6CM5 / EL36: Applications in Audio



Triode Mode

Application to Audio, by John Hunter, August 2002, March 2003

My Experiences with Single-Ended EL36 Output Stages, by Graham Dicker, March 2003, October 2004

A PP Amp using EL36's and the Altronics-M1115 transformer, by Grant Wills, May 2003

Comparison to EL34

Two EL36 Designations

· “European” EL36/6CM5, XL36/13CM5, PL36/25E5, also CV2940

· “Mullard” EL6, EL12, EL36, EL54

Related EL36 Family Members

· EL360 · E130L · E235L · 6π31C

Existing Circuit Applications

· Philips EL6405, EL6415, EL6425

· Melody EL36S

Substituting an EL36 for an EL34

· EL36 – EL34 Comparison Chart



The EL36 pentode was originally intended for use both as a “line output”, (or “horizontal deflection”) valve in television receivers, and as a Class B audio power amplifier. It found very widespread use, (in Australia and New Zealand at very least) in the former application, but is found only extremely rarely in (almost exclusively  Philips) audio applications.

Ratings for use in Class B push-pull are readily available from Philips datasheets available on the internet. An anode voltage up to 300V, screen voltage of 150V, and a relatively low anode to anode load of 3.5kOhms are suggested for a pair of EL36’s, for an output power of 44.5W, albeit at 7.2% distortion! The actual rated anode dissipation is relatively low, at 12W, but the EL36 can sustain moderately high anode currents – to around 100mA – at relatively low anode voltages.

EL36 DataSheets 1 (ex Frank Philipse)

EL36 DataSheets 2 (ex Frank Philipse)

Canberra based design engineer Patrick Turner is one who has investigated audio applications for the 6CM5. He offers a 6CM5/EL36 based 2 x 22 Watt Stereo Class AB1 triode amplifier, which he describes as follows. You can visit his website article and pic of this amplifier.

"This triode amplifier was built to be able to lend out to people who wished to discover what all the fuss was about tube sound, while their solid-state amp's are being repaired. They are always sad to have to return it when their solid state amp is fixed. The amp is by no means a state-of-the-art high-end unit in terms of style and finish, but it is amazing how much better it is than most other more powerful solid state amp's. In each channel, the 6CM5 TV line output output valves are connected as triodes and are driven by a long-tail pair using a 6CG7. The input tube is a paralleled 12AT7. The output transformers were made in 1960 in Sydney and the power transformer and chassis is a new one by T.A. The pale coloured aluminium central box is a channel with solid-state components for regulating the anode supply instead of using a choke. The sound is clean, accurate, detailed, emotionally involving, with the typical smooth treble and precise bass with a warm midrange. There is a perforated steel cover which fastens down over the top of the amp to prevent anyone coming into contact with high voltage terminals".

6CM5, Application to Audio,
John Hunter, August 2002

John Hunter hails from the picturesque Blue Mountains, to the west of Sydney, NSW. I am most grateful for John's contribution of time, experience and knowledge as evidenced below. John contacted me in August 2002 - it quickly became evident that he had a much much greater knowledge of this interesting valve than I, and therefore I am most appreciative of his willingness to allow me to repeat his advice to me here.

"I've always been into valve technology. For someone who grew up in the 1970's, it's ironic that I learnt about valves before solid state. I'm into collecting, restoring and building monochrome valve TV sets. So the 6CM5/EL36 is a valve I'm definitely familiar with. As I build a lot of valve projects requiring an audio stage the 6CM5 has been of interest, especially as I have a huge quantity of them!"

John Hunter

"It is quite amusing to get boxes full of 200+ TV valves for free or under $10, at auctions and swap-meets, because no one thinks they are of any use except for the few eccentrics like me into old televisions. Most of the valve stuff I've built has used TV valves whether or not it's TV, audio, RF or test gear".

"As an example I designed a TRF receiver that is electrically 1920's but with three 6BX6's and a 21A6 (another line output valve). I've built numerous other AM and FM radios using only TV valves rather than comparatively rare and expensive GT and G style radio valves, which I'd rather keep for equipment made in that era. There must be at least 1000 6BL8's in my valve collection, and what a useful valve that is, for example. You get a triode and pentode that work anywhere from DC to VHF in this small package. It's usefulness goes way beyond the mixer oscillator use mentioned in the data books. The other thing with TV valves, particularly the European ones, is they are good performers with low voltage anode and screen. This is an advantage when you want to build up a power supply with modern transformers. For someone who wants to get into valve technology these days, TV valves are ideal!"

Regarding the 6CM5/EL36, John writes;

Outside Australia the 6CM5 valve is virtually unknown. US, UK and European data books often do not show it. However, the series heater version, PL36, is as common in Europe/UK as the 6CM5 is here. As a point of interest, one of my colleagues did the curves for the 6CM5 at AWA when they started making the valve. The Telefunken PL36 (the RMA number for PL36 is 25E5), is what the local AWV production was technically compared with, that company being the European AWA affiliate).

It is important to remember that series heater TV sets were standard in the UK and Europe, whereas parallel heater designs were pretty much unique to Australia and about half of US sets. (New Zealand followed the series heater designs to a much greater degree than here in Australia). The very few parallel heater TV designs in the UK used the 6CD6 or 6DQ6. (I'm talking 1950's~1960's here). In the US, the standard line output valve of the era was the 6BQ6 or it's higher rated version, the 6DQ6. The 6CM5 is not a plug in equivalent, although it might have the same or similar connections.

What is unique about the 6CM5 is the low anode voltage. Again, the series heater TV design was responsible for this. When the B+ comes only from the 220Vac European power mains, the filtered DC supply is only about 200Vdc, (and maybe even lower for DC mains supply), for the whole TV set. So, any line output valve is going to need a very small voltage drop from anode to cathode when it is turned hard on. When you've got 250Vdc of B+ from a transformer supply or a voltage doubler running off 120Vac, inefficient valves like the 6DQ6 are okay.

This means that for Class A audio, the 6CM5 is one of the worst valves you could use. Class C is what it is most efficient at, and also Class B (I have one of the Philips public-address amplifiers with 6CM5's operating in pretend Class B). Not to be deterred, I have experimented just to see what I could get out of a single-ended Class A output stage. (Actually I've got a car radio project coming up . . .) 

I tried three topologies: conventional pentode with drive to the control grid, pentode with drive to the screen grid, and triode. The conventional pentode connection was the worst! I couldn't really get good linearity with decent output power. Screen grid drive was a lot better, even though about 42Vpp was required to drive it.

Click on the image to enlarge it.

John's Philips  push-pull amp, intended for public address application,  model EV4435. Click on the image to enlarge and view the circuit properly.

The best was triode connected. Although sensitivity was low, again requiring 42Vpp, the output was the greatest and distortion the lowest. Maximum output was about 2.25W into 5kOhms with a B+ of about 260Vdc at 50mA. These tests were done with a regulated valve power supply, (a BWD model 215). The results aren't really surprising, and after looking at the STC manufacturer data on the 6CD6 when used as a single-ended Class A output stage, it just confirms what I suspected. (For triode connection the output is 1.5W and for pentode it's 4.7W - as a point of interest, they also mention that 6CD6's used for audio are liable to have a large spread in their characteristics and they're basically aren't recommended, except as a point of interest for low anode voltage supply).

So, yes it can be done, but it's terribly inefficient when you consider that the heater current is 1.2A (6.3V @ 1.2A = 7.5W heater power alone, for 2.25W useful audio output!). A 6AQ5 or 6V6 will give nearly double the output power for less than half the heater current! I have seen a $3000 amplifier made in Western Australia using a colour TV line output valve (PL509/40KG6). It only gave out 7W!

I intend to go ahead and use the 6CM5 in single-ended Class A applications but only where heater current isn't a problem and a couple of Watts is acceptable output. But there are better valves for the job, and it's only because of my quantity of 6CM5's, (and wanting to be different), that I'll use it. Having said all that, the low anode voltage is of interest, as it may permit a valve amplifier to operate on modern low voltage power supplies. This I have yet to play around with.

I guess I should emphasise that my disappointment with this valve was with its application in single-ended mode. For push-pull audio, transmitter output stages, and as a B+ regulator series pass valve, I think it has a lot of potential.

John Hunter, 26 August, 2002

Several months later, John emailed me further, to announce that some significant headway had been made in this regard. John writes;

"I have, with much perseverance, got the 6CM5 to work as a single ended Class A audio amplifier. The operating conditions are most bizarre (no bias and 25V for the screen grid), but it really does work, giving about 4W output. This is with 240V on the anode and a 5kOhm load, so it compares favourably with 'proper' audio valves like the 6AQ5 and 6V6. My circuit yields somewhat more output than say a 6BM8. The original bias setup was sensitive and critical, as the non-linearity of this valve is unbelievable. However, the DC stabilisation circuit I have subsequently used seems to have solved that problem.

Driving via the control grid (g1) is not the way to use this valve for Class A audio, as much as I have tried. You can only get about half the output power that you can get from screen grid (g2) drive. Remember we a forcing a switching valve to work as a linear amplifier. Triode mode is the best way to use this valve when conventional g1 drive is used. It is no surprise that measures need to be taken to avoid self-oscillation issues, with a transconductance (gm) of 14mA/V.

My circuit provides about the theoretical maximum audio power from a 250V anode supply and 5kOhm load. Apart from the horrendous heater power, it probably works as well as a 6V6. I say 'probably' as I haven't done any actual distortion tests but only visual inspection of distortion as seen on the CRO.

Click on the image to enlarge it.

At first it seemed to be a lost cause, trying to use this valve as an audio amplifier. The valve-data does include Class A operation, but that is for a 100V supply at 100mA. This and the required 1000 Ohm speaker transformer don't tend to match in well with other valve circuitry and parts availability (i.e. a 5kOhm transformer is easier to get than 1kOhm). I could see that the 6CM5 might be a good output valve in a 32V amplifier or in other low B+ applications.

The original aim of this design was to use a conventional 5kOhm transformer and 250V B+ supply. Alas, at this voltage the 6CM5 is very non linear! Designed for Class C work, this is hardly surprising. Looking at the data, we see the anode dissipation is about 10W max (yes, it can be slightly higher depending on screen dissipation, but I prefer to be conservative).

This means about 40mA anode current at 250V which is similar to the likes of 6M5, 6V6, 42, etc. The rule of thumb is that the power output from such an audio stage will be about 40% of the anode dissipation; i.e. 4W. In reality the actual power to the voice coil will be less due to transformer losses. Additionally, the method of bias and voltage regulation also bears some importance for maximum power output. This is why in a small mantle set or a car radio, a 6V6 may only give out 2W before distortion becomes evident.

As I've mentioned before, initial attempts at getting a 6CM5 working in the conventional triode or pentode circuits only provided about 2W. Screen grid drive seemed to be a better option, going on the theory that the gain of the valve would be less, and possibly more linear. In practice it did seem to work better and a bit over 2W was available. The method of drive used was simply RC coupling with the B+ fed into the screen via a resistor and the signal coupled in via an isolating capacitor.

My latest and successful experiments came from the theory that RC coupling was probably not such a clever idea for screen grid drive. For ordinary control grid drive it's fine as the grid is of infinite impedance. However, screen grids draw current, which does not suit the relatively high impedance of RC coupling. Either transformer or cathode follower drive would have to be used. The practicalities of using an audio transformer meant that a cathode follower stage would be preferred, and so a test circuit was tried.

The results were looking promising but still less than the requisite 4W was forthcoming. At this point the 6CM5 had the screen grid at about 150V and the grid bias set to draw about 45mA anode current. What did become apparent was the less bias the 6CM5 had, the more power it would produce without distortion. So, why not get rid of the grid bias and just control the anode current with the screen grid voltage?

That's where the breakthrough came! With no bias and 25V on the screen, the anode was drawing 45mA at 240V. I couldn't believe it when I was getting 3.8W before clipping into the 15 Ohm load resistor, (and at least 4W into the transformer primary).

Looking at the circuit now, to start with the output stage, the supply used was 265V @ 50mA. The anode voltage should be 240V when the 6CM5 anode is drawing 45mA. Allowing for the DC across the speaker transformer, the total supply will be more than this (hence the 265V in my circuit). The ideal load impedance is 240/0.045 = 5.3kOhms, but obviously in the real world 5kOhms will do.

Note that the control-grid and cathode are connected together so there is no bias; the cathode resistor is used for a DC stabilisation circuit. The screen-grid voltage is set by the 6CG7 cathode follower; about 25V is what is required for correct 6CM5 anode current. Due to such a low cathode voltage, it is necessary to take the load resistor to a negative supply otherwise the waveform becomes clipped on the negative swing. The -30V supply is not critical at all; anything more negative than about -20V is adequate.

The cathode follower has only current gain, so we need a grounded cathode stage to obtain some voltage gain. This is the purpose of the other half of the 6CG7. The anode is DC coupled into the grid of the cathode follower by a voltage divider which sets the 25V on the cathode. However, such a divider would also reduce the AC component if it were not for the 0.47µF capacitor.

The reason for using DC coupling becomes apparent when we see that the grid of the 6CG7 voltage amplifier is connected to the 6CM5 cathode via an adjustable voltage divider. This is where the DC stabilisation comes in. If the 6CM5 should start to draw more current, the voltage across the 50 Ohm cathode resistor will rise, making the 6CG7 control grid more positive. This in turn makes the anode voltage decrease, thus reducing the DC on the cathode follower output and therefore reducing the 6CM5 screen voltage which in turn reduces anode current.

Notice that adjustment of DC feedback is achieved with a 100kOhm preset pot (potentiometer). The way this is connected means that if it should go open circuit (which is the normal method of failure of a preset pot), the 6CM5 will be throttled back. The pot is simply adjusted for a total current consumption of 50mA, or 45mA for the 6CM5 anode. The bypass across the preset pot is optional; it was included to prevent feedback, but in practice it doesn't have much effect. If omitted there will be slight negative feedback from the 6CM5 cathode. I probably won't bother including it in future amplifiers.

Prior to the 6CG7 stage is a simple voltage amplifier using a 6AV6 triode. This increases the sensitivity of the amplifier to 56mVrms or without the cathode bypass on the first 6CG7 triode, to 100mVrms. Negative feedback is provided in the shunt form by the 470kOhm resistor, with the 0.1µF used for DC isolation. Feedback isn't applied to the 6AV6 as it's a triode handling low voltage and in my opinion it isn't necessary. That's a matter of personal preference; in fact I probably won't include any negative feedback for my first application of this circuit; that of a car radio for my Model T Ford. With the road and engine noise in an open car, Hi-Fi is scarcely worth trying to attain!

There are points to note regarding construction. Firstly, the 6CM5 cathode resistors could obviously be replaced with a single 47 or 56 Ohm 1/2W resistor. Secondly, if the component or valve tolerances are so far out that there's not enough adjustment in the 100kOhm preset, the 150kOhm resistor can be reduced if the current can't be reduced, or increased if the 6CM5 can't draw the requisite 45mA.

The components marked * are for decoupling. I didn't need them in my prototype as a regulated power supply was being used (a BWD 215) with very low output impedance. The values aren't critical, with the resistor being from about 2.2kOhm to 10kOhm and the capacitor being 8µF or more.

The 6CM5 could obviously be replaced with a 25E5 if it's convenient to obtain 25V @ 300mA for it's heater. The 6CG7 is the modern equivalent for 6SN7 so this valve could be used along with others in this family such as 12AU7 and 12BH7.

The 6AV6 is really half a 12AX7, but just about anything could be used for this stage. More gain could be obtained by bypassing the 6AV6 cathode resistor. And finally, keep in mind that the heater current for this amplifier is 2.1A, which is one of the main disadvantages of using the 6CM5 in the first place.

John Hunter, February/March, 2003


Recently, I've further been privileged to have highly esteemed South Australian designer Graham Dicker offer the following most valuable and extensive research report. You can see links and background for Graham on my main page, in the section relating to the "ETI480" amplifier, and get a taste of his current output as a writer by visiting AudioXpress. Says Graham;

Graham Dicker

"I hate to throw out anything, and EL36's are in that category. I suspect very strongly that there are some hidden aspects to the EL36 design that may be still to be uncovered. I do not think that the full stories surrounding both this valve and the 6DQ6A, have come to light. So it would be good to hear from some engineers at AWV and Philips to find out more!"

My Experiences with Single-Ended EL36 Output Stages
Graham Dicker, March 2003

The EL36 is a plentiful output valve in Australia as discussed by many others. These valves were mostly manufactured at the Philips Hendon works in South Australia or at the North Ryde AWV works in N.S.W. Many of these valves made in Australia were re-badged for other vendors. The Philips cathodes were usually imported from Philips Holland in lots of no more than 1000 at a time to reduce the risk of radiation, while I have been told the AWV cathodes may have come from RCA in the USA.

Conventional push-pull EL36 implementations are covered in detail elsewhere, so my comments in this article below are related to using the valve in a single-ended output stage. The EL36 has however been applied in quite unusual fashion from time to time. An ex-Racal, UK, and later AWA engineer here in Australia, commented to me that AWA did some application experiments in the early 1960's using the EL36 as a transmitter modulator. A pair of valves were used in Class AB1 push-pull, with most of the output coupled to a modulation transformer and some driving another 4 in Class C (yes, you read it correctly). The total output power was in the order of 200 Watts RMS. When the driver valves lost emission they just swapped them for a pair of the Class C valves.

For single-ended application, the EL36 linearity is problematic to say the least, and available output power is somewhat less than a 6BM8. I have found other characteristics as well that do not lend the valve to audio applications normally. One of these is the propensity for the valve to self oscillate, which has been traced in the case of my prototypes to the screen grid. In a pentode output circuit the screen can be adequately bypassed, but in triode-connected or screen-drive circuits this is not possible. I found that by winding a few turns on a toroid core an RFC (radio frequency choke) can be made and placed in series with the screen, close to the socket pin. This makes the valve stable under all kinds of loads, and even with output transformers with large leakage inductance.

The second characteristic that I found problematic was that when using a fixed bias supply the grid voltage was critical within a few Volts, from trying to set an anode current of 50mA to being completely cut-off. A self-bias connection also produced variable results at best.

Click on the image to enlarge it.
An early direct coupled test circuit. Click on the image to enlarge.

When using a low supply voltage of 150 Volts the screen can be tied to the anode supply, thus not requiring its own regulated supply voltage. Under low supply voltages you need more anode current and a lower anode load, and this increases the THD to around 20%. No amount of feedback can fix this. In this configuration with low Volts and low impedance anode loads you are fighting the dynamic internal anode resistance of the valve, making it almost impossible to match an anode load with reasonable output power and efficiency.

For a higher anode voltage, a 22kOhm 2 Watt series resistor and two 75V Zeners in series to ground form a nice regulated screen supply. Best output power was achieved in the triode-connected mode, using a high anode supply voltage of 450 Volts and a 10kOhm anode load. With a transformer driven grid the output was 10 Volts RMS across an 8 Ohm load or about 12 Watts RMS. The drive voltage was 24Vpp which does drive the grid into grid current. If you also do the sums on anode dissipation you will find that my experiment was conducted with over double the normal anode dissipation.

The anode dissipation limit of this valve also brings some interesting questions to mind. Data obtained from different sources indicate different limits; from 10 to 18 Watts dissipation for the anode and 4 Watts for the screen. After having taken a few of these apart it appears that the Philips made ones have similar anode dimensions and cathode size as the 6CA7 (not the EL34). Looking at how much power you need to get a cherry red anode (around 40 Watts) all I can conclude is that the 12 Watt rating is conservative. This is confirmed after repairing many Philips public address amplifiers, in never ever having to replace an output valve.

I also tried screen drive, via transformer coupling, and also zero bias mode where you drive the screen and control grid via a 22kOhm series resistor at the same time. The results here were no better. After having thought about some of the problems for a while I next tried a low impedance direct driven grid, with fixed bias, the cathode earthed, and the valve in triode mode. This produced a pleasing low-power stage with insufficient drive and problems of setting the bias Q point for the EL36. The distortion was less objectionable but still around 14%.

My last method tried was to raise the HT to 310 Volts with a 5kOhm anode load, and run the EL36 in grounded grid mode. (I have done this many times before with other valves with great success; try an 807 and you will never use any other method again). This time I used a BD679 Darlington, which not only drives the EL36 but also provides self-regulating bias by way of the BC547 and the BD679 wired up as a constant current source. This not only works a treat but produces an extremely linear output stage with low distortion (below clipping) of around 1/4% and a nice sonically performing amplifier. The down side again is low output power in the order of 3 Watts RMS. Why use an EL36? Well, firstly, in this mode it sounds a lot better than a single-ended 6V6 or EL34, and better even than a single-ended 2A3, and secondly, its low cost. I built five of these on the same chassis for a home theatre amplifier for the five main channels; the sound is glorious and a lot easier to listen to than a solid-state amp!

Click on the image to enlarge it.
Grounded-grid mode with constant-current drive. Click on the image to enlarge it.

Click on the image to enlarge it.

Here is a further advance in the EL36 story, with another breakthrough in R&D! I have developed a completely new simple topology for the EL36 (I have also tested it with 6V6 and 6L6 valves as well), that I have never seen before. The circuit is a simple cathode follower using an easy to wind autotransformer. The transformer core was from a discarded microwave oven transformer.

The input transformer is from Altronics and is a nice low cost mu-metal screened one. The big difference is that the EL36 is in what is really a grounded cathode design, except that the load is in the cathode with all of the advantages inherent in locating it there. The input drive is still effectively from grid to cathode. This makes a nice simple power amplifier. I do however personally prefer the sound from the 6V6 and the higher output power from the 6L6. An 807 would be worthwhile trying in the circuit here as well.

Note that if you build a stereo version then use a separate filament supply for each output valve and do not ground the supply. This helps ensure that you do not exceed the valve's filament-cathode potential rating limits.


Grounded Grid

Following a number of reader enquiries, Graham noted that there seemed to be very little published on the internet on the subject of grounded grid amplifiers in audio. He comments "I have a particular love of the 807 as an audio valve, so here is my simple grounded grid amplifier. It is equally adaptable for the EL36."

Graham Dicker's Grounded-Grid amplifier; click here to enlarge the circuit.

The EL36-6CM5 Story Continues!
Graham Dicker, October 2004

It has been a while since I was last able to find some time to play with the EL36. Intrigued by the large number of people who are now turning to low power single ended amplifiers for high quality audio reproduction, I discovered there was a definite trend in customers and others asking about SEPP (Single Ended Push Pull) designs. The big advantage here, like other SE (Single Ended) designs that use a shunt-fed output transformer (OPT), is that there is no DC component through the transformer.

In the case of the normal SE design this is handled by the anode choke, making it a lot easier to build and design. For the home builder there are still plenty of 50+ Henry filter chokes on the second hand market that can be used (and are cheap). These days people tend not to use filter chokes in power supplies as there are plenty of high capacity electrolytics available at reasonable prices. This has made the filter choke redundant, and many of these have been thrown out.

The Public-Address field has helped us enormously, with good well made 100-Volt line to voice coil transformers available at low cost ($30-$40 each), from Altronics, Jaycar and the like. These are designed to work with no DC component in the windings and are ideal for these applications. I have tried unpacking these transformers and adding an air gap; this is okay if you want a small 2-3 Watt output transformer, but as the primary inductance falls from around 50 Henries to around 10 Henries (at best when tested and optimised), it will effect the low frequency response markedly. If however you wish to build a single-ended guitar amplifier, they will work up to 10 Watts quite well for lead and rhythm use.

Back to the Project

The 6CM5-EL36 is a good, low cost choice for an SEPP (single-ended push-pull) amplifier as it has a low average anode voltage and is capable of passing around 200mA of anode current. The big problem with these amplifiers is low efficiency due to fighting the internal anode resistance of the valve (Ra). In this case you have an overall dissipation of around 40 Watts, and only 2 Watts of really nice, usable output. More valves in parallel will give more output power, but you will need to take out shares in the local electricity supply utility. Regardless, the circuit in Figure 1 yields a nice, high quality, low cost (let's make that "really, really cheap"), amplifier.

Figure 1

How it Works

Audio from the balanced line input transformer provides around a 10:1 step-up voltage to drive the grid of the lower valve (V2). This provides plenty of voltage gain from a normal +4dBm or +8dBm source. Most CD players and modern hi-fi music sources these days have op-amp output stages which are low impedance and can also drive the transformer in unbalanced mode.

Figure 2

The bias is set by the 68 Ohm cathode resistors to around 120mA. This gives an anode dissipation of around 18 Watts per valve. Drive for the upper valve (V1) is from the anode of the first. Grid and screen grid stopper resistors are a must if you want the EL36-6CM5 to be stable and not go into parasitic oscillations.

I have spent a lot of time on this point, and can't argue enough to add these to every valve amplifier you make. Very little is published on this, so take heed! The output is capacitively coupled through the 100µF output capacitor to the output transformer.

A photo of the prototype is shown above for an idea of layout. The 68 Ohm resistors each dissipate about 1W normally, so need to be say 2W types. The 100µF output capacitor needs to be a 200V type as a minimum, 250~300V rated preferably.

An alternate drive scheme for V1 is shown in Figure 2. Here, the drive for V1 is taken from the cathode rather than the anode of V2, and so needs to be phase-inverted to provide the correct drive phase polarity for V1. Q1 provides this function. An MPSA42 was selected, as it has the requisite relatively high Vceo required. R8 needs to be a ˝W type preferably.

A Push-Pull Amplifier,
featuring the 6CM5/EL36 and the Altronics-M1115 transformer
Grant Wills, May 2003.

Grant Wills wrote to me in early 2003; "My new amplifier, based on M1115 transformers sourced from Altonics, works a treat! The final amp topology used the front end of the Leak Stereo 20 amp, as I had heaps of twin-triode valves in the junk box. The output stage initially used 6L6's with a B+ supply of 300V, as again I had some on hand. But after quite a bit of experimentation with other output valves I used 6CM5's, based on the large amount of information now available on these dime-a-dozen valves. As is well understood, they are an odd valve to attempt to apply to audio applications, and I had lots of unsuccessful attempts to use them."

Click here to read Grant's most informative article!

Grant Wills

Comparison to EL34

For comparison, it’s instructive to consider the EL34 manufacturer’s data for Class B application. Using an anode voltage of 775V, a screen voltage just under 400V, and an anode to anode load of 11kOhms, a pair of EL34’s are rated for 100W at 5% THD. Note that far more practical and realistic is application in Class AB, rather than Class B, where about one third of these power levels can be expected. But the comparison between the completely different nature of the two valves is apparent; the EL36 is about half the power rating of an EL34, but this is achieved using directly comparable currents and substantially less voltage; the ra of the EL36 is relatively low. The comparison chart lists the parameters of both valves in Class A and Class B applications.

Amongst the reasons the EL36 is not commonly seen in audio applications is the prevalence of extremely good and cheep 6V6’s, 6L6’s and EL34’s, all of which are still in production. “The wonderfully cheap (and good!) Russian, Eastern European, and Chinese audio valves mean that we don't need to make do with old non-audio types unless we want to.” (1)


Two EL36 Designations

When considering the EL36 it is important at the outset to be aware that there exist two types with the same designation. The two types are similar in characteristics but not directly interchangeable. The one referred to in this text primarily, is that with a top-cap anode connection. To differentiate it, it is commonly referred to as the “European” type. Members of its family include the XL36 (13V, 600mA heater), the very common PL36 (25V, 300mA heater), and the directly equivalent US types for all three, the 6CM5, 13CM5 and 25E5, all of which have top-cap anode terminations. The CV2940 is an alternate designation for the EL36/6CM5.

PL36 DataSheets 1 (ex Frank Philipse)

PL36 DataSheets 2 (ex Frank Philipse)

The Mullard EL36 is distinguished by the absence of the top-cap; the anode is at pin 3 at the base. (Pin 3 in the top-cap version is designated “internal connection”). The characteristics of the EL6, EL12, EL36, EL54 family from which this alternate valve comes, are somewhat different, although similar to that of the “European” type. Notable is that the anode power dissipation is specified at 18W; considerably higher. In many respects the data for this valve indicates general similarity, but clearly the two are not immediately interchangeable.

EL6/12/36/54 DataSheets (i), (ii), (iii) and (iv) (ex Bernhard Wybranski, per Duncan’s Amp Pages)


Related EL36 Family Members

Variants of the EL36/ 6CM5 include the EL236, and the EL360, the later reportedly originally intended for regulator applications. “The EL360 is also a member of this family and probably a little better than the EL36 as an audio valve. Ratings are similar; same top cap and pin out. Octal base.” (2) “EL36 and it's cousins PL36, EL360, E235L, are pretty nifty valves. Lots of power at low anode voltages. Their big brother E130L is pretty slick, too, with 300V Va, 150V Vg2, a pair will put out 60W across 1600 Ohms anode-to-anode.” “EL36, EL35 [a 6L6 variant rather than an EL36 variant], or E235L are unusual, but ass kickers! They'll put out 30W~44W a pair with only 300V on the anode”. (3)

EL236 DataSheet (ex Bernhard Wybranski, per Duncan’s Amp Pages)

6π31C Data ex Tubes.RU

• 6π31S (a.k.a. = 6P31C, 6p31C, 6pi31C) = 6n31C (Analogue of EL36)

General: Beam tetrode, used in output stages of line scanning of TV with 1100 deflection. Envelope: glass. Mass: 45g.

Lead diagram

General characteristics: Type 6n31C

Limited operating values: Type 6n31C

Anode-grid and anode curves


Triode Mode

Although a plot of EL36 curves in triode mode appeared reasonably linear, the EL36 was found to perform relatively poorly in a simple single-ended triode-strapped configuration, exhibiting very obviously poor linearity. Such is not uncommon for most power pentodes used this way, apparently. However a test-bed variant of Bob Danielak’s EL509/6KG6 circuit using direct coupling and screen drive exhibited very very clean output waveforms indeed; quite the case of the ugly duckling and the beautiful princess! Such performance merits further investigation.

“I was fooling around with some old EL36's, and hooked one up in enhanced triode mode, (screen driven), and the sound blew me away. It was so totally different from my EL84 amp, (comparing mono to mono). And I was using the EL84 amp as the EL36-driver! The sound was endlessly more transparent/brilliant. Can't be the brand of the valves; ECC83, EL84 and EL36 were all Philips.”

"I triode-strapped three EL36/6CM5 NOS samples, in the middle of a rats nest of some power supplies and a few DVM's to establish detailed anode curves for this mode of operation. Here is the resulting set of curves. They show data derived and averaged from three Siemens, Philips and RFT EL36 or PL36 samples."

"Using a 100 Ohm resistor between anode and screen, there were no signs of excessive stress on the screen, even at the fairly high voltages used to plot these curves. The samples showed no immediate stress signs (not even a faint dark anode-glow at complete room darkness), at a combined Pd (anode plus screen dissipation) of 15 Watts. So, that's why I set the Pd line to 15 Watts on this sheet."

Tom Schlangen, Germany, 12/04

“I'm not sure if I actually prefer the 'new' sound of the EL36's. Listening in mono, the instruments didn't come from the speaker, but rather from somewhere around it. In stereo, it may work out great or it may result in unlocalisable positions. I'm waiting for the transformers to try that. Also, the sound was very bright. Could become fatiguing. Furthermore, this is my first experiment with enhanced triode mode. It seems to be working for the EL36, maybe EL84's would like it too. I don't know. Enhanced triode mode is harder to drive, but also much more linear than conventional triode mode. I came to try it with the EL36 since I had some lying around and they look like the baby brother of the PL519/EL519, which a friend of mine is experimenting with. I wouldn't say right away that the EL36 is better than the EL84. Especially since the EL84's were used as the drivers for the EL36, and so must have had their part in the signal shaping. (4)


Existing Circuit Applications

“A lot of Phillips PA amp's sold here in Australia used EL36's, some with eight valves in push-pull parallel.” (5) “Philips put out a couple of PA amp's based on the EL36. They were of unusual construction and very compact. The EL6425 was a 70Wrms amp using four EL36's in push-pull parallel. The EL6415 was a 35Wrms amp using two EL36's in push-pull. Not HiFi but great robust workhorses that ran for decades without attention. If you don't mind anode caps, the EL36 is a super cheap valve worthy of attention. Philips got around the height problem by mounting them horizontally with generous ventilation.” (6) “My powered mixer; a Phillips EL6425, has four EL36 outputs.” (7)

“The Philips EL6425 amplifier is the one in the nutty typewriter case with sliders up front. There was also a 19" rack version, the EL6405. The primary impedance of the output transformers would be something like 3.4kOhms for the push-pull variant. You can use these with a pair of EL34's in pentode, or indeed try four in push-pull parallel triode strapped. There was also a variant with 4 x EL36 push-pull parallel with Ra-a of c. 1.5kohms. Puts out about 70W, according to Philips. The transformers are mediocre at best, though. Lots of secondary taps for various impedances, including 70V and 100V line driving windings. Also, separate feedback windings. Be aware of the fact, that the power transformer is wired rather oddly in these amp's. This is done to get 150V for the screens, a voltage which, under no circumstances, may be exceeded. Also, you can't simply plug in EL34's, this won't work.” (8)

“The EL36, the sweep valve with top-anode type, makes a fine audio amplifier. A pair in push-pull makes great sound in my guitar amp! Philips used them in PA amplifiers as well. My second amp, I still have. It's a Maton Powerpack Studio amp. Really nice reverb, 2 x 12 speakers, uses two EL36 output valves, (about 40W), combo style, but not open back - the speakers are fully enclosed.” (9)

“I discovered some interesting articles in an old Wireless World, (February 1957) - OTL theory only and Czech Amate'rske' Radio: 6/1958, 11/1960 and 3/1961 - OTL amplifiers building and start-up. Very interesting amplifiers with little distortion and power of 10W, using EL86, to 30W using EL36. High impedance loudspeakers were an engineering problem, and the authors recommended 100V/4 ohm or 8 ohm lines. Peculiar but interesting single-ended/push-pull amplifiers .” (10)

While trading, Melody Valve HiFi, of Melbourne, Australia offered an EL36 based integrated stereo amplifier. The valve compliment for each channel is 4 x EL36, 1 x 12AX7, 1 x 6922 and 1 x 12AU7. 60W/8 ohm, THD 1%, 20Hz~30kHz +/-0.5dB and an SNR of 90dB was claimed. The price was around A$2000.

“Philips made a couple of rather nice looking PA amp’s using these valves. The EL6415 used a pair to produce 35W in Class B (!) and the EL6425 had 4 valves in push-pull parallel to produce 70W. Interesting they were, HiFi they weren't!” (6)

“There was also an OTL amp made for 600Ω speakers sold with them by Philips. I wanted to try them in a tiny regulated power supply, and I just measured the triode characteristics of one specimen. They don't look bad at all! The μ is about 6, and ra is 450Ω. This would do some nice things in single-ended configuration I guess, or in push-pull it would be quite efficient for triodes, while requiring a modest drive. Just some ideas, I think in single-ended it will be good for about 3 Watts.” (11)

“There is a PDF file of almost 2Mb about the EL6425 at both Frank Philipse's Polish site and his USA site. The language is mostly Dutch, but that shouldn't be a problem.” (12)

“I got some EL36’s on Ebay; they are Svetlana's and look of very decent quality. Year of manufacture of my valves is 1978 and data for them is at Tubes.RU, where it says that it is a full EL36 analogue. I got them quite fast and for a good price. I remember a previous thread on getter-less valves, which was about these valves as well. There is clearly a getter inside, but it is not flashed as it is of a type that doesn't need to be, (and cannot be), flashed, but outperforms traditional getters. I tested all valves and they are fine. Also, triode-mode usage looks quite promising, as I measured triode characteristics and they don't look bad at all! μ of around 6 and a low ra of around 450Ω looks promising, they don't need shocking high voltage to operate efficiently. So they might be very good for an el-cheapo triode project that still sounds nice”. (6) 2/1/01

I made some very crude Ia-Va curves for the EL36 in triode mode - using a NOS Trigon sample and an AVO 160 tester. I was only able to determine a minimal number of test points but when I plotted them graphically I got a slope, the ra, of 822Ω, which I note is about twice the figure you mentioned. I used the curves for Vg = -15V and Vg = - 20V which seem to be in the more linear, central area of the characteristics. It sounds like the Svetlana's you have are a bit beefier than my odds-and-sod's so that is good for you!

“I figured out that the ra should be around 450Ω by using the formula μ = ra.gm, as gm in triode-mode is not really different from that in pentode mode, it is even slightly higher. All is determined at which current you measure ra, as this can make a difference of a factor three with no problems. At very low currents, ra rises. I measured the characteristics very rudely with a stabilised power supply and two digital voltmeters, so I didn't measure many points as well. But they indicated that triode mode has potential”. (6)

At this point, so as not to exceed the 11~12W anode dissipation it would appear Va needs to be no more than 100~150V, which is indeed relatively low.

Selling Russian EL36 equivalents on eBay is Gintaras Sakenas, [a.k.a. lempos on eBay]; you can email Gintaras or view his website. The postal address is: P.O.Box 1367, Vilnius, LT-2056, Lithuania. “6CM5/EL36 Russian equivalents, marked 6π31C, which come from military surplus in Russia.”


Substituting an EL36 for an EL34

Firstly, hardwire the socket pin 1 to pin 8 if such a link isn’t already present. The EL34 has separate g3 (pin 1) and cathode (pin 8) terminations, whereas the EL36 does not; in the EL36 both g3 and cathode appear at pin 8, while pin 1 is “n/c”. If such a link doesn’t already exist, the circuit to the g3 (suppressor grid) should be inspected, to determine the effect of linking this point to the cathode, first.

Then hardwire the former anode, EL34 pin 3, to the top-cap for the EL36. The socket pin 3 can still stay connected as it was, since for the EL36 pin 3 is “n/c”


EL36 – EL34 Comparison Chart

Amplifier Class Parameter EL36 / 6CM5 EL34 / 6CA7
Ih(Vh = 6.3V) 1.25A 1.5A
A Va 100V 265V
Ia 100mA
Pa 10W 25W
Vg1 -7.7V -13.5V
Vg2 100V 250V
Ig2 7mA 15mA
Pg2 0.7W 3.75W
gm 14mA/V 11mA/V
Ra 5.3kOhm 15kOhm
Vin(pk)   12.3V
RL   2kOhm
Po   11W
Cga 1.1pF 1pF
THD   10%
B Va 300V 775V (from 800V anode supply)
Ia (zero signal) 2 x 18mA 2 x 25mA
Ia (max signal) 2 x 100mA 2 x 91mA
Pa (zero signal) 2 x 5.4W 2 x 19.4W
Pa (max signal) 2 x 30W 2 x 70.5W
Vg1 -29V -39V
Vg2 150V 395V
Ig2 (zero signal) 2 x 0.5mA 2 x 3mA
Ig2 (max signal) 2 x 19mA 2 x 19mA
Pg2 (zero signal) 2 x 0.075W 2 x 1.2W
Pg2 (max signal) 2 x 2.85W 2 x 7W
RL (anode-anode) 3.5kOhm 11kOhm
THD 7.20% 5%
Vin (pk) 20V 23.4V
Po 44.5W 100W
Cga   1pF