what is the best way to switch between taps on a transformer?

[carbon dude oxide]

hello,

if i had a transformer with multiple taps such as this 12-6-0-6-12 and would like to use a single tap at a time with a shared output, so i can select 12V or 6V. which would be the best way of doing it?

would it be best to have relays on each matched tap output and select them that way? or would using triacs be another way of doing it? is there any other way which would be better for this situation?

it would also need to be able to handle 4A on the highest voltage tap so i am unsure weather or not this will effect some of the solutions due tot here being too much current. currently the relay solution is seaming like the best idea so far but i would like to know some other solution to this

[Jebnor]

I'm thinking some form of relay.   Double pole, double throw (DPDT) type.  Position 1 could be 6V and Position 2 could be 12V. Or vice versa, I'm not your mom.

Check your friendly local Auto-parts store.  Relays for 12-24 at >4A are not that uncommon. Also they come with convenient (albeit big) spade connectors for wiring.

Kind Regads,

[carbon dude oxide]

i know your not my mum...

i just wanted to know if there were better solutions.

i need to find one that can be switched by a microcontroler, it should not be too hard to find

[mariush]

Any relay can be switched by a microcontroller, just slap a npn transistor or a darlington in the circuit and you're good to go.

In fact I just discussed about this on another forum a few hours ago so I cropped a picture from a datasheet, where a relay is switched without using a microcontroller :



picture from a 18v 1A linear power supply schematic ...

78l12 - 12v 100mA ldo, 1008 is probably ksc1008 but basically any similar npn transistor will do.. k1a+k1b is a 12v relay...
if you use a microcontroller, you just connect the base of the transistor to a microcontroller pin through a resistance so you won't burn up the npn ...

[SeanB]

If you have a half bridge and want a switchable rail I have seen 2 thyristors and 2 diodes being used. Thyristors turned off it connects the 6V via the diodes and if you turn the thyristors on it turns on the 12V. The thyristors need a control that floatd at the positive rail though, so needs a higher voltage rail to drive the 2 gates together, using an optocoupler to switch them on and off a needed. Was used in a few power supplies to switch between multiple taps very fast with no noise. Just needed thyristors rated for operation at reverse bias with gate current, otherwise they do suffer from degradation with time.

[IonizedGears]

Why hasn't anyone suggested a rotary switch yet? Seems perfect for the application...

[carbon dude oxide]

I firgot to mention that it needs to be controled by a micro in my original post, i did say in a post after it. If it was manual selection then yes a rotary switch would be quite good

[Paul Price]

Use a N-chan power MOSFET and PWM the negative return side of the bridge rectifier to give you the exact pre-regulated voltage to feed your regulator chip that you want for proper operation of your regulator.  A simple MCU circuit could use two channels to A2D the output voltage and input voltage and set  the voltage difference between the input an output of the linear regulator to get just the right amount of Vin to the reg. chip above dropout.

So no relays, just use the MCU to set the optimum voltage.

You would use an optocoupler to turn on/off the N-Chan MOSFET driven by a bjt NPN-PNP two-transistor complementary pair. This circuit would also require a resistor-diode-zener diode clamp to protect the MOSFET gate from over voltage.

[Paul Price]

The circuit diagram supplied by Mariush will not work. The 7812 portion of the schematic is not only floating..it is really floating!

[mariush]

I didn't build that personally, but it's cropped from a datasheet of an existing power supply, so I doubt it's wrong. Maybe you could explain why you think it's really "floating" instead of just saying.

It certainly looks much better and simpler compared to your suggestion... way to go involving mosfets and pwms and all that crap just to switch between two secondary windings. 

It's a "beginners" section of the forum after all.

[Paul Price]

Here's where the confusion starts:

Why R7 and C4 are connected to -VSS at all is somewhat of mystery to me, perhaps because of the lack of a designated return path connection.


The 78L12 circuit has only one connection to the  adjacent circuitry, so it is inactive, floating, really floating. You mention a connection to a MCU, but connecting an output pin of an MCU directly to the electrolytic cap would not work well with the MCU without a current limiting resistor and the resistor values you have set here are quite low and would require a even lower resistor from an output pin to the base, overloading the MCU output pin current limit, but you still have the floating problem, you haven't designated the return path (MCU ground) to the MCU control pin circuit clearly(though it is easy enough for me to guess.) 

R6 would, without MCU intervention, activate the relay (if there was a designated ground reference point..I assume it is negative terminal of the 78L12) and it would be the job of the MCU to pull down this current current  and this requires a small-valued current limiting resistor to the MCU pin and yet it must pull the base voltage below .6V and this is also troublesome  because of the large capacitor on the base.

If you have approx 11.3V across the 561 ohm resistor the MCU pin would have to sink >20mA and also discharge the capacitor through a current limiting resistor. So the resistor would be very small an value to pull the base below.66V and the MCU pin would be overloaded trying to discharge the capacitor upon instantaneous activation of the pin(probably causing it to switch from 0 to 1 state), but trying to fix this by increasing the MCU output control pin resistor would not pull the base below .7V to shut off the relay.

The relay could be just as easily be connected to the main high power transformer rectifier-filter cap + power point with perhaps just a current limiting resistor and a simple npn transistor to activate the relay,  base of npn transistor to MCU output pin, emitter to ground.

Perhaps I am just a beginner, but I cannot see there is really a need for the 78L12 or two secondary transformer windings just to have a power source to operate the relay.

Also, C1 is a 104  that just .1uF  not enough capacitance to accomplish anything to prevent arcing of the relay where it is placed, but there should maybe be two, both on the transformer side of the relay to help absorb the inductive kick, because that is where the inductance is.

R5 is probably not serving any useful purpose(except as a bleeder) and it must be several watts in size to prevent it from burning up.

C3 is only .1 uF, a rather very small output filer capacitor for a power supply.


R1 and R2 are attenuating the gain of the regulating or control circuit you are attempting to create here. Rather ingenious but perhaps not optimal strategy to accomplish something that is perhaps not quite obvious to me.

If R1 is to connect to a circuit to control the output of the power transistor, then it must be supplied with a voltage that is always slightly higher than the output of the supply..not quite clear how efficient this circuit is to do this, I donno how well this circuit configuration might accomplish this with what I see here, considering the R1  attenuating resistor. I can't see how you get this control voltage.

Maybe I am just a beginner here, not seeing the big picture, the whole schematic.

[edavid]

Use a N-chan power MOSFET and PWM the negative return side of the bridge rectifier to give you the exact pre-regulated voltage to feed your regulator chip that you want for proper operation of your regulator.  A simple MCU circuit could use two channels to A2D the output voltage and input voltage and set  the voltage difference between the input an output of the linear regulator to get just the right amount of Vin to the reg. chip above dropout.

This is a bit tricky, because if you turn on the MOSFET at a random time there is no current limiting, and things tend to blow up. All the pre-regulated designs I've seen use a largish inductor to limit the current.

I suppose you could turn on the MOSFET at the zero crossing, and then turn it off when the filter cap is at the right voltage.

Is it really simpler than tap switching?

[rsjsouza]

The 78L12 circuit has only one connection to the  adjacent circuitry, so it is inactive, floating, really floating.

Yep, and poor Q3 will be really saturated at power up with 20mA on its base ((12-0,6)/561 = 0,02A)!

There must be a mistake on the drawing... 

[Paul Price]

Edavid,

Power MOSFETs can handle large instantaneous currents and the internal resistance of the transformer secondary wiring combined with the resistance of the bridge rectifier and ESR of the filter capacitor itself will definitely limit the current through the the MOSFET to a safe value the MOSFET can handle.  The MOSFET RDSon might be chosen to be as small as one to a few milli-ohms and an inexpensive MOSFET can be easily found that can handle 180amp surge currents with a 18-70Amp continuous Source-Drain current rating with RDSon <.002 ohm. From the pass transistor shown here and seeing the 3300uF filter cap. I assume the power transformer would be relatively small in size, hence have a fairly low current max. it could deliver.  Also, duty cycle would limit MOSFET power dissipation and instantaneous current as the capacitor is always charged near optimal voltage and charging could begin at zero-crossing and with PWM duty-cycle limiting.

[Paul Price]

edavid,

IMHO this is much better than tap switching because a relay is a mechanical device that has contacts that will erode over time or possibly fuse together, the relay controller might find a point that the relay oscillates or else make noise and chatter, it will use extra wasted power to activate the relay, certainly not always provide the optimal lowest voltage for lowest power dissipation of the linear regulator to reduce size and heat and the ability to achieve the largest regulated current output, require a much larger heatsink for the pass transistor, and create EMI switching the transformer voltages on and off.

Finally, and maybe even more important limitation of a relay is the activation time of a relay, too slow to adapt to a sudden load change and cause the power supply to momentarily fall of regulation.

[edavid]

IMHO this is much better than tap switching because a relay is a mechanical device that has contacts that will erode over time or possibly fuse together, the relay controller might find a point that the relay oscillates or else make noise and chatter, it will use extra wasted power to activate the relay, certainly not always provide the optimal lowest voltage for lowest power dissipation of the linear regulator to reduce size and heat and the ability to achieve the largest regulated current output, require a much larger heatsink for the pass transistor, and create EMI switching the transformer voltages on and off.

Relay oscillation doesn't seem to be a real world problem.

PWM will create more EMI than tap switching (and more noise in the PS output).

Otherwise, good points, but I wonder why all of the linear power supply makers seem to use tap switching.  I think the repeated surge current is a real issue.

Finally, and maybe even more important limitation of a relay is the activation time of a relay, too slow to adapt to a sudden load change and cause the power supply to momentarily fall of regulation.

This is not an issue unless the load changes so much that the supply switches from constant current mode to constant voltage mode, which is not the usual case.

Anyway, PWM has the same problem.  If you've missed this cycle's peak, you can't charge the capacitor until the next cycle.  If this mode is important, you need a real switching regulator.

[Paul Price]

Edavid,
I really doubt that PWM should cause any noticeable noise problem at all with a simple r-c snubber and perhaps a small inductor core wound with a few turns going from the bridge to the MOSFET and and its snubber, just a core like you see on some power supply wallwarts that are used to prevent EMI.

In contrast the large spark of relay contacts opening and closing can be quite an EMI event. There is no reason to have any switching PWM noise to reach abd affect the output of a properly designed power supply.

Surge current is also a problem with relay contacts.

PWM noise can be easily squelched because it is at high frequency and this means that it can be effectively dealt with using small inductor/cap filters and by controlling rise/fall times of the switching elements.

In a lab bench supply it is easy enough to filter voltages with small inductors and bypass caps and to physically separate or shield the switching and linear circuits to prevent any significant noise from reaching the output.

The time for a relay to activate/de-activate can be in the 10's of millisceconds to respond to a surge. A PWM supply can react instantaneously in the same AC or next AC cycle. In the same situation, the relay has missed several cycles and peaks to react. A MCU controlling the PWM can detect and adapt to this situation, increasing the hysteresis points , a relay cannot without some intelligent adaptive controller or by setting the hysteresis point so as to be increasing the regulator wasted power dissipation greatly.
         
The power supply MFG's use relays because they are cheaper or are just copying antiquated designs created by large power supply manufacturers in the hayday of high tech infancy of the 50 to the early 70's. So, some Chinese want to not pay an engineer to design a circuit that is only being sold to a small audience of hobby oriented customers, so to save development time and perhaps PCB real estate, like everything else, they persist in making everything by just copying older technology, using circuits created long ago, long before the age of cheap and powerful MOSFETS.

 The fact is that most equipment today is powered by switching power supply circuits to save power, increase reliability, reduce size, and both decrease weight and cost.

Also, the fact is that a power supply can often be used in situations where it is often switching between constant I to E  so this is not necessarily the oddball operation mode for some applications during some interval of use. A good example of this would be the common use of regulated power supplies to charge batteries.

And I forgot to mention that small animals can crawl into relay contacts and squish so as to keep the contacts from closing at all.
 

[edavid]

Also, the fact is that a power supply can often be used in situations where it is often switching between constant I to E  so this is not necessarily the oddball operation mode for some applications during some interval of use.

Can you name some?

A good example of this would be the common use of regulated power supplies to charge batteries.

In that application, the supply switches mode only once, and it doesn't matter how long it takes to get to the set voltage.

And I forgot to mention that small animals can crawl into relay contacts and squish so as to keep the contacts from closing at all.

Perhaps an integrated relay/mousetrap could solve that problem

[megajocke]

The 78L12 circuit has only one connection to the  adjacent circuitry, so it is inactive, floating, really floating.

Yep, and poor Q3 will be really saturated at power up with 20mA on its base ((12-0,6)/561 = 0,02A)!

There must be a mistake on the drawing... 

It might be a bit excessive, but the Vce(sat) rating is given at 50 mA base current so the transistor can take 20 mA without any problems.

However, it looks like the connection between the 78L12 GND pin and the OUT node (or similar) is missing in that schematic. The relay contact NO and NC connections also seem to be backwards.

[BravoV]

is there any other way which would be better for this situation?

Better ? Not sure, but there is an idea circuit with simulation by amspire (Richard) using multiple mosfets to switch between transformer taps to reduce power dissipation for linear part, read the details here -> https://www.eevblog.com/forum/projects/lab-psu-design-ideaquestions/msg205205/#msg205205

For sure, if this really works, it will be a true solid state transformer tap switchers, no more mechanical relays. 

[Zero999]

How about using transistors to switch taps, according to the output voltage, rather than relying on the MCU to do it?

Here's how it could be implemented with the LM317, of course it would also work with an MCU controlled regulator.

[Paul Price]

Hero999, 

The 470K (no R designator) from B-E of the PNP switch transistor would have no effect.

 The tap switch transistor woiuld get quite need a farily large heatsink and the regulator would probably not be able to output  the  max.regulated output current or volts at Vout >12V.

[Zero999]

It's a good idea to have a base emitter resistor to ensure the transistor turns off fully.

The power dissipation of Tr1 will still be less than what it would've been, if the tap changer were not used. Tr1 could be a MOSFET for lower switching losses but you'll need to ensure the gate voltage is in between 10V to 20V.

I didn't specify the voltage of the transformer, a 15-0-15V transformer should be fine.