RH6080HE (Headphones Edition)

It’s been years since I have posted any new projects on my blog – it’s not due to a lack of inspiration or ideas, rather I was too busy to build any new projects – or publish what I was working on. In the meantime my interests and experiences have evolved towards fields that I have been less involved in during the past years, and the general lack of time has made me rethink some of my building practices towards more time-efficient solutions.

Headphones and Speakers

While like most of my generation I have experienced first hand the advent of mobile audio in the form of the first “Walkman” devices (cassette players with and without FM radio) and thus grew accustomed to using headphones, I was never particularly taken by headphones listening – always preferring speakers. Over the years, this preference has basically cost me a lot in terms of musical pleasures not experienced: everyone can hear what (and how loud) you are listening when you use speakers, which implies that your family willing or nilling takes part of your musical and audiophile life. If your tastes include rather hermetic music, or hard bop, post bop, modal, classical – you name it, basically whatever is not mainstream commercial… in other words not suitable for everyone and not necessarily acceptable to others – you will likely listen to less music than you might have had, or will have to avoid some of the music you like. That’s where headphones enter the life of the regular music lover and audiophile.

The main reason why I prefer speakers to headphones is space – the music otherwise being confined to one’s head and thereabouts. Furthermore, the subjective sound quality being (perceived as) far higher with speakers – which might be a surprise to those who are aware or at least convinced that headphones are far superior in terms of transducers (no need for crossovers, little or no limitations due to available power, etc). Well, most probably that was due to the fact that we used to listen to headphones connected to the headphones connectors of our (integrated) amplifiers and receivers, or CD players and cassette decks: simply put, driven by a handful of small transistors, or op-amps. The advent of the new generation of often USB capable DACs, of which some were marketed as headphones amplifiers, did little or nothing – for me, at least – to change the perception of headphones sound quality as inferior to speakers in a good system. Well, I guess the words “good system” are the key to this perception – just like trying to compare some integrated solid state amplifier with a good (or much better than good, for that matter) tube preamplifier matched to some SE tube amps, with good cables to boot and all the possible amenities – on the same set of speakers. I needed to listen to more music, and it became obvious to me that there must be a way to equalize the perceived sound quality between speakers and headphones – let’s search for it in the missing link, the amplification.

Design Choices

Headphones do not need a lot of power – so far I have not had any planar magnetic headphones, but even these should probably do well with a couple of good watts… while most headphones are actually rated for less than 2W of power (i.e. they can handle that much power before being destroyed – assuming that your ears are not involved in the ordeal as they would make it painfully obvious that the power is too high).

What headphones do need is a high quality amplifier with low (lowish, actually) output impedance. Besides the most obvious choice of active element (tubes and solid state), and the quality of the power supply, passive elements are also relevant to the perceived sound quality of the amplifier – most notably capacitors and transformers. I guess it’s more than obvious that I would not choose solid state active elements to build an amplifier for my headphones, and with tubes there are basically 2 design options – with or without output transformer.

While there are lots of amp designs and ideas all around us on the net when it comes to using tubes to drive headphones without an output transformer, almost all reasonably feasible (and repeatable in terms of performance) alternatives must use a coupling capacitor between the headphones and the amplifying stage. I prefer to avoid coupling capacitors in the signal path, and my designs are centered on using the least number of gain stages and therefore coupling caps. Furthermore, when it comes to headphones, the values of the coupling caps have to be rather high if one is after a reasonable response in the lower frequencies: this is particularly true for most “modern” headphones which tend to be anywhere between 32 and 150 ohms impedance. The large value implies the use of electrolytic capacitors: while it is possible to parallel film caps to some extent, this will almost always end up in a very costly and physically large compromise. Caps will tend to leak, and that’s just another problem one is facing with output caps in a headphone amp. I have the feeling that the headphones are not really safe with a capacitor coupled output, although that is probably irrelevant to most: you only live once, so who cares…

As I already mentioned, most currently available dynamic headphones are between 32 and 150 ohms impedance, where above 80 ohms the choice becomes rather slim, at least in terms of variety. While these lower impedance headphones are relatively easily driven with solid state devices, they represent a taxing load for tubes and the possible solutions are either push-pull circuits, or the usual highly inefficient cathode follower compromises that are quite good at driving higher impedance headphones (300 - 600 ohms) but cover the fact that they are struggling to drive lower impedance headphones regardless of the inefficient high idle current draw and relatively high output power (relatively high – that is in terms of headphones power handling capabilities).

Some well-known amps and kits resort on 6080 or 6AS7 in cathode followers drawing high currents and running their cathode resistors rather hot: I have considered this option with 2k 25W cathode resistors and was not happy with the high current that would have to be drawn, the heat generated by these cathode resistors, the compromise when driving anything else than high impedance headphones – and last but not least, the output capacitors. OK, let’s put this straight – amplifier size or power consumption was the least of my concerns; I do not like unnecessary heat in an amp, and I do not like output coupling capacitors – but what I hate most is the compromise involved. While I own several headphones, most are low or lower impedance, and I guess most DIY-ers and audiophiles are facing the same choice of headphones. Last but not least, I like universal solutions (does the word ring some bell?) and dislike compromises that can be avoided.

RH Amplifier – Headphones Edition

Once we leave the output capacitor path behind, the only way to go is with an output transformer. While some would consider the output transformer as another compromise in quality, similar to the capacitor, I would beg to differ. The transformer does not leak, unlike the capacitor, and headphones are safe unless abused: the only regular problem that comes to mind is the pops and clicks when powering up or powering down the amplifier, but that can be avoided in more ways than one. The limitations of output transformers are known, and in this case due to the fact that headphones do not require a lot of power, can easily be solved with larger cores than strictly necessary. A core that would be good for a decent 5W SE amplifier is more than good for a headphones amplifier that will in most cases deliver less than ½W of power. On the other hand, transformers can be rather universal devices that offer flexibility – just like the 4, 8, and 16 ohm taps on some output transformers, there can be 32, 64, 128… ohm taps on the output transformer for headphones, matching the impedance of the headphones at hand, or better – the impedance groups of headphones.

Now that the decision is made – go with an output transformer – an amplifier is needed. Well, in my case that can only be an RH amp, otherwise I would not be true to myself. Over the years I had several times been asked if the RH84 could be used as a headphones amp, provided the appropriate output transformer is installed – and I have replied that it most certainly can, but it was not designed, or optimized, for this role. The purpose of the RH84 is to drive speakers with as much power as can be had from the EL84 while not sacrificing the sound quality – actually, trying to solve that design task better than most other amps.

For the headphones RH amp, I chose to design something “new and different”, although the resemblance to my other designs is striking. This amp does not need to produce more than a couple of Watts of power, but needs to yield excellent sound quality, and absolutely needs to be quiet – i.e. no noise is allowed because the headphones being on the listener’s ears are quite sensitive to noise and hum. Last but not least, the optimization needs to be easy, taking into account the intrinsic differences between the imperfect devices that tubes are, and the variations possible even within the same batch.

One additional element that needed to be taken into account is the lowest sensitivity possible without resorting to additional volume potentiometers, while the amp needs to be able to easily accommodate various use cases with simple modifications. Lowest sensitivity possible – this implies that while it might be beneficial for a low power SE amp to get to its full power of 3W with an input of 0.775mV RMS, this would be totally impractical for the same SE amp used as a headphones amp. As I said several times already, this amp will not be used at 2W output power by most users!

The design centers around the 6080/6AS7 tube which is a dual triode just like the ECC81 or the 6SN7, but the two triodes inside the envelope have 12W anode dissipation each, and mu is very low – lower than 2A3 or 300B. Obviously, the low sensitivity requirement excludes all the usual pentodes that come to mind -  EL84, 6V6, and the smaller “siblings” like EL85 or EL95 which can still be found and had at low prices – although all of these tubes would have more than enough power for the task. Even if those pentodes were used as triodes (something that I personally don’t do) their resulting mu as pseudo-triodes would still be too high. The 2A3, 6BG4, or 300B might be used, and would probably be capable of interesting results – but they are all directly heated and thus would require additional attention (and circuitry) to avoid filament hum. The 6080/6AS7 tubes are rather plentiful and therefore still cheap, and will not require any additional attention to heating – being indirectly heated triodes.

The driver is the usual ECC81 that I use for all my other RH amps – I guess by now it does not need any introduction. Let me just reiterate that this tube has a relatively high mu (reasonably high internal impedance) combined with a relatively high transconductance – two characteristics that make it especially suitable for the role it is supposed to have – driving the output tube to desired results.

As mentioned at the beginning, the lack of time has driven me towards different solutions than previously adopted: installing sockets and doing a hard-wired installation might be the best solution for most people, provided they had the time to do it. Additionally, while a hard wired installation is expected to yield better sound quality and last longer, it is quite impractical for servicing and adaptations, let alone re-use or recycling in other projects. On the other hand, my recent preamplifier revalidation has shown that even PCBs I have drawn and etched manually have lasted more than a decade without any problems whatsoever, and continue to perform well in spite of the servicing and modification changes (change of component values to suit other tubes). Thus when building a new preamplifier I have redesigned the PCBs and this time had them etched instead of doing it myself.

In accordance with the experience gained using the old preamplifier for more than a decade, and having just built a new version, I have decided to design an amplifier board that I named RH6080 – generally suitable for application of most noval tubes as drivers, and the 6080/6AS7 family of tubes in the output position octal socket. The two channels are basically dual-mono except for the common ground on the board, and the fact that the tubes are dual triodes: in theory, even the B+ could come from a different source or be of different value, although that is highly unlikely.

I can use the same board to build other projects that I have in mind – an 8W SE amp for loudspeakers, PP RH amps – and even a driver amplifier that can push large output tubes into class A2… projects awaiting to happen in the future.

Performance and Optimization

As shown in the simulation, this amplifier is capable of 3.2W output power with an input of 2.5V RMS. As such, with an appropriate output transformer it could be used to drive speakers – rivaling classic SE amps with 2A3 output tubes, usually specified as 3.5W output power. The 6080/6AS7 output impedance is lower than that of the 2A3, and combined with the RH specific feedback – this translates into quite low output impedance for an SE amp: what DIY-ers who have built RH amps have previously experienced with the amps they built would be adequately replicated by this amp. Simulations are just as good as the models used, and the models I am using both for ECC81 (accuracy comparison shown in the RH813 post) and the 6080/6AS7G are as accurate as possible. By that I mean – the models were created based on a sets of curves shown in the datasheets – to get the results with 100% accuracy one would have to generate curves from the batch of tubes used, and create models based on these curves. The accuracy achievable with datasheets curves should be considered as “as accurate as possible” – unless one is in the “laboratory measurements” hobby as opposed to the “designing and building amplifiers” hobby.

This amplifier is supposed to be used as regular amplifier for loudspeakers – the only difference being that headphones are connected to it instead of the loudspeakers. Therefore, there is no volume potentiometer except for the one on the preamplifier. My current preamplifier has 20dB of line stage gain, giving it a reasonably ample range of volume adjustment with my power amplifiers into my speakers, where the amps are not particularly sensitive, and the speakers are reasonably efficient. When operating the RH6080HE I have found that the volume adjustment range on the preamplifier pot is twice as limited – basically, at 9 o’clock the volume with most headphones is at the limit of my comfort zone in terms of loudness, i.e. output power. In practice, this means that with the addition of a volume potentiometer and eventually a selector it can be used without a line stage preamplifier, or connected to the tape output of the preamplifier (without line stage gain).

A slight modification to the input circuitry where the cathode resistor coupling cap is removed decreases input sensitivity further to almost 3V RMS needed for full output, while the remaining parameters are the same – this increases the range of volume adjustment on the potentiometer, and makes it easier to find the adequate volume for listening. I guess in this case use without a preamplifier as explained above is still borderline possible and will depend on the output voltage of the source used – a CD or DAC at 2V RMS output should be amply sufficient to exceed comfort zone volumes with headphones, but lower output sources might lead to insufficient volume and/or a lack of fullness in the sound.

This is a single ended, i.e. SE amp – thus absolutely suitable to drive any headphones without the need for any strange or non-standard connectors. Nevertheless, as the output transformers totally isolate the headphones from signal or power ground, and even the left from the right channel, I have chosen to adopt as standard a 4-pin XLR socket, basically the same as used by some manufacturers to connect headphones in balanced mode. Needless to say, I have chosen to implement the same pin connection standard as used by the manufacturers, mainly to simplify potential issues with cables compatibility.

Balanced headphones connection requires a separation of ground returns between channels, hence the need for 4 connections instead of the usual 3 (common ground). Thus a pair of headphones configured for balanced connection can easily be connected to this amplifier, as the grounds are galvanically isolated – although obviously they will be driven in single-ended mode since the amplifier is SE. In order to connect headphones with regular 6.3mm jacks and 3 connections (common ground) an adapter is sufficient where the ground connections are shorted: this might be even dangerous with a balanced amplifier, but is absolutely normal with this SE amplifier. Therefore I have decided against having an additional 6.3mm jack terminal next to the 4-pin XLR and to use a custom-made adapter instead.

The knob visible on the facia of the amplifier is not a volume potentiometer – rather a selector. The purpose is to choose the secondary of the output transformer based on the impedance of the headphones connected to the amplifier. While the type pictured is good enough for the task and works fine in practice, it is sub-optimal and I suggest a better selector is installed – besides mechanical stability the clue regarding it’s adequacy for the task is declared current/power capability.

The Power Supply

The initial version was built with a rather simple CLC power supply that would be suitable for an SE amplifier that drives speakers – basically, if this amp was used to power loudspeakers, there would be no audible hum or noise audible at 1m and even less from the speakers. But, this amp is used with headphones, and absolutely all sound effects can be heard – down to the eventual propensity to microphonics of some tubes (even the output tubes), and this includes all sorts of hum that come to mind. As usual, I have had no problems with the wiring or ground loops, but in the absence of music clean low frequency (100Hz, twice the mains frequency) hum could be heard. At that point it became clear that the power supply is either going to be large (marginally too large for the box in which I was building it), elaborate, and expensive – or regulated.

Having had excellent results with the power supply I have recently developed for a new iteration of my classic preamplifier (some might remember the RPA, not available on the net any more as the old site has disappeared), I decided to modify the voltage setting resistors ratio to match the needs of the RH6080HE and populate a spare PCB etched for the preamplifier project that I had at hand. It took me literally minutes to build it, as opposed to hard-wiring capacitors and finding a way to keep them fixed…  This power supply works perfectly with the phono stage, both in terms of hum and noise, as in terms of perceived sound quality.

Needless to say, as expected, the amplifier became dead quiet – it all boils down to tubes quality now, as some driver tubes (ECC81) might be less quiet than needed – and even some 6AS7 can be noisier than a perfectionist would accept. The power supply uses a hybrid bridge composed of a dual rectifier tube and two solid state diodes – the rectifier tube is slower and dictates the behavior of the solid state diodes, with the result being sonically equal to what the rectifier tube would yield in a pure tube rectification circuit. The active pass element is a TL783, a device similar to the LM317 but with a much higher input-output voltage differential of 125V. Another difference is the pass element which in the TL783 is not bipolar rather FET. The same circuit can be built with an LM317, but in some circumstances the input-output voltage differential might be higher than 35V (highly unlikely but possible) most probably killing the LM317 instantly… no harm would happen to the tubes, obviously, but the hum would immediately rise to unacceptable levels pointing out that something needs to be done about it… Another difference is the quality of the TL783, which unlike most LM317 is not noisy. The difference in price is irrelevant in DIY terms and the TL783 should be relatively easy to come by.

Some would probably object to the solid state regulator, as the solid state pass element is expected to mask the “sound” of the tube rectifier. As I already said, this power supply has proven its worth in the preamplifier project, and rectifier tube rolling with this power supply is more than effective in fine-tuning the sound. The only issue for tube rolling might be the possibly large difference in output voltage between different rectifier types (actually, the difference in diode voltage drop) – precluding the use of 5R4 and 5Y3 tubes in this circuit (input-output differential too low and the regulator is not working properly, with high hum as a result). Of course, that can be circumvented with a higher secondary AC voltage (for instance, 360V instead of 330V) but in that case when using more efficient rectifiers like the 5Z4 the input-output differential across the regulator element will increase to probably 30 or 40V, which combined with 100mA current draw for the amplifier circuits yields 3-4W of dissipation – precluding the use of more efficient rectifiers and limiting reasonable operation with a higher voltage secondary to the 5Y3 or 5R4 rectifier tubes. In that case, a good heatsink capable of at least 5W dissipation is necessary: I prefer keeping the dissipation below 1.5W. It goes without saying that both the TL783 used in the power supply, and the regulators used as current sinks below the cathodes of the output tubes must be heatsinked, although the heat-sinks do not need to be particularly large (adequate for up to 2W dissipation).

The Output Transformers and Sound Quality

Well, so far so good – it’s just a regular RH amp with output transformers that are adequate for headphones, and can be build on PCBs – making it esier for DIY-ers… but most SE output transformers are actually designed and manufactured for loudspeakers, with 4 and 8 ohm taps. They are not adequate for 32 ohm headphones, let alone high impedance types at 300 or more.

The output transformers used for this project were manufactured by Heyboer in the US based on project requirements – and provided by a fellow DIY-er: Larry Granger. I would like to thank him for finding the subject interesting, and for his kindness in providing the output transformers – without those, the project would just be dead drawings on (electronic) paper.

The primary was chosen to be 5k – this is perfectly suitable in general for tubes drawing 40-50mA, and a value generally suitable for most low power pentodes and tetrodes, like the EL84 or the 6V6. While this value seems too high for use with 6080/6AS7, the assumption is not particularly correct. The output tube in this circuit is used at a rather unusual operating point, constantly drawing 50mA and having just above 200V across the tube (cathode to anode). The output tube in practice operates at around 10W anode dissipation, which is absolutely acceptable for this type of tube and guarantees a long operating life. On the other hand, output power maximization is not necessary – and it is already done with the particular feedback applied which includes the characteristics of the driver tube.

The core of the transformer would be absolutely suitable for a high quality SE amplifier in the range 3-5W, and most manufacturers would market it as a 10W core: thus it is expected to behave very well in terms of bandwidth. The higher than usual primary impedance for  the tube, and the low output impedance of the circuit mean that with this transformer there should be no bandwidth constraints, particularly at 1W output power and below. While not having any planar magnetics to try, I expect that even the slightly higher power requirements of such headphones would be served nicely.

The secondary windings were chosen to be multiples of 32 ohms – 32 ohms being the de-facto standard value with headphones nowadays. Thus the values are 32 – 128 – 256 – 512 ohms, and as such will accommodate a wide range of headphones from 32 to 600 ohms. Connecting 600 ohm headphones to the 512 ohm secondary will result in lower primary impedance seen by the output tube as 4.2k instead of 5k – but as explained above, this tube and the circuit can easily handle 20% differences in primary load, while 450 ohm headphones will again fit the 512 ohm tap very well. Similarly, connecting 300 ohm headphones to the 256 ohm tap is absolutely fine. Most 70 or 80 ohm headphones actually show impedance charts around 100 ohms, and they can be well served by the 128 ohm tap, just like the 150 ohm headphones. I am deliberately not mentioning the brand names of the “usual suspects”, and I expect most headphones enthusiasts know well which brand still manufactures 300 and 600 ohm headphones, and eventually proposes new “improved” models of 150 ohm impedance…

Last but not least, the sound quality: much better than expected, actually. While my amplifiers perform very well in my room driving the reasonably efficient speakers that I use, with the RH6080HE power is not a relevant topic, at least in terms of loudness and dynamics. Having much more power than needed imparts an ease and effortlessness in the presentation of music, and the most important characteristics that I have found is what I was missing most with headphones – space, or rather the sense of space. It is common knowledge that open headphones convey a better sense of space, frequently at the expense of less depth and definition in the bass notes. With this amplifier, I have found improvements on both: closed headphones have an excellent and unexpected rendition of space, while even those bass-heavy among them show a very controlled low register. On the other hand, the sense of space with open headphones is amazing, while the rendition of bass notes is so effortless and well defined that it easily rivals listening on loudspeakers.

More importantly – how does this amplifier compare to commercial alternatives? Well, first of all, I am not aware that there are many commercially available tube amps with output transformers and multiple secondaries – and not having listened to any such amps, I can only believe that the difference is proportional to what can be had when a “regular” loudspeaker driving RH amp is compared to commercial alternatives of similar power. On the other hand, there is literally no comparison with most if not all commercial solid state device powered alternatives that I have had the possibility to try: the difference in sound quality is quite pronounced and it becomes quite obvious that the tube amp is in a league of its own. But, as a good friend nicely puts it – one can take along most of these amps and DACs in one’s pocket and carry them along enjoying music everywhere, while I cannot take this amplifier, rather heavy and the size of a regular SE amplifier, and use it on the go. Still, the purpose of this design has never been portability, but sound quality rivaling listening to a good system with loudspeakers – and that goal has been amply achieved.

RH-TTA – Tube Tester Amplifier

The RH300B project has spawned a schematics variation for the 2A3 tube, and from there several similar variations derive, RH2A3/1619, RH6B4G/6L6 – as presented in the previous blog entry. Indeed, the 1619 is very much compatible with the 2A3, with its 2.5V heaters, and the overall compatibility of the octal socket pin disposition as found in the special octal version of the 2A3, manufactured originally for the Audio Innovations amplifiers during the 90s. On the other hand, the 6B4G (by which I intend NOS types) and the more common Russian 6C4C (6S4S) have an identical octal pinout (5S) that is basically compatible with the 7AC pinout of the 6L6... not to mention that the 8EP pinout of the EL34 is compatible as well if we consider grounding the third grid (g3) instead of connecting it to the cathode (as a matter of fact, I prefer grounding g3 i.e. the beam former – to connecting it with the cathode, since a gradual increase of voltage on g3 leads to the forming of a kink or wave similar to that of a tetrode).

Why should we not have the best of both worlds? All we need are additional heater secondary windings for 6.3V tubes: plural, since each direct heated tube should have its own, and indirect heated tubes will not mind having their own heater secondary. Now here we are facing two possibilities – either a single 6.3V 2.5A secondary where the voltage can be reduced across resistors, or separate 6.3V 1.5A secondary and 2.5V 2.5A secondary. The latter solution is what I have chosen, since adding resistors both clutters the interior of amplifiers, and generates unnecessary heat by burning down voltage (not to mention an increase in power consumption).

The Best of Both Worlds

The result is an amplifier that can use a very wide array of tubes – from 2A3 octal and 1619, to 6A5G, 6B4G and most of the pin compatible indirect heated tubes (even 6V6 and 6F6 can be used if the rectifier applied to the power supply has a higher voltage drop, like the 5Y3). The basic schematics is as published under RH6B4G/6L6 – but for a small detail: in order to avoid the switch that selects cathode type (well, let’s get rid of at least one switch), I have chosen to connect pin 8 (cathode in 7AC and 8EP pinouts) directly to the “virtual cathode point” that is created on pin 6 (this pin is unused in all the pin-outs mentioned). Incidentally, pin 8 is the mid-filament point to which the indirect heated cathode is connected on the 6A5G.

The connection of pin8 to pin 6 is the only relevant schematics difference between the RH6B4G/6L6 schematics, and the RH Tube Tester Amplifier schematics… so far.

This is basically an amplifier that can use a wide array of tubes capable of at least 15W dissipation… now that brings to mind a few other similar tubes with different bases. The 307A that I have already designed for (RH307A), and the 2E22 come to mind, as well as the 1624, and the evergreen 807 – all of those require the UX5 socket and an anode cap. While the compatibility of these tubes with the amplifier could be solved with adapters – UX5 to octal socket – they all require anode caps which would complicate the wiring of the adapters, and the anode caps are not even the same size… Thus I have chosen to install parallel UX5 sockets.

The UX5 Alternative

The parallel UX5 socket is wired by connecting the appropriate pins – 1 and 5 on the UX5 are the cathode (filament) connections and should be connected to pins 2 and 7 on the octal socket with twisted wires (AC heating!), pin 2 of the UX5 is the screen grid (g2) and should be connected with pin4 of the octal socket.

Now for another difference: the maximum g2 voltage of most tubes that can be used in this amp is actually at least 300V, and thus the 1N5370B (56V 5W) zener diode in the schematics gets exchanged for a 22V zener of the same series (1N5358B – 22V 5W), which will just allow for the voltage drop across the primary of the output transformer, keeping the anode voltage at a slightly higher potential than the screen grid voltage.

If you are planning to use 2E22 tubes in your RH-TTA (like I do), you should connect pin 2 of the UX5 socket to pin 4 of the octal socket via an adequately biased zener diode (i.e. the cathode represented by the line on the zener diode should be connected to pin 4 on the octal socket). Of course, the zener in this case should be the already mentioned 1N5370B – 56V5W zener – since the maximum g2 voltage for 2E22 is 250V (g2 to cathode).

The grid or pin 3 on the UX5 socket should be connected via a grid stopper resistor (anything between 200 and 500 ohm would do) to the same point to which the grid stopper resistor leading to the octal socket is connected (i.e. the connection between coupling cap and grid bleeder resistor). Connection by wire is a possibility as well, but I think that this solution will provide better protection of the circuit from oscillations.

Last but not least, pin 4 of the UX5 socket – this pin is either g3 in the pentodes (307A, 2E22) or NC for the 1624. Thus, depending on whether you are planning to use the 807 or not, you may connect it directly to pin 6 on the octal socket (virtual cathode point). If you are not planning to use the 807 in this amplifier (like myself, since I have got no 807 tubes), you can also connect this pin simply to ground, ensuring that g3 on 307A and 2E22 is always at 0V potential.

A perfectly safe solution for the anode connection is a 4mm “banana” plug like the one you might use for the loudspeakers – if it is well isolated. Make sure that the plastic isolation is good enough for at least 400V DC if you are installing the UX5 socket and the anode “banana” jack on a metal sheet: if you are using wood or some other isolating material (like Plexiglas) no particular care has to be taken to isolate this jack. Having a jack instead of a fixed wire allows the use of sets of cables with anode cap on one side (9mm for 307A, 1624, 807; 14mm for 2E22) and a 4mm plug on the other, so they can be removed from the amp when you are not using them, while the socket and plug that you are not using can be covered (protected from dust and fingers).

What About UX4 Tubes?

Indeed, the 2A3 was originally meant for the UX4 socket – and as such it was manufactured in a wide array of versions – double-plate, bi-plate, mono-plate. The 2A3 is currently produced on UX4 socket by several manufacturers, while the special edition octal 2A3 is probably not produced anymore – so what if your main point of interest is the classic UX4 2A3?

The first option would be the standard RH2A3 schematics, with UX4 sockets. Or, if you would like to use other similar tubes, like 6B4G, 6A3, or 6A5G – and you are not all that into odd direct heated pentodes and tetrodes – you could build the RH-TTA but with an UX4 socket instead of the UX5 socket. Indeed, all those UX5 tubes are NOS only, and while prices may still be low enough to be intriguing – you might not be interested in buying any if you do not own some already. Having an UX4 socket in parallel with the octal socket is still allowing for a wide range of relatively common tubes, most of which are still being produced, or available at quite affordable prices.

Finally, you can use UX4-to-octal adapters. Besides buying the adapters, you could build your own, just like I had to do. To build an adapter you will need an empty octal base (preferably new, but you can also remove them from dead or shorted octal tubes) and a suitable UX4 socket – by suitable, I mean the round body type that can be removed from its metal retainer. Since the base will most probably be plastic (phenolic or similar), plastic sockets might be preferable for this use – but I chose to use ceramic sockets since I had those available.

Adapters are rather easy to make – all you need to do is solder short insulated pieces of wire to the UX4 socket lugs (making shure that once you insert the socket in the octal base the lugs will not short with the pins of the base) and remove just enough insulation from the wire pieces as to fit the length of the octal pins: the pin will be filled with wire which should protrude slightly on the other side, while inside the base the wire remains insulated. Once the wires are in place, apply solder with rosin and paste, ensuring that it flows inside the pins and makes good contact with the wire. When you finish soldering and the check shows that no mistakes were made connecting, and the connections are sound, you can glue the socket to the base. While cyan-acetate glues might seem appropriate, they are rather conducive to messy work and do not guarantee good connection unless the fit is tight. Two-component epoxy, on the other hand, will hold perfectly regardless of material type (ceramics on phenolic plastic or similar) and will fill the empty space if the fit is not tight. Do not forget to cover the exposed parts with tape that you will remove later to prevent ugly spills of glue or epoxy.

Once you make your adapters (or buy them) – you are ready to use UX4 based tubes on the TTA! And this means some unexpected guests, like the 45…

Enter the RH45

The 45 is highly coveted as one of the best sounding tubes of all times. A predecessor of the 2A3, it shares socket type, pinout, and heaters voltage with that much wider known and nowadays more used DHT. It seems that, unlike the 50 – the 45 is not yet extinct, and that holds particularly true for SE amps applications. After all, you do not need a matched pair – similar tubes are good enough – while with the current setting arrangements in RH amplifiers the only thing you need to worry is finding some of these triodes in pristine condition.

The 45 family can be basically divided as older globe types and the later ST shaped types. Interestingly enough, while many call ST shaped tubes “Coke bottles”, and find them rather sexy looking – in this case form only follows function. While the globe shape allowed for larger (and easier to manufacture) inside structures, the ST or Shoulder Tube (Type) boasts the shoulder that helps better arrange and fix the internal structure. Globe shaped tubes of the same type probably sound different than ST shaped tubes due to the different structure and possibly being more or less prone to micro-phonics.

What is the maximum anode dissipation of the 45 - or, for that matter, the maximum anode dissipation the 2A3? This is rather difficult to find out. While it is mostly assumed that maximum anode dissipation of 2A3 types is 15W (some datasheets actually state that value as “design center values”, which means they may be exceeded), data on the 45 is even more difficult to find. Guessing from several sources, the “historical” 45 had probably 10-11W of maximum anode dissipation. Now, those were certainly not “absolute maximum ratings”. This is not a story about constructing better than necessary – rather about technology being not precise enough. Just like the old skyscrapers were over-engineered and more massive than it seems necessary in today’s terms, the clue is material tolerances. The standards for steel and concrete were less tight than nowadays, while they had to build skyscrapers that would not crumble with the change of wind direction, or stormy weather, or medium intensity earthquakes – tubes were built to do their task, for instance amplify music in a home radio receiver. 2W of output power during many hours (years) of use meant having to use a given size or thickness for the anodes… and so on. Whether the 45 is really a 10W anode dissipation tube, or more powerful than that – this really depends on what you expect to get from it, and for how long: that is the real tube power equation.

 

Starting from available UX4 sockets or adapters in an RH2A3 (or RH-TTA), there are several important issues for implementing the 45 (i.e. in the RH45 amplifier):

a)      current draw should be set at 36mA – either by adding a 15 ohm resistor in series with the current setting resistor of the RH2A3 (TTA), or using a 36 ohm resistor in an amplifier to be used exclusively with 45 tubes (the RH45 amplifier);

b)      lower current draw will cause a rise in the B+, thus it is advisable to use 5R4 rectifiers with higher voltage drop (5R4GY and 5R4WGY) or 5Y3 rectifiers – another alternative that comes to mind is an 80 type rectifier (to be swapped for 5Z3 in the same UX4 socket when other higher dissipation tubes are used drawing more current); this will ensure that the B+ stays below 340V;

c)       36mA current draw and 330-340V B+ means a bias voltage (cathode to ground potential) of 55V approximately, which leaves about 270V across the tube, for slightly below 10W anode dissipation – so far so good, but the LM317 will need to be replaced with a TL783 as the voltage across the rectifier will probably exceed the maximum voltage rating of the LM317 regulator.

The TL783 is a totally transparent replacement for the LM317: same pinout, almost identical reference voltage, same packages available – which means that it is very easy to implement. While some think of the TL783 as an LM317 with higher voltage MOS type pass element, the differences actually exceed the initial expectations. Nevertheless, for setting a precise current draw at higher voltages (good for up to 125V across the regulator) the TL783 is perfectly suited, and will add a measure of reliability to any RH amplifier, regardless of expected bias voltage. The only issue with the TL783 is its price (almost triple that of an LM317) and availability (not all resellers have it on stock – but if you live outside of Serbia you will source it rather easily).

If the intended version is RH45 only, the TL783 is not strictly necessary, since the 470 ohm resistor can be increased to about 1k ohm, increasing the voltage drop across it and keeping the LM317 safe. But increasing the resistor in an RH-TTA (or any of the RH2A3 versions) is not feasible since the voltage will be either too low for current regulation, or too high for LM317 implementation, depending on the current draw.

Last but not least, with the TL783 the DIYer does not need to worry too much about the switch that excludes the voltage dropping resistor from the cathode circuit – leaving it excluded for the 2A3 or 45 will not harm the TL783 (but would kill the LM317), particularly if the heat-sink used allows for 2-3W of dissipation. With the TL783 the main purpose of the voltage dropping resistor remains reducing the dissipation of the regulator, keeping it as cool as possible.

Output Power

The RH-TTA is a further development of the RH2A3 presented earlier, thus output power and distortions are as already shown. While 2A3 and 6B4G tubes will allow about 5W output power, more can be had from the 1619 and the other DH pentodes and DH beam tetrodes mentioned, due to the higher efficiency of pentodes and beam tetrodes in particular. The 307A is a 15W dissipation pentode, similar in output power to the 1619 – while the 1624 is a 1619 in ST shaped envelope with increased dissipation to 25W. The 2E22 is a 30W dissipation pentode, but output power remains basically the same as the other pentodes/beam tetrodes mentioned, since the output power is limited by the voltage across the tube and the fixed current draw. The 1624 and the 2E22 can be operated as well with GZ34/5AR4 rectifiers, increasing the available B+ for 35V approximately. The same is true of other indirect heated tubes that might be used in the RH-TTA and have higher than 15W anode dissipation ratings, like the EL34 or the modern 6L6 types: output power is limited by the available B+ and current, as well as the necessity to apply higher primary resistance to keep distortions at bay. Output power will generally range between 5 and 7.5W, which is more than enough for serious listening even on relatively inefficient speakers (88-90dB/W/m). The RH-TTA is not about an important increase of power using pentode and beam tetrode tubes – but about the possibility to choose in accordance with taste and availability.

The novelty in this case is output power with the 45 tube (RH45 amplifier) – almost 3W. Whether you like the simulation results or not, listening to the amp clearly shows that output power is slightly lower than with 2A3 tube types, but the result is unexpectedly loud and satisfying. Basically, it is almost the same power output possible with a classic “no feedback” 2A3 SE amp. I have not tried to push the 45 in the same operating conditions applied to the 2A3, which would result in 15W dissipation: while I am curious enough to try it (and see whether the anodes would develop red sports or, probably, not) – that would be totally unnecessary and irrelevant, just like I consider pushing the 307A to 25W dissipation an unnecessary pass-time: if you need more power, there are tubes fitting the same socket that might be used for that purpose, like the 2A3 instead of the 45, or the 2E22 instead of the 307A. The 45 is too rare to be squandered for power at all costs. That said, 3W is almost 50% more than your average 45 SE amplifier, without excessive stress for the tubes.

I noticed the RH300B output power being discussed on a forum – while some were questioning the feasibility, others were quick to point out the merits of their designs. All in all, between those who understand what it takes, and those that do not understand, there was no mention of the particular sound quality achieved by the various RH amplifiers. Design and engineering are not meant to be the purpose, rather the means to achieve sonic excellence – but  those who never built one, or listened to an RH amplifier cannot discuss the sonic merits, being confined to accepting or negating design accomplishments. Anyway, for those interested, here is a simulation of the RH45 without Rfb at approximately 1.2% distortion...

I guess this example is illustrative enough - while correct application of feedback is paramount in RH amplifiers, some merit should be given to distortion cancellation as well.

Output Transformers

Just like I suggested when writing about the RH2A3/1619, for the operation of the RH-TTA with a wide array of tubes you will need a flexible transformer. While you could get away with a usual 3k primary with 8 or 4 ohms secondary – if you use the 4 ohms secondary with your 8 ohm speakers you will get approximately 6k primary loading – such arrangement is valid only if you do not change your speakers. Assuming 4 ohm speakers, the 8 ohm output will become approximately 1.5k – which is not a value usable with the RH-TTA: if that is the case you would obviously need to change your output transformers as well.

The Lundahl LL1623 is extremely flexible, and besides the possibility to configure from 1.6k to 5.6k primary and a choice of 4, 8, and 16 ohm secondary configurations – you can easily arrange the “neighboring” values to be changed with switches, like I did. Operating 3 switches simultaneously (it is difficult to find a two way 12 contact switch, or a rotary switch that has adequate current capability) I can choose between 3k and 5.6k primary loading into 8 ohms (on both channels at the same time), and if in the future I change speakers (for instance, 4 ohm units) I would just have to reconfigure the wiring harness. I regard this characteristic as paramount, particularly in a DIY project – since transformers are virtually “forever”: they do not change their characteristics appreciably with time, and when built with modern isolation materials they should have a much extended lifespan. Therefore, if nothing else, you should be able to reuse your transformers in some new project one day, and that is where flexibility means good investment.

If the RH-TTA is to be realized as a more limited solution – like just using 2A3, 6A3, 6B4G, and 6A5G, for instance… any good 3k primary output transformer would do, including the one I mentioned already. If the choice is to use only the UX5 pentodes and beam tetrodes, any good 5k-6k primary output transformer would do, again – including the one I mentioned.

The RH45 (standalone version) would require a 5k-6k output transformer. Since the output power is rather limited at almost 3W, it could even be built with output transformers salvaged from EL84 consoles, the type of output transformers that DIYers often use to build RH84 amplifiers with success. But I guess that most would opt for higher quality alternatives, since the 45 is a highly coveted and rather rare tube.

Amplifiers, Tubes, Lifetime, and Boredom

On the other hand, expensive boutique tubes and rare NOS tubes are not something I would recommend. Good examples for this might be the (original) NOS WE 300B tubes, including the later manufactured WE300B tubes (no comments necessary here I believe) or the EML tubes. While the various Emission Labs tubes might be of exceptional manufacturing quality, viewed from a historical perspective I cannot see a rational reason to use such tubes at exorbitantly high prices (take a look at the price list and ask yourself are you buying historical rarities or tubes made yesterday “to exacting standards”: the prices are ridiculous since you probably cannot use them as a means to avoid taxation, unlike investments in real estate). To rephrase the previous tought, while they might last a long time, they will not last forever, and only time can tell whether they will last as long as the “original” WE300B tubes were reported to last in the theater amplifiers (decades, or tens of thousands of working hours). Even if they do last 20 years of everyday operation, I guess one might get tired of listening to the same amp and the same tubes for 20 years? (while WE was probably happy for not having to replace the tubes in the amps they rented and which represented a source of income and business venture). Well, life is not a permanent condition, too – while you might reuse your transformers in a new amp one day, with different output tubes, what good is a tube that might last 20 years and it costs more than several complements of tubes for several amplifiers that you might use in those 20 years? Besides that, high precision and attention to detail is today represented as a path to great sound, while just throwing a glance on the interior of some of the tubes renowned as great sounding (45s, particularly the globe versions) should make you wonder how is it possible to show such low attention to detail (was it, really) and still get lasting quality and great sound? Last but not least, even if a tube is capable of exceptional sound quality, it will not provide said sound quality without a good output transformer, good passive components, maybe good drivers… not to mention good schematics exploiting the quality of the components: a tube is just a component that we use for a given period of time in our lives.

I am aware that manufacturing tubes in small volume nowadays, and eventually starting from scratch, is not cheap or investment efficient, I frequently ask myself how is it possible for the Chinese factories to produce acceptable replicas of “extinct” tubes? While those tubes can hardly be considered exact replicas, they indeed tend to work just fine in their own right. I am not questioning the quality difference between boutique and Chinese production, although it is often perceived as higher than it actually is, but addressing the quality vs. price ratio, as an obvious function of the perceived value. Sometimes marketing is used to mix true and useful information with not-so-true statements backed by assumptions…

Anyway, this type of “boredom” with common places in life and all the repetitive experiences of our everyday routine (and the same tubes for 20 years) is something that I am addressing with the RH-TTA: you may as well build several amps, but finding place for them on your shelves might be a problem - or you might not have the time to build new amps as often as you might want to. A flexible amplifier will let you use and explore many tubes, changing nuances and enjoying your music without too much effort: not even having to remove cables to connect another output, or another amplifier. While the RH Universal was directed at those requiring power and simplicity, the RH-TTA is directed at those who are interested in exploring the possibilities while output power is an issue relegated to the background.

The Tube Tester Amplifier

This amplifier is capable of using indirect heated tubes without modification, but most tubes I have used or tried it with are direct heated types. At first I was thinking about “DH-Universal”, but the presence of a number of switches that modify the characteristics to the tube, and the paralleled sockets are reminiscent of a tube tester.

First of all, there are 3 switches necessary to choose between 3k and 5.6k primary resistance – those should not be operated while the amplifier is operating, of course, just like all the other switches: all those choices are not something that is made “on the run”, but a choice between conditions required to operate. Located between the output tubes are two additional switches: one is used to exclude the voltage dropping resistor from the cathode circuit (necessary when operating the pentodes and beam tetrodes, since their bias voltage is less than 20V and the resistor would preclude correct operation of the current draw regulator), having a “triode” and “pentode” setting. The other switch is used to choose current draw, 60mA or 36mA (for the 45), which is achieved by bypassing the additional 15 ohm resistor.

Using the switch to bypass a resistor means that if the switch loses contact, the worst case scenario is the most benign – tubes will be operated at 36mA, and the voltage dropping resistor will remain in the cathode circuit at all times. The operation of the amplifier will be compromised in terms of performance for the duration of the fault, but neither the tubes nor other components will be at risk.

On either side of the rectifier tube, near the cathode connections of the output tubes, two larger switches (25A contacts) are used to choose between 2.5V or 6.3V secondary. This obviously allows to use i.e. 2A3 or 6B4G tubes. The unused secondary is left “open”, thus no current is drawn and no additional heat dissipated. Indeed, trying to operate 2A3 tubes on 6.3V might damage them… but I guess the DIYer is going to be aware of that – on the other hand, 6B4G tubes will not light their filaments at 2.5V and there should be no sound output or current draw since the cathodes will not be able to emit electrons: this condition might damage the tubes as well, contrary to what many believe.

The last switch is located out of sight on the back side near the power transformer, since it is rather rarely used. The purpose of this switch is to allow 5.5V heaters operation with the correct voltage – it is an 8 contacts 25A switch which is basically bypassed from side to side with 0.39 ohm resistors. When the switch is closed, the resistors are excluded and 6.3V are directed to the heaters voltage switch. When the switch is open, the current flows through the resistors: at 1A current draw each resistor drops 0.39V, thus 5.52V are delivered to the heaters voltage switch, to be used for the 307A.

If built with UX4 sockets instead of UX5, or if UX4 to octal adapters are used, the 300B tube can be used as well – the 8 contacts switch has to be fitted with different resistor values (0.56 ohm) to provide 5V to the 300B heaters. It is questionable whether it makes sense to use the 300B tube in the RH-TTA, since the output power is going to be limited by the available voltage and current draw, in a similar way in which other more powerful tubes are limited. The 300B will yield just 5.5W at 1% distortion – but the DIYer who would build the RH-TTA is probably satisfied to get 5W from the 2A3 and might be interested in getting approximately the same power from 300B tubes – without building another amplifier.

While the 300B can actually yield a lot more output power, operating it at low anode dissipation will without doubt guarantee long tube life and trouble-free operation. Those for whom 5W might be enough would probably enjoy the 300B and its characteristic sound in this circuit…

Last But Not Least – The Sound

All this text, and no mention of how does the amplifier perform in the music reproduction scope… basically, the sound quality issue was covered in my post on the RH2A3/1619 amplifier, thus the points worth mentioning are both comparison between the sound of various tubes, and the implication of having a finalized well laid-out box against a breadboard amplifier.

The RH-TTA ended up being quite large by my standards – it almost dwarfs the RH300B. I was unable to fit all the necessary parts in the rather small box in which I have fitted the RH300B, but I used the extra place to ensure that there is no interaction between the various elements.





All transformers are hidden inside the box, and the toroid transformers are vertically mounted, further minimizing the effect of their field on the amplifier. With 2.5V output tube types no noise or hum is audible even with an ear on the woofer or midrange of my 88dB/W/m speakers – although the heating is AC. Enough said: no need for DC on the heaters in this amplifier, and that is probably one of the reasons it sounds as good as it does – but this is a different topic, one that may require further research and discussion.

The sound has improved in comparison to the previous impression – maybe the output transformers have gone through some break-in period, and probably the remaining components have also benefited from a break-in period. I am not a big fan of the breaking-in theory: either it works or it does not… but it is a matter of fact that improvements in time can be heard. Good quality stuff usually sounds good from the first note, although it might improve with the passage of time. The basic tone quality of the amplifier remains unchanged regardless of output tube used, but each tube brings its intrinsic sound quality to the mix, like a distinct flavor.

The overall winner in the triode class is the NOS 6B4G (double-plate, black anodes), showing a margin of midrange quality above the Shuguang special octal 2A3 (bi-plate, black anodes) version: the new generation of mono-plate current production 2A3 tubes probably sound slightly better and can be directly compared to the NOS 6B4G. By the way, I was using this same pair of 6B4G tubes for years in a classic no feedback SE design, and they have probably worked at least 2000 hours – but they still look and test like new… and their price when manufactured was nowhere near the asking price of the current production boutique tubes. As a rule of thumb, the 2A3 family tubes are slightly forward in the midrange, but the midrange nevertheless shows warmth of tone.

The 1619 is my favorite in the direct heated pentodes/beam-tetrodes class. While being a rather ugly metal tube (no heaters to warm up your sight), it has no cap and the envelope is grounded, thus very safe. The sound is very liquid, with an overall quality that goes a long way towards beating the particular qualities of the 307A or the warmth of the 2E22. I have not tried the 1624 in this circuit since I haven’t got any – and it would be interesting to assess whether it is better or worse than it’s lower power metal sibling. As a rule of thumb, the pentodes tend to leave the impression of better extension and smoother frequency response, without midrange forwardness – but without the midrange warmth shown by the DHTs. This is more than anything else a matter of taste – the choice between rich overall tone, or pronounced yet warm midrange (more or less).

Finally, the highly coveted 45: I am not mentioning it in the triode class since it is operated at lower current and has lower power dissipation. The 45 tube is in a class of its own. My expectations were indeed very high, and while this is not “sound like I have never heard before”, the 45 does, even in the newer ST shape (the only 45s I have are Sylvania ST shape, provided by a DIY friend) confirm the intrinsic qualities that it was coveted for.

The particular sound of the 45 is rather different than the 6B4G some consider as its descendant (in particular, the bi-plate version is often considered as two 45s in a single tube). The 45 shows none of the mellowness expected from DHTs – on the contrary, it is lightning fast and detailed. While bass is well defined, fast and quite strong (unexpectedly so), the mids and highs are extremely detailed, liquid, and fast at the same time. The closest approximation to this type of sound is actually the “fake mesh” globe shape 300B of current production. Just like this particular 300B might seem bass shy, the 45 seems to lack some bass volume – at least until the listener understands that it is more about a lack of oomph, while the vibrations are very much present. But the particular quality of the mids and highs is what fascinates the listener – making it difficult to change the 45 for another tube, just like many prefer the sound of the “fake mesh” 300B to other more solid types, due to the transparency and liquidity of mids and highs. The less pronounced bass, the power limitation (on rather inefficient speakers in a rather large room) – it all fades away against the quality of the mids and highs. There is neither the particular richness of tone like with the pentodes, nor the forward but warm midrange characteristic of the DHTs: the strength of the 45 seems to be the speed, and liquidity, combined with a forgiving distortion pattern reminiscent of dust showing in rays of sunshine. On the other hand, the 3W provided by the 45 in this amplifier can go quite loud, and the soundstage thrown is very large and airy.

Credits

Just like on previous occasions, I would like to thank all those who support my work – in particular Mr. Per Lundahl, and a group of DIY-ers based in England who post on the audio-talk forum. New designs and developments would remain just ideas and afterthoughts without the help of friends from all over the world who follow my work and my blog.