RH Universal v.2 – Totally Universal

The RH Universal is a derivative of the pilot 2^nd generation RH amplifier project – the RH88. When I was working on the 2^nd generation design, my main goal was stunning simplicity and lowest possible parts count. The concept was tested on 6550 tubes, both tetrode and triode connected, confirming in practice the auspicious simulation results, and showing that there is no difference in sound between triode and tetrode connection, except for the lower power obtainable in triode mode. While this makes operation of tetrode/pentode tubes in triode mode just a waste of available power, it does however open the possibility to adopt the 2^nd generation circuit with triodes, as in the pilot project RH300B.

The 2^nd generation circuit however requires relatively high voltages, basically excluding many possible driver tubes and making it unsuitable for many output tubes. The original RH Universal pushes this concept to the limit by making it possible to create adequate operating conditions for a range of tetrode/pentode tubes: with a fixed current draw, anode voltage is simply adjusted by means of changing rectifier type. Thus a wide range of output tubes can be used to full output power – at the expense of strict driver tube limitation (ECC81, eventually ECC82), and low input sensitivity. While low input sensitivity should not be a problem when a good active preamp is used, the driver inflexibility is in stark contrast with the output tube universality.

With a couple of modifications to the original RH Universal circuit, here comes a version 2 – now (almost) totally universal! The basic differences are:

1. Application of 1^st generation driver circuitry – the classic anode resistor returns, allowing more freedom in operation to the driver tube which is not limited by the swing of the output tube.
2. Adjustable Rfb allowing the circuit to be perfectly tuned for the ECC81 family of double triodes (high-ish mu, relatively high transconductance) – or for the ECC83 family of double triodes (high mu, relatively low transconductance).

The advantages of this version 2 can be summarized as:

• Total universality – a very wide range of output tubes can be successfully implemented: at the flip of a switch the circuit can be optimized for either of the two different tube types, making it possible to use a wide range of tubes which fit the same socket (and pinout) type, even adding the possibility to combine tube-rolling with different operating points and feedback. Some of the now usable driver tube types may be odd, rare, or low cost types – a personal favorite of mine is a 12BZ7 used as a substitute for ECC83. 
•  Technical perfection (as far as possible) for the two optimized tube types, and near perfection for similar driver types, allowing for 10-11W output power at approximately 1% distortion levels (with KT88/6550). It goes without saying that maximum output power is higher… and depends on tube type.

A further advantage of the driver universality is the possibility to make an “all octal” version of the amplifier, as 6SN7 and 6SL7 can be alternated with excellent results – and similar tubes like the Mullard ECC35 can be used as well.

While modification of original RH Universal amplifiers is quite simple – the basic difference is one switch and two added resistors per channel – the v2 does not supplant the original. Besides being simpler, maybe more elegant, the original has a slightly different sound due to the 2^nd generation driver circuit. The v2 sounds like an RH84 with lots of power, while the original version offers a slightly different nuance to the sound. It goes without saying that nuances are to be heard and judged based on the same passive and active components – the output transformers, caps, and resistors have an important sonic character which cannot be circumvented by simply changing the active components. The perfect combination may be completely individual, and the RH Universal allows for a wide range of active components alternatives on the same passive platform – a range now considerably widened with the version 2 update.

RH-307A

(the DHP RH amp)

An RH amplifier with the 307A (VT225) output tube has been a long delayed project for me, a standing promise since the time of the original RH807. The resemblance between the two tubes actually ends with the 5-pin socket and cap, among other reasons because the 307A is a true pentode – unlike the 807 which is a beam tetrode tube. At the time, datasheets for the 307A were unavailable on the net, and the only reference while I write is a WE datasheet: strangely, because I have never seen a WE 307A, not even a picture of it – the most common 307A (VT225) are Ken-Rad and National Union.

 

While the promised amp was obviously to be an RH type, when I finally decided to design it I was already over the “old” type schematics, but the relatively low anode voltage at which it would operate as compared to the relatively high bias required (i.e. cathode voltage: the voltage swing that the driver tube has to produce will be significant) meant that the tried, tested, and faithful ECC81 family might not be entirely suitable for this task, unless used in the original RH type schematics. This is where the ECC88/6DJ8 family enters: the very high trans-conductance allows this tube to obtain the necessary voltage swing (approximately 60V p-to-p) while being operated at relatively low anode voltage (low, that is, to ECC81 standards).

The RH-307A is largely based on the RH Universal schematics – the main differences are the different driver circuitry and the fact that being a direct heated pentode, the 307A has a slightly complicated “cathode circuit”. The anode of the driver tube is connected in RH 2^nd generation style (Rfb being at the same time the anode resistor of the driver tube), and g2 (the screen grid) is connected to B+ through a 51V zener diode. While providing a referenced voltage for g2, the zener diode value is chosen to ensure that in normal operating conditions, whichever rectifier is chosen, the voltage across g2 will not exceed 300V.

As the 307A is a direct heated pentode, its heater wire is at the same time its cathode. Besides the known AC hum related issues, this poses as well the voltage referencing problem – there is a definitive need for a reference point substituting the cathode connection in indirect heated tubes. This reference point can be created with a wire-wound potentiometer which can be used as well for hum-nulling, but besides the fact that a pair of resistors costs less than a pot, I tend to have more faith in a soldered connection than a pot slider. Since the hum-nulling spot will inevitably be the center position, the best solution is to match two resistors (100 ohms) and use them to connect in series the two heater-cathode terminals: in  this manner the mid-point between the two resistors becomes a cathode reference point.

Being a true pentode, the 307A has a suppressor grid (g3), and being a direct heated tube, this grid has its own pin. Since the suppressor grid in pentode mode should be connected to the cathode, in this case the connection should be performed between the g3 pin and the “cathode reference point”.

The current setting circuitry comprises the usual LM317 and a current setting resistor, in this case 27 ohms. This means a constant current draw at the cathode of approximately 46mA, of which 43mA are anode current, while 3mA are g2 current (estimates based on the datasheet, and on mathematical operations subtracting driver current draw from the voltage drop across the transformer primary). To bypass the cathode circuitry, each heater connection is decoupled to ground via a 100uF 63V (or 100V) cap, providing a direct AC path from cathode to ground.

The heating of the output tubes could warrant a debate of its own – choices vary between AC and DC, where DC can be un-regulated, regulated, or fixed current. Whichever choice is made, it is very important to keep the cathode circuitry intact: I can only stress so much the importance of the “cathode reference point” in the connection towards ground.

My choice is always the simplest solution – AC heating. AC heating has several theoretical advantages and some shortcomings. The most important theoretical advantage is having the same potential (DC voltage) across the entire length of the cathode (heater). The most important (and audible) shortcoming of AC heating is hum (AC mains 50Hz or 60Hz low hum) which can be heard oh-so-much without the proper hum-nulling potentiometer, or the cathode reference point solution which I advocate. Even nulled, hum can still be heard, depending on the efficiency of the speakers. On my “common real world speakers” of approximately 90dB sensitivity – hum can be heard from the vicinity of the woofer, but is completely inaudible a couple of meters away (listener’s position), even during the quiet late night hours. AC heater hum depends as well on the heater voltage and is definitely lower than with 6B4G tubes, due to the 5.5V heater voltage as opposed to the 6.3V required for the 6B4G: I would guess it is comparable to the hum produced by an AC heated 300B amp.

A further shortcoming of AC heating is maybe more theoretical and seldom mentioned: due to the sinusoidal nature of AC current, the heaters (cathode) are not at the same exact temperature all the time… which temperature changes 50 (or 60) times per second. With DC heating, the temperature is constant since the voltage is constant as well… but I guess this is more metaphysics than real world experience.

The main advantage of DC heating is the lack of hum – but the rectifying and regulating circuitry is directly connected with the cathode and thus very much in the signal path (while the cathode current source made with the LM317 and current setting resistor is not, since it is bypassed with the two caps providing a separate AC path). Frankly, for high efficiency speakers (96dB and more) DC heating is the only reasonable option. In that case, whichever solution is chosen (regulated, current draw…) the output terminals should be connected to the heaters terminals on the schematics, and the rectifying/regulated DC circuitry MUST NOT be connected to ground.

Last but not least, the output transformers issue… The operating point is largely chosen on the recommendations given in the datasheet, at 43mA anode current and anode voltage slightly in excess of 300V (across the tube), taking care not to exceed the 15W anode dissipation rating. The chosen anode load is 5k – basically the same load used in all the 1^st generation RH amps, which is very similar to the datasheet recommendation. The 9W output power seems too high if compared with a 15W anode dissipation – but it can be expected that usable power will reach 7W. While I have not performed any measurements on my amp, it does go very loud – way above 6B4G levels, and definitely louder than the RH84. It is thus safe to say that the “usual” console amp output transformers that can be used for an RH84 can be used for the RH-307A as well, but a decent pair of output transformers is definitely in order if the full potential of the amp is to be unlocked.

The power supply should foresee at least three low voltage secondary windings, since each output tube must have its own, and the driver tube must be heated separately from the output tubes. Another low voltage secondary is needed for the rectifier tube – if one is used (which I always advocate). I would recommend the choke input solution as documented – after all, due to the DH nature of the output tubes, this amp immediately commends more respect than a humble EL84 tube. Still, the most important requirement is achieving between 360V and 380V at the B+ point. This is the variation between 5Y3GT and 5U4G in the proposed schematics, the latter touching 15W anode dissipation (in my amp – slight variations are possible with different DCR values of chokes and output transformer primary).

The choice of tubes is in this case limited to the driver and of course the rectifier tube – the 307A having no compatible replacements that I know of, including those with a different socket. Thus, while the driver circuit is designed having in mind the ECC88 family of tubes (any ECC88 type will do, since expected anode voltage is 85-90V, way lower than the 130V limit for common ECC88/6DJ8). E88CC, CCa, 6921… all those tubes will perform flawlessly in the driver task. Besides those, direct replacements (same pins) that will allow the amplifier to achieve full power are 6BQ7, 6BQ7A, and ECC85 (6AQ8). The latter two tubes have lower (but still high) trans-conductance than the ECC88, but make up for it with a higher mu (the ECC85 has 58). While I have not tried it, I know that with a different socket wiring the 5670 (2C51) would do just as well as a 6BQ7A. All the tubes mentioned require 6.3V at the heaters and have internal shields: the shield should be connected with one of the heater connectors and directly to the ground, thus grounding the driver heater winding.

 

The rectifiers proposed are the usual 5R4 and 5U4, but due to the low current draw of this amp (110mA) it is also possible to use the 5Y3GT. The proposed power supply is choke first, and with the hybrid Graetz configuration each anode will “see” half the secondary voltage… at 260V and 110mA current draw, the 5Y3GT can be used in choke first power supplies without any worries or problems. If a different socket is used, 5Z3 and 80 are viable alternatives to 5U4 and 5Y3.

For the 307A heater windings, since 5.5V is an odd number, there is a simple and cost effective solution – 6.3V windings (again those 6B4G amps!) are perfectly suitable if 0.44 ohm resistors are placed in series with each terminal (at 1A current draw, 2x 0.4 ohms means 0.8V difference, i.e. exactly 5.5V – most 6.3V secondary windings give slightly more than 6.3V, and 0.4 ohms is not a standard value). 5V windings for 300B heaters might be probably used with success, since the 300B needs slightly higher heater current, thus the output will most probably be at least 5.3V at 1A.

The 307A (VT225) is a relatively odd tube which is poorly documented on the net. Besides a very popular high quality headphone amplifier, there is almost no other amplifier with this output tube -proposed or documented. Most amplifiers mentioned or seen use the 307A in triode strapped configuration, as some sort of “poor man’s” DHT – which is a shame, given the huge difference in power in comparison with the 15W anode dissipation DHTs (2A3 family including 6A3 and 6B4G). Actually, while the 1^st generation of RH amps was mainly challenging comparative designs with EL84, 807, EL34 – the 2^nd generation challenges further the more expensive and coveted DHTs, like 300B, 211/845, and the 2A3 family of tubes. As always, I prefer letting others judge the sound of RH amps, but in this case I will allow myself a hint: this amp sounds disturbingly good…

RH84 amplifier - revision 2

After more than a decade since the original RH84 was designed, I feel that this 1st generation RH design could benefit from a revision updating it to my current design views and correcting some decisions made in a previous time with solutions that are, perhaps, more matured.

   
The schematics of this revision 2 does not stray too far from the original design, as all the most important design choices and component values are still there. What are the revised issues?

1. g2 power supply

The original g2 resistor is now replaced with a zener diode, allowing for a more constant voltage on the second grid, and actually correcting a slight issue with Ug2 which has been observed with several builds. Depending on the DCR of the primary of the output transformer chosen for the build, the voltage of the 2nd grid tends to be higher than the anode voltage. While this is not a particular problem in practice, it is rather annoying and imprecise.

By replacing the grid resistor with a zener diode (in the schematics 21V) a stable voltage drop from B+ to Ug2 is made possible, where the zener diode value is chosen to be higher than the expected voltage drop across the primary of the output transformer. Most builds were made with smallish output transformers (sometimes from old console amplifiers) with a DCR of 200 – 270 ohms. As the anode current is approximately 40mA, the voltage drop lies between 8 and 11V: thus it is safe to assume that with a 21V drop from B+, the resulting Ug2 will always place the 2nd grid on a lower potential than the anode.

The lower the Ug2, the less will be the total current draw of the tube, and in particular less current will be drawn by the anode – if automatic or fixed bias is applied to the cathode or the 1st grid, respectively. This sets some limits as to the value of the zener diode, since too low Ug2 could cause the output tube to draw less current than intended (or needed for a given performance). The potential current draw issue is addressed by the next revision.

2. Automatic bias vs. fixed current draw

The second important revision in the RH84 schematic is the replacement of the cathode resistor (automatic bias) with a CCS (actually, current sink) fixing the current draw of the tube. Fixing the current draw of the tube with a simple circuit (LM317 + current setting resistor) allows a constant current draw regardless of the tube inserted. Besides the fact that no tube matching is necessary (particularly in SE amplifiers), this relieves the user or DIY builder from all concerns, including tube aging – as the current draw will have to remain the same throughout the life of the output tube. Of course, there are limits to everything – thus tubes reaching the end of life should be replaced with new or less aged ones.

The current setting resistor is chosen for a standard value (27 ohms) and needs to be checked for exact value, as well as reasonably matched between channels: 1% metal film 1/4W units should be just fine. The cap bypassing the current regulator needs to be at least twice the “bias” value (voltage drop from cathode to ground): for an EL84 22V is a good starting point among standard values. Needless to say, the LM317 needs to be isolated from ground and from the other channel’s LM317, while some basic heat-sink can be provided but is not strictly necessary, since the dissipation of each unit will be approximately 0.41W, a value most LMx17 in TO-220 package can handle without a heat-sink.

Power Supply for the RH84

When I designed the RH84 I was more interested with the audio part of the amplifier, than with the power supply the DIY builders would choose. While some understood and applied this “free to choose your own” concept, some had problems calculating correct values or choosing adequate components. Some mistakes in the power supply schematics part (lapsus calami) did not help in the process…

The first proposed power supply is a classic CLC type resembling the original proposal. The amplifier draws (both channels) slightly less than 100mA and the voltage foreseen at the B+ point (actually, C2 voltage) is approximately 315V. This time, all values are given – and are to be regarded as the necessary minimum – DIY-ers are free to build the amp with oversized power transformers, although oversized power supplies are not particularly necessary in SE amps since there is a very slight difference in current draw during operation.

 
The second power supply schematics refers to a higher quality choke input power supply which should yield the same B+ as the simpler one. Component values are given and should be regarded as the necessary minimum. While there is no important difference in current draw during the operation of the amplifier, I feel that choke input power supplies allow for better sound overall, and particularly in the bass section. This is probably due to the lower output impedance of the power supply, or to the intrinsic “play” of the impedances of the power supply components.

Particular attention should be given to the choice of the input choke – it will be prone to vibrations, and thus needs to be a potted or enclosed unit, as stiff as possible. Care should be given as well to the positioning of this choke, since it will have an important magnetic field due to the difficult task. On the other hand, the second choke does not need any particular attention – it sees low voltage swings and its task is just smoothing the ripple on the rectified AC. It can be the same choke as used for the CLC power supply version – an open frame unit will do just fine if it can handle the current (100mA).

Overall Improvements

Compared to the original RH84, this second revision should improve both the sound and the general operation. While RH84 amplifiers did not pose particular operation problems due to the simplicity of the design and overall stability – the fact that the output tube current draw is fixed, as well as a fixed g2 voltage, contribute to a general improvement of stability in operation, as well as predictable performance throughout the lifetime of the output tubes (the lifetime of the driver tube being longer by default).

When it comes to sound, both revisions lead to improvements, particularly in the reproduction of lower frequencies. While some might criticize the adoption of a solid-state current source, or a solid state zener diode, both devices are readily available and cheap – and used as in this amplifier can only help the performance while not introducing any superfluous noise or distortions to the sound. The current source (or sink) under the cathode is effectively bypassed since the capacitor provides an effective AC path.

The proposed revisions have already been implemented in the RH Universal, a 2nd generation RH amp. Basically, those design points have transformed the RH88 (for KT88 and 6550) into a universal amplifier accepting a wide range of output tubes – most of which are still in current production. While the RH84 is not universal, the proposed revisions add universality to the schematics, allowing its direct application to other similar but not identical tubes, like 6V6, 6F6, SV83 (6P15P), and 6L6G (6P3S). Basically, the RH84 revision 2 can be built using any of the above tubes without any modification – except for a choice of g2 zener value necessary to provide the safe voltage. For use with SV83 a 100V zener is recommended, while for most of the other tubes mentioned a 51V zener would be a fine precaution.

All existing RH84 amps can be easily modified to the revision B version – it is necessary to replace the g2 resistor with a zener diode, and to replace the cathode resistor with an LM317 with current setting resistor. The proposed power supplies are meant for all who intend building an RH84 from scratch (or any of its derivatives with 6V6, 6F6, SV83) – it is not necessary to replace existing power supplies with those, but a B+ of 315V (or less) is necessary for the safe application of the revisions.

RH Universal

After the success of the 1st generation of RH amplifiers (RH84, RH807, RH34), I have decided to create a new generation of amplifiers: simpler to build, less simple to understand (maybe) - and better, possibly.

Initial work and ideas were done in early 2005, creating the basis for two amplifiers - RH88 and RH300B. The RH88 was immediately transformed into breadboard, and it remained in function for the better part of the year. Due to objective reasons, work on the 2nd generation of RH amplifiers was halted, and it was not until the summer of 2012 that the breadboard emerged from a box.

Initial publishing of the schematics for the RH88 has shown that DIY-ers are interested, but fear the high prices of tubes, OPTs, chokes, and other parts associated with this amplifier. In particular, NOS tube prices have soared during the last decade, making both KT88 and 6550 NOS tubes almost collector's items with high price tags.

It is difficult to judge the prices involved when building a 5W class amplifier against the prices of building a 10W+ class amplifier. One should keep in mind that while the RH84 could be built with great results using parts from old console amplifiers, and improved using new parts - it's competitors are maybe some 2A3 or 6B4G amplifiers. The RH88 was conceived to compete with 211/845 amps (in class A1 - which most of those are, although many commercial products are marketed claiming class A2 output power) and easily exceed the power offered by the 300B SE amps... as such, it is still a relatively cheap amplifier to build, depending on the parts used.

The RH88 was initially conceived as a 6550 or KT88 amplifier, and later some DIY-ers asked for a customization to use 8417. While the latter could be inserted in the socket and used without problems, the output power would be lower, and the sound inferior, due to a lower current draw in the same circuit. I was also considering whether EL34 and 6L6 types could be used as well, and eventually decided that the amplifier should become universal - allowing the user to insert almost any given tube which is compatible with the pinout.

The principle is quite simple: regardless of tube type, the cathode current draw remains fixed at 100mA. Since output tubes vary in anode dissipation, with a fixed current draw the voltage across the tube must be variable to allow for different tubes to be used. The variable B+ is achieved by changing rectifier tubes (in this case, to simplify everything and avoid mistakes, rectification is done by hybrid Graetz bridge). Basically, a 5AR4 (GZ34) will give approximately 45V of DC more than a 5R4, which creates a difference of, roughly, 4.5W of dissipation. Some tubes draw less current through the anode, since they draw more current through the second grid (i.e. EL34)... which creates a further difference in anode dissipation (anode current x voltage across tube). Last but not least, tubes will differ in the cathode voltage at the same current draw (EL34 and 8417 approx. 10V while 6550 will have approx. 22V) which creates a further difference in anode dissipation (different voltage across tube).

As a rule of thumb, in the above circuit EL34 can be used only with 5R4 rectifiers, while GZ34/5AR4 should be used only with tubes capable of more than 30W anode dissipation (6550, KT88, 8417). It goes without saying that nothing forbids the user to combine the latter tubes with a 5U4 or even 5R4 - where of course the lower voltage across tube will result in a slightly lower output power. It can be thus said that output power varies between 8W per channel and 12W per channel.

The above picture: 7027A RCA tall bottle in action with 5AS4 GE and 6201 Philips driver.
In order to use EL34, it is necessary to connect pins 1 and 8 - which does not represent a problem to most tubes, except the 7027 which has a second connection to grid No. 1 on pin 1. A switch will allow the user to connect or disconnect the two pins on the output sockets (part of the switch can be seen on the above picture, choosing between the "7027" and the "EL34" setting.

Besides allowing the user to use a wide variety of tubes (basically, what is at hand), the amplifier is particularly suited for playing with tubes (tube rolling). While the basic character of the sound remains the RH signature, additional nuances can be had from different combination of output tubes, rectifiers, and even different drivers (ECC82 can be used with 8417 and EL34, but not with 6550 and KT88).

I prefer to let others speak about the sound of the RH amps...