Friday, 16 January 2015

The Legacy of Tube Preamplifier Designs – The Driving Output Stage II

Hello to everybody for the first entry in 2015!

I ended my last entry about the output stage of tube preamplifiers with the words: "the role of the tube is highly overrated". In a way it is and in some other ways it is not. In terms known from classic tube preamplifiers where "tube rolling" is often practiced method for filter like compensations in order to equalize typical sonic errors which are quite normal in tube audio sets from mixed brands. If a tube ideally matches the preferences of a output transformer, the two parts are melting together to be seen as one functional unit and the typical effects from different tube makes are getting quite negligible.

But before I try to refer about the benefits and disadvantages of output transformers in a preamplifiers, it might make sense to describe the general technical advantages of such inductive coupling. Again, it can't be said with enough emphasis, the preamplifier is sound wise one of the most important electronic devices in a elaborated audio chain. In early small signal stages, even more with high gain phono stages, the preamp has a tremendous impact to the attitudes of the finalized sound stage of a complete system. In real life almost everybody takes a lot more care about power amplifiers and loudspeakers as to be the most decisive components for the sound, a common and well established position in the world of hifi since generations.
In history manufacturers have long time given preamplifiers a underrated position, just a functional necessity to equalize different sources, that was it. Almost every well named manufacturer in the golden age of audio did power their units from dedicated power amps. I do not know one example in history till the 1970s years that a preamp has got equipped with a complex power supply, i.e. with regulation. In these days cheap transistor regulators got introduced for voltage regulation and stabilization, since the benefits of tube rectification already had got completely forgotten. Back in these days the transistor has been seen as a universal remedy to any sort of disease and it was cheap. A inductive iron choke or even  better two have been hardly found in any commercial audio component designed in the western world after 1950, even in preamplifiers breaking the 15k $ pricing line like AR, CJ or MFA in the 1980s. But instead each tube got regulated alone by one own transistor in order to distinguish the last touch of typical tube life...
Within expensive government amplifiers custom made to order at that time period you can still observe all the goodies for perfect signal shaping, – chokes, tube rectification, stabilization and regulation. As in Japan inductive couplers in combination with tubes have always been involved more or less as standard solution from the beginning of the rediscovery of tubes for audio amplification around 1970.

Western Electric amplifier, all stages are transformer coupled

Today we have learned to use transformers and chokes again as indispensable parts within tube amplified devices, a development which is given way by the communication throughout the web. Their influence is a improved linear performance of tubes, i.e. their refinement of micro dynamic detail, their ability to open up for holographic performance, their positive effect to keep the music in a natural flow with unsurpassed sonic completeness and as well their ability of involvement, – all these attributes culminate in the small signal stages of a preamplifiers as to be seen as little sound machines.

Within this low impedance any sort of resistive loads like from different cables or their different length is a negligible factor. Without such low Z standardization, like common in the tube equipped hifi world, it is impossible to drive coming tubes stages without loss or it will come to very common difficulties between connecting impedances and as result inadequate dynamic losses. In particular with tube amplified audio components this topic is a major issue for the "never ending process of device exchange mania" in the audio world. It is as well a welcomed perfect instrument to keep the market in a permanent flow of search …?

One instrument to keep the losses as low as possible is a defined low output impedance within each amplifying stage. Almost my complete life with audio components it was said that transformer coupling within tube amplified components is a necessary must, but will show up with several inherent restrictions and limitations. The loss of bandwidth and typical adding effects like the increase of dynamics from bridging transformers (input - output: input - output: etc.) are the most well known limitations named. The arguments did sound logical to my understanding, if every unit will add two coupling transformers at the input and output stage into a chain of several components, it cannot be better like without them. It has to be seen that the omission makes additional components necessary, like dc blocking coupling capacitors. These are soundwise some of the most shaping parts within a amplifier design. As well resistors for the dedicated anode load of the tube will bring obvious effects into the soundstage like roughness and phase shifts. Simple RC-coupled circuits (anode or cathode followers) are much simpler design solutions with almost no extra costs comparing the result seen from a technical point of view, where the transformers is the most expensive part.

Within the rediscovery of tube amplifiers within audio in the late 1970ties and 1980ties, transformer coupling within small signal stages was almost forgotten and was seen as a expensive and as aged historical defeated solution in the western world in these years. Instead US american designers did work hardly to introduce new silicon devices to regulate each tube separately as brave hearts new invention. As in Europe, the british, french and german designers preferred simplest traditional rc-coupled designs from the RCA handbooks to skim the cream of profit from the new growing interest at tubes in the audio markets. So the advanced transformer coupling was reintroduced mainly in Japan in preamplifiers. In Japan inductive phono equalization was seen as superior already in the 1970ties years, a topology which make lowest possible impedancies a must. So such 15K:600Z preamplifier transformers haven been produced by all well known transformer brands (Tango, Tamura, Hashimoto) and have been widely used in advanced designs from that time on.

The implementation

As already described a perfect internal resistance of the driving tube is essential for the best possible transformer matching. Values of 3 to 4 K are good, lower values are even better.  If you match a given tube to a output transformer it can be helpful to charge the secondary for improved frequency response. Transformers like the Lundahl LL1660 with a 4.5:1 ratio don't have a specific primary impedance, their primary is a reflection of the charge at the secondary from the following stage. With a resistor parallel to the secondary winding the output impedance will change the primary as its transmission. Together with a capacitor some unwanted peaks in the frequency response of the amplifying stage can be adjusted for perfect flatness.
Another main important question is the inductance of the primary winding of such a transformer. It is the important value for the low frequency extension of a tube. As you can see in the table, a low internal resistance is very helpful for a good match of low frequency response (there can be seen loads over 5 kilo ohms bring several problems into the design), were the transformer itself is the limiting factor. A air gapped transformer like the Lundahl is set for maximum dc in its primary winding within its necessary air gap. A Lundahl LL1660/18mA is set for 18 mA dc in primary and within this configuration at 100 henry inductance before the core saturation will limit its performance.
Now with such preferences we have already three values in mind, a maximum internal tube resistance for 100 H is around 3 kilo ohms with a maximum current of 18mA limit. A internal resistance of 3000 ohm divided through the transmission of the transformer will give the wanted output impedance.

From the RCA handbook
Now we do look for tubes with a internal resistance of almost 3 kilo ohms and a average current of 18mA at a class a working point. Paralleled 12AY7/ECC82/6SN7 are true candidates and have been historically used in such designs. But there are other and less known tubes like the 6S4, 6386, 5687. A very clever alternative are the multiple system tubes like ECL84/86 etc., they combine a complete amplifier system with perfect output parameters in one glass envelope. And there are the hard to get ancient direct heated triodes, like 3A/110B and WE equivalents WE101, the french, dutch and german post tubes, all with 4 kilo ohms internal resistance are promising candidates for the extra delicate soundstage only direct heated tungsten cathode filament tube are known for.

I made some tests with 6SN7GT/5692 and with 5687, the latter is for me a new tube in the field of experience, since I did always rely on the famous RCA red base types for decades. The 5687 is a typical space age tube developed for computers in 1950 but mainly used in military and industrial applications of the 1970ties.
One tube, two transformers, two resistors, two capacitors together with a stepped attenuator form a perfect line stage able to drive any tube amplifiers input stage. Working in pure class at 18mA current this tube (I had several different brands like Sylvania, Raytheon and RCA in test) sings with exceptional natural fluidity, micro detail and perfect dynamic attributes.  Homogenity, perfect phasing, a unknown depths and wide open soundstage show up with a balanced dynamic performance. All results are benefits from the inductive load of the tube instead resistive loads with more common designs. In theory it is said that the tube works more linear at a inductive load. The transformer if perfectly implemented, brings several benefits into the performance, which is a sort of sonic complete- and naturalness with refinements unknown from RC-coupled or SRPP designs.

Typical japanese designed preamplifier interior with transformer coupled tube stages and inductive RIAA equalisation, stepped attenuattor and massive pio capacitor use, here a design made by P+C company, Japan.

Only a few designers do have understand the lessons of history about the general improvements of transformer coupling and its alternative less improvements for the soundstage. Like Sakuma who did couple almost any tube at any stage with transformers in all his life, for him it seems to be a mandatory question. He brought this idea to a further point since he did built always the complete amplifying chain into one case, from cartridge step up transformer to the speaker out put transformer within three or four stages all coupled with interstage transformers. To the common theory such a system with five to six transformers in one chain will perform with a hardly limited bandwidth of a bakelite telephone, but in reality it seems to be very different?

Sakuma intergrated amplifier design from MC-input to speaker output in one monophonic designed chassis with some transformers on top.

As well Ken Shindo did always implement transformer coupling into his more elaborated amplifiers, since 15 years he did install them even into his simpler designs, a sign of indispensability? Or Kondo who did use widely a combination of SRPP (here one tube regulates the other, it can be seen as one tube acting as electronic choke) topology and the benefits of transformer coupling in mixed exchange within his high priced complex units. In general you will find in Japan loads of diy kits and custom built preamplifiers from the past 40 years were coupling output transformers are a indispensable standard. As already named the inductive equalized phono stage (LCR) is it a preferred topology since 40 years as the final word for RIAA deemphasis. Originally designed for the use with professional NBA (japanese broadcast) components, this topology got very famous and is highly preferred by japanese enthusiasts, even if the implementation of such unit without complications seems to be impossible. Here a low z output behind the gain stage it is a must to couple the inductive equalizing filters in between. A 600 Z transformer output is the only possible way to see the red sun again, but this topic will be discussed at its own in some future time.

Read on soon about alternative ways of inductive transformer coupling for output stages, like the parafeed or choke loaded design topology, Volker