I'm was offered with design services but (call me crazy) I refused. Designers insist that toroid audio transformer must have only one impedance output tap (e.g. 5 or 8 Ohm), the same like Plitron. Reason is very simple - not enough space on toroid's window to fit all layers necessary.
So, after all, I sent to factory core size, material, schematic & winding layout diagrams which I have chosen/calculated myself.
I do not need 100KHz upper frequency limit, the sine-wave generator with 5 Ohm output has it, so I tested transformer at entire bandwidth available.
The PP 6L6GC amp I'm about to build has primary split into 2 parts - main + CFB (cathode feedback), it has upper limit of 50 KHz. This kind of output stage design is very sensitive to the phase shift (vulnerable to oscillation). Phase shift and upper frequency limit are depending on leakage inductance.
50W was actually OK for 6L6GC. A little reserve yields 60 - 70W (and better handling of DC offset). As I said before, selection of cores was rather limited (in fact it was more then enough for power transformers, but building audio-frequency units dictates some additional requirements). 15sm toroid core had effective cross-section area to build 150W unit with 1.5T flux density. Since I do not need 150W, and 1.5T is nowhere optimal value, objective was set at 70W and 8500G of max flux density.
The secondary was calculated by 2 target parameters - DC resistance <= 0.4 Ohm at 8 Ohm tap and current density <= 3A/mm^2.
Target primary Rdc was set at 150 Ohm (in finished unit measures at 136 Ohm).
The program I wrote calculates leakage inductance/upper frequency limit based on transformer winding data and geometry. Unfortunately, ONLY for EI or double C cores, in other words, for fixed bobbin length and fixed number of turns per layer. I wrote a subroutine which simulated toroid winding process (incl. swelling factor) - each next layer has smaller number of turns/shorter "bobbin" length, so I can check if winding and insulation between layers will fit or not. Unfortunately, formulas I have require fixed values for calculating leakage inductance. IMHO its possible to take averages derived from winding process simulation, but who knows how correct it will be ??? I am was unable to measure leakage inductance with high-precision professional L-meter owned by my friend because it suddenly get broken.
Anyway, I'm now one step closer to building real amp. Today evening I finished 350-550V regulated stabilized power supply. On the weekend I probably will have draft assembly of one channel of that PP unit.
Hi George,
You rock!
It will be very interesting to see how these do in the real amp compared to the bench measurement.
Stay tuned...
I have a pair of really big (12 pound) Plitron toroidal OPTS. They are rated for "400 watts @ 20Hz" They also have a multi tapped secondary (2,4,8 ohms). Someday I will build an amp to use them to their fullest potential.
Are you sure they are only 12 lbs? 12 lbs is not that big, its 5.5 kg. Main weight 8 kg, that's 17.63 lbs.
Correction to earlier post:
No, I was refering to Rl. Load resistance. Plate Z, Output Z, Output R, Load R,... I lump these terms, perhaps a bit loosely. Plate R can't be used to match a tranny primary Z.
I still don't follow this. Are you saying that you lump "Plate Z" and "Load R" together? How do you define damping factor of a power amplifier?
Also, why do you believe it's a goal to match the tranny primary Z? Lundahl, for example, specify their 162x output transformers using a 650 ohm driving impedance (=plate resistance). That seems to make sense if you have to pick some value, because it is more representative of actual circuit conditions.
Thanks,
Michael
What everyone says is right about using a realistic load.
But why not take advantage of the fact that transformers are symmetric? Put your signal generator on the secondary.
You should be able to use a pair or diodes and resistors to make a good simulation of the vacuum tubes in push-pull. I'd think a diode and 5K resistor would act a lot like a tube
The next thing, in an output transformer frequency response is influenced by power. At a few milliwatts you will measure very good response but maybe you care to know the response near the transformer's power limit. You might be able to use any old solid state power amp on the secondary.
But why not take advantage of the fact that transformers are symmetric? Put your signal generator on the secondary.
You should be able to use a pair or diodes and resistors to make a good simulation of the vacuum tubes in push-pull. I'd think a diode and 5K resistor would act a lot like a tube
The next thing, in an output transformer frequency response is influenced by power. At a few milliwatts you will measure very good response but maybe you care to know the response near the transformer's power limit. You might be able to use any old solid state power amp on the secondary.
Re: driving the transformer from the secondary
On further thought, a 5K resistor won't act like a tube, but it will act as a reflected load, which is what you want.
I think one would drive the transformer secondary with approximately the amplifier's expected output impedance (output resistance), (e.g. 1 ohm) and load the primary with approximately the expected reflected load impedance (e.g. 1K25 for a 5K OPT in class AB).
Finally a use for the old Hafler
Hmm power testing needs power 5K (or 1K25) load resistors
Cheers,
Michael
Last edited:
This thread started as a troubleshooting query of a transformer FR test. My understanding of the setup for that is to determine the Primary Z first and then place that R in series to simulate the tube's Load R.
I believe it's a goal to match the tranny Primary Z for all the basic electronics theory that says impedance matching is necessary to provide the highest level of circuit efficiency. That is the goal of the FR test.
If you want to put any tube in the socket then shack your booty till the woofers blow, go for it!
I think when they say driving impedence, they mean Load Resistance. Which is not the same as Plate R.
For the sake of argument lets say you were building a class A PP 2A3 amp and decided on a 5K transformer.
If you measured the transformer as you suggest and then build the amp and measure it again you may see somewhat different results. If you then remeasure the transformer on the bench with balanced drive and series resistance to equal 800 ohms with proper attention to grounding your results will be much closer to the actual in circuit performance.
You can choose to measure transformers in any method that you wish, but unless you intend to drive a 5K transformer with a 5K source, the testing method you suggest will not give you meaningful results.
dave
If you measured the transformer as you suggest and then build the amp and measure it again you may see somewhat different results. If you then remeasure the transformer on the bench with balanced drive and series resistance to equal 800 ohms with proper attention to grounding your results will be much closer to the actual in circuit performance.
You can choose to measure transformers in any method that you wish, but unless you intend to drive a 5K transformer with a 5K source, the testing method you suggest will not give you meaningful results.
dave
He tested his on the bench.
Tranny winders build them and give you a Z rating and a bandwidth rating and they haven't the foggiest notion of what kind of circuit you might hang on their perfect iron.
They should give you an impedance ratio, a value of induction for a given voltage/frequency couple, a primary inductance for a given (preferably the same) V/F couple, a leakage curve (no, its not a pure inductance !) and a summary of DC resistance of the winding . . . at least.
All things that do not depend on how you use it.
That given, you will be able to evaluate which load your tubes will really see when loaded by a perfectly caracterized loudspeaker system. 😀
Yves.
All things that do not depend on how you use it.
That given, you will be able to evaluate which load your tubes will really see when loaded by a perfectly caracterized loudspeaker system. 😀
Yves.
Measuring Leakage Inductance & Stray Capacitance
I took RLC meter from my friend, and made several measurements listed below:
Primary resistance - 136 Ohm
Secondary Resistance < 0.4 Ohm at 8 Ohm tap
Primary inductance:
100 Hz 1V - 102 H
100 Hz 50mV - 77 H
1 KHz - out of range
Primary inductance, secondary shorted (leakage inductance):
100 Hz – out of range
1KHz 1V - 14.69 mH (CFB disconnected)
1KHz 1V - 17 mH (CFB in series with primary)
Calculated was 9.5 mH, but formulas were for fixed coil length and did not take into account variable "coil length" of toroid.
Capacitance, primary - secondary
3.68 nF (CFB disconnected)
4.91 nF (Primary and CFB in series)
So 40 KHz roll-off is still puzzling... Could it be property of magnetic material core made with at very low signal level (btw, core is really huge)? Someone can explain that behavior with the data listed above? I can make additional measurements if necessary.
I took RLC meter from my friend, and made several measurements listed below:
Primary resistance - 136 Ohm
Secondary Resistance < 0.4 Ohm at 8 Ohm tap
Primary inductance:
100 Hz 1V - 102 H
100 Hz 50mV - 77 H
1 KHz - out of range
Primary inductance, secondary shorted (leakage inductance):
100 Hz – out of range
1KHz 1V - 14.69 mH (CFB disconnected)
1KHz 1V - 17 mH (CFB in series with primary)
Calculated was 9.5 mH, but formulas were for fixed coil length and did not take into account variable "coil length" of toroid.
Capacitance, primary - secondary
3.68 nF (CFB disconnected)
4.91 nF (Primary and CFB in series)
So 40 KHz roll-off is still puzzling... Could it be property of magnetic material core made with at very low signal level (btw, core is really huge)? Someone can explain that behavior with the data listed above? I can make additional measurements if necessary.
For 4 nF and 15 mH you should have 20 KHz resonance.
Right in a ballpack. A tranny is a complex system, so you should have both parallel and series resonances. I would recommend once more to try the tranny with real tubes powered by real voltages.
Right in a ballpack. A tranny is a complex system, so you should have both parallel and series resonances. I would recommend once more to try the tranny with real tubes powered by real voltages.
One problem is the leakage numbers are quoted at 100hz and the values will be completely different at 40Khz where the resonance is being measured. The OP could remove the core and measure the leakage... wait.... oops darn toroids 🙂
We can pretty much be assured that the leakage at 40khz will be substantially less than at 100hz and if we cut it by 1/4 the series resonance gives a nice dip at 40khz.
dave
We can pretty much be assured that the leakage at 40khz will be substantially less than at 100hz and if we cut it by 1/4 the series resonance gives a nice dip at 40khz.
dave
- Status
- Not open for further replies.