# Transformer winding disparity

1. Dec 2, 2016

### GeorgeCKR

I am attempting to reverse engineer a small 1960's vintage power transformer that is employed in a vintage audio amplifier. It contains no marking to indicate operating voltage or current ratings, however I do have several sample transformers as well an an operational example of the amplifier that employs it.

I have dissected a sample transformer to determine as much as possible about it's construction to estimate what its actual rating might be. It has three windings, including a single 120V primary winding and two secondary windings consisting of a high voltage winding and low voltage filament winding, which measured 640V and 12.7V respectively unloaded. I counted the number of turns of the low voltage secondary and primary windings, but not the HV since that proved to be impractical given the large number of windings involved, along with extremely thin wire embedded within layers of insulation. The 12.7 volt secondary was found to have 128 turns which implied about 10 turns/volt. Given this, I had assumed the number of primary turns should be in the vicinity of 1200 turns. However, the actual number turns I counted was only 810. Given this turns ratio, the low voltage secondary voltage should have been closer to 19V rather than the measured 12.7V (unloaded). I realize that there are losses in real transformers, however, the disparity in this case seems far to great. Am I missing something obvious here?

Using the core area I had measured, and applying the fundamental equation to determine the associated number of primary turns, assuming a flux density of 10,000 gauss, resulted in about 1100 turns. More in line with what I would have expected. I am just considering the transformer unloaded at this time, so secondary copper loss should not be a concern. The primary DC resistance is 41 ohms, and no load current is 31ma. I'm kind of at an impasses now as I can't seem to reconcile the disparity. Any ideas?

2. Dec 2, 2016

### The Electrician

Are the windings concentrically wound, or is this a split bobbin wound transformer? What is the secondary resistance? You might need to subtract the DC resistance of your ohmmeter leads if the secondary resistance is low, as is typical for a 12 volt winding. How much does the transformer weigh?

It would be good to determine the coupling coefficient. Apply voltage to the primary (120 VAC, presumably) and measure the exact value of the applied primary, and at the same time measure the exact value of the no-load voltage at the 12 volt secondary.

Now, apply a nominal 12 volts to the 12 volt secondary and measure it, and then measure the voltage appearing at the primary. If you don't have a variac to derive the 12 volts to apply to the 12 volt secondary, use one of your other samples to derive 12 volts.

Post and identify the results of your measurements.

3. Dec 2, 2016

### GeorgeCKR

The transformer is layer wound with the primary on the core, of course, then the HV winding on top of that, and finally the 12V winding on the outside layer. The 12V winding has a resistance of 1.09 ohms, the meter has the capability to null out the lead resistance. The transformer weighs 1.20 lbs complete, and 1.05 lbs with the bells removed.

With 120 VAC applied to primary, established with a variac, and measured, the measured voltage at the secondary was 12.54 VAC.

With 12.0 VAC applied to the secondary, again established via variac and measured, the output at the primary was 114.36 VAC
With 12.54 VAC applied to the secondary the output at the primary was 119.22 VAC.

Separate meters were used to monitor the primary and secondary voltages simultaneously. The variac was adjusted as required during measurements to maintain the required voltage. Line voltage fluctuates quite a bit in my area.

The voltage transformation ratio, of about 9.5, seems to hold in all cases.

4. Dec 2, 2016

### The Electrician

The transformer's weight would indicate, to a first approximation, a rating of 20 VA.

Is there enough room around the outer insulation to thread a few turns of small magnet wire around the center leg? Then with a known voltage applied to the 120 volt winding, and possibly the 12 volt winding, measure the volts/turn on the additional few turns of wire.

Perhaps you lost count of the primary turns while unwrapping it?

5. Dec 2, 2016

### GeorgeCKR

I agree with your rating assessment based on weight. I had already done some investigating in that regard. Based on the core area, though, the rating would be somewhat less. The reason I am going through this process is that it has often been claimed, on vintage audio forums, that the transformer is undersized in this application and is being overly stressed. The temperature testing I have done does not support that claim however, so I have proceeded with trying to determine an actual rating.

In order to try adding extra windings around the center leg I would have to partly disassemble another transformer since the one I dissected is totally disassembled right down to a pile of laminations. Not a huge problem though, I have several, so I will investigate that.

Could I have miscounted the turns? Well, it's a possibility since the primary winding was a tedious chore going through numerous layers. The disparity in the turns ratio makes me think that may have occurred, and yet, going by the wire guage, resistance, and bobbin size, I estimated the number of turns, which still came out reasonably close to the count. It could be that I will go through the counting process yet again.

6. Dec 2, 2016

### The Electrician

What really determines a transformer's rating is the hot spot temperature. You probably don't know the insulation class, but for consumer transformers, class A is typical: http://www.stancor.com/pdfs/Catalog_2006/Pg_g.pdf, with a max hotspot of 105C. The presence of two secondary windings makes it a little harder, but if you load the transformer (both secondaries) for a couple of hours and calculate the internal temperature rise by measuring the resistance of the windings hot and cold, you can get an idea of what load will cause the winding hot spots to reach 105C.

You ought to be able to snake a few turns of 30 gauge through there. What I do when it appears to be too tight is to take a bamboo chopstick and rasp the big end to a thin wedge. Then use that wedge to force open some space between the outer paper insulation on the windings and the outer leg of the core. Bamboo is strong enough, but not hard enough to damage formvar if it should come into contact with the wire.

7. Dec 2, 2016

### GeorgeCKR

I had already done the temperature rise by wire resistance change procedure, with the transformer operating fully loaded in its intended application. At room temperature, with the covers removed from the amplifier, and then again with covers installed. Two thermocouples were used to monitor both the internal ambient temperature, as well as the case temperature of the transformer. Short, heavy wires were connected to the primary winding and brought out externally to monitor the winding temperature. With a room temperature of 21.7C the internal ambient temperature (covers on) rose to to 37.3C and the transformer case was 52.2C. The primary winding temperature, based on the resistance change, was 62.32C. Extrapolated to 40C, the winding temperature would be 65.35C. So.with a 10C hotspot allowance, that would be 75.35C in its usual operating environment. I had assumed a 105C insulation class, with an allowable 55C rise above 40C ambient plus 10C hotspot margin. So there should still be a fair margin before the transformer was overly stressed. At room temperature, with the amplifier covers removed, the transformer is not hot to the touch, which I would have thought it would be if it was really being pushed hard. I would consider it to be more of a pleasant hand warmer on a cold day.

8. Dec 2, 2016

### The Electrician

Sounds like you've done all the right testing. Now the only question remaining is the primary turns discrepancy.

Another test you could do along those lines is to measure the inductance of the primary and the 12 volt secondary. Can you remove the laminations from one of the sample transformers without ruining the windings? If you can do that, a measurement of the inductances with no core in place gives you a way to determine how many turns are on the primary, assuming the number you have for the 12 volt secondary is correct. The measurement would need to be made with an LCR meter that uses sinusoidal excitation for the measurement at, say, 120 Hz.

I have tried to make measurements of inductance with the core in place, but the results are not good unless you use a meter that allows a wide range of excitation voltage. By adjusting the excitation voltage to be proportional to the number of turns of a winding, the flux density in the core can be the same for the two measurements. You might say, when the number of turns in a winding is what is to be determined, how can the measurement voltage be adjusted to be proportional to the turns? A guessed value would have to be used, and there's the problem with trying to make the measurement with the core in place. Or, when making the measurement on one winding, measure the actual voltage applied to the winding by the LCR meter, and also measure the voltage seen at the other winding. Then apply the same excitation voltage measured at the other winding when making the measurement there.

Last edited: Dec 2, 2016
9. Dec 2, 2016

### The Electrician

Do you have a true wattmeter so that you can measure the core loss with no load on the transformer? Some consumer grade transformers can get almost too hot to touch, but they are probably using a low grade transformer steel for the core. Transformers using M6 laminations and class A insulation shouldn't get too hot to touch in normal room ambient.

What is the DC resistance of the HV secondary, and the expected load on it?

10. Dec 2, 2016

### GeorgeCKR

Unfortunately, I do not have a wattmeter. I know that it is common to employ one when trying to determine core loss since the reactive component, and phase shift, involved makes it a bit ugly to deduce otherwise. Just never had the need to go this far into that particular aspect of a transformer, until now perhaps. I did wonder about core loss though, but then thought that surely it could not be so great as to account for such a winding disparity. Plus the transformer does not get that hot. But then again, transformers are not my particular area of expertise, which is the main reason I am here looking for some insights. Basic transformer theory is quite straightforward, but this one seemed to be defying physics.

As to the HV secondary. The DC resistance is 3K and it is center tapped, so 1.5K either side of the tap. It is employed in a CT full wave rectifier, with capacitor loading, followed by another simple RC filter. The DC load current is 6.8mA. The load on the 12V secondary is 1.15A.

I actually had tried measuring the inductance, with the core in place, and did not come up with results that felt I could trust. I do have an LC meter, which did not read consistently. I also tried things like creating a resonant circuit with external capacitors, and signal generator, but again, the results were not conclusive. What I did not do was try to measure the inductance after removing the core. For some reason, I just did not think to do that before removing the windings. Thank you for bringing that up. I can remove the laminations without ruining the windings, and did so with the transformer I dissected. Tomorrow I will first attempt to add the additional windings as you suggested, and take some measurements. Then I'll remove the core and measure the inductance. I expect I will get proper readings with the core removed. The final step, after exhausting all other options, would be to proceed with counting the windings again. I'm beginning to suspect that will turn out to be embarrassing.

11. Dec 2, 2016

### The Electrician

What is the model of your LC meter?

12. Dec 2, 2016

### The Electrician

I wasn't thinking that core loss had anything to do with the turns discrepancy, but your finding that the transformer doesn't get very hot under load made me think that it's a high quality core, which could be confirmed by the core loss measurement.

I'll dig out a small transformer of about that weight and measure the unloaded core loss, just for grins.

13. Dec 3, 2016

### GeorgeCKR

It would have been nice to be able to determine the core loss, just to fully characterize the transformer, since things have gone this far. But I'm not sure how practical that would be given the equipment I have to work with. I can measure RMS current and voltage, I have a scope (analog) and a phase meter (Wiltron). Also various signal generators up to 1GHz and a function generator.

I just removed the casing and bells from another transformer sample. The window is fully jam packed edge to edge, not even a pin hole to try to get a few turns of wire around the center leg. I suspect I will damage something if I try to force open enough space. The construction is very solid and sturdy. I may just have to go on to removing the laminations so I can try the measure the inductance.

The LC meter I have is actually a board level product (from China), so no model number to so speak. I had purchased it specifically for an RF related project in order to measure very small inductance (nH - uH) were it seemed to be very accurate. I had never used it to measure large inductance, although it has a claimed range of 100H. The largest inductor, of known value, that I have on hand is 1mH, which it measures correctly. When I tried to measure the inductance of the transformer (with the core) the readout was unstable so I decided to discount it. The meter obviously was not happy trying to measure it. It does display the frequency being used, which is variable depending upon the range, but I don't know that it's sinusoidal. I could check that with a scope.

Last edited: Dec 3, 2016
14. Dec 7, 2016

### GeorgeCKR

It's time to fess up. It finally did come down to performing a recount on another transformer sample. It turns out I had, indeed, made an error in the count, and on the winding with the least turns no less. Still not sure how it happened, but it did. The real count for the 12V secondary is 94 turns, not the 128 turns I had originally thought. That's quite a difference, but I am certain the latest count is correct. I was extremely careful this time. Likely was rushing too much the first time, and some kind of brain fade happened along the way. Now the numbers work out much better. Thanks Electrician for your time and input, much appreciated.

I did measure the inductance, core removed, for all windings, before doing the last turns count, but I'm not sure it matters at this point, now that the error has been discovered. I am now working with yet another transformer sample to see what I can do to determine what the core losses might be using the equipment I have on hand. I looked at the voltage and current phase shift using my scope. The intent being, if I could determine the phase shift, I could calculate the no load power using Vrms x Irms x Cos shift. However, the current is not sinusoidal, so the phase shift relative to voltage is not constant. The shift at the peaks is much different than the shift at zero crossing, for example, so I'm not quite sure exactly how to consider that. I guess this is what makes a wattmeter so desirable.

15. Dec 7, 2016

### The Electrician

The typical core loss for a 20 VA transformer will be only 1 or 2 watts, so a wattmeter that measures rather low values is needed.

This is a low cost possibility: https://www.amazon.com/P3-P4400-Ele...id=1481166413&sr=1-1&keywords=killawatt+meter

but it can't measure such low power well unless you open it up and change the sense resistor.

If your scope is new enough it should have trace math and you can measure the product of voltage and current and average to get true watts.

Sometimes you can find one of these: http://www.ebay.com/itm/Vintage-NIB...226896?hash=item2378202110:g:iEMAAOSwo4pYP8gV

with a suitably low full scale range. I have one that's 15 watts full scale I got in the last year or two.

Last edited by a moderator: May 8, 2017
16. Dec 8, 2016

### GeorgeCKR

It would seem that the typical low cost consumer grade power meters will have poor resolution at low power levels. There's lots of them available, and some really low cost ones on ebay. Some use a sense resistor and others a current transformer, but all are intended for higher power than what I would be measuring. Changing the sense resistor to increase the sensitivity seems like something worth trying though. I will consider going this route since it's not a big expense to give it a try.

My scope is not new enough to include trace math, so all I can do is observe the waveforms. Had both been sinusoidal, I could have come up with a reasonably close estimation of the phase shift, but since the current is not, well, coming up with a meaningful average would be a challenge.

The mechanical type wattmeters, for low power ranges, would seem to be difficult to find. Just not enough demand I would guess.

17. Dec 8, 2016

### The Electrician

18. Dec 9, 2016

### GeorgeCKR

Thanks for the links. The comparison of your scope related readings to the modified Kill a Watt was very interesting. I found one on Amazon for 10.99 and ordered it. From reading the comments of others who have performed the modification, it would seem there is some variation in the value of the shunt resistor, so I would verify that value as a first step. I do have access to a good milliohmeter for that purpose.

I'm jealous of your scope though. I've been putting off getting a digital scope, but getting real tempted now. My trusty old Tek 465 still works great for most of the stuff I need to look at, but those math functions would be wonderful to have at times.

19. Dec 9, 2016

### The Electrician

You shouldn't have too much trouble with the modification. I do recommend the comment of one of the posters to put a couple of 1N5400 series diodes across the new shunt so that if you forget and plug a kilowatt load in, the shunt doesn't get burned up!

I wrote on the face of mine with a marker that it is the high sensitivity modified unit.

20. Dec 9, 2016

### GeorgeCKR

I've seen internal photos of the Kill a Watt, and it looks like it should not be too difficult to modify. I just happen to have some 5408 diodes on hand, which should do just fine. I will be in wait mode until the meter arrives, then I'll assess the shunt resistor value. Looking forward to playing with this, and thanks again for the useful info. I'll report back when I get the meter.