Can anyone explain or point me to a web explaination of how the impedance of the primary side of an output transformer is chosen?
The only stuff I've been able to find so far seems to leave out any reference to the inductance of the primary winding, or refer to "rules of thumb" in relation to the Ra of the output tube(s)
- I'm sure it can be more scientific than that........
Cheers,
Pete McK
The only stuff I've been able to find so far seems to leave out any reference to the inductance of the primary winding, or refer to "rules of thumb" in relation to the Ra of the output tube(s)
- I'm sure it can be more scientific than that........
Cheers,
Pete McK
A few points that more-or-less nail it down. Hmm, seems I can't remember all of them.. someone fill in for me... 
Inductance - obviously, this changes with frequency, so you could have a transformer rated for 5k at 20Hz good for 10k, but only to 40Hz. (Similar to this is power handling, but that involves power levels, not impedance.)
Basically, you set it for a given bandwidth.. the inductance and capacitance (I'd mention C like L above, but I don't have enough to say about it right now for a whole paragraph ) are constant, whatever the applied voltage and current - (well, L changes with DC bias, but nevermind that for a minute ) - it depends on construction. So for 5k, you have 20-20,000Hz BW... at 10k, it drops to 40-10,000.. eventually, it poops out entirely, in this case it would be about 105kohms, parallel resonant at 450Hz or so (other frequencies are shunted by either L (below 450Hz) or C (above)).
You can go the other way - 2.5k on the above tranny ideally would give 10-40,000Hz - but I forget exactly what's wrong. One thing is resistance of the windings, it becomes more significant at the lower impedance. Say the primary resistance is 250 ohms. At 5k, that's 5%, not much. At 2.5k, it's 10% and at 1.25k, it's 20%!
I think power handling goes down with impedance, since current levels go up, leading to easier saturation of the core...but I don't remember. Need sleep.
Tim
Inductance - obviously, this changes with frequency, so you could have a transformer rated for 5k at 20Hz good for 10k, but only to 40Hz. (Similar to this is power handling, but that involves power levels, not impedance.)
Basically, you set it for a given bandwidth.. the inductance and capacitance (I'd mention C like L above, but I don't have enough to say about it right now for a whole paragraph
) are constant, whatever the applied voltage and current - (well, L changes with DC bias, but nevermind that for a minute ) - it depends on construction. So for 5k, you have 20-20,000Hz BW... at 10k, it drops to 40-10,000.. eventually, it poops out entirely, in this case it would be about 105kohms, parallel resonant at 450Hz or so (other frequencies are shunted by either L (below 450Hz) or C (above)).You can go the other way - 2.5k on the above tranny ideally would give 10-40,000Hz - but I forget exactly what's wrong. One thing is resistance of the windings, it becomes more significant at the lower impedance. Say the primary resistance is 250 ohms. At 5k, that's 5%, not much. At 2.5k, it's 10% and at 1.25k, it's 20%!
I think power handling goes down with impedance, since current levels go up, leading to easier saturation of the core...but I don't remember. Need sleep.
Tim
Hi Pete,
Many of the 'rules of thumb' wrt loading tubes have in fact been worked out scientifically, and experimentally proven over a long time. For example the oft quoted maxim of between 2 and 5 x Rp as a load works, depending on what you're wanting to do. Lower values give more power but more distortion, and higher values, the opposite.
The primary impedance of a transformer is usually referred to the load impedance and the turns ratio of the transformer.
Primary reflected impedance (Zpr) = Secondary impedance (Zs) x (number of turns in primary winding/number of turns in secondary winding)^2
Zpr = Zs x (Np/Ns)^2
An OPT with a Zp of 5k designed for an 8 ohm load has a turns ratio of
sqrt(Zpr/Zs) = sqrt(5000/8) = 25
This is normally what people talk about when talking about the Zp of an OPT. However, it's not quite that simple. The primary inductance (Lp) is effectively in parallel with the Zpr. (remember two resistances in parallel will always be smaller than the smallest of the two resitances). At lower frequencies this might be significantly lower than the Zpr, which means the tube sees a lower load impedance, and gives more distortion. 'Good' OPTs have enough inductance to do the job properly. VoltSeconds page linked below explains this graphically.
The primary winding also has a resistance that is simply the resistnce of the copper wire in the winding. This is in series with the parallel combination of Zpr and Lp.
The Western Electric 300B page has a ton of experimentally derived data. Look at the first 3 graphs to see how the distortion amount and the spectrum changes with different load impedances for the "classic" 300B operating point. On page 5 of the pdf datasheet it also has the same data condensed and related to the power output in the one chart. Whilst this is for the 300B, it can be roughly extrapolated to similar behaviour in other triodes, and slightly less accurately to triode connected pentodes.
To find out the technicalities in choosing a loadline, download the Norman Crowhurst articles in the audio classroom series. Additional information can be found in John Broskie's excellent and clearly written grounded (common) cathode amplifier article. This is written for resistive loading, but the same applies to transformer loading, with some minor changes. Lots of very interesting other articles on his site too.
Steve Bench also has a series on loadlines. At this site too are a number of other excellent technical articles to browse as well.
Finally, to relate the primary inductance and reflected impedance of an output transformer to the performance of an amplifier VoltSecond has developed a page using a single ended 211 amplifier as an example using spice. Excellent and with graphs. Can be found here
My suggestion would be to read these, digest, and then ask further questions for clarification.
This give an overview of the technicalities, but how to apply these to give the best sound quality in a given topology is a whole 'nother can of worms.
HTH
Cheers
Many of the 'rules of thumb' wrt loading tubes have in fact been worked out scientifically, and experimentally proven over a long time. For example the oft quoted maxim of between 2 and 5 x Rp as a load works, depending on what you're wanting to do. Lower values give more power but more distortion, and higher values, the opposite.
The primary impedance of a transformer is usually referred to the load impedance and the turns ratio of the transformer.
Primary reflected impedance (Zpr) = Secondary impedance (Zs) x (number of turns in primary winding/number of turns in secondary winding)^2
Zpr = Zs x (Np/Ns)^2
An OPT with a Zp of 5k designed for an 8 ohm load has a turns ratio of
sqrt(Zpr/Zs) = sqrt(5000/8) = 25
This is normally what people talk about when talking about the Zp of an OPT. However, it's not quite that simple. The primary inductance (Lp) is effectively in parallel with the Zpr. (remember two resistances in parallel will always be smaller than the smallest of the two resitances). At lower frequencies this might be significantly lower than the Zpr, which means the tube sees a lower load impedance, and gives more distortion. 'Good' OPTs have enough inductance to do the job properly. VoltSeconds page linked below explains this graphically.
The primary winding also has a resistance that is simply the resistnce of the copper wire in the winding. This is in series with the parallel combination of Zpr and Lp.
The Western Electric 300B page has a ton of experimentally derived data. Look at the first 3 graphs to see how the distortion amount and the spectrum changes with different load impedances for the "classic" 300B operating point. On page 5 of the pdf datasheet it also has the same data condensed and related to the power output in the one chart. Whilst this is for the 300B, it can be roughly extrapolated to similar behaviour in other triodes, and slightly less accurately to triode connected pentodes.
To find out the technicalities in choosing a loadline, download the Norman Crowhurst articles in the audio classroom series. Additional information can be found in John Broskie's excellent and clearly written grounded (common) cathode amplifier article. This is written for resistive loading, but the same applies to transformer loading, with some minor changes. Lots of very interesting other articles on his site too.
Steve Bench also has a series on loadlines. At this site too are a number of other excellent technical articles to browse as well.
Finally, to relate the primary inductance and reflected impedance of an output transformer to the performance of an amplifier VoltSecond has developed a page using a single ended 211 amplifier as an example using spice. Excellent and with graphs. Can be found here
My suggestion would be to read these, digest, and then ask further questions for clarification.
This give an overview of the technicalities, but how to apply these to give the best sound quality in a given topology is a whole 'nother can of worms.
HTH
Cheers
Stolen current
Considering current is an even more graphic way of viewing the effect of finite output transformer primary inductance Lp. As frequency falls, the reactance XL falls, so current intended for the load is stolen by the output transformer's primary inductance. This causes considerable distortion...
Since XL = 2 pi f L, and it is in parallel with the reflected load resistance Ra-a, we could set the two to be equal at our lowest frequency of interest (perhaps 20Hz). Thus, Lp = Ra-a / (2 pi f).
In practice, you might want to double the above figure in order to ensure that 20Hz can be passed with acceptable distortion.
Considering current is an even more graphic way of viewing the effect of finite output transformer primary inductance Lp. As frequency falls, the reactance XL falls, so current intended for the load is stolen by the output transformer's primary inductance. This causes considerable distortion...
Since XL = 2 pi f L, and it is in parallel with the reflected load resistance Ra-a, we could set the two to be equal at our lowest frequency of interest (perhaps 20Hz). Thus, Lp = Ra-a / (2 pi f).
In practice, you might want to double the above figure in order to ensure that 20Hz can be passed with acceptable distortion.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.