Assuming you start at a zero crossing you have 8 to 10 milliseconds of cycle, 60 or 50 Hz, to the next crossing. Since I am using 2.4 to 4 KVA toroids and large CLC storage I like to hold the in-rush back for at least a full 3 to 4 seconds for which you need thermistors or resistors, at least that is what I like to use as they are not expensive and are reliable. I really hate to have to walk to the power sub-panel to reset a circuit breaker every time I fire up an amp. Even though I have run 10 gauge copper wire to each amp AC outlet they are still only backed by a 20 amp breaker each.
That's true. You need to charge those cap banks downstream anyway, and that takes a bunch of current and some time.
The larger the cap banks and the beefier the transformer, the more in-rush there is and it really has to be tamed to keep it within acceptable limits.
I've looked at many commercial applications, and there are many ways to do it. Some don't even bother and do nothing at all to keep that down under control.
I found the way Carver does things a bit complex, and I'm not sure there is a real benefit there. It should certainly keep it down though.
My main project goal will have several power amps in a single rack, plus other things, and I want a global comprehensive power up system that makes it sequential and progressive.
I'm not a fan of electromechanical relays, but they can be used in addition to other means, like I've seen used in addition to Triac, when the Triac has done its job, a relay can additionally short it.
The tricky thing is that in-rush currents from a transformer and cap banks downstream aren't in sync, so it takes compromises.
A zero crossing power up may not be the way to go for a transformer alone, but with the cap banks after it, which have a different behavior, it's not too far off.
Perhaps using a ready made zero crossing detection to trigger a Triac may not be quite optimal, but still, doing the zero crossing detection can help, if it's then used to sync things up, perhaps by programming from a micro-controller that would trigger a Triac at a better time, using the zero crossing as the basis, plus a determined delay, which could be adjusted for best performance by programming.
In any case, I plan to use a micro-controller anyway, so it can do much more than a basic analog function. I want to sense the rail voltage after the cap banks to see when they've reached their steady state, and only after an extra small delay, would the sequence move on to the next set of PSU...
Something like this can actually monitor if something has gone wrong at some point and stop the power up sequence, reporting on the error, knowing where it happened, so it can then be easily and quicky tracked to be fixed, and nothing would be used if there was such a fault. If a channel with 8 or 10 amps has one of the amps faulty, I wouldn't want it to finish powering up, I would want it to stop where it failed and report it to me.
The larger the cap banks and the beefier the transformer, the more in-rush there is and it really has to be tamed to keep it within acceptable limits.
I've looked at many commercial applications, and there are many ways to do it. Some don't even bother and do nothing at all to keep that down under control.
I found the way Carver does things a bit complex, and I'm not sure there is a real benefit there. It should certainly keep it down though.
My main project goal will have several power amps in a single rack, plus other things, and I want a global comprehensive power up system that makes it sequential and progressive.
I'm not a fan of electromechanical relays, but they can be used in addition to other means, like I've seen used in addition to Triac, when the Triac has done its job, a relay can additionally short it.
The tricky thing is that in-rush currents from a transformer and cap banks downstream aren't in sync, so it takes compromises.
A zero crossing power up may not be the way to go for a transformer alone, but with the cap banks after it, which have a different behavior, it's not too far off.
Perhaps using a ready made zero crossing detection to trigger a Triac may not be quite optimal, but still, doing the zero crossing detection can help, if it's then used to sync things up, perhaps by programming from a micro-controller that would trigger a Triac at a better time, using the zero crossing as the basis, plus a determined delay, which could be adjusted for best performance by programming.
In any case, I plan to use a micro-controller anyway, so it can do much more than a basic analog function. I want to sense the rail voltage after the cap banks to see when they've reached their steady state, and only after an extra small delay, would the sequence move on to the next set of PSU...
Something like this can actually monitor if something has gone wrong at some point and stop the power up sequence, reporting on the error, knowing where it happened, so it can then be easily and quicky tracked to be fixed, and nothing would be used if there was such a fault. If a channel with 8 or 10 amps has one of the amps faulty, I wouldn't want it to finish powering up, I would want it to stop where it failed and report it to me.
I think you were thinking about a transformer all by itself, not loaded with big cap banks afterwards. Not quite the same.
Hey spookydd,
Your plans are ambitious and interesting but I wouldn't want to do a wiring harness for that, though a central controller and minimum wires to each amp is probably doable. You would probably want to concentrate your amps for easy control wiring but not required. I like my amps right next to the speakers for short speaker cable and long interconnects, but that is another whole discussion on it own about preferences.
Me, I prefer to walk to the amps and power them up or off one at a time, I know, very minimalistic and old school but it doesn't happen more than once a listening session. Now volume control and mute, give me a remote, which is what I designed in another posting.
Have a good one.
Your plans are ambitious and interesting but I wouldn't want to do a wiring harness for that, though a central controller and minimum wires to each amp is probably doable. You would probably want to concentrate your amps for easy control wiring but not required. I like my amps right next to the speakers for short speaker cable and long interconnects, but that is another whole discussion on it own about preferences.
Me, I prefer to walk to the amps and power them up or off one at a time, I know, very minimalistic and old school but it doesn't happen more than once a listening session. Now volume control and mute, give me a remote, which is what I designed in another posting.
Have a good one.
Me neither, and I hate too much wiring, so it's been the plan all along, to minimize wiring as much as possible, so my aim for amp design is each, with everything (PSU/Bridges, Caps...) on its own pcb, with the only wires being the short ones that are inevitable from the toroidal transformers to the pcb. Input and output plugs all on the pcb, so no wires. Making a single module with heatsinks and toroid, all-in-one.
Very much so and that's also the plan. There is no need for a bunch of wires, just the command wires from the controller to each amp module, so they can be turned on from "remote". I guess I will have a few wires in each of those links because I also want to return monitoring info back to the controller.
That's exactly what I'm aiming for. A single rack, with 4way xover and 4 amps, on wheels that can lock in place, so it's easier to move around when required, then locked in place once in its location, which I want right next to, or right at the back of, each 4way speaker system, with the shortest cables going to each speaker.
I have planned to not have ANY buttons/knobs or whatever on that rack, which is meant to be next to the speakers and somewhat far away from the mixer/equalizer. So the link from the equalizer has to be balanced, and the plan is to have a signal detection right at the entry point in the rack, so it can sense the presence of a music signal and proceed to power up everything properly in the rack.
That rack would not require any human intervention at any time once in its place and plugged in. Nothing to adjust or turn on. It would turn itself on when signal is there, and then after a determined delay of dead silence, it would proceed with a power down sequence, to end up with only a tiny micro-controller on a small button battery to stay alive to look for incoming signal to go ahead with a power up.
Well, it's a matter of preference and also the conditions on location.
I want it automated completely, so it's plug-in-and forget, at least for what's back there with the speakers. The only thing I want to turn on and off are the mixer and equalizer.
I've considered if it was worth putting the soft start stuff on the pcb for each amp (not the centralized controller, just what it remotely controls), which would also reduce the wiring, which is already minimized a lot, but I figured it' not really that much to have a small extra pcb right next to the toroid, and keep those glitchy signals away from the amp...
A lot more work to do, but hey, this project dates back to the early 80s, so I'm in no hurry. I already designed a 4way xover to test things before I design that xover as an integral part of the rack. I designed it as a stand alone "device" for now, so I can test things with it, see how it turns out, and still it would be useful independently afterwards as a regular xover. I'm done with a pcb, with the same design philosophy of no wiring if possible, which was achieved, and I'll be building it in the next few months. I do have the pcbs already, looking not too shabby, so far...
Spookydd, a transformer's inrush current (whether the very first time of connection is at a zero crossing or not) will typically be a peak with exponential decay over many cycles. The zero crossing aspect is a "red herring".
Any subsequent rectifier and large filter capacitor will also take a few mains cycles to charge up, as the transformer secondary is not an ideal low impedance path for high charging currents during the initial cycles where its primary winding and core flux are trying to settle to some semblance of symmetry.
I am all for the simple technique raised by Mona in post #20. That technique has been around for decades, and has a nice positive latching aspect to it. It can also be used for secondary side circuits needing surge management, such as huge capacitor banks and valve heater circuits. As always, the NTC's need to be selected for continuous fail-safe operation, and the relay contact has it pretty easy (even less stressful than AC1 rating).
In essence, you want an in-rush profile that just steps up to idle operating current and doesn't really peak in such a way as to require any upstream fuse or cb to get anywhere close to its blow/trip region. And that is mainly achieved by NTC selection.
Why an amp would want what appears to be a humongous power transformer (although the amp audio output rating is not disclosed), or a humongous level of filter capacitance, with no valid reason, is another matter. I'd be happy if there was a valid reason, with some test distortion results to show up the need, but this thread is up to 50 posts with no sign of a technical base.
Any subsequent rectifier and large filter capacitor will also take a few mains cycles to charge up, as the transformer secondary is not an ideal low impedance path for high charging currents during the initial cycles where its primary winding and core flux are trying to settle to some semblance of symmetry.
I am all for the simple technique raised by Mona in post #20. That technique has been around for decades, and has a nice positive latching aspect to it. It can also be used for secondary side circuits needing surge management, such as huge capacitor banks and valve heater circuits. As always, the NTC's need to be selected for continuous fail-safe operation, and the relay contact has it pretty easy (even less stressful than AC1 rating).
In essence, you want an in-rush profile that just steps up to idle operating current and doesn't really peak in such a way as to require any upstream fuse or cb to get anywhere close to its blow/trip region. And that is mainly achieved by NTC selection.
Why an amp would want what appears to be a humongous power transformer (although the amp audio output rating is not disclosed), or a humongous level of filter capacitance, with no valid reason, is another matter. I'd be happy if there was a valid reason, with some test distortion results to show up the need, but this thread is up to 50 posts with no sign of a technical base.
Last edited:
Hi,
That it is the way I do when do the ramping. The triac it is fired 6 times in sequence at the same phase angle to allow the capacitors to slowly charge up. Then the next face angle it is increase and repeat the cycle until it reach the AC voltage peak. The voltage will be slowly raise up until it reach the voltage peak. At that moment when it reach the peak the triac it is turn on full time. Also at that moment will wait for delay of few milliseconds then the triac is bypass with a relay contact because it no longer need it. The reason of bypassing the traic it will keep firing at the zero crossing and may caused electrical noise in the amplifier. It did the job of bringing the voltage up and longer need it. Also by doing the ramping you will minimizes the inrush current that occur when power on a device with a component like a transformer and a bank of discharged capacitors. This method as been working in my amplifier flawless for more than 4 years using an Arduino uno with zero failures. I advice anyone to built one and report back the finding. It does not take too much effort to built one. You will not regret building it.
Last edited:
to TRobbins: Quote(Why an amp would want what appears to be a humongous power transformer (although the amp audio output rating is not disclosed), or a humongous level of filter capacitance, with no valid reason, is another matter. I'd be happy if there was a valid reason, with some test distortion results to show up the need, but this thread is up to 50 posts with no sign of a technical base.)
The large toroid and capacitor bank has an audible effect, more effortless dynamics, I wish I had a distortion analyzer to show the reduction in higher order harmonics that more bias current brings, but there are articles on that out there. BTW my amps are 160 watts Class A into 8 ohms, has a 2.4KVA toroid, and .760f in a CLC config.
Also see:
Power Supplies | Pass DIY
But the thread is about soft starting the amp, the big power supply is just the reason why.
The large toroid and capacitor bank has an audible effect, more effortless dynamics, I wish I had a distortion analyzer to show the reduction in higher order harmonics that more bias current brings, but there are articles on that out there. BTW my amps are 160 watts Class A into 8 ohms, has a 2.4KVA toroid, and .760f in a CLC config.
Also see:
Power Supplies | Pass DIY
But the thread is about soft starting the amp, the big power supply is just the reason why.
After looking at the various methods and how far to go with the soft starting, I think I'll do it this way:
Since I'll be using a micro-controller to handle all the housekeeping and sequential power ups for everything, I might as well go all the way and do s few more simple steps.
- I'll insert a resistor after the bridges before the cap banks, to put a damper on the charging, which should make the transformer's inrush profile far more dominant at first
- so the initial inrush current will be much more transfo based and can be dealt with by powering it up with a short delay programmed in the micro-controller after it detects the zero crossing
- a resistor in the transfo's primary would greatly limit the inrush current, but the timing of the power up would also do so
- zero crossing is easy to detect and provide to a micro-controller, which info can then be used by the program to add a proper delay, and I'm thinking about also having at the same time a threshold level set by a divider and given to the micro-controller as well, such as is done by bryston on their amps. they don't use any limiting resistors, they just power up with a triac using zero crossing and a threshold
- after the initial power up with the delay, an other short delay added and short out the primary's limiting resistor, which will speed up the cap banks charging, still limited by the resistor there as well
- and then, after perhaps some 3 seconds or so, the cap banks limiting series resistor can also be shorted (by relay) to finish the charging
- a detection of the output rails levels on the cap banks can then be used to send a signal back to the micro-controller that it's all done, which allows moving on to powering up the next amp psu
The toroids won't all be the same size, but at the moment I think the largest will be something like 800VA or so, and probably 2 of the amps will have similar sized psu/cap banks.
This is a 4 way multi-amp, with a xover that would be powered up even before the amps, and all the amps would have to power up muted, and nothing would be unmuted until the whole chain would be fully powered up, with no failures. All fully automatic, no human intervention, triggered by the presence of a signal, and an absence of signal for several minutes would cause a complete shutdown of everything.
Since I'll be using a micro-controller to handle all the housekeeping and sequential power ups for everything, I might as well go all the way and do s few more simple steps.
- I'll insert a resistor after the bridges before the cap banks, to put a damper on the charging, which should make the transformer's inrush profile far more dominant at first
- so the initial inrush current will be much more transfo based and can be dealt with by powering it up with a short delay programmed in the micro-controller after it detects the zero crossing
- a resistor in the transfo's primary would greatly limit the inrush current, but the timing of the power up would also do so
- zero crossing is easy to detect and provide to a micro-controller, which info can then be used by the program to add a proper delay, and I'm thinking about also having at the same time a threshold level set by a divider and given to the micro-controller as well, such as is done by bryston on their amps. they don't use any limiting resistors, they just power up with a triac using zero crossing and a threshold
- after the initial power up with the delay, an other short delay added and short out the primary's limiting resistor, which will speed up the cap banks charging, still limited by the resistor there as well
- and then, after perhaps some 3 seconds or so, the cap banks limiting series resistor can also be shorted (by relay) to finish the charging
- a detection of the output rails levels on the cap banks can then be used to send a signal back to the micro-controller that it's all done, which allows moving on to powering up the next amp psu
The toroids won't all be the same size, but at the moment I think the largest will be something like 800VA or so, and probably 2 of the amps will have similar sized psu/cap banks.
This is a 4 way multi-amp, with a xover that would be powered up even before the amps, and all the amps would have to power up muted, and nothing would be unmuted until the whole chain would be fully powered up, with no failures. All fully automatic, no human intervention, triggered by the presence of a signal, and an absence of signal for several minutes would cause a complete shutdown of everything.
Simple is nice.
this thread is reminding me that my last concern before my next build is a soft start.
Though new to the art of soft start. A thought that passed my mind too. If a relay is going to be involved. then might as well have a circuit that would place very low current on the main on off switch. So its a interesting approach and i enjoyed the Original Posts design
Then again im lazy and just want simple. dirt dirt simple. But seems relay is more professional. just leaving a NTC thermistor in circuit would be horrible.
then im wondering if dealing with say 250 to 330 VA transformer. would it really be that incredibly horrible. I think 150 to 220 VA might be the limit on just leaving thermistors in circuit.
Anyhoo. If a triac was used with a Diac. Im curious if this automatically gives ideal or non ideal 0 crossing
this thread is reminding me that my last concern before my next build is a soft start.
Though new to the art of soft start. A thought that passed my mind too. If a relay is going to be involved. then might as well have a circuit that would place very low current on the main on off switch. So its a interesting approach and i enjoyed the Original Posts design
Then again im lazy and just want simple. dirt dirt simple. But seems relay is more professional. just leaving a NTC thermistor in circuit would be horrible.
then im wondering if dealing with say 250 to 330 VA transformer. would it really be that incredibly horrible. I think 150 to 220 VA might be the limit on just leaving thermistors in circuit.
Anyhoo. If a triac was used with a Diac. Im curious if this automatically gives ideal or non ideal 0 crossing
Hi,
Can someone please explain to me why with these soft start circuits they all use low ohm resistors? I modified all my amateur radio power supplies with the soft start resistor as 10k 50w, and all have not had the cracking thump at switch on (13.8v 25A). A relay bypasses this resistor when the voltage goes above 10v, and these have worked for twenty years with no problems. The question is, have I been doing this wrong?
regards john
Can someone please explain to me why with these soft start circuits they all use low ohm resistors? I modified all my amateur radio power supplies with the soft start resistor as 10k 50w, and all have not had the cracking thump at switch on (13.8v 25A). A relay bypasses this resistor when the voltage goes above 10v, and these have worked for twenty years with no problems. The question is, have I been doing this wrong?
regards john
A 10k resistor will have zero effect on mains.
However on the secondary of a valve supply will have an effect.
I suspect your getting mixed up between mains side soft starts and secondary side soft starts.
Hi, and thanks for answering. I actually have these 25amp amateur radio power supplies that are in use daily. To prevent problems with switch-on arcing and replacing mains switches, I put a 10k 50watt aluminium clad resistor in series with the mains transformer with a relay that bypasses this when the secondary volts goes over 12v. These have worked like a charm for best part of twenty years. I had thought of using a picaxe chip to switch relays, and as a voltage adc with timing, but so far I haven't needed to. I worked this out roughly to 24mA at 240v, and even at this current the transformer still gives a barely audible hum at startup. Without anything the power supply would almost move at switch on. They are Diamond gsv3000 linear power supplies, and are quite heavy. With this resistor the relay kicks in after about 2 seconds. I suppose for the sake of safety I should fit a thermal fuse in contact with these resistors, but they are never left unattended, and I did check on one recently and the resistor has not discoloured in any way. They are bolted onto the metal back plate of the power supply as a heatsink. Even if the relay failed to trip, the wattage amounts to just under 6 watts - well within limits.
The thing that interests me though, is why both systems appear to work ok, low value resistors, and the 10k I've been using.
Thanks again
john
The thing that interests me though, is why both systems appear to work ok, low value resistors, and the 10k I've been using.
Thanks again
john
Hi John20851,
The reason for using low ohm resistors is simply ohm's law. You want to limit the current to some maximum value. For 240v mains a 20 ohm resistance will limit the current in-rush to 12 amps, this is the value I choose, you can use different if you like.
The initial current in-rush will drop off exponentially as the capacitor bank is charged so you are able to bypass the resistance after a few seconds. A large resistance would limit the current but too much, charging the capacitor bank too slowly putting a strain on the toroid.
Thanks,
John
The reason for using low ohm resistors is simply ohm's law. You want to limit the current to some maximum value. For 240v mains a 20 ohm resistance will limit the current in-rush to 12 amps, this is the value I choose, you can use different if you like.
The initial current in-rush will drop off exponentially as the capacitor bank is charged so you are able to bypass the resistance after a few seconds. A large resistance would limit the current but too much, charging the capacitor bank too slowly putting a strain on the toroid.
Thanks,
John
Last edited:
Hi,
I agree there is quite a difference between 10 and 10r, thing is, it works. Four of them have been in almost daily use since around 1986 when I got my amateur radio licence. I would call 12amps at 240v quite a nasty surge, and not going to do switch contacts any good over a short time, the only other thing , these power supplies don't use a toroidal transformer. I did some experiments, seeing as I have quite a few 50w clad resistors of 1k to 15k in value. They all seemed to work, with obviously the 1k switching the relay the fastest. But it also made the transformer thump more when switched on. I measured the ac current with a 10k in series with the psu mains, and it was 25mA for about 2.5S then the bypass rely kicks in as the output voltage goes over 12v. I haven't had chance to test these on some 650watt audio amps I have with toroidal transformers, funny thing is they don't "crack" when switched on. Maybe they already have some system. I still don't like the idea of 12A surge, that's over 2.8kW albeit a short time.
regards
john
I agree there is quite a difference between 10 and 10r, thing is, it works. Four of them have been in almost daily use since around 1986 when I got my amateur radio licence. I would call 12amps at 240v quite a nasty surge, and not going to do switch contacts any good over a short time, the only other thing , these power supplies don't use a toroidal transformer. I did some experiments, seeing as I have quite a few 50w clad resistors of 1k to 15k in value. They all seemed to work, with obviously the 1k switching the relay the fastest. But it also made the transformer thump more when switched on. I measured the ac current with a 10k in series with the psu mains, and it was 25mA for about 2.5S then the bypass rely kicks in as the output voltage goes over 12v. I haven't had chance to test these on some 650watt audio amps I have with toroidal transformers, funny thing is they don't "crack" when switched on. Maybe they already have some system. I still don't like the idea of 12A surge, that's over 2.8kW albeit a short time.
regards
john