Is a triac on the primary side of a transformer a good idea?

[daubmaso]

Hi there, I've been wanting to make a variable power supply for some time now (above an lm317), and was considering either using the 12v output of some spare computer power supplies I have laying around and using a buck-boost converter or a buck converter on the 55v output of a nice big audio transformer I salvaged a while back. So I got to thinking if it were at all practical to just use a triac on the primary side of the transformer to step down the voltage.
So would this be at all practical? Could it produce a stable output voltage? Would it take too much filtering if I tried to go down too low (say 3.3v)? Does this idea have any merit at all or should I just stick with a buck boost on the 12v supply?
I'm sorry if the answers to these questions are obvious, or if it is a completely terrible idea. Any input would be greatly appreciated, thank you.

[Circlotron]

the 55v output of a nice big audio transformer I salvaged a while back.

Do you mean the 55VAC output of a mains transformer that came out of an audio amplifier?
You can use a triac but make sure both the positive and the negative half cycle are exactly equal or you will get a net flow of DC in the transformer primary and possible burn it out. Use a phase control IC, not a simple light dimmer.

Secondly, use a choke input filter after the rectifier so that it utilises the whole of the AC waveform coming out of the secondary. If you use only a capacitive filter the resulting short sharp current peaks of the chopped output waveform means the transformer will be greatly loaded for not much DC power output, e.g. a 100 VA transformer might only be able to supply 30 watts DC. Even with no waveform chopping a 100VA transformer can only supply typically 60 watts DC when feeding a capacitive input filter.

[daubmaso]

Yeah it was a mains transformer and I usually get around 55V after it's rectified, it has a few other taps but most of them only get a couple of volts which I imagine was for the filament of the VFD.
And for the control I was planning on just making a 0 crossing detection circuit and control the switching with one of the PICs I have laying around here. The whole plan was to make a digitally controlled power supply with rotary encoder controlled voltage and current limiting, so no the use of a light dimmer was never the intended method.

I have plenty of beefy inductors and stuff from old TVs I took apart, so at least I could make use of those finally.

And would this really be any different than a half bridge rectifier? And if so, would I be better off using switching transformers?
I guess what I'm really asking here is not so much if it's possible to do this, but more if it's practical over traditional switching supplies. I have plenty of high power mosfets, igbts and thyristors just collecting dust in my junk bin, but I thought the triac route might be interesting to see if it was worth looking into.

[oldway]

Hello, as this is your first post on this forum, I welcome you.

To answer your question directly, no, it's not a good idea.

Controlling a transformer through the primary causes a lot of problems:
1) Safety: working on circuits connected to the main's supply is dangerous, even lethal. As a beginner, you must avoid unnecessary risks.
2) highly inductive load: the phase control of highly inductive loads poses great problems of dV / dt when switching the triac. This can lead to a loss of control of the triac which goes in full conduction.
3) saturation of the transformer: any assimetry in the control causes a continuous voltage which saturates the transformer and will blow the fuse and perhaps the triac.
4) the very bad waveform of the current: with a normal transformer, the shape of the current waveform is very unfavorable: they are pulses of great amplitude and short duration ... The rms value is very high for a low average value ... It is an operation to avoid because it will heat up very hard the transformer, the triac and the rectifier.
For older battery chargers that used this principle, the transformer was "special" with increased short circuit impedance. (20 to 25% instead of 7%)

A far better solution would be to supply the transformer primary with a much lower voltage than the nominal voltage.

[Ian.M]

Alternatively, working with what you've got, a switched mode buck converter (after the bridge rectifier and main reservoir cap) could be used as a preregulator to drop the secondary side DC bus to a few volts above the desired output voltage, followed by a good filter and large reservoir cap to clean up the switching hash and provide a sufficient reserve to supply load transients then finally a linear regulator to get the exact voltage and current limit required.

However the devil is in the details, and unless you've got a *LOT* more PSU design experience than most of us, it wont be an easy project to get right, and the mistakes along the way are likely to involve a lot of smoke and expensive parts.

Can you post a photo of the transformer with a ruler in it for scale, and measure *ALL* secondary voltages (AC V RMS please) + the primary voltage the measurements were taken at.  There may be other options like commercially available DC input digitally controlled bench PSU modules.

[madires]

There are PSUs with a TRIAC regulating the primary side but they require a special transformer. A standard AC mains transformer wouldn't really work. This design was rarely used, mostly for high power lab PSUs or audio amps to reduce the power dissipation.

[oldway]

I read on Internet that main's voltage is 110V in Canada....Then the solution is perhaps quite easy....Your actual transformer has a secondary voltage of about 40Vac. (55V when rectified)

If your transformer has a multi-voltage primary (universal multi taps transformer), choose the 220V tap of the primary anf feed it with 110V.....Secondary voltage will be 20Vac, which is perfect for a LM317.

[lordvader88]

Here on The Rock it's 120V

[daubmaso]

 Yeah here in Canada mains is closer to 120.
Anyways there are 3 taps on the primary side, one is for a 220v input, and there 5 on ther other side.
Between the first two is 19.5 VAC, between the second and the third there is another 19.5VAC for a grand total of 40VAC, which is very hand for either making a fairly decent 55v DC or a +-27V supply. The other two center taps have anywhere between 1 and 15 VAC depending on where you use as a reference, and are not terribly interesting to me.

The dimensions of the transformer are 60x55x70 (mm) approximately.

I thought about making a tracking buck converter with a linear output regulator to get rid of the switching noise, but I wasn't sure if that was a good idea. I have no idea how to make a switching converter efficient over a range of voltages, and most application notes for the switching regulators I've looked at only really talk about making it most efficient at 1 output voltage.

[daubmaso]

Oh and I didnt want to make an LM317 supply because I wanted some more output current capability, and wanted to be able to go up to 30v at least, which I can't do with an LM317

[Ian.M]

LM317HV goes up to 60V, and with external pass transistors, as many amps as you can afford to heatsink away the dissipation for!

However LM317 based 'lab' PSUs are madness over about 500mA - they are just too inefficient, with excessive voltage drop. 

I suspect the winding on the middle two pins of the secondary side is an independent secondary giving 15V AC at a lower current (from the thinner wire gauge).  That could be very useful to power the regulator control circuit floating with respect to the +output terminal. 

If you measure the primary and both secondary winding resistances^*, and also the core thickness, one of our transformer experts can probably give you an estimate of its current capability on the main and auxiliary secondaries. 

When you said "55v output of a nice big audio transformer I salvaged a while back", I had visions of a brute of a transformer up around 1KVA or even higher that takes two hands and a bit of 'grunt' to pick up, not a little thing you could sit on the palm of one hand with your arm out straight.   I'm fairly disappointed! 

The mean DC current available after a bridge rectifier and reservoir capacitor is only 62% of the RMS current into it, and that's a fairly small transformer.  I'd guess somewhere around 10-15VA, so it may only be good for 150mA continuous DC output and maybe 300mA for a 5 minute overload rating from cold.  When the output's low enough you could tap-change down to the center tap, reconfiguring the bridge as a two diode full wave rectifier, and get 50% more output current.   If you want to get the most out of it a LM723 floating regulator based design should allow you to build a 0-50V PSU with current limiting.  Adjustable limiting is more complex - it may be worth compromising on the maximum output voltage and using a pair of LM317HV regulators in the LM317 datasheet 'Lab' psu circuit with the negative bias rail + a cooling fan run off the auxiliary secondary.  That may actually be the best simple option for turning it into something usable on the bench.

* Reasonably accurate (+/-5%) measurements of what may be very low resistances are *ESSENTIAL* if we are to estimate the relative wire gauge and thus the VA distribution between the main and auxiliary secondaries.   If you dont have a bench meter that does four wire (Kelvin connection) low ohms measurements, set up a LM317 as a 100mA current source, put 100mA DC through the secondaries (you can wire them in series) and measure the DC voltage drop across them in mV.   With that setup the resistance is ten milliohms per mV measured.

[oldway]

So it's a transformer with a center tap 20V-0V-20V .... This is perfect for making a dual power supply +1.5 to + 20V and -1.5 to -20V adjustable.
Use a LM317 and a LM117.
No need for current limitation, the LM317 and 117 have their own internal protection.
The current limitation value will depend of the heatsinks you have used and of the input and output voltages with a maximum of 1.5A.

[Ian.M]

I think you mean a LM317 and a LM337 for the negative rail. 

1.5A on either rail would fry that transformer - that's 30VA and its just not big enough.  If its got a thermal fuse in the primary, a sustained overload will kill it dead.  If you dont want to implement current limiting, a thermal trip would be a good idea - a sub-miniature glass bead thermistor tucked into the secondary with a dab of heatsink grease  and a comparator circuit that senses the temperature and if its over the setpoint, latches, lights a LED and drags both regulators ADJ pins to 2V the other side of the 0V rail.  If you do want to implement current limiting that's appropriate for the transformer, a LM723 based design would be simpler. 

If the O.P's interested in logic circuits, it may be worth adding an EBAY CC/CV buck module to derive an independently floating fixed +5V output from the Auxillary secondary.  My best guess at the moment is it would be good for about 200mA.

[oldway]

Yes, indeed...
Sorry 
I used a LM7805 and a LM7905......

[daubmaso]

By a big transformer, I mostly meant it's the biggest I've found so far. I have access to a much bigger one, probably 5 times the volume, but it's in an old receiver that I'm still using.
I was hoping to get at least 2-3A out of the supply, and that's why I brought up the use of a buck-boost converter with some computer power supplies I have (I somehow managed to come across 4).
I was hoping that transformer could dish out more current though, I didnt realizes it would only be about 30VA.
I am just a second year physics student who likes to mess around with microcontrollers and robots, so I haven't learned too much about AC circuits, so I'm sorry that I'm not too savvy with this stuff.

But anyone have any recommendations for a buck boost converter that would be able to handle 3-5A, or at least have the option of an external switching transistor? I think I might be better off with the computer power supply if I want this type of current capability.

[Ian.M]

30VA is far too optimistic until one of our transformer gurus has run the numbers for you given exact core dimensions, winding resistances, and voltages etc.   I'm not that level of expert (although I can hum the tune if I've got a crib-sheet), but its a similar size to a Hammond Transformers 12VA single secondary model with the same primary configuration, but with an extra lower current auxiliary secondary  and I therefore estimate you may get 10VA out of the center tapped (main) secondary.   One way forward without the experts would be to measure the exact DC primary resistance, then hook it up with a 160 ohm >10W resistive load across the outer terminals of the CT secondary, and switch it on, briefly switching it off every 5 minutes at first then every 15 after the first 15 to measure the primary resistance again so you can estimate the average winding temperature using the temperature coefficient of resistivity for copper (see http://hyperphysics.phy-astr.gsu.edu/hbase/electric/restmp.html ).  Keep it under 115 deg C as you are are measuring an average for the whole winding and there may be hot-spots. If after a couple of hours its still running cool enough, you can estimate how many more VA it can handle, from Newton's law of cooling, and the knowledge that resistive losses are proportional to the SQUARE of the current.  If it starts to overheat, STOP, plot your readings, estimate the final temperature and scale back the VA rating accordingly.    Finally retest with a load appropriate to the revised VA rating, finishing with an eight hour soak test, measuring the resistance hourly.

Caution: if it has a thermal fuse in the primary with a low trip point this procedure may kill it.

[oldway]

By a big transformer, I mostly meant it's the biggest I've found so far. I have access to a much bigger one, probably 5 times the volume, but it's in an old receiver that I'm still using.
I was hoping to get at least 2-3A out of the supply, and that's why I brought up the use of a buck-boost converter with some computer power supplies I have (I somehow managed to come across 4).
I was hoping that transformer could dish out more current though, I didnt realizes it would only be about 30VA.
I am just a second year physics student who likes to mess around with microcontrollers and robots, so I haven't learned too much about AC circuits, so I'm sorry that I'm not too savvy with this stuff.

But anyone have any recommendations for a buck boost converter that would be able to handle 3-5A, or at least have the option of an external switching transistor? I think I might be better off with the computer power supply if I want this type of current capability.

As a student, you must learn that the method you follow to analyze and solve a problem is completely wrong.
Before specifying what you want to do, you must first analyze what you have at your disposal to know the possible limits to your desires.

You do the opposite, you say what you want, then, you try to match what you have to your specifications ... This is completely wrong.

A transformer has very specific limits and especially with regard to its heating.
If this is not respected, the transformer may be damaged and become unusable.

EDIT: Another remark: someone who studies in the field of science or technology can not be satisfied with an imprecise and subjective language.

We must adopt a scientific and technical language .... a huge amount of gas, a large transformer, a huge current, a high temperature, ... all this is not scientific and does not mean anything! .... It should be specified: for example, a quantity of gas of 10L per minute, a transformer of 500VA, a current of 2000A, a temperature of 250 ° C, ....

[Circlotron]

Stick the tranny on a set of scales and weigh it.
That should give us a good starting point as to guessing it's VA rating.

[daubmaso]

Yes, I know I haven't been very precise, and I never was really intending to be very formal here with the way I present my questions. The reason I've been doing this is because I really was just curious if, as the name suggests, it was possible to drive the primary of a transformer to make a crude power supply, not so much discuss the details of the end product.
But yes you're right I should be more precise with my wording, especially in a more technical setting such as this forum.

I originally planned on using the transformer to make a 1.25-25v lm317 based power supply where current and voltage control would be done with an atmega328p microcontroller, quite similar to the design Dave made for his micro supply. I later changed my mind and wanted to use a switching regulator so that I could make it more efficient and increase the current capability at the lower voltage outputs. I never thought 5A at 25V was possible, but maybe down at 3.3-5V. Ideas that I thought of to design a switching based power supply were a tracking buck regulator with a linear output stage. This got me wondering if I needed the transformer at all, or if I could make use of some 300-400W PSUs I have. Using a tracking boost converter and a linear output stage came to mind, with either the 12V or 5V output of the PSU. Other ideas included a fixed boost regulator to around 35V and a buck regulator to change it from there, or a boost-buck switching regulator to try to reduce complexity.
The main control IC I've decided on is a dsPIC33EP16GS506 because of its 12 bit ADC and 2 integrated 12 bit DAC outputs. Beyond this, I haven't decided on any switching or linear regulators.
I'm aiming for 30v out max, although a greater output voltage would be more preferable, and a 3-5A output current.
Do any of the afformentioned topologies posses the capability of creating the desired output without any major issues? Or should I abandon the higher output current and stick with an lm317?
And thank you if you are still willing to answer my questions.

Stick the tranny on a set of scales and weigh it.
That should give us a good starting point as to guessing it's VA rating.

I've heard this as a method before, but was never clear on the numbers. I'll certainly weigh it in the morning, thanks.

[Jwillis]

Just a quick look you have approximately 4 square centimeter core .That will get you around 12 VA maybe . That,s around 300mA at 40VAC.
If you want 3 amps at 20 volt you'll need around 150VA .The core area would be around 14 or 15 square centimeters .

[daubmaso]

Oh okay, thanks!

[james_s]

Here on The Rock it's 120V

It has crept up over the years, I believe to extract a bit more capacity from the grid. US and Canada are the same spec, at one point it was 110/220V, then 115/230V, then 117/234V, then the current 120/240V. Many people colloquially refer to it as "110" or "220" even though that hasn't been the standard in decades.

For anyone confused by the double voltages, we use a split phase that comes off a transformer with a grounded center tap, 120-0-120V. The standard residential branch circuits are 120V but major appliances like ovens, clothes dryers, water heaters and other large loads are 240V. 

[Circlotron]

For anyone confused by the double voltages, we use a split phase that comes off a transformer with a grounded center tap, 120-0-120V. The standard residential branch circuits are 120V but major appliances like ovens, clothes dryers, water heaters and other large loads are 240V.

So the transformer secondary out in the street has a centre tap and 6 free ends? Effectively a star config and a second star shifted 60 deg. That's be cool! 6 different phases to neutral and a further not sure how many from phase end to adjacent phase end.

[oldway]

Regarding the use of your transformer to make a DIY symmetrical +/- 1.5 to 20V adjustable power supply with LM317 and LM337, I think it is possible, despite the reduced power of the transformer.

You should know that a transformer has a fairly long thermal constant, it takes between 15 minutes and 30 minutes for it to reach its equilibrium temperature ... It can therefore been overload for a short period of time without risk.

On the other hand, you can increase its power by improving its cooling, for example by forced ventilation or cooling in the oil.

And, finally, it is a small transformer of low value, the possible overheating will reduce the useful life of the transformer ... But who is worried about it?

[james_s]

For anyone confused by the double voltages, we use a split phase that comes off a transformer with a grounded center tap, 120-0-120V. The standard residential branch circuits are 120V but major appliances like ovens, clothes dryers, water heaters and other large loads are 240V.

So the transformer secondary out in the street has a centre tap and 6 free ends? Effectively a star config and a second star shifted 60 deg. That's be cool! 6 different phases to neutral and a further not sure how many from phase end to adjacent phase end.

No, it only has a single secondary with a center tap that is tied to earth, so 3 wires coming out. From one side you get 120V to neutral, from the other side you get 120V to neutral and from live to live you get 240V. The disadvantage of this setup is that 240V circuits have two lives but in practice it does work fine.