Dear Forum participants,
It has been quite a while now that I've been playing with the thoughts of making a new power supply,for
my ancient Quad405 amplifier.
I was thinking of replacing the bridge rectifier by a thyristor version.Ofcoarse one would require a controller,
to fire the thyristors,but that's not the issue I wanted to discuss here.
First of all,let me point out the benefits of a thing like this.
1) soft start : Firing the thyristors every period of the 50Hz a little earlier, the DC voltage will
ramp-up gently.This prevents the Thummmm at power up and avoids unnesesary stress for transformer and capacitors.
2) Auto on/off : since there is a microcontroller in control of it all,it will be easy to make a signal detector
and shut down the amp when there is no signal for a while,and power it up again when signal arrives.
3) emergency shut down : in case of a fault condition (DC on the speaker) the whole bridge can be shut down,limiting
the amount of destructive energy to the contents of the electrolytic capacitors.
4) a thyristor is hardly more expensive than a diode.
So that's for the benefits I personally see,but what about the downsides? what about EMI? I thougt of that too,
and it can indeed be a problem when cutting the sine, but that's only during the first seconds at startup.
In the "ON" phase i will have the thyristors fed with a continuous gate current,effectively turning them into
plain rectifier diodes.
The fact is, I see a lot buzzing around this forum about rectifier bridges,building them with stealth or
soft recovery diodes,but nothing on thyristors.Has anyone done this before?
I made a simulation of the results see attach.
It has been quite a while now that I've been playing with the thoughts of making a new power supply,for
my ancient Quad405 amplifier.
I was thinking of replacing the bridge rectifier by a thyristor version.Ofcoarse one would require a controller,
to fire the thyristors,but that's not the issue I wanted to discuss here.
First of all,let me point out the benefits of a thing like this.
1) soft start : Firing the thyristors every period of the 50Hz a little earlier, the DC voltage will
ramp-up gently.This prevents the Thummmm at power up and avoids unnesesary stress for transformer and capacitors.
2) Auto on/off : since there is a microcontroller in control of it all,it will be easy to make a signal detector
and shut down the amp when there is no signal for a while,and power it up again when signal arrives.
3) emergency shut down : in case of a fault condition (DC on the speaker) the whole bridge can be shut down,limiting
the amount of destructive energy to the contents of the electrolytic capacitors.
4) a thyristor is hardly more expensive than a diode.
So that's for the benefits I personally see,but what about the downsides? what about EMI? I thougt of that too,
and it can indeed be a problem when cutting the sine, but that's only during the first seconds at startup.
In the "ON" phase i will have the thyristors fed with a continuous gate current,effectively turning them into
plain rectifier diodes.
The fact is, I see a lot buzzing around this forum about rectifier bridges,building them with stealth or
soft recovery diodes,but nothing on thyristors.Has anyone done this before?
I made a simulation of the results see attach.
Well Toino, this is not what I had in mind,but this design uses a plain diode bridge for rectification and a triac for softstart,besides this it requires a small mains transformer (TX1) to power the softart circuit.
I was hoping to find anyone who had some experience with thyristor bridges for power amplifiers,Or else if it behaves badly I'd like to know why.
I was hoping to find anyone who had some experience with thyristor bridges for power amplifiers,Or else if it behaves badly I'd like to know why.
Cowboy, it is possible to do the same in a bridge with thyristors. You must use a similar phase control scheme, and it’s the reason I have posted the link.
I use that kind of bridge in a big Pa. amplifier and it works without any problems.
Before that I have tried all the conventional ways with lots of problems: exploded NTC or PTC, burned contact relays, etc.
Anyway don’t forget that you have the power capacitors charged and turn-off or shut-down is never instantaneous.
See page 68/69 of this one: http://www.meyersound.com/pdf/products/legacy/ms-1000a.pdf
Page 68 the thyristor here is only on-off switch but is possible to simplify, modifying the bridge with thyristors; then you have full phase control… if you need or want.
On last page you have a circuit to force capacitor discharge.
More on soft-start here: http://www.diyaudio.com/forums/showthread.php?t=89576
I use that kind of bridge in a big Pa. amplifier and it works without any problems.
Before that I have tried all the conventional ways with lots of problems: exploded NTC or PTC, burned contact relays, etc.
Anyway don’t forget that you have the power capacitors charged and turn-off or shut-down is never instantaneous.
See page 68/69 of this one: http://www.meyersound.com/pdf/products/legacy/ms-1000a.pdf
Page 68 the thyristor here is only on-off switch but is possible to simplify, modifying the bridge with thyristors; then you have full phase control… if you need or want.
On last page you have a circuit to force capacitor discharge.
More on soft-start here: http://www.diyaudio.com/forums/showthread.php?t=89576
Looking at the Meyersound schematics,they use a diode bridge AND thyristors.
So either they didn't bother OR they have a good reason to use a plain diode bridge instead of a thyristor bridge.Heat dissipation is worse, and the thyristors carry the double amount of current.
I think I'll just have to build it and see how it performs.At least in simulation it performs very well.
So either they didn't bother OR they have a good reason to use a plain diode bridge instead of a thyristor bridge.Heat dissipation is worse, and the thyristors carry the double amount of current.
I think I'll just have to build it and see how it performs.At least in simulation it performs very well.
A simple approach is to place a saturatable induction between your output rectifier and your output diodes. During initial start up the induction will have a high impediance that reduces the sudden inrush current. Once its saturates the impediance will drop to near zero. it will also help reduce output noise.
Another option would be to use buck regular with a soft start option that is between your rectifier and your output caps:
http://cds.linear.com/docs/Datasheet/3724fc.pdf
The Buck regulator would also add voltage regulation to your output. I think the Buck Regulation option would be much simpler to implement than something using Thyristors.
Best of Luck to you!
Another option would be to use buck regular with a soft start option that is between your rectifier and your output caps:
http://cds.linear.com/docs/Datasheet/3724fc.pdf
The Buck regulator would also add voltage regulation to your output. I think the Buck Regulation option would be much simpler to implement than something using Thyristors.
Best of Luck to you!
Yes star882, i am familiar with the NTC,but it lacks the feature of being able to turn the amp off.
and TechGuy neither of your suggestions is viable in my opinion.When starting up the current will be high,so the inductor saturates immediately. So the effect id the opposite of the desired.And the buck... will require an electrolyte at its input: will we fill it up with a regular bridge without softstart ?
Everybody gives me alternatives, but there is really no argument against the thyristor bridge.Complex ?At first I thought so too,but i was able to simplify a lot.It needs only 4 small transistors and 2 dual diodes for a single bridge and 6 transistors for a dual version.
I do not count the thyristors themselves for they are substituting a diode anyway.
In attachment is the schematic I used for simulation.Please note that the firing circuit is the same for either branche except for the polarity.
and TechGuy neither of your suggestions is viable in my opinion.When starting up the current will be high,so the inductor saturates immediately. So the effect id the opposite of the desired.And the buck... will require an electrolyte at its input: will we fill it up with a regular bridge without softstart ?
Everybody gives me alternatives, but there is really no argument against the thyristor bridge.Complex ?At first I thought so too,but i was able to simplify a lot.It needs only 4 small transistors and 2 dual diodes for a single bridge and 6 transistors for a dual version.
I do not count the thyristors themselves for they are substituting a diode anyway.
In attachment is the schematic I used for simulation.Please note that the firing circuit is the same for either branche except for the polarity.
You've got it backwards. The inductor resists change in current, until it saturates, during which time, the inductor acts like a choke to prevent excessive current flow. The only issue would be if you have a huge cap that swamps the inductor so that after it saturates the cap still isn't charged enough to prevent excessive input current A inductor with a higher inductance can offset a larger cap. The only issue is that a really large cap bank would require a very large and expensive inductor.
For the buck regulator, you would need a relativily small capacitor to prevent excessive ripple on the output. I suppose during the initial charge up of your filter caps their would be an elevated ripple on the output if you use too small of a soft-start RC value. Once the filter cap is charged your using the caps to power the equipment and the buck is just keeping the caps topped off.
Another option is to use a one-shot PWM that starts off with a low duty cycle and finished off with a 100% duty cycle and remains at 100% until the power supply is switched off. You would need a PWM controller, and inductor, a MOSFET and a high-side mosfet driver unless you can find a PWM with an integrated mosfet controller. You would use a RC circuit to program the PWM duty cycle ramp up, an easier option might be to use small PIC microcontroller to avoid designing a RC circuit to program the PWM duty cycle.
The problem I see with a thyristor bridge is that it will difficult to turn them on after the peak voltage is reached. If you look at a typical thryristor bridge. You can design a simple circuit to turn on after the AC zero voltage crossing reaches a selected voltage and before peak voltage is reached (see gif image below). But you want to turn on the thyristor well after the peak voltage so that the input voltage remains low during startup. If you turn on at the peak or before the peak, your excerbated your problem because the input voltage at initial start up will be higher causing even higher current flow. You need a circuit that turns on with a low voltage, and turns off when the voltage rises to point it causes excessive current flow. Unfortunately you can't turn off thyristors, you need to wait until the voltage drops to zero before they can be turned off. Hopefully my writing is clear enough for you to grasp my point.
http://upload.wikimedia.org/wikipedia/commons/archive/0/07/20070324214637!Regulated_rectifier.gif
My take is that each thyristor needs a floating power supply. Too, it might be advantageous to regulate the primary side of a transformer, especially in the case of a toroid.
I have decided to use MOSFETs instead. They can be switched off as soon as a current threshold is reached. I have found it simpler to set up a control circuit and feedback loop under that circumstance where there is no turn-off delay as with thyristors. MOSFETs also need only very low current gate control supplies.
The best way I have found is to use 4 MOSFETs total. I have set up a circuit that uses two pairs of two MOSFETs that are connected source to source to permit dual polarity operation of each pair.
One pair of MOSFETs operates at a time. Usually, the pair that connects the transformer primary winding to the mains hot wire is operating by itself unless current limiting is operating. When the main pair of MOSFETs is off, so as to cut off the AC input, the other pair turns on to short the transformer primary. It serves to reduce noise and spikes.
Just another possibility. I see how thyristors can be a good option as well. My project, as it presently is, uses them.
I have decided to use MOSFETs instead. They can be switched off as soon as a current threshold is reached. I have found it simpler to set up a control circuit and feedback loop under that circumstance where there is no turn-off delay as with thyristors. MOSFETs also need only very low current gate control supplies.
The best way I have found is to use 4 MOSFETs total. I have set up a circuit that uses two pairs of two MOSFETs that are connected source to source to permit dual polarity operation of each pair.
One pair of MOSFETs operates at a time. Usually, the pair that connects the transformer primary winding to the mains hot wire is operating by itself unless current limiting is operating. When the main pair of MOSFETs is off, so as to cut off the AC input, the other pair turns on to short the transformer primary. It serves to reduce noise and spikes.
Just another possibility. I see how thyristors can be a good option as well. My project, as it presently is, uses them.
Using a set of MOSFET for rectification is know as "synchronous rectification". It typically used in high current\low voltage power supplies to reduce losses over standard diode rectifiers. You will probably need to use a snubber to protect the MOSFETs from voltage spike if the load is inductive.
I think it still going to be a complex to get the timing right using a rectification soft-start mechanism but going with MOSFET should make your job considerably easier than with Thyristors.
Best of luck!
I have compared that bidirectional MOSFET way and another way that uses just one MOSFET along with a bridge rectifier. The first way does good at preventing spikes when an added snubbing capacitor can absorb the spikes until the primary transformer shorting MOSFET pair turn on. I used just a capacitor without a series resistor and used pretty slow gate turn on. It worked well and the toroid did not hum much. The capacitor value could be as low as about 1/2 uF while the way using just one MOSFET needs about a 20uF capacitor to absorb the spikes, and that is with just a resistive load on the transformer. I don't think the second way is good for audio.
The second way with the single MOSFET and bridge rectifier isn't able to keep primary winding flyback under control, as far as I know, but it was simpler and fed the toroid with AC very evenly. It might be OK with purely resistive loads at the transformer output, which isn't the case with an audio power supply which charges filter capacitors.
I am now building and planning to use the second circuit because it is simpler and I am using it to power a resistive heating element.
Thanks!
Hello TechGuy,
Thyristor triggering is not after the zero crossing,but a bit before.The simulation results
in the attachment start with a firing angle of 177 degrees,and every cycle 100uS earlier.
Until the moment that the duty cycle of the trigger pulses is 100% (permanently triggered).
So this is actually the PWM that controls your mosfet.The full output voltage is reached at the point
where the duty cycle is 50%.This is hardly complicated,when implemented in a uP.
When the Thyristor gates are permanently activated,the thyristors act like a diode,so actually at normal
operation they work like diodes in a graetz bridge.
Thyristor triggering is not after the zero crossing,but a bit before.The simulation results
in the attachment start with a firing angle of 177 degrees,and every cycle 100uS earlier.
Until the moment that the duty cycle of the trigger pulses is 100% (permanently triggered).
So this is actually the PWM that controls your mosfet.The full output voltage is reached at the point
where the duty cycle is 50%.This is hardly complicated,when implemented in a uP.
When the Thyristor gates are permanently activated,the thyristors act like a diode,so actually at normal
operation they work like diodes in a graetz bridge.
If you want to softly charge up the caps, than you want to turn on the bridge near the zero-crossing voltage and turn off the bridge as the input voltage rises. See my very terrible drawing below:
An externally hosted image should be here but it was not working when we last tested it.
your idea to turn off past the the peak voltage is going to be difficult as I discussed in my previous post. Normally you could just use a voltage compartor or a zenor to turn on\off the input voltage, but you need to include some timing so that it turns on only during the downslope, which is probably going to be more complicated then you want to be. I think an with an analog circuit you will run into problems with input line noise that false trips the softstart, or the inductance spikes when you turn on the Thryristors before the zero crossing. If you get it wrong, then you're going to exacerbate your problem.
Yes you can go with a Microcontroller, but you did say you wanted something simple right? Now you need into include a 3.3V/5V PS and components to isolate or protect the ucontroller from EMI, that can fry it. The uController Code is very likely going to me more complicated than you think it is because your code must accomidate voltage spikes created by input noise, the transformer and the load itself. You haven't simplified your effort over any of my suggestions.
The only last option I can come up with is to just turn on at the zero crossing, using a zero crossing detector. Perhaps that would be sufficient to minimize current inrush surges.
Well, I am done covering the subject. Best of luck to whatever you decide.
Nice lively conversation.
I wanted to avoid controlling the primary side of the transformer,some do it with triac's others do it with solid state relays like electrone (back-to-back mosfets) since this has the downside of requiring an auxilary power supply to feed the control circuit.This may be a capdrop circuit or a small transformer,anyway I wanted to avoid this,and all safety issues that come along with it.
I want to keep the transformer permanently on the grid,which in my case is a good thing since it
consumes only 1.4Watt in idle.For others with worse performing transformers this might not be an option.
With the transformer permanently powered,I'll have a low voltage permanently available to power the uP.
I want this uP to monitor the amp's input signal in order to decide whether to turn it on or off.
And about programming uP's :This may be a hurdle or barrier for one,and a breeze for another,it all depends on the amount of experience you've got with those components.Since it is my daily job,I've got access to all kind of tools,and basic software frameworks,that are ready to go.
The only really valid argument against it, is the EMI type of trojan horse that one might drag into one's audio equipment.Careful PCB design,decoupling and filter techniques are the tools that combat this kind of evil.
I mean,almost every commercially available audio product,being a CD player,preamp,AV receiver has a uP aboard. They all solved this problem (in more or less extend).
A MOSFET between the bridge and the capacitor may be an option,In my case I would require 2: of them
1 for the positive and 1 for the negative rail.TechGuy's solution with terrible drawing may even be temptive,
It requires a voltage comparator and a ramp generator.But these are just the beginning.You will
require certaily some flipflops and gates too,and some timers too,to decide wheter to turn the amp on or off.
And as we all know:these circuits are as flexible as a brick.This requires breadboarding (with poor layout as a consequence)
and rewiring.All this to avoid a highly flexible and cheap uP.Implemented in a uP the thyristor triggering is simple:
Wait for the zero crossing and start a timer,then trigger the thyristor. and then evry cycle decrease the timer value,
until the full voltage is reached,after this turn on the trigger permanently.
If you want to avoid a uP at all cost,my scheme can be difficult to implement indeed.
In my humble opinion,it is a good thing to avoid components carrying heavy currents.
They dissipate heat,require cooling,wide PCB tracks and so on.In a Graetz bridge there are 4 diodes carrying heavy current: you just can't avoid that.If I just substitute these diodes by thyristors, I will not generate any extra heat,and have no extra components carrying heavy current.All I need is to add a humble control circuit.
And electrone - referring to the schematic I posted on 26th august.this works,and no floating power supply
is needed.Simulate it and you'll see.Since Q3 and Q6 are configured as a current source,they will provide
the same trigger current regardless of the height of the voltage at Vpos and Vneg.
It might be a good idea to add a transformer to the simulation model, in order to see what kind of spikes will be generated.
I wanted to avoid controlling the primary side of the transformer,some do it with triac's others do it with solid state relays like electrone (back-to-back mosfets) since this has the downside of requiring an auxilary power supply to feed the control circuit.This may be a capdrop circuit or a small transformer,anyway I wanted to avoid this,and all safety issues that come along with it.
I want to keep the transformer permanently on the grid,which in my case is a good thing since it
consumes only 1.4Watt in idle.For others with worse performing transformers this might not be an option.
With the transformer permanently powered,I'll have a low voltage permanently available to power the uP.
I want this uP to monitor the amp's input signal in order to decide whether to turn it on or off.
And about programming uP's :This may be a hurdle or barrier for one,and a breeze for another,it all depends on the amount of experience you've got with those components.Since it is my daily job,I've got access to all kind of tools,and basic software frameworks,that are ready to go.
The only really valid argument against it, is the EMI type of trojan horse that one might drag into one's audio equipment.Careful PCB design,decoupling and filter techniques are the tools that combat this kind of evil.
I mean,almost every commercially available audio product,being a CD player,preamp,AV receiver has a uP aboard. They all solved this problem (in more or less extend).
A MOSFET between the bridge and the capacitor may be an option,In my case I would require 2: of them
1 for the positive and 1 for the negative rail.TechGuy's solution with terrible drawing may even be temptive,
It requires a voltage comparator and a ramp generator.But these are just the beginning.You will
require certaily some flipflops and gates too,and some timers too,to decide wheter to turn the amp on or off.
And as we all know:these circuits are as flexible as a brick.This requires breadboarding (with poor layout as a consequence)
and rewiring.All this to avoid a highly flexible and cheap uP.Implemented in a uP the thyristor triggering is simple:
Wait for the zero crossing and start a timer,then trigger the thyristor. and then evry cycle decrease the timer value,
until the full voltage is reached,after this turn on the trigger permanently.
If you want to avoid a uP at all cost,my scheme can be difficult to implement indeed.
In my humble opinion,it is a good thing to avoid components carrying heavy currents.
They dissipate heat,require cooling,wide PCB tracks and so on.In a Graetz bridge there are 4 diodes carrying heavy current: you just can't avoid that.If I just substitute these diodes by thyristors, I will not generate any extra heat,and have no extra components carrying heavy current.All I need is to add a humble control circuit.
And electrone - referring to the schematic I posted on 26th august.this works,and no floating power supply
is needed.Simulate it and you'll see.Since Q3 and Q6 are configured as a current source,they will provide
the same trigger current regardless of the height of the voltage at Vpos and Vneg.
It might be a good idea to add a transformer to the simulation model, in order to see what kind of spikes will be generated.
Hi Everyone..
What I need to do is have a diode bridge circuit that will rectify 110 VAC to ~100VDC. They part im struggling with is that after 5 seconds or so I need to have to voltage dropped to 55VDC. I think I can do this by bypassing one of the diodes in the bridge circuit and going to half wave but i'm not sure how to do that. I need the package to be small and compact. The current draw of the circuit is about 10A. Any ideas. Thanks in advance.
What I need to do is have a diode bridge circuit that will rectify 110 VAC to ~100VDC. They part im struggling with is that after 5 seconds or so I need to have to voltage dropped to 55VDC. I think I can do this by bypassing one of the diodes in the bridge circuit and going to half wave but i'm not sure how to do that. I need the package to be small and compact. The current draw of the circuit is about 10A. Any ideas. Thanks in advance.
Had once a application note from motorola from the 70th for a regulated power supply with thyristors. Cant remember the details other than it was 50V 10A used forced thyristor turn off and was highly efficient. Back then, I wanted to build me a lab supply like this but I never found the time. Losses in thyristors are somewhat higher in SCRs compared to diodes.
dv/dt is you enemy here, while SCR controlled power supplies dominate the high power scene the high dv/dt when the SCR switches needs to be adequately filtered. For a soft starter they are unnecessary, if it is good enough to start a 400 kVa supply with 100kJ of filter capacitors using a contactor and resistors then a relay and resistor going to be adequate for any home equipment.
On a side note some of the highest performance SCR controlled power supplies are the dipole magnet power supplies for synchrotrons they can hold their constant current output stable to a few parts per million. Ripple is also in the parts per million helped in no small part by the magnet inductance.
Primary side control is rare in industrial supplies, there are some welders and old computer equipment power supplies which use primary side control, the transformers are usually specials with a magnetic shunt to increase the leakage inductance.
On a side note some of the highest performance SCR controlled power supplies are the dipole magnet power supplies for synchrotrons they can hold their constant current output stable to a few parts per million. Ripple is also in the parts per million helped in no small part by the magnet inductance.
Primary side control is rare in industrial supplies, there are some welders and old computer equipment power supplies which use primary side control, the transformers are usually specials with a magnetic shunt to increase the leakage inductance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.