Hybrid

home
back to amplifiers

Hybrid amplifiers are amplifiers with both vacuum tubes(valves) and solid state components such as transistors, FET’s, MOSFETS, op-amps. The reason for making hybrid amps may be a commercial one where one wants to attract people that love the nostalgic glowing glass bottles that exist in many shapes and topologies.

Above a DIY hybrid amplifier that runs the tube of 100V and uses BUF-634 output devices.

Technically there are simple reasons for hybrid amplifiers.

A: 12V to 48V power supply voltages can be used instead of > 100V for ‘all-tube’ amplifiers
B: To ceate smaller sized/light weight amplifiers.
C: To increase output power for LOW impedance headphones without the need to use output transformers.

In most cases hybrid amplifiers use tubes in the voltage amplification stage and solid state devices in the output stage.To get the most ‘tube amp effect’ this voltage amplification stage should not use local nor overall feedback as that ‘kills’ the reason why tubes sound different. When feedback IS used there should only be a small amount of it, just enough to reduce the gain of this stage. The reason for this is the NON- linear behaviour of the tube.
Because of this imperfect behaviour relatively large amounts of DISTORTION are introduced. This may seem a bad thing (distortion …yuck !) but because of the nature of this distortion it doesn’t SOUND like distortion at all. In fact the ‘flattening’ of the original waveform, which simply IS distortion, makes the tube PRODUCE harmonics BUT in a different way transistors do.
In hybrid amps not much gain is needed, in tube amps the output voltage created by the tubes MUST be a LOT higher because the output transformer lowers it again.
This is needed to increase output current. Therefore multiple tubes are needed for all tube amps while hybrids can work with just one tube already.

With solid state components the induced harmonics do not get much smaller in amplitude with increasingly higher harmonic frequencies. Tubes, however, exhibit large amounts of lowest order harmonics and the amplitude of the next harmonic is lower in amplitude. In nature most sound sources (instruments e.t.c.) ALSO have harmonics and for MOST of these instruments the decay in amplitude with each higher harmonic frequency is quite similar to that of a tube a tube. For this reason a tube is often said to sound ‘more real’ or warmer because of the ADDED harmonics.

Above typical distortion products of a tube amp (in this case the OPA88, the predecessor of the ‘Project Ember’ at -20dB) where you can see the fundamental of 1kHz (the big spike that is supposed to be there) and the first harmonic at 2kHz and the second one at 3kHz. The higher harmonics are drowned in the noise floor of the sound-card that was used. The THD (Total Harmonic Distortion) is 0.1%, but as these harmonics are very similar to those of real music sources it is NOT perceived as distortion but as ‘natural’ sound. Sometimes this type of pleasant sounding additions is referred to as ‘Euphonic distortion’.

When larger amplitudes are present the distortion increases (see below). The 1st harmonic is higher in amplitude and more harmonics appear. It starts to sound brighter at higher volumes and at low volumes it sounds more ‘correct’. Below is at -10dB and the distortion is up to 0.5% already. Still sounds ‘natural’ though and to most people it will even  sound MORE natural than the actual recording is. Similar to the ‘distortion’ present in vinyl recordings due to compression.

Solid state amplifiers have lower harmonic amplitudes but they do not become smaller in amplitude (much) the higher those added harmonics are. This is heard as ‘harsh’ sound and ‘unnatural’ as such an harmonics spread is not present in actual instruments.

Below a plot of a solid state design. It is made with a ‘regular’ op-amp and is for illustrative purposes. Most (headphone) amplifiers will MUCH perform better (have lower amplitude harmonics). These lower amplitude harmonics could well be far below the audible limits under normal listening conditions. THD (which means the total energy of all the harmonics is added) amounts to just 0.01%. Better amps can easily reach 0.001% or even lower. The harmonics would ‘drown’ in the noise of the used sound card so are pointless to show. Yes… I know I should buy a sound-card that has a better noise floor.

The clear conclusion is of course that the PERCEIVED increase in sound quality with tubes is ‘artificial‘ but pleasant to the ears of many. Signal purists, who feel the output signal should be as CLOSE as possible to the recorded waveform, strive for unaltered sound NOT use this kind of amplifiers, unless it has more than 3 amplifying tubes in it and loads of (overall) feedback so these added harmonics are reduced in amplitude considerably but what would be the fun in that ?

So we can use a tube to ALTER the sound in a pleasant way and preferably NOT use feedback that lowers these generated harmonics.

Tubes usually work on high voltages (as they are designed to function optimally on those as higher voltages) and this could be problematic certainly for DIY’ers. The fun part of a LOT (but certainly NOT all) is they DO work on low voltages and even surprisingly well. A disadvantage of these low voltages is the currents in the tube are quite small and thus NO power can be delivered, most certainly not in most of today’s low impedance headphones as these require relatively high currents and low voltages. This means ALL tube headphone amplifiers on low impedance headphones on LOW operating voltages is simply IMPOSSIBLE. One of the few hybrids that use tubes in the output stage I know of is the Musical Fidelity X-cans v1. This one has an op-amp for voltage amplification and 2 tube ‘halves’ as output devices (along with a solid state current source). For this to work overall feedback is ESSENTIAL. Also the output CURRENT capabilities of this design are rather small and thus it is really only suited to drive higher impedance headphones and is FAR less suited to drive LOW impedance headphones. This version was followed by other versions but the later ones all used solid state output devices (and extra gain stages) and a tube in the input stage and it was used in LTP (Long-Tailed Pair) configuration which ‘cancels’ harmonic distortion to a degree.

Enter the hybrid part… We have the tube for the voltage amplification part (without local nor overall feedback) and simply add semiconductors to deliver the needed CURRENT. These can be discrete components (transitors, MOSFETS’s for example) or op-amps with LOCAL feedback.

So hybrid amps work on relatively LOW power supply voltages, boast tube sound (added harmonics in the right ‘palette’) and can drive low impedance headphones as well as higher impedance ones.

Hybrid amplifier designs can range from VERY simple (but are usually not the best performers) to quite complex commercial designs. On e-bay there are plenty of offerings around ranging from cheap Chinese amps to more expensive ones. These cheaper Chinese products (Indeed, Bravo, Miridiy, Muse e.t.c.) all have certain things in common. They work reasonably well in the beginning, are cheap, are fun looking and do even sound good. BUT eventually (and this may even be within 1 year) will fail due to cheap components used and bad thermal management (they get too hot locally). They are highly modifiable though. Modification guides of the ‘Indeed G2’ and a-likes from e-bay can be found here. Shown below a picture of a heavily modified Indeed G2 amplifier.

I designed the schematics for a few Hybrid amplifiers and they can also be found here. A few designs are not commercially available, but some of them were designed for Garage 1217. The Project Sunrise and Horizon are all discrete designs and the Starlight and Ember have op-amps as an output stage. In the Starlight these can be ‘rolled’ just like tubes can. The most capable one, the Ember, also has an op-amp output stage but with a component that can deliver higher amounts of power. Other (quite similar) commercial designs are sold by CustomCables, Zigis, Nixiekits e.t.c.

a bit about tubes in hybrids.

There is two kinds of power (voltage x current) when we are taking about tubes.
The plate current & Voltage and the heater current & Voltage.
There is 3 voltages on a triode tube, grid voltage (very low, just a few Volts), plate Voltage (high Voltage) and heater Voltage.
The plate (or anode) current is determined by the plate voltage, the bias voltage on the grid (is NOT the same as the bias voltage of these amps) and the load provided by the CCS (constant current source)

heaters, filaments

The heater needs to heat the cathode so electrons can fly away from it easily.
In all the tubes used in most hybrids the filaments inside the tube sections are all 6V.
Also the 12V tubes (at least the 9-pin Noval tubes) all operate on 6V per heater.
In 6V tubes the heaters are in parallel (between pin 4 and 5) and in 12V tubes they are in series (between pin 4 and 5) making them 12V.
The fun thing about it is that for 12V tubes the heaters on one tube section is connected between 4 and 9 and from the other section between 5 and 9. So when 4 and 5 are connected together and 6V is connected between 4-5 and 9 that 12V tube runs on 6V.
In G1217 amps you can switch between these heater arrangements using a jumper. The Project Ember does it automatically.

Usually we speak of 6V tubes but in reality it is 6.3V (or 12.6V) . The currents at that voltage can vary depending on the type of tube between say 300mA and 1A in general.

The Pxxx and 7xxx tubes run on 300mA currents and are designed to be set in series with each other.
When there is 300mA flowing through it there should be around 7V over it (7.6V actually).

Light bulbs work longer when used on a dimmer but these things need to emit light and to get as much as possible light from it they sort of design the filament a bit too thin. More light (and heat) but shorter life span.
In electron tubes there are similar filaments but all they need to do is heat the cathode and don’t have to be stressed nearly as much as light bulbs. As a side effect they also glow. This can range from orange glow to almost white light.

There are basically 2 types of filaments, those that take a relative long time to heat up and thus also to start ‘working’.
Those are the tubes that last longest (filament wise).
When switched on it takes a while before you see them glow.

Then there are tubes that, when switched on, give a bright ‘flash’ and then glow softer quickly.
Because of this (intentional) behaviour  the tube is much faster at its operating temperature.
The filament, however,  draws much higher currents on start up.
This sometimes causes problems with the heater supply in some G1217 amps.
The heater voltage regulator goes in ‘protection’ and switches off and on again after a short moment to try again.
Some tubes keep ‘flashing’ because the flash is too short to really heat up (when the heater gets warmer the current draw gets less) and some heaters do get a chance to warm up and ‘flash’ a few times and then stay on.

These ‘quick heater’ tubes are more likely to blow a heater than the slow-on types, certainly when switched on and off very often during the day.

The output relay (that mutes the output during start-up) timing in G1217 amps  is set slightly shorter than the time it takes for normal heaters to reach their operating temperature.
‘Fast heater’ tubes thus are of little benefit, when a quick ‘on’ action is required as the amp may work already but the sound is muted anyway.

You can run these tubes at a lower voltage and expect a longer life time (of the filament) BUT in reality it is never the heater the goes dead as it isn’t designed like a light bulb. Tube filaments are basically under-dimensioned on purpose for longevity.
Tube filaments are thus designed to have a long life (longer than the cathode itself).
For this reason lowering heater voltages doesn’t do anything for the lifespan of the tube itself as it is NOT determined by how long the heater will operate during its life span.

In all G1217 amps without the supercharger you will find the heater voltage is always slightly below 6.3V (somewhere between 6.1 and 6.2V depending on the heater current) but this is not because of longevity but has to do with the used heater voltage regulator which puts out 6.5V. This is lowered with a Schottky diode to the mentioned voltages.
7V tubes will thus always run on undervoltage (on their heaters) BUT the tubes itself will nNOT last longer.

6.3V tubes should not be run at voltages below 6V and above 6.6V. The tolerance of heater voltages is about +/- 0.3Vfor most tubes so with 6.1 to 6.2 volt one is already on the low side of voltages ensuring longevity of the heater.
But as mentioned a slightly lower heater Voltage says NOTHING about the longevity of the tube itself which, in my experience, is never determined by the heater giving up its ghost.
In all the years I have worked with tubes I never had to replace a tube because the heater was dead.
This doesn’t mean it cannot happen and some people may have needed to replace a tube because of a broken heater though.
In G1217 amps (without Supercharger) all 6V and 12V tubes thus run at around 95% of their nominal voltage.

7xxx and Pxxx tubes have a nominal voltage of 7.6V actually tolerance of around +/- 0.6V so these run optimally between 7V and 8.2V.
Running of 6.2V is thus too low (runs on 80% in that case) and less electrons will leave the cathode.
Not a big issue for low voltage hybrids though as the tubes have very low plate currents anyway BUT it won’t live longer.

Longevity of a tube is predominantly determined by the ability of the cathodes to emit electrons. This becomes slowly less over time.
Those remembering old CRT TV’s and perhaps the ‘green indicators’ on tube radios know that the amount of light coming from them slowly dies with age. This is because less electrons leave the cathode and this is what determines the life span of the tube.
It is impossible to predict how much ‘life’ there still is in a tube, vintage, NOS (New Old Stock) or used nor for new tubes.
Even a measurement is still a point in time and ONLY says something about the emission at the point of measurement.
It could be the same after a year or usage or be almost dead.
When running tube heaters on too low Voltages the tubes actually are emitting less electrons to start with and also gets less over time.
Running heaters on under-voltages thus doesn’t prolong longevity but actually shortens it as the point where too little electrons are generated is coming sooner !
This is why tube manufacturers have specified a minimal voltage (to ensure the cathode is hot enough) and a maximum (to avoid the heater from burning out).

Voltage levels

First is the grid voltage. This is really only important for all tube amps running on plate voltages > 80V and on higher plate currents.
In low voltage hybrid amps the grid voltage is very low (between 0V and 1V depending on tube type) and is ‘self regulating’ in all these types of amps and cannot be set manually.
I just mentioned it because grid voltages in all tube amps determine the bias point and are of importance there.
In hybrids the bias voltage point is determined by the current source setting in the anode (plate) circuit and needs to be around half the power supply voltage to ensure maximum output voltage swing.
In high voltage designs the voltage swing is not (less of) a problem.

Then there is the plate voltage and its maximum plate voltage.
Most people think tubes are supposed to work on voltage well above 100V and wouldn’t even operate on voltages below that.
All nonsense… they do work more linear over a larger voltage range though and thus has less distortion.
That distortion, however, is just what these hybrid amps (and their sound) is all about unless they have lots of overall feedback (think Musical Fidelity amps)
It is true that higher anode voltages make the tube perform better as they can attract the emitted electrons from the cathode a lot faster and thus higher bandwidths can be reached. For audio the bandwidth doesn’t have to be much higher than 50kHz though.

Some tubes are even specifically designed to run of car battery voltages, like the ECC86/6GM8.
These have max plate voltages of 30V and thus cannot be used in the Ember and Horizon for instance (having a48V power supply).

The G1217 amps all work on either 24V (Starlight, Solstice, Sunrise) or 48V (Ember, Horizon).
Higher voltages aren’t really needed aand are safe for open frame designs as you can touch voltages/components.
For indoor situations (not bathroom) is 60Vdc the maximum allowed voltage.
These hybrid amps are designed around are the 6DJ8/ECC88 type (and its many equivalents).
Their plate voltages are rated at 90V max (and run at 22V or 46V max depending on power supply voltages).
This certainly isn’t too low as some think.
These tubes can still even amplify on 12V power supplies (5V plate voltage nominal) though not really optimally.
ECC88/6DJ8 types of tubes thus run comfortably below their maximum ratings and still work very well, even on 24V power supply voltages

Tubes like the ECC83/12AX7 are intended for 250V power supply volatges and indeed perform far from optimal operating points and may be quite noisy, but to some, the sound (being rolled-off well within the audible band in this case) is pleasant.

6SN7 tubes are designed for plate voltages of 450V and thus not really suited.
Given the fact that its sound is liked by many and even amps running on 24V do sound excellent shows that even 450V tubes can still do their job at  24Vpower supply voltages.
Not only the plate voltages is just 3% of its maximum voltage also the plate current is about 7% of its maximum so the total plate ‘power’ is even below 1% of what it is supposed to work on.

Schematics of the Garage 1217 amplifiers are freely available but may NOT be used for commercial designs. We have no problem with the amplifiers being built by DIY’ers in the hobby area though but don’t like commercial enterprises to shamelessly copy the designs. Kits and build amplifiers are available. These designs can be discussed in our FORUM.

Reviews of the EMBER can be found here: http://diyah.boards.net/thread/143/project-ember

Project Ember…. This amplifier has some really nifty features that are not found in other designs. Below an under the bonnet picture:

The Ember is an off-spin from an earlier design that was made from an experiment with a prototype Sunrise board where Auto-bias and auto heater select and the ‘op-amp’ output stage were added, pictures below.

home
back to amplifiers