Transformer Input Voltage Problem

You have a 26% increase in input voltage. If your load is not designed to handle this much of an over voltage then you can cause considerable over draw of current and power. The common but not sole culprit of excessive current draw are electrolytic capacitors that have too small of an voltage margin. Without knowing the load circuit I cannot offer any more advice.

Thanks for your reply. The secondary is connected for 30V and that feeds a 4 diode bridge rectifier and a 2200uF/50V

Aluminum electrolytic capacitor. The regulator is a LM317 to provide a 30VDC to the rest of the circuit. I have measured

the voltage on the cap with an input of 140VAC, it is 51.3VDC. This exceeds the working voltage of the cap and this could

be the problem. I will change it it a 63V cap and measure what happen.

Do you think the 26% increase in input voltage is not excessive? The transformer should be able to handle that?

Classic over voltage scenario. The capacitor voltage rating is only slightly exceeded so that a catastrophic failure does not happen but excessive current gets drawn.

Now the LM317 is a linear voltage regulator. The current drawn through this supply will be the same as with the proper AC voltage being applied to your transformer. If we ignore the added current going through under-voltage rated capacitors then the power dissipation of your whole system should go to 126% of the original power draw. Virtually all of this 26% power increase must be dissipated by the linear regulator. With proper heat-sinking and ventilation this may not be a problem. Without knowing how your unit cools itself, this heat maybe indirectly be causing the built in thermal breaker of the transformer to operate.

Getting finally to the question in your reply, the 26% voltage increase is likely not a problem for the transformer itself. The insulation leakage current will be slightly higher but this should be such a small number in the first place that it should be a negligible effect.

As others have mentioned, the real root of your problem is your main power is providing 126% of the anticipated voltage. This supply is not the only thing that might be getting cooked by the higher voltage. Fix the real problem.

Electrolytics can be 'conditioned' to a higher voltage rating by gently stressing them over their rated voltage with appropriate series resistance to limit current. They will build up a thicker oxide layer and end up with a higher voltage rating (at the expense of capacitance value of course).

TANSTAAFL

RedFred is right, tell us what is the load?

Specifically: what type load - motor, lighting, or ?

And is what is the rating of he load: volts, amps, Ac or Dc?

Does the PCB have a full of half wave bridge on it to output DC?

Several years ago, a 240-115V transformer was inadvertently connected across two phases instead of phase and neutral, leading to excessive voltage on the 115V side. The possibility of this being the case with the high 110V input upstream of the transformer in question needs to be eliminated from the enquiries.

^^ GA ! ^^

Whether the transformer functions properly over a long period of time with higher voltage impressed on it more than what it is designed for will depend on the class of insulation used in the transformer. As the VA rating is only 24 VA, the most probable cause of transformer overheat is because of insulation stress. This is not a good operating condition. The best is to ensure correct rating of input voltage for which the system is designed. This can be done by tapping supply from UPS etc.

As you say that the load is fairly constant, the next best feasible solution is to provide suitable rating resistor before the transformer to drop the excess voltage.

Assuming that the insulation class is more than adequate to tackle the excess voltage, another solution could be to provide for forced ventilation to take away the excess heat to keep the system in continuous operation. While this is not ideal way to operate, it may just serve the purpose. Of course liability is yours.

Are you able to utilize a buck boost transformer on the primary side of your transformer?

Your best bet is a custom transformer shop that will wind you a 140v to 110v autotransformer, cheap and easy to do at your power levels.

Putting a dropping resistor in series with your existing setup may cause safety and heat buildup problems, but if not a concern, it is the cheapest way to go.

If it was me I would swap out the transformer for a universal type switching power supply that has a proper secondary voltage in the range you need.

Given your nominal 30 VAC output once rectified you are working around a nominal DC voltage of ~42 volts with an apparently capable peak working voltage of 53 VDC or so. From that I would expect that any small 25 - 30 VA universal power supply with an output of 42 - 48 VDC would work for you.

you overlook the $dollar here

Not really.

A quick online search shows that a decent quality 24 VA 110/15/15 transformer can be bought for around $15 - $20 whereas a universal input 25 watt 48 VDC output SMPS costs around $17 - $25.

Factor in that with the SMPS it has the universal 88- 265 VAC input plus all secondary side power is already DC and regulated eliminating the added components and circuitry that a transformer based power supply would require and it's clearly the cheaper easier and more forgiving design to use.

ups, however if the Gizmos are only manufactured not yet distributed that like cut's some 60% of the losses . . .

i don't see the ops concern getting components off the PCB intact

The components on the board are secondary but if the main power transformer fail they serve no purpose either however given that the OP admits that the present design is pushing the on board components at or above their limits I see them as being very relevant to the issue at hand.

As far as my search went I just did quick general comparison based on off the shelf items I could find in one of the major electronics component suppliers online catalogs. Digikey to be honest.

To me the transformer is not the only issue in this design. Everything in the power supply section is. Being it is not designed around a more universal type power input even though clearly the device is being used where such a capability should have been obvious. Even more so given that a universal type switching power supply could have bee incorporated into the design for the very close to the same overall component costs.

I completely agree but to play devils advocate here there maybe a legacy and/or noise bandwidth issue that precludes a complete redesign. Broad input voltage range power supplies are not unique to switching supplies, many analog power supplies have been designed to handle a broad input range. Clearly this particular design did not anticipate a broad input change.

As usual Winfield Hill brought up an excellent analysis and quite plausible another added non-linear effect in the transformer saturating that could additionally draw excessive current like the already discovered capacitor problem. In my experience a 50% over voltage input condition would not saturate an on the board power transformer. Then again those were typically my circuit design...

Possible but given the basic info so far everything points me to it being a far from fussy or high precision/sensitivity manufactured circuit.

Speaking of swapping, another solution, if there's room and a chassis spot, would be to add a small 20 to 30V transformer and wire its secondary in series with the PCB's ac-line input connection, to reduce the PCB's voltage.

To know whether the problem is due over voltage do following:

Take a small autotransformer (Variac).

Feed power supply to variac and in output of variac connect your Supply kit via AC Ammeter of 0 to 250mA/ 0-500mA range.

Slowly raise the voltage, starting with 110V, and check magnetising current corresponding to applied voltage.

If the problem is due to over voltage, you will find that suddenly the increase in current is non-linear. That means current increment is more than voltage increment - indicating the core of transformer is saturating.

If the core is saturating, it will result in:

1. Draw of extra current (above saturation point, it is almost like an air core transformer) - overheating.

2. Distort the waveform of voltage being supplied to rectifier, resulting in more harmonics and heating of capacitor used in filtration circuit of rectifier.

if above is established by simple test then adding series resistance is not the solution. Reason being that resistance in series also distorts the waveform of AC induced in the secondary winding - so again harmonics and overheating.

The solution then is to make a small inductor (choke) and connect in series to regulate the over voltage condition.

I have done the following test. Connect a variac to the input and measure the AC current and the DC current just after the bridge rectifier to the cap. The result is in the graph attached. As can be seen the load current is 170mA and stable over the entire input voltage range.

I see following results from the current vs voltage curve:

Up to 110V the curve is linear and is the usable range of transformer.

You are operating at 139V, where the magnetising current (from curve) is 0.24A

For a 24VA Transformer, at 90% efficiency (conservative figure), The load component of current is 0.24mA.

Hence at 139V, assuming that magnetising current is at 90° phase to load component of current (again conservative assumptions), the current in input to transformer at 139V works out to √(2*0.24²) = .34A.

The same current at if supply is 110V works out to √(0.1²+0.24²) = 0.26A

The Cu loss of winding or its heating is proportional to square of currents.

Hence effect of saturation is that when operating at 139V the heating in winding is 70% more than when operating at 110V.

Hence the only ECONOMICAL solution, for the production lot already manufactured, I see is to reduce the input voltage from 140V to 110V.

Input impedance of your transformer works out to 110/0.26 = 420 Ohm (assuming this to be at 0.7PF), The resistive and inductive components work out to 294Ω and 300Ω respectively.

When we divide 140V into 110V +30V and considering 30V is dropped across series resistance, the additional resistance works out to 110Ω, 7.67W.

So suggest to do following:

- Connect in series with primary of transformer a resistance of 110 Ohm and 10W.

- Connect load.

- Measure Applied voltage at primary of transformer, See if it is between 110V to 115V.

- If it is, soak your assembly on load for 72 hours at 140V. If it stays go ahead with remaining already manufactured units.

Since I have made some assumptions as Power Factors etc. 99% you shall get the voltage within limit. In case of some deviation, you can slightly manipulate with value of resistance.

As said before, we do not drop voltage to transformer via resistance as this adds to distortion and harmonics, but since yours is a case of further rectification and filtration - this may work - Soak test will confirm. Pl let me know if it works.

You almost brought up the point that has been bothering me in this problem that until now I could not put my finger on. This transformer at the rated input voltage is only about 70% efficient. This is also with the transformer producing less than a third (7W=[30V*2^0.5-1.4V]*0.17A)of the rated output (24 VA). Your supposition of 90% efficiency at rated power is a good design rule of thumb because this is usually the case. This transformer should not be anywhere near saturation. Even with a gap less transformer, the magnetic core should be less than half of its design flux @110V input. I now think that there must be a parasitic resistive loss somewhere in the transformer primary circuit. Circuit board trace leakages, uncleaned flux, something is robbing power on the primary side. The numbers do not add up.

In the 8.3 mS that a full diode bridge generates current cycles, a 2200 uF capacitor will only have a 0.64 volt drop. So most of the time the capacitor is providing current to the regulator not the transformer. (I doubt that the graph accurately displays the current from transformer to capacitor through the diode bridge. I suspect this is regulator input or output.) The ESR of the capacitor will produce further self heating in this capacitor than just the over voltage. Presumably the voltage regulator is configured to generate 30V. If this is the case then the capacitor is under sized for the voltage applied to it and over sized in capacitance since the voltage regulator has nearly 10V of over head with a 110VAC input voltage.

Let's keep in mind that the primary magnetizing current, and the associated core and copper losses have nothing to do with the output power consumption on the transformer. The primary magnetizing-current losses are related to the input voltage, and not the load, and are present with no load at all.

So transformer "efficiency" calculations have two contributing factors, and the efficiency value under one specific condition, like rated power, is of limited usefulness in determining the two factors. I agree that a large transformer, one boasting 98% or 99% efficiency, etc., will probably be operating far from saturation. But most of the small transformers I've measured were designed to operate uncomfortably-close to saturation. The 10% loss value for full-load operation for these transformers might be dominated by the fixed load-invariant magnetizing-current core-related losses.

Yes, the primary magnetizing current and the associated losses has nothing to do with the output power consumption. However, core saturation happens from the total magnetic field in a core not just the primary winding current but also secondary winding. At the rated 110 VAC input the output load is drawing less than a 1/3 of this transformer's rated output power. So with the primary magnetizing current and this load the magnetic field should be less than a 1/3 of the saturation field. I agree that many transformer manufacturers do and should design their product to operate close to saturation when at full load but that's not what we have here. There should be plenty of magnetic field head room.

I still believe that the root of the problem lies in the capacitor choice. With the 0.6 voltage drop I calculated earlier that should appear across the 2200uF capacitor, the diode bridge should be ON for just 10° of the 180° half cycle. (Come to think of it there are always two conducting diodes. This is getting less linear with every thought.) So the secondary windings will be adding a significantly nonlinear, higher frequency magnetizing field in the core. This significantly higher frequency will produce higher eddy currents in the core laminations. I believe that this is why the transformer has a poor efficiency value @ 110 VAC. The frequencies of the loading current are significantly higher than the linear load the transformer manufacturer likely used to achieve 24VA. Now it is possible that the core material is optimized for 60HZ and the nominal 1080 HZ of the secondary current frequency is a different permeability value that saturates at a lower field. (Remember all of my calculations have been at nominal voltages not the 27% over voltage the OP is having a problem with.) When the primary voltage increases, the secondary voltage increases but the time that the diodes are ON decreases due to this higher voltage. This further increases the lamination eddy currents. So a lower capacitance value should reduce the eddy currents in the transformer, at the price of a larger voltage swing across the capacitor. Since this load down stream of the voltage regulator will be nearly constant, the voltage swing on this capacitor should drop the voltage to 3 V above the output voltage of the LM317 regulator with the lowest input to the transformer.

"Yes, the primary magnetizing current and the associated losses has nothing to do with the output power consumption. However, core saturation happens from the total magnetic field in a core not just the primary winding current but also secondary winding."

Correct.

"At the rated 110 VAC input the output load is drawing less than a 1/3 of this transformer's rated output power. So with the primary magnetizing current and this load the magnetic field should be less than a 1/3 of the saturation field."

No, no. The primary magnetizing current is 90º out of phase with the primary voltage. As I understand it, the fields created by the primary and secondary currents due to the load cancel each other, and do not significantly contribute to the core's magnetizing flux (also they're in phase, for resistive loads). The full 90-degree-shifted primary magnetizing current must be present, to hold off the primary voltage, even if there's no load at all. Otherwise the low-resistance primary winding would simply look like a piece of wire. And as we've observed, the core's peak magnetizing field is uncomfortably close to Bmax.

BTW, all of my ac-line small-transformer saturation-characterization measurements I mentioned earlier were made with no load.

Yep, you have an inadequate magnetizing-inductance core-saturation problem, consistent with my small-transformer measurements, outlined in post #11, and as suggested by several other posters here. powersolutionsFBD analysed your measurements and suggested you drop your primary voltage by 30 volts using a 110-ohm 10W resistor, dissipating about 8W.

You describe an excessively-high ac-line voltage, but if the power company fixes this and returns to more usual line voltages, you may then have an under-voltage condition for your modified system. A 15 or 20V reduction may be more advisable, with a 50 to 68-ohm 10W resistor (using FBD's numbers). But given that you have a rather tight compromise, a more precise load-independent fixed-voltage-drop active-zener circuit like the one I suggested may be better than a resistor. Here's the circuit.

The diode bridge's 2 V_D and the transistor's V_BE together drop about 2 volts. You choose the zener voltage Vz for the rest of the drop, the total will be Vz + 2V. The 2SC6082, $0.76 qty 100, an insulated-package power transistor I suggested can easily handle peak currents of 10A in this circuit. You can bolt the transistor to the chassis or case with its tab hole, and in a pinch the circuit's three other parts, which won't heat significantly, could be soldered to the transistor's hefty leads. The transistor's insulating plastic is convenient, but it is a thermal insulator, reducing the transistor's power rating to 23 watts (the pnp complement 2SA2210 is rated at 30W).

If you like, you can use an electronic zener diode to get an easily adjustable voltage drop. The TL431 is a popular and inexpensive shunt regulator, $0.19 at Mouser, that makes an excellent programmable precision zener diode, up to 36 volts. Its voltage drop will be 2.5V + 2.5 R2/R1, and you could use a pot for R2. For example if R1 = 910 ohms, then a 10k pot would make a 2.5-volt to 30-volt zener. In the circuit I have it running at 5mA, but it can deliver a peak current of over 150mA to the transistor, allowing for peak active-zener circuit currents of at least 8A (the transistor has a minimum beta of 50 at 10A). I imaging your peak currents will be well below that!

One last point, you may find a power transistor to be cheaper and easier to mount to heat-removal surfaces than a power resistor. In fact, like to use power transistors as heating elements, in place of resistors.

While suggesting the resistance to drop 30V, under voltage condition appeared to me, but it did not bother me as in the Voltage vs current curves, the output (load) current is constant over wide range of voltage 90V to maximum AC voltage applied. The curve presented starts at 90V and I assume that similar conditiion may be existing even at 80V supply voltage.

Probably the output circuit has a Zener like voltage stabilizer in DC output and the whole design is such that under voltage conditions won't create any problem.

Install a UPS to feed the equipment so that the voltage is regulated and stable.

You could easily get customer transformers made.

You could use Switched Mode Power Supplies that would accept the input voltage.

You could add an auto transformer to reduce the input voltage to 110 volts.

I am sure that there are at least 100 possible alternatives....

Thanks for all your comments and help. I will do some tests and some thinking about this problem and your solutions. I have already started to change the design to use an AC/DC wide input convertor, but the main problem is the 200 or so already manufactured PCB with transformers that cannot be changed without damaging the PCB. Will give you feedback as I go along.

i donno with this >> 110/220V to 15/15V 24VA
very specification - i would say your thing should work eigther ~110VAC/~15VAC or ~220VAC/~15VAC both at 24VA(watts) max load at 2-ndary's side

now if it's such a qute thing the ~140VAC shouldn't be a problem for it ?????????

I suspect what you do not grasp is that the input of the transformer can be wired or switch selected for 110 or 220 VAC. This is not a range but two options.

???? i suspect there can't be "something unconnected winding" at the primaries' side
anyway just curious ... i was

(nothing to to =) pg.4 & 56

Thanks again for all the feedback, I really appreciate it. Just a few comments, a correction and action plans. First some the correction: I mention the voltage of between 139V and 140V, it should read 130V and 140V , the 0 and 9 are to close on the keyboard, but it should not make any difference on the discussion, the input voltage is still to high.

Some comments:

Winfield Hill: the circuit you propose looks very workable, just one problem, my nearest electronic shop is a good two hours drive from here, thus a round trip is one day including some shopping time and it is not a guarantee I will get all the stuff but I will still pursue your solution.

powersolutionsFBD: Just give me some time, I have some 100ohm and 50ohm aluminium resistors which I can use. I will do some measurements and post the results.

I also have some 2200uF/63V and 1000uF/100V caps I will try. The UI395 series transformer from Hahn_Eng.pdf is the same that is used in this design. If there is more information required, please ask and I will try to supply it.

I stumble across this web page/blog:

http://blogs.hbr.org/2013/09/when-youre-innovating-resist-l/

Regards.

I would like to thank you all for your comments and help. By staying on the problem longer, I have discovered the real problem with this transformer. The real problem is that this transformer is a poor quality and rubbish transformer. This is a transformer from our second batch of PCB. The first batch transformers perform much better. At 150Vac input the max primary current is < 150mA, the second batch was > 450mA. The second batch transformers was done by a different manufacturer and I believe he does not have a cooking clue how the make good transformers. There are only two workable solutions for this problem, dump the PCB with the rubbish transformers, and learn a expensive lesson, and the 2nd solution; use 25W resistors in series to lower the input voltage. The reason for using resistors is because there is a space problem and any circuit as suggested will occupy more space and secondly the output of the transformer goes directly to a bridge rectifier and cap, thus no problem with waveform distortion. I have tested the resistor solution and the max temp of the transformer at 150V is only 60Dec C.

I would have bought "good" transformers, removed the "bad" ones and soldered "good" ones in their space.

Nowadays there are people changing out huge SMD chips, with loads of really tiny, close together, legs on multilayer PCBs, a transformer should be REALLY easy.....

What am I missing?

The main problem in removing the transformer from the PCB is that the transformer has 8 pins and you need to heat up at least 4 pins simultaneously and then wedge these out, followed the the other 4. I am sure it can be done, but one need to have the skills and patiences to do that. The danger is that the THP barrel will come out with the pin and then you are stuffed because some tracks are under the transformer. We will try a couple of PCB and see what the success rate will be, and if it is economical to carry on. The PCU PCB is not that expensive and it is a matter of economics at the end.

SMD Devices are easy to remove and replace and the chances of PCB damage are relatively small compare the THP devices, at least that is my experience.

With a reasonably priced de-solder gun and light flexing of the PCB, I would remove it in about 3 minutes, but I am experienced.

If you can get access to the "legs" (not always possible), cut them through and remove each one separately. This may become more possible once the transformer has been raised slightly.

It may even be possible to carefully cut through the transformer with say a bolt cutter, (who needs the old ones! NOBODY!!) and then remove pin for pin separately.

Even if in the worst case happens and a land is damaged, its easy to secure the transformer (I have even used hot glue for such repairs!) and then wire each leg with solid insulated copper wire.

If you do not have the skills, search out a small company to take on the job.

For a true professional (or even many good amateurs!), its no big deal.