输出变压器设计

输出变压器设计

It has been a while since we posted. Today, let's talk about output transformer design (输出变压器设计).

Output transformer is the heart of tube amplifiers (OTL - output transformer-less designs are exceptions of course). The quality of the output transformer determines the overall quality of the tube amplifier. It is much easier to get a power transformer made correctly than an output transformer with wide flat bandwidth.Good tube amp manufacturers often places a great deal of focus on their output transformer design, such as McIntosh, AN and ARC, which are famous for their high quality output transformers. For those 2nd or 3rd rate manufacturers with unknown transformers, or transformer designs, the transformer quality is something to be examined further and purchased with caution. 

So, let's think about what qualities should a high quality HIFI output transformer possesses? 

1. High primary inductance - yes, on a surface,for better low frequency response and loading, matching the driving tube and circuit, but to a certain limit so as not to create other problems.
2. Low leakage inductance, and capacitance, sometimes more accurately described as distributed inductance and inductance, because they are evenly distributed across all the winding of the transformer. 
3. No resonances across audio bandwidth.
4. No distortions.
5. 0 phase delay across audio bandwidth (usually happens at both ends of the frequency spectrum.

For those that knows a bit or read about magnetic properties, it is very obvious that the above are very hard to achieve, and the parameters actually work against each other. 

In output transformer design, you get some, you loose some, and you can never have all. If someone tells you he can, think twice.

Magnetic cores, such as SiFe, or any cores, will introduce its' own distortion due to different and imperfect BH characteristics (curve). The input and output will be different due to the transformer core BH behavior, and therefore some sort distortions have been introduced.

We are often asked about having unusually high primary inductance for Single Ended output transformers for their orders. Yes, high primary inductance is good, but over-design is never ideal in output transformer design. In order to have high primary inductance, we need much longer wires, thinner wires, and more turns. These have the side effects of higher primary DCR, higher copper loss, higher leakage (higher distributed inductance & capacitance), and higher phase shift. This will cause narrowing of the bandwidth, phase shift at both audio frequency extremes, and decrease of dynamic range. 

That's why, often, for those with high primary inductance requests, we will usually ask them to trust us in our judgement to select the best design for their applications. If one blindly follows the "high primary inductance" rule, they are free to suffer from their own decision. Some just focus so much on the primary inductance, that they do not really care about the high frequency response, making their tube amplifiers sounded very smooth, warm and mellow. At first, it may sound soothing and nice, but after a while, listeners with good ears will start to feel something amiss - the HIFI component, i.e., the high frequency components and dynamics will be missing. It may sound dull and lifeless, but some may like it, mistaking it as fake warmth or vocal (mid-range) superiority. Such kind of tube amplifiers may be good for speakers with high frequency focus loudspeakers, with a drift that goes upwards in the frequency response, compensating this design defect.

We strive to design a wide bandwidth (frequency response) output transformer that excel at both frequency end, with very flat response at the standard 20-20kHz region, at +/- 1dB flatness, or better.

The way to achieve this is to fulfill the highest primary inductance requirement according to the design at stake, and try to lower the distributed capacitance and leakage inductance to achieve the high frequency response for modern playback/recording devices usages. The application of high quality transformer core materials such as high grade SiFe, amorphous cores, Nanocrystalline cores, Permalloy cores, and with complex winding methodology, we try to get as close as we can to a perfect output transformer. 

As we have mentioned, the parameters are contradicting each other and therefore the bandwidth will not be as wide as an output transformer-less tube amplifier or Solid State amplifiers. Still, we love and we are addicted tube amplifiers with output transformers!  

Sometimes, bandwidth limitation can be a blessing! You have heard of deployment of high quality vintage input transformers from the golden age of tube amplifiers, such as input transformers from Western Electric (which have attained a cult following, with insane sky high prices) makes the sound better, smoother, despite their limited audio frequency bandwidth. Such bandwidth limitation acts as a filter for the high frequency noises, interference from surrounding or playback electronics, or even high frequency component of music signals. Therefore, making the sound smoother, and less harsher compared to direct connection or equipment without coupling transformers. McIntosh, for one, uses such output transformer technology on their solid state amplifiers. 

The leakage inductance, distributed capacitance, the primary inductance, the tube plate impedance, the primary current, the plate load impedance, the primary/secondary DCR, primary/secondary turns ratio and the desired flatness in audio frequency range, has a semi-fixed relationship, and there is a formula (cannot disclose here, J&K proprietary, but can be found and derived from various magnetic formulas and textbooks) to ensure the design meets a certain HIFI criteria or requirements. The primary/secondary DCR cannot be too high, or else the dynamic range, resolution and transient responses will suffer. These all actually sums up to be the primary impedance of the output transformer, i.e. the load for the tube in use. Varying components at difference frequencies within the audio frequency range will affect the flatness of the response curve, and also the phase shift over the frequency range. Deviation of the phase response and response curve flatness generates distortions, and power loss or variations across frequencies. 

Furthermore, these inductance and capacitance that exist in the transformer forms a LC (RLC) circuit and causes resonances at certain frequencies. At resonances, the impedance at such frequencies are either 0 or infinite, causing peaks and dips, abnormality/gradients on the frequency response curve. It may cause premature roll-off at both ends of the frequency spectrum if resonances fall at those 2 extremes. Therefore, one other factor to consider is to design the output transformer so as to reduce these artifacts, or make them fall FAR outside of the audible frequency spectrum. 

We often stress on the quality of transformer cores because of the effect is has on the dynamic ranges (of course there are other effects). The signal amplitude, or the responsiveness to response to minute or major signal changes, highly relies on the induction density and permeance rate of change, which varies according to the signal amplitude, and therefore affects the dynamic range, or how well it responds to small/big signal changes. 

Another factor that is related to signal amplitude is the transformer core saturation. If the signal is large, it may be working at the saturation region of the B-H curve and therefore further increase in signal amplifier will be saturated and output be distorted. That is why some magnetic cores are for small signal usages, and some are for power output usages due to their core magnetic flux density limitation. 

We have not talked about DC magnetization of the cores, such as in single ended tube amplifier application, where we need to add air-gap in the cores to prevent saturation. The control of the air-gap size will affect the magnetic properties of the core. Sometimes, some air-gap is used on push-pull output transformers too due to mismatch in both arms of the push-pull output transformer caused by mismatched tubes or unbalanced loading (difference in DC bias). 

We've discussed a lot in this session. Till then. 

Hear the unheard!
J&K Audio Design.
7/23/2016