Transformer tap switcher for bench supply

[diy didi]

HI Guys.
I'm in the process of building a variable bench power supply. 0-50V @5A.
I have wound the transformer secondary for 0-6-12-22-44 taps. By doing this I can geep the dissipation through the output transistors to a minimum.
I would like to build a circuit that can automatically switch the taps in and out as I advance the voltage pot etc.
Obviously only one relay should be on at any given time, otherwise two of the taps may be shorted. BOOM!!
I will have to power whatever circuit and reference voltage from one of these taps as I don't have space for another auxilary power supply to power this circuit too. I was thinking 4 comparators driving npn transistors, switching 4 relays in and out.
Any ideas? Anyone build something similar before?

 

[Les Jones]

You can get round the problem of shorting taps by using relays with single pole change over contacts. The first relay (RL1) will have it's common to the rectifier and its NO contact to TAP 1 on the transformer. It's NC contact goes to the common of RL2. RL2's NO contact goes to TAP 2 on the transformer. Just repeat this sequence for the number of taps required. When more than one relay is energised the tap connected to the lowest number relay will be selected. The relays can be driven from comparitor chips and transistors connected to a potential divider chain connected to the voltage setting pot (Via a buffer.) You could also use the difference between input and output voltages to select a higher or lower tap but this might have a tendency to jump between taps.

Les.

 

[crutschow]

....................
The relays can be driven from comparitor chips and transistors connected to a potential divider chain connected to the voltage setting pot (Via a buffer.) .................

The voltage divider can also be directly connected to the output voltage.
That would not require a buffer.

 

[ronsimpson]

In my power supply (home made) I have a light that lights when the regulator is not in regulation. Another light for too much voltage across the regulator. There is a rotary switch that must be turned by hand. It take a little operator support.

 

[crutschow]

Here's my take on Les's idea for connecting the relays.
You only need 3 relays for your 4 taps.
With no relays energized, tap 6 is connected. You then just turn the relays on in sequence (U1, U2, U3) using three comparators, to sequentially connect each of the other taps without having to turn the previous relay off.
Thus there's no concern about having more than one relay on at a time and shorting two taps together.

 

[ronsimpson]

The peak current in the relays will be high. Don't use small relays.

 

[dr pepper]

Also if your supply has current limit you need to consider depending on design you might end up with a situation like the limit circuit reducing the o/p to a few volts to maintain a max current, but the voltage control pot is set for 50 volts meaning the tap selector is on a high voltage and the actual o/p voltage is low - cooking the pass device.
So you need to switch taps depending on the o/p voltage not the setpoint voltage.
I used to use such a supply, it was a cheapie eagle one I think.

 

[crutschow]

Also if your supply has current limit you need to consider depending on design you might end up with a situation like the limit circuit reducing the o/p to a few volts to maintain a max current, but the voltage control pot is set for 50 volts meaning the tap selector is on a high voltage and the actual o/p voltage is low - cooking the pass device.
So you need to switch taps depending on the o/p voltage not the setpoint voltage.
..........................

That's a good point.

 

[crutschow]

The peak current in the relays will be high. Don't use small relays.

Yes, there will be a large surge current to charge the filter capacitors whenever the tap is switched to a higher voltage, which can weld the contacts of a small relay.
A small resistor in series with the relay output to the rectifier-capacitor would help reduce this surge.

 

[Les Jones]

crutschow, Your schematic is exactly what I thought. I agree with all the above comments. dr pepper makes a very good point about the situation when the power supply was in current limiting mode. If the relays were controlled by the input / output differential (Rather than the desired output voltage.) then this problem would not occure. Overnight I was thinking about the contol methods using comparitors for window detectors (So some hysteresis could be put in the system to avoid relays clicking on and off near the changeover levels.) I decided if I was doing it I would use a micro controller. I would use 3 analog inputs. (Via potential dividers.) One to voltage setting pot . one to the regulator input and one to the regulator output. This would make adding hysterisis simple and would allow changing from the set voltage method to the input / output differential method just by changing the firmware.) One other point is I don't think the OPs idea to use one of the transformer taps to provide power for the contol circuit is posible unless only half wave rectification is used to power the control circuit.

Edit.
I have just read crutschow's post #9. One other way to limit the problem with the high current when the relays switch to a higher tap would be be to have a well rated SCR on the output of the bridge (Or SCRs used as two of the bridge diodes.) and remove the drive to the gate for a few mS when a relay was switching. (Easy if a micro controller was used.)

Les.

 

[crutschow]

......................
One other point is I don't think the OPs idea to use one of the transformer taps to provide power for the contol circuit is posible unless only half wave rectification is used to power the control circuit.
Edit.
One other way to limit the problem with the high current when the relays switch to a higher tap would be be to have a well rated SCR on the output of the bridge (Or SCRs used as two of the bridge diodes.) and remove the drive to the gate for a few mS when a relay was switching. (Easy if a micro controller was used.)

I see no problem with using a half-wave rectifier to generate the small current needed to power the control circuit.
Edit: My simulation of a bridge rectifier connected to a higher tap along with a single rectifier connected to a lower tap showed an interesting effect.
The current through the single diode had a full-wave component, not half-wave

Edit: Using the SCR in the bridge (not at the bridge output since it needs AC to turn off) would work to limit the surge but you need an isolated drive for the gate, such as with an opto-coupler.
You could also use a MOSFET at the bridge output which wouldn't require any isolation.

 

[ronsimpson]

Back to post #5.
Replace the relays with solid state relays or MOSFETs. Time the switching to happen at the zero crossing of the power line. (or at least not at the peaks) This will help with the high current switching.
OR
Put one MOSFET at the bottom of the page. (I don't know what the switching time for the relays are, lets say 100mS) When the switching starts, open up the MOSFET for time = 50 to 150mS then close again. The relays will have zero current while switching.

 

[Les Jones]

Hi crutschow,
I am almost certain that an SCR will turn off if connected to the putput of a bridge rectifier so long as there is no capacitors on the bridge side of the SCR The current through the SCR will go to zero every half cycle. As the output of the SCR will be connected to a capacitor the current will be zero for quite a large proportion of each half cycle. Putting the SCR on the negative connection to the bridge would make driving it simpler. (But it would probably cause a problem with the half wave control supply.) I think the output from the half wave rectifier would be one half cycle with the peak voltage of the tap to one end of the full Winding (Assuming the full winding was connected to the bridge at the time.) and for the other half cycle the peak voltage from the tap to the other end of the winding. (Ignoring diode forward voltage drops.)

Les.

 

[crutschow]

Hi crutschow,
I am almost certain that an SCR will turn off if connected to the putput of a bridge rectifier so long as there is no capacitors on the bridge side of the SCR The current through the SCR will go to zero every half cycle. As the output of the SCR will be connected to a capacitor the current will be zero for quite a large proportion of each half cycle. Putting the SCR on the negative connection to the bridge would make driving it simpler. (But it would probably cause a problem with the half wave control supply.) I think the output from the half wave rectifier would be one half cycle with the peak voltage of the tap to one end of the full Winding (Assuming the full winding was connected to the bridge at the time.) and for the other half cycle the peak voltage from the tap to the other end of the winding. (Ignoring diode forward voltage drops.)

You are correct about the SCR in series with the bridge output. The current through the SCR will into the capacitor will stop each half cycle, turning the SCR off. I wasn't thinking clearly.

But the simplicity of driving two SCRs in the ground side of the bridge may make that approach preferable.
That would also reduce the voltage drop to the filter capacitor by one diode forward drop.
As far as a problem with the half wave control supply, you could use a large enough filter capacitor for the half-wave circuit to continue to power the control circuit during the 100ms or so the SCRs are off during relay switching. The control current should be small enough that a reasonably sized capacitor should work.

 

[MikeMl]

Good application for an Arduino. I have a commercial power supply that has a micro in it...

I have a motorized high-current rotary switch (out of an antenna tuner) that would work...

I have built a couple of linear supplies like this. One uses a Variac in the power transformer primary; the other has manually switchable taps...

 

[ronsimpson]

One uses a Variac in the power transformer primary

My first big FM transmitter had a variac and motor. If the power got over +2.5% the motor turned the voltage down. If the power got under -5% the motor turned the voltage up.

 

[schmitt trigger]

I remember those old motorized variacs! Made by Superior Electric if my memory serves me well.
Really expensive stuff, specially the three phase units.

But to seamlessly regulate AC power back then, that was the way to go.

 

[diy didi]

Hi guys.
Thanx for all the help. Got a little comparator circuit going with Crutschow's relay diagram.
Works a charm!!
Instead of starting s new thread, i have another question.
On the last tap 44VAC, i might be dissspating 30V @5A which is 150Watt. How many 2N3055' s will i need? What size heatsink?
Here in South Africa heatsinks are sold in K/W and not C/W. Any difference?
Thanx again!!

 

[ronsimpson]

On the last tap 44VAC, i might be dissspating 30V @5A which is 150Watt.

I thought the idea of your project is to not dissipate heat.

How many 2N3055' s will i need?

There are several different packages for the 2N3055, each having a different wattage. Don't use an insulator as that will reduce the movement of heat. The "wattage" gating is for a infinite heat sink at room temperature. Even in South Africa you can not get one.
I have used a fan and sometimes a thermal switch to cool off a heatsink. With out air flow the heatsink has problems. Most datasheets show a graph of watts & temperature rise & airflow. There is a difference of temperature rise from standing up to laying down of the heatsink.