ voltage is cheap - about $2 if you buy the thermistor and the cinch terminal strip to solder it to. New vacuum tubes are not cheap.
As an aside, I think 1n4007 are too short for B+ voltage rectifiers. Give them a good coating of dust, locate near the coast on a foggy night, see what happens. I can't buy longer package rectifiers, but there are plenty of dead computer displays on the curb to salvage them out of. Or old CRT TV's.
Last edited:
Thanks for everyone's responses, much appreciated.
JMFahey's reply was particularly helpful - I hadn't expected that the winding resistances might be as high as tens of ohms. Today I measured the winding resistances of the Triad N68X - and I found 16.7 ohms on the secondary, 25.6 ohms on one primary, and 27.8 ohms on the other primary.
Those two primaries get wired in parallel for 120V operation, so there's roughly 13 ohms there, and another 17 ohms in the secondary. Thirty ohms total? Okay, now I'm no longer worried about that peak inrush current!
By the way - based on Tubelab (George's) measurements, the Triad N68x is not quite 1:1. It has fairly poor load regulation (15% according to the data sheet), and the manufacturer apparently decided to compensate by adding a few more turns to the secondary, so it still produces 120V at near its rated max load - but only when used the way the manufacturer intended, i.e., primary and secondary not swapped.
I considered the back-to-front (interchange primary and secondary windings) approach, but decided against it in order to gain a little more DC voltage. If the secondary has even 10% more turns on it, then we're talking about a roughly 21% difference in DC voltage depending on which way the transformer is used. If there really are 15% more turns on the secondary, then we're looking at about a 33% change in DC voltage.
Thanks again to everyone who replied.
-Gnobuddy
JMFahey's reply was particularly helpful - I hadn't expected that the winding resistances might be as high as tens of ohms. Today I measured the winding resistances of the Triad N68X - and I found 16.7 ohms on the secondary, 25.6 ohms on one primary, and 27.8 ohms on the other primary.
Those two primaries get wired in parallel for 120V operation, so there's roughly 13 ohms there, and another 17 ohms in the secondary. Thirty ohms total?
Okay, now I'm no longer worried about that peak inrush current!By the way - based on Tubelab (George's) measurements, the Triad N68x is not quite 1:1. It has fairly poor load regulation (15% according to the data sheet), and the manufacturer apparently decided to compensate by adding a few more turns to the secondary, so it still produces 120V at near its rated max load - but only when used the way the manufacturer intended, i.e., primary and secondary not swapped.
I considered the back-to-front (interchange primary and secondary windings) approach, but decided against it in order to gain a little more DC voltage. If the secondary has even 10% more turns on it, then we're talking about a roughly 21% difference in DC voltage depending on which way the transformer is used. If there really are 15% more turns on the secondary, then we're looking at about a 33% change in DC voltage.
Thanks again to everyone who replied.
-Gnobuddy
This is exactly what I'd been thinking, until JM Fahey's post prompted me to measure my transformer, which turned out to be crappier than I was expecting. With 30 ohms (!) of effective DC series resistance, my inrush current is limited to 4 amps, so maybe I don't need to do anything further.
I bought 450V caps as well.
Interesting idea - I wonder if the warmup time for an NTC thermistor is long enough to "soften the blow" for the tubes? I would think that would require tens of seconds of delay.
Tell me about it! They were pricey when I lived in the USA, then I moved to Canada, where my income dropped, the dollar is weaker, and prices for many things are higher.
I am very glad I moved to Canada, for reasons we're not allowed to discuss on this forum. But I don't see myself buying a lot more new tubes from now on. I'll just use DIY amps using old NOS tubes from the $1 list at ESRC Vacuum Tubes.
Mebbe the "two 1N4007 in series with voltage-sharing small ceramic caps in parallel" idea will help?
To my surprise, I have had no luck finding one - people have been diligent about recycling them here (toxic metal in the CRT), except for the occasional twit who thinks his old CRT TV is worth $300 on Craigslist.
I want a cheap or free old CRT TV because a friend wants me to make a Jacob's Ladder for demo use in the physics program at a local college, so I've been looking for a flyback transformer to generate the HV with. But that's another story, and another project!
-Gnobuddy
I see you got it. But for "scale":
120V limited by just the 13r is 9 Amps.
While pri and sec are normally different voltage and different resistance, the total resistance is normally minimized (at some price-point) so the relative resistance pri/sec is about equal. Which means a secondary "short" pulls like 4.6 Amps at most.
There's a thread in Lounge about heating coffee water. 600W small pot? 1,000W MrCoff? 1,600Watt microwave? These suck 5A to 13A, and for one to many minutes, not cap-charge time. Yet when you need coffee you do not fear "surge".
Instead of minutes, 30r charging 220uFd is 63% done in <10mS. Not even a eye-blink. No way a wire will hot-up that fast. mS response in a fuse is rare except at gross overload. (0.5A fuse at 10A.)
At an extreme: my well-pump is 120V and sucks 44 Amps for a part-second until the motor gets spinning some back-EMF. Yeah, I know the pump just started by the lamp-dim (too-long line to the house) but 44A surge is not a problem.
It's probably time to insert an image showing what your breakers are supposed to be doing. Note that for less than one second not even the fastest blows below twice rated current. Note that you can pull 1A from a 1A rated breaker all day and it should not trip.
Most breakers are C curve. For one cycle (1/50 or 1/60 seconds) you can pull a whack of overload without the breaker even getting mildly interested. And these breakers are there to protect your wires (not your semiconductors). Within a normal house the most you're going to be able to pull from a GPO will be around 1kA and even that won't (shouldn't) cause things to burst into flames (household stuff is usually rated to a few kA for short-circuit current - Iscc*)
The audiophile world had also discovered sag-induced IMD (though I doubt that anyone understood it) and had a phase where the search for a "stiff power supply" resulted in amp's filled with capacitors and toroids (e.g. "The power supply of the SA/12e consists of twin 1200VA toroidal transformers (with a short-term capability of twice that value) and 250,000µF of total filter capacitance") At the extremes, some of these required soft-start circuits but mostly it's not an issue for normal wiring and from-the-grid power (run off your UPS, well that's a whole different thing
)*this all varies from country to country and also on your line voltage.
Last edited:
Thanks to everyone for their input.
In retrospect, my use of the word "scary" was a very poor choice on my part.
I have not been lying awake wondering if the electrical wiring in my apartment was about to burst into flame - it was more a case of wondering what sort of long-term damage large inrush transient currents might inflict on filter caps and rectifier diodes.
-Gnobuddy
In retrospect, my use of the word "scary" was a very poor choice on my part.
I have not been lying awake wondering if the electrical wiring in my apartment was about to burst into flame - it was more a case of wondering what sort of long-term damage large inrush transient currents might inflict on filter caps and rectifier diodes.
-Gnobuddy
It's amazing how far beyond absurdity audiophile products tend to go.
From a little bit of reading I did on the subject of inrush currents, apparently a lot of today's switching power supplies use inrush current protection, because they contain large-value filter caps that are charged up rapidly by rectifying the incoming AC line voltage. There is no transformer in between, so there is negligible series resistance to limit inrush current.
-Gnobuddy
Diodes can be a concern. Many appreciate that valve diodes definitely require care in selection. And the UF4007 is not as well rated as the 1N4007, although a valve amp would need to be pretty potent to require more. ss amp peak currents are a little harder to define as circuit and part resistances are much lower.
Filter caps are typically only a concern from a ripple current perspective. i've not come across any degradation data relating inrush for common filter caps.
Yep - depending on the design, a valve rectified power supply that delivers steady current well below the maximum specified in the data sheet can nonetheless draw peak plate current well above the specified maximum during the first half second or so after switch-on. A bit of modelling using PSUD2 can reveal this even though the power supply seems basically OK.
I have seen this with guitar amps - the new rectifier works well for a number of gigs and then "phut" it dies on start up. Guitarists take to carrying a spare. One possible solution is to include a suitable NTC resistor on the transformer primary to reduce the inrush current at switch-on. Now I can't say for certain that this works, as I've not checked to see if the rectifiers still go "phut". But, I've not heard back and assume that either 1) it worked and the guitarist has no need to talk to me again or 2) it didn't work the guitarist doesn't want to talk to me ever again.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.