• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Realistic inductance value in Output Transformer

What "formula" does these PP output transformers use ?

From piemme-elektra:
1. Push Pull 25W 6,6k: Total Primary inductance: > 80H @100Hz 5V
2. Push Pull 25W 6k UL tap 20%/4-8 Ohm: Total Primary inductance: > 175 H @50hz 5V
3. Push Pull O.T. 20W 8000/150 Ohm: Primary inductance: > 180H @20hz 5V
4. Push Pull 25W 10k UL tap: Primary inductance @20hz 5V: >130H

which have impressive inductances.

On the other hand,

5. Dynaco A470 from Robert McLean's spice transformer Model:
LP1 7.29664888730406, LS1 1.77011598090312

seems to have small inductances (in spice).
 
5. Dynaco A470 from Robert McLean's spice transformer Model:
LP1 7.29664888730406, LS1 1.77011598090312
Transformers are modelled in spice as a series of coupled inductors, which is somewhat non-intuitive. In reality, 'LS1' is reflected across to the primary by the turns ratio, bringing the 'effective' LP1 up to the expected value (I'm guessing about 50H)
 
Last edited:
  • Like
Reactions: Chris Hornbeck
The older (pre-computers) model of a central ideal transformer surrounded by parasitic elements seems both intuitive and complete (once we've accepted that the parasitic elements are variable with DC, signal level and frequency) so looks to this "innocent eye" to be do-able in an iterative computer model. Then again, don't really know.

Inductance varies with turns, so their ratios must vary with turns ratios.

All good fortune,
Chris
 
In reality, 'LS1' is reflected across to the primary by the turns ratio, bringing the 'effective' LP1 up to the expected value (I'm guessing about 50H)
Reviewing how Spice works it seems I was wrong about this. So yeah, I don't understand Robert's numbers either. This is why I use Tina TI instead of Spice; I'm a dumdum.
 
Last edited:
  • Like
Reactions: Chris Hornbeck
When you measure inductance with a cheap LC meter have in mind what has been said before.
Inductance of iron core coils drop with rising frequency and also drop with low excitation.
My cheap meter e.g. tests at 500 Hz and no more than one or two Volt amplitude.
The value you get with such a meter can be much lower than what you actually have at - say - 50 Hz and 100 V amplitude.
A simple alternative methode to get more relevant numbers could be to apply a 50 or 60 Hz AC voltage of your choice from a mains transformer and simply measure Voltage V and Current I with your multimeter.
Inductance can then be calculated from L = V / ( I * 2 * pi * F )
... ignoring DCR which should be much lower than inductive impedance anyhow ...
I have the same problem as the author of this post. I'm even looking for a similar inductance value for a single ended amplifier. I made a coil with 3120 turns on a core obtained from a common 60Hz transformer, standard EI84 lamination, sheet thickness of 0.5mm, central leg area of 1200mm2. According to calculations, I should have obtained around 60Henries, but using a cheap RLC meter I only obtained 18Henries, with practically no air gap.
I thought that my problem might be related to the type of iron I used but now I'm going to try to measure it the way you suggested, applying 100Vac or some other value.
 
I use a 12-volt transformer to provide 60-Hz excitation for the ohm's Law calculation. This is fine for push-pull or parafeed output transformers (with no DC). But for SET outputs or chokes, you need to add some DC to the test current. I: know of two simple ways to get this:

1) Some 25(?) years ago I acquired a few television vertical sweep output transformers, which are airgapped. Mine are 10:1 stepdown, so 12v output with the primary connected to the 120v powerline. They can handle around 150mADC on the secondary. I just put a DC source in series with the secondary. It's crude but good enough for most purposes - probably within 20% or so.

2) Even simpler is to rectify a variable AC source (e.g. a variac), whose output contains DC and AC. The AC is not sinusoidal, but is dominated by the second harmonic, 120Hz. This is even cruder, but often good enough. The AC/DC ratio is not adjustable but is close to saturation for SET outputs. I have not done this myself, but have seen the trick mentioned several times.
 
Last edited:
  • Like
Reactions: andredebacco
Hi Paul! Great ideas, I'll try them. I can use a variable DC source connected in series with the output of a variac and even see the effect of DC polarization! I will test and post the results. To measure the current you would need a true rms multimeter capable of measuring AC current superimposed on a DC level. I don't know if what I have can measure this. But I think you can already have a good idea of the value of inductance using L = V / ( I * 2 * pi * F )
 
here using my values of V=346 , A=3.68 , f1=20 , B=14KGauss , a much more reasonable value of 1302 turns is shown.

If the E/I transformer is for HiFi then I personally there is no way would I select Bmax at 1.4T; the situation is a little more complicated when it comes to asking how much F2 harmonic distortion one tolerates at LF cutoff freq. The Mullard 20W pp ( a large core) uses a Bmax of 0.7T for that very reason; hence the number of turns and excitation, impedance, and core size are all inter-related. For a very long time, the industry build standard for Si-Iron is 1.4% F2 distortion at the designated LF cutoff frequency. Naturally with global nfb that figure is reduced. So halve that 1.4T, and the core size basically doubles.
To cap it, I known the formula used for mains fransformer calculations used for MI transfomers. One can get-away with where harmonics and racuous distortion is preferred.

However, good luck if one is doing this for the first time. One needs the physics.
Bench baron
 
  • Like
Reactions: andredebacco
I decided to take the measurements on my single ended transfomer, taking all precautions with insulation, I used two 12V transformers connected in opposition to the Variac output! Well, I came to some conclusions:
1 -The inductance measured using a cheap LRC meter is approximately half the value measured by applying an AC voltage, measuring the current and calculating the inductance using L = V / ( I * 2 * pi * F ).
2 - I placed a DC polarization in series with the AC voltage using a 12V battery to see what happens to the inductance. BUT it is difficult to measure the result because my multimeter does not measure AC voltage with DC level, or do I only need to consider the AC part? And it also probably doesn't measure AC current with a superimposed DC current. BUT, taking only the AC part into account, the inductance increased slightly with DC polarization.
3- The good part is that I discovered that after applying DC polarization, my core was slightly magnetized 🙂
Anyone who wants can see the measurements in the attached pdf file. The value measured with a cheap meter was 18H (without gap)
 

Attachments

Hi Paul! Great ideas, I'll try them. I can use a variable DC source connected in series with the output of a variac and even see the effect of DC polarization! I will test and post the results. To measure the current you would need a true rms multimeter capable of measuring AC current superimposed on a DC level. I don't know if what I have can measure this. But I think you can already have a good idea of the value of inductance using L = V / ( I * 2 * pi * F )
Two problems with this. As kevinkr said, you would need an isolation transformer. And the variac does not have an airgap, so you can't run DC through it. The rectifier provides the necessary DC without running it through a winding.

As I said, I haven't tried this. But I dug up a document on it:

https://www.dalmura.com.au/static/Choke measurement.pdf
 
  • Like
Reactions: andredebacco
Two problems with this. As kevinkr said, you would need an isolation transformer. And the variac does not have an airgap, so you can't run DC through it. The rectifier provides the necessary DC without running it through a winding.

As I said, I haven't tried this. But I dug up a document on it:

https://www.dalmura.com.au/static/Choke measurement.pdf
Good news!

I performed measurements using a variation of the suggested setup. I used a variac, two transformers in opposition as isolators, a car battery to provide the DC polarization, which in my experiment dc bias was fixed, with a value of 35mA. I carried out voltage measurements on resistor Rs using a digital oscilloscope, reading the RMS values, I forgot to take photos but I will do so as it could be useful!

1735584096074.png



It was necessary to change the cell in the spreadsheet where the inductance is calculated because my frequency does not double, as I did not use the rectifier bridge

I used a fixed DC bias current of 35mA and made measurements with different AC voltage levels. Apparently the inductance increases as the AC voltage increases.

There was no problem putting a small DC current circulating through the isolating transformer, perfect waveforms. There was some noise level, so perhaps the value of 28.7 is incorrect.

In the last measurement in the table, I had 11mA AC and 35mA DC which resulted in 36H. The smaller inductance values may be due to the large difference between the DC bias and the AC current.

Thank you very much for the information! At the end of the this document: https://www.dalmura.com.au/static/Choke measurement.pdf there are very good references!
 
Last edited:
Late last year I compared different output transformer iron using the same winding, the test amp was a quickly knocked up single channel amp - I wound a copy of the Dynaco Z565 output transformer - stacked it with 'ordinary' 0.5mm power transformer iron and then ran some frequency response square wave and distortion tests. I ran the same tests after restacking it with M6 GOSS 0.35mm laminations - while I couldn't carry out any A/B listening tests, as I only had the one mono amp both sounded perfectly fine - there was a minor difference but it certainly wasn't night and day, most of the difference was in the bottom end performance below around 50Hz, increasing the stack size of the ordinary laminations, would certainly improve it's low end performance.
Your observations don't surprise me at all. Anyway, laminations of 0.35 mm thickness usually are stamped from GOSS sheets, usually M6X (or VM111-35, as it is called here), and allow higher flux densities than plain 0.5 mm dynamo steel laminations. Hence, turns counts can be reduced accordingly, wire diameters can be increased, resulting in lower DC (and also iron) losses, increased efficiency, lower inter and intra winding capacitances, ergo in extended high frequency response.

Best regards!
 
  • Like
Reactions: andredebacco