Reducing Power Dissipation in Back-to-Back Transformers

[Rob Strand]

But I suspect I'm overthinking it. Seems to me all you're trying to do is put a cap in parallel with the magnetising inductance, selected to resonate at 50/60Hz. This can be a big cap on the low voltage side, or a small cap on the high voltage side -the latter is therefore easier since it will be a more accurate cap.*

I don't think you are over thinking it.   It is a valid point.   It's going to come down to specifics. 
The way I look at harmonics is they are orthogonal to the fundamental because of the frequency. However reactive power in the true sense is orthogonal because of the phase and the energy is store and released.   When resistances are present both increase losses.

Anyway even though the current pulses are more or less aligned with the fundamental, they aren't perfectly aligned and this might make all the difference to stuffing up the resonance.   When the diodes are not conducting the circuit does look like the  just the L and C sort of thing.   However when the diode conducts it's whole different story.   If you (roughly) think about the circuit as a switched conductance filter the primary resistance of tx2 (the one at the HV rectifier) is *permanently* connected to the filter cap through a larger resistor, one which has value Rp2' = Rp2 * (T / t_on).   So that could have an effect.   If it does is probably means the larger LV cap versions is slightly better.   The other way to "fix" that would to add series resistance to Rp2.   These circuit have a light load so that won't be a big deal.

Anyway Re using the smaller cap.  My second longer post covered that.   Without the rectifier present it is largely the same behaviour.

There's plenty of details to consider and things that might need to be massaged to get it to work in practice.   It might even fall in a heap  .

[PRR]

> The 12V 150mA one on this page is close.

I must say I have not seen that type, with the fully plastic-enclosed primary, sold loose in the US. (I may have seen them inside products.)

Your examples are both Australian. I have not seen mag-current specs on most US-market parts. Do your electrical boffins have to account for this current? (And demand the full-enclosed winding?)

The specs look weird. 1.9VA, but 8% regulation?? I would expect >20% sag in such a tiny wind.

The low sag suggests fairly fat copper. The plastic in primary space says not a full dose of copper. This suggests not so many turns, and low inductance. Which agrees with the specified magnetizing current.

Primary load current: 2.05VA/240V= 8.5mA
Max magnetizing current: 15mA
Assuming mag-current is "typical", most of the primary current is reactive. Assuming "Max" is twice typical, load and mag currents are similar.

This does suggest a 1:1 config (no load on low-volt) might indeed run hot in the first primary.

Most of our uses for this hack are for small tubes where heater power is greatly in excess of plate power. 12AX7 runs say 1mA/plate, 0.6W of B+, and 1.8W of heater. The first transformer should be selected >2.4VA. A second tranny at 0.9VA might not be a problem; but we are at the limit of what can be wound cheaply. Simplicity suggests two of same 2.4VA transformers. Indeed the heat from double magnetizing current may be more than designed.

My path is to pick the first as 2X the estimated total. Prices in this range are soft, 2X VA will not cost 2X bucks. However a buck is a buck and another quarter inch may not fit. And if there is a "close-out", the "smart" buyer will opt for two of the same of whatever is on-sale.

OTOH--- note the Altronics PDF page wants a primary fuse for 6VA to 15VA, but not on the 2VA to 4.5V parts. This suggests they are "intrinsically safe", can be shorted forever. (All have thermal fuses too, in accord with local regs, but a bad-enough *small* transformer can't hurt itself from external shorts, only internal flaws.)

[R.G.]

For some reason my post didn't show up here. I did some sim-ing and got odd results.

The effect of leakage inductance gets huge on small transformers. It is especially bad on side-by-side wound transformers, and it seems to be the style to do that on all smaller VA transformers. Spacing side-by-side primary and secondary windings increases it even more. At some illogically large spacings, the secondary isn't coupled to the primary at all any more. The leakage between the two coils means that all the field leaks before it gets into the secondary.

I suspect that the truly tiny transformers down in the units of VA range rely on both the primary resistance and the high leakage of side by side windings to decouple the primary and secondary under shorted-secondary conditions.

Silly me. I was sim-ing a 12V/1A trannie.

Hmmm. A 12AX7 needs 12V at 300ma for the heater, and about 1-2ma per plate, max at 100-300V. So if you do a primary to 12Vac for heaters, you need 12*0.3 = 3.6VA for a heater. The high voltage end needs at least 100V at 3ma, so with rectifier RMS losses, that runs out to about 0.48W out of the second trannie. That goes up linearly with the output voltage on the second trannie, so if you're using 240V primaries, you'd get about 1W out of the second trannie.

So at a minimum, a two-transformer setup needs to supply something like 4-5VA per 12AX7. Maybe an asymmetrical setup with a 4-6VA first transformer and a dinky 2VA second transformer would work, but in these sizes, you're long past smaller VAs making the transformers cheaper. All you'd save is likely to be a little weight, and the weight saving for tiny transformers isn't that big, either.

Here's a thought: if you have to use two transformers anyway, why not make one of them be 120:120 (or 240:120) for the high voltage, and the other 120 (or 240) to 12V or 6v for the heaters?

[PRR]

I too just lost a post, while the forum did an endless TLS handshake. Is it raining somewhere?

> A 12AX7 needs 12V at 300ma for the heater

No. 12V @ 0.15A, or 6V @ 0.3A. 1.89 Watts per bottle.
http://www.mif.pg.gda.pl/homepages/frank/sheets/127/1/12AX7.pdf

Other than the usual 2X mis-count, your math seems fine.

> why not make one of them be 120:120

While the catalogs go up that high, it often turns out there is No Stock of the 120V end of the range. 6V/12V iron is always in deep stock.

[R.G.]

Doh. Yes, on 12V it uses 150ma. It's when it's when it's on 6V that is uses 300ma.

I guess it's been too long since I've designed in a 12AX7. 

[Rob Strand]

Excuse my absence.   Give me a couple of days to get back on track.
I've got the flu + throat infection.
Computer is broken and I am broken - life is real great atm.

[R.G.]

My condolences. Flu is no laughing matter. Get better - the transformers will be waiting for you.

[PRR]

Have weak honey and vinegar tea. (Manuka Honey is costly but special.) I'll have one with you. Get better!!

[Rob Strand]

Thanks for the well wishes.  I'm nearly back on track.   It's not a good when it takes 40 mins to eat breakfast cereal you would normally scoff down in 3 or 4 mins if you had to rush off!

Just got my original drive back up and running with a lot of effort, tricks and help from some in-depth pages on the web (I'm indebted to the guys who wrote those pages!).   There was no way I was going to attempt to fix that problem with a foggy head.

[Rob Strand]

> The 12V 150mA one on this page is close.

I must say I have not seen that type, with the fully plastic-enclosed primary, sold loose in the US. (I may have seen them inside products.)

Your examples are both Australian. I have not seen mag-current specs on most US-market parts. Do your electrical boffins have to account for this current? (And demand the full-enclosed winding?)

The way these things have been build over the last 45 years has varied enormously.     The plastic around the primary is a recent addition.   So is quoting Imag.   This definitely was not quoted in the old days, not even by the more reputable brands.

These days I do see Imag being quoted here and there.  RS Components were quoting Imag on a number of transformers.  Other specs are indirect in that they quote power dissipation at no load and maximum efficiency.

The specs look weird. 1.9VA, but 8% regulation?? I would expect >20% sag in such a tiny wind.

The low sag suggests fairly fat copper. The plastic in primary space says not a full dose of copper. This suggests not so many turns, and low inductance. Which agrees with the specified magnetizing current.

Yes that's exactly what is happening.   Over history, not all versions had 8% reg.  The trick to get better regulation is have a low turns and high Imag.  It actually works well because the load current is out pf phase with Imag. Suppose the full load output reference to the primary is Is' =Imag then the drop across the primary is only Rp * sqrt(Imag^2 + Is'^2) = 1.4*Rp*Imag, so the change in the voltage drops is 0.4*Rp*Is'.  If Imag was small we would see a drop of Rs * Is'.   A further trick used by some era's of the transformer was to drive the iron close to saturation.  That pushes-up the no load Imag.  However when a load is placed across the transformer the drop across primary resistance Rp cause the depth of saturation to be reduced and that less the current shifts from Imag to Is' with lower change in the drop across Rp.

This does suggest a 1:1 config (no load on low-volt) might indeed run hot in the first primary.

I'm not sure how hot the Altronics ones get.   For the others in the past the smaller core-sizes got hot and the larger core sizes were not very hot at all.

Most of our uses for this hack are for small tubes where heater power is greatly in excess of plate power. 12AX7 runs say 1mA/plate, 0.6W of B+, and 1.8W of heater. The first transformer should be selected >2.4VA. A second tranny at 0.9VA might not be a problem; but we are at the limit of what can be wound cheaply. Simplicity suggests two of same 2.4VA transformers. Indeed the heat from double magnetizing current may be more than designed.

My path is to pick the first as 2X the estimated total. Prices in this range are soft, 2X VA will not cost 2X bucks. However a buck is a buck and another quarter inch may not fit. And if there is a "close-out", the "smart" buyer will opt for two of the same of whatever is on-sale.

IMHO, connecting one TX to another back to back is pot luck.  You don't know how much free capacity is left because Imag of the second transformer can use-up or exceed the capacity of the first one.   That's what inspired the capacitor solution.  It's a means of gaining enough capacity back so a small load can be placed on the second transformer.

This suggests they are "intrinsically safe", can be shorted forever. (All have thermal fuses too, in accord with local regs, but a bad-enough *small* transformer can't hurt itself from external shorts, only internal flaws.)

I think they rely entirely on thermal fuses and cannot be shorted forever (eventually they will overheat an blow the thermal fuse).   There's be a movement to use thermal fuses for small Tx's.   However, yes, there are some which are designed for indefinite shorts.   They tend to be the ones with poor regulation.   IIRC the 1.9VA units don't quite make it; not unless you drop the ambient temperature.

[Rob Strand]

FWIW,  I did some back of the envelope calculations regarding the small 2VA transformers.   Based on the typical DC resistances found on these transformers it seems if we try to optimize the number of turns in order to produce the least power dissipation from the transformer the magnetizing current comes out around 10mA to 12mA or so for 2VA.   It depends a lot on the iron but at least these numbers point to some sort of optimum close to what we see in practice.   I'm still stretching to  believe the high value Imags from Altronics though.

[Rob Strand]

Just to get ball rolling: To the output side of Tx2 I added a bridge rectifier, 10uF cap and approx 1mA load using the circuit with the HV cap.

I then plotted the total dissipated power from a transient analysis:

Click to Enlarge:


From these results it shows:
- pretty clear the added cap does in fact reduce the power dissipation on Tx1 quite a bit.
- the added cap required for minimum power dissipation in Tx1 is quite close to cap required when Tx2 is unloaded.
- changing the value confirms we are close the optimum for minimum power dissipation in Tx1

So at this point I'd conclude the idea of adding the cap does in fact work when a rectifier load +filter is present.

[Rob Strand]

The effect of leakage inductance gets huge on small transformers. It is especially bad on side-by-side wound transformers, and it seems to be the style to do that on all smaller VA transformers. Spacing side-by-side primary and secondary windings increases it even more. At some illogically large spacings, the secondary isn't coupled to the primary at all any more. The leakage between the two coils means that all the field leaks before it gets into the secondary.

I'll have to have a look at that.

[Perfboard Patcher]

Guys,

I have a problem, my preamp is working wonderfully well, am I required to post the results of some measurements?     

I've added a cap at the high voltage side of the second transformer of my preamp but unfortunately didn't gain much from it. When adding a 100nF cap the voltage went up from 238 Volts to 240 Volts. But it might as well be the change of voltage from the supply line over time.

Then I did some measurements.

Both of my transformers are Amplimo ringcores, the first one is a 2x12V, the second one is a 2x15V.
The single ECC83 in the preamp was disconnected from high voltage but filaments and other s.s. circuitry on -/+ 12V were still connected. The secondary of T2 was still connected to the rectifier (4 s.s. diodes,cap) + r.c. filter (10k,44uF) + resistor for slow discharge (1M1).

Voltage
Primary T2: 28V AC
Secondary T2: 182V AC
Rectified high voltage: 245V

That's lower than I expected, 182 *sqrt(2) -2 = 255 or 230 x12/15 *sqrt(2) -2 = 258.

Current
But the thing I don't understand is the AC current through primary T2: Too low to be bothered! Some inconsistent readouts way below 1mA in 20mA and 200mA position of the multimeter. Actually there's a more substantial flow of current through primary T2 right after the preamp is switched on but my digital multimeter doesn't allow me to figure out how much peak current exactly.

Will the voltage up transformer (T2) cause a higher peak current to occur?
Is there a relation between lower rectified voltage than expected and low current through primary T2?


Some other reasoning to get some peace of mind...
Neither one of the transformers seems to heat up when the preamp is on. I don't get a sensation of power dissipation when I touch the transformers.
I've measured the 2x12V when I purchased it, it was 28 volts unloaded at the secondary.
The datasheet for T1 mentions: 2x 12V/ 1.25A and "All voltages are under full ohmic load". I interpret this information
that in case the 24V secondary was loaded by a 19.2 ohms resistor 1.25A would flow and the voltage across the secondary should most likely have dropped from 28V to 24V. As shown, the primary of T2/ secondary of T1 measured 28V AC, not a sign of a stressed out T1.

cheers,
PP

[Rob Strand]

I've added a cap at the high voltage side of the second transformer of my preamp but unfortunately didn't gain much from it. When adding a 100nF cap the voltage went up from 238 Volts to 240 Volts. But it might as well be the change of voltage from the supply line over time.

The aim is to reduce the input current and stop Tx1 overheating, not to increase the output voltage.

The cap value needs to be tuned to suit your transformer.  The optimum value could be vastly different to what I have posted.   I only posted an example.   (For the sake of others,  a 120V transformer equivalent to my example would required a cap size will be 4 times without considering different transformers.)

A simple way to tune it would be to measure the input current then adjust the cap size until the minimum current is found.   This requires mains measurements.

But the thing I don't understand is the AC current through primary T2: Too low to be bothered! Some inconsistent readouts way below 1mA in 20mA and 200mA position of the multimeter. Actually there's a more substantial flow of current through primary T2 right after the preamp is switched on but my digital multimeter doesn't allow me to figure out how much peak current exactly.

Maybe measure the DC current on the HV side first to get an idea what load current you are dealing with.  The rms AC current is typically just under two times the DC current.

Neither one of the transformers seems to heat up when the preamp is on. I don't get a sensation of power dissipation when I touch the transformers.

That's a good sign but you often have to wait 1hr or more for things to heat-up.

I've measured the 2x12V when I purchased it, it was 28 volts unloaded at the secondary.

That's a fairly typical result.

The datasheet for T1 mentions: 2x 12V/ 1.25A and "All voltages are under full ohmic load". I interpret this information
that in case the 24V secondary was loaded by a 19.2 ohms resistor 1.25A would flow and the voltage across the secondary should most likely have dropped from 28V to 24V.

Yes transformer ratings apply to resistive load.  Normally the voltage is higher than the nominal voltage with no load then drops to the rated voltage when the maximum resistive load is applied.   Some don't quite work this way.

As shown, the primary of T2/ secondary of T1 measured 28V AC, not a sign of a stressed out T1.

Given you measured the no load voltage then yes it looks like it's not dropping much and being overloaded.


--------------
Edit:
It just occurred to me your transformer is 30VA.   When you get to larger transformers the back-to-back loading issue becomes less of a problem because the magnetizing current to the rated current ratio becomes smaller.

Is this the transformer?

https://www.amplimo.nl/images/downloads/ds%20standardrange/17012.pdf

(or maybe
https://www.amplimo.nl/images/downloads/ds%20standardrange/18612.pdf
)

[Rob Strand]

I found this data on toroid transformers, see page 7
http://www.nuvotem.com/en/products/pdf/Trafo%20Catalog%20-%20English%20Feb-08.pdf

In the table look at the "No Load" current (presumably for 230V configuration) which is effectively the magnetizing current.   15VA: Imag = 2mA,   30VA, Imag = 2.8mA.    These are extremely low in comparison to the 2VA EI core transformer I posted with Imag = 15mA.

[Rob Strand]

I found these as well, see series 62000 (not 70000 as the file name implies) on p24 (p25 of PDF).  Unfortunately no longer stocked by digikey,
https://media.digikey.com/pdf/Data%20Sheets/Amveco%20PDFs/70000_Series_Cat.pdf

They are small toroids and all have very low no-load current (Imag).

[Rob Strand]

I did another test with a rectifier and load.   This time I increase the output current to around the maximum the transformer can take (6.5mA on the DC output).

Click to Enlarge:


This test shows the cap value that minimizes the power dissipation in Tx1 is smaller than that for the unloaded (and lightly loaded) cases.   The conclusion is: as you increase the load the rectifier and filter does have an effect and it does affect the resonant tuning.  However all we need to do is use a smaller cap.   The power dissipation in Tx1 is reduced with the added cap.

[PRR]

Antek tests their stuff and the criteria are published.

1500VA Toroidal Transformers
Open Circuit Test (core loss test):
TEST CONDITION: Apply variable voltage to primary coils (in parallel). Set voltages 120 and 140VAC at 60Hz. No load on secondary coils. Measure the primary current and input power.
Voltage input -- 120V   140V
Current input -- .09A    .12A
Power lost ----- 12W     17W
http://www.antekinc.com/an-15445-1500va-445v-transformer/
http://www.antekinc.com/content/AN-15445.pdf

400VA Toroidal Transformers
Voltage input -- 120V   140V
Current input -- .04A    .09A
Power lost ----- 4W     7W
http://www.antekinc.com/an-4tk400-400va-400v-transformer/
http://www.antekinc.com/content/AN-4TK400.pdf

25VA Toroidal Transformers
Voltage input -- 120V   140V
Current input -- .01A    .01A
Power lost ----- 0.5W    0.5W
http://www.antekinc.com/an-0215-25va-15v-transformer/
http://www.antekinc.com/content/AN-0215.pdf

(no no-load data on the 10VA part)

Note that 60:1 range of rated VA is only 9:1 range of no-load current.

[Rob Strand]

25VA Toroidal Transformers
Voltage input -- 120V   140V
Current input -- .01A    .01A
Power lost ----- 0.5W    0.5W

Effectively the same values as the Amveco  (25VA, 230V, 5mA => equivalent to 10mA @ 115V).

I remember looking at ratings, resistances, temp rise, and weights of toroids in the past and they do tend to fall into a much tighter box, in terms of parameters, than EI cores.

Note that 60:1 range of rated VA is only 9:1 range of no-load current.

Yes, it's not proportional.  Small Tx's always have relatively higher Imag.