Tube Rolling – The Big Mystery Part 1

One of the biggest subjects within the permanently growing tube scene is dealing with tube rolling…. And it is one of the most misunderstood elements in the whole High End audio field. Now you will shout at me…”of course I hear the big difference if i change a Telefunken ECC83 Smooth Plate against a Mullard MC1…” – and yes, you are right, there might be a difference, perhaps a big one. Confusion starts when we study thousands upon thousands of webpages or HiFi forums, were people share their tube rolling results with such an absolute claim, that some technically experienced people will raise their eyebrows – and the poor people who asked for some help will have to spend a lot of money buying rare NOS stuff on the basis of these indications to learn, that most of these spot on advices are not worth a single penny. 

The task of tubes:

One of the most important things with internet community advices regarding the usage of a certain electron tube, is to make clear that the asking person and the helpful community speak about the same thing. If I would like to get some advice in choosing a specific 12AU7 to be used in a splitter position on my amplifier it is completely wrong to ask stupid questions like ” what is the best 12AU7″. As all “what’s best” questions are obsolete in our beloved hobby – the answering community might have recently tested some 12AU7 in an USB DAC, where the task of the 12AU7 is – if we are lucky – to be a gain stage and not only a device to add some harmonic distortion to the otherwise clinical sounding DAC. Such an application area is far away from what a 12AU7 must achieve in a splitter stage on an power amplifier.

To make things even more complicated – there is no such thing as THE splitter stage, as a lot of amplifiers combine in a single double triode per channel a splitter AND a driver stage. If two people share their experience about a good splitter tube the one with separated splitter and driver tubes might come to a completely different solution than the other person who has a combined splitter and driver stage in one single tube. To cut a long story short…. the minimal basis on which such discussions should take place is some knowledge about the circuit in which a tube does its job!!!!

A little test:

Lets try to get things simplified….In the following example we use a pre amplifier which has just one double triode of the ECC83/12AX7 type in its gain section – and the only thing we will change is exactly this single tube – we change nothing else. To start with our little test session we have to choose some suitable test signals…It is very interesting to start with just white noise generated by your laptop via an online noise generator. Therefore we connect a laptop to the preamp via some cable adapters. Please adjust a fixed volume and do not touch it anymore during the following test:

Now we will start the white noise generator with our first test subject – lets say we use a Telefunken ECC83 smooth plate. After a proper warmup period (10 minutes) we will then listen to the most “exciting” signal on this planet – pure white noise…it doesn’t get anymore stupid than listening to white noise – but for our short experiment we have to survive this! 

While we are listening to boring white noise we should concentrate on the sound of the test signal as much as we can… After we have gotten an idea of the sound and have inculcated it as hard as we can, we will switch the amp off and should wait 2 – 5 minutes to pull the Telefunken (use gloves – the tube might still be hot) and change the German tube to our second test example – the RCA 12AX7 Black Plate…. switch the amplifier on and sit down on the same spot you heard the Telefunken just 5 minutes ago….

….As white noise implies all frequencies with the same energy, we have a tremendous broad band signal without any tone, any melody, anything which can distract our attention from the frequency response of those two tubes. But what is this? …The difference is very, very subtle, so that we have to do this little test more than one time. If you have a chance to do this with two people – try to complete some blind tests. Can you really hear a difference???

Maybe there is a slight difference in broad band static frequency response – the RCA might be a little bit darker with a tad more upper bass extension and the Telefunken might be a little bit slimmer in the upper bass register and possibly shows more energy at the absolute top of the spectrum – but the difference will be very, very minor! – I bet that in a blind test you will have a 50% chance to name the right tube – not more!

Now we change the test setting…

We will listen to a digital medium like a CD and we choose some type of music with a very boring drum pattern which would be repeated in an endless loop – electro tracks are very, very good for such a test. The track should have some good dynamics – not one of the mid 90ties productions that are crushed by the mastering limiter…. Now we start the same procedure as with the white noise session – and remember – do not touch the volume potetiometer once the test has begun. 

Now the difference is much clearer… (depending on how well you choose your test song). With the Telefunken we have a clean and clear picture of the music. We have a near to neutral frequency response and a tremendous clarity in the whole midband as well as a good macro and an outstanding micro dynamic. Now we switch to the RCA Black Plate…. The first – maybe most obvious – difference is the deep bass the RCA is producing. While the Telefunken was fast and very differentiated in the deepest register the RCA has more punch and less clarity and speed but it seems that there is a higher dynamic in the deepest bass. The mids are thicker and a little bit less clear – the dimension of the stage is huge – very wide – but not deep. The Telefunken has a slightly tighter stage width but is able to tell us something about the distance of each source, which the RCA is not able to do in an equal fashion. The presence of the RCA is higher as with the Telefunken – it sounds more dynamic around 3 – 5Khz. A snare will hit us directly and the snare has slightly less mid body and sounds sharper. The attack shifts upward with the RCA. On the other hand the RCA doesn’t have the resolution in the upper spectrum around 12Khz and above, which the Telefunken can provide. The RCA sounds more muted as the Telefunken and that makes this tube sound more warm, more like people think a tube should sound. The localization accuracy of the Telefunken compared to the RCA is a different class….were the RCA tells you something about the position of an instrument in a more relaxed fashion, the Telefunken produces a pinpoint accuracy. 

Maybe your observations may vary slightly – but the basic direction will surely be comparable. The big question is now – why do we hear such a slight difference if we listen to a static white noise test signal and a more or less huge difference if we listen to real music? The answer is, that the tubes are all relatively equal in frequency response if the signal is static. All the differences we determine are more or less a result of the dynamic response each tube gives us. 

Time for a more drastic test procedure:

In this third test we need to have some special gear, which not everyone has at home. We need a digital workstation to record a waveform of the dynamic response our two test tubes produce. For this experiment we connect the input of our workstations digital interface to the output of our tube preamp and vice versa (be careful: don’t use the same input and track or you will create a feedback – so some routing experience would come in handy). Now we can record the signal the preamplifier normally would send to our power amplifier. And at the same time we can do much more with such a hard disc recording system – we can design our own test signals. Therefore we will prepare the workstation to send some burst signals in different frequencies to the amplifier – which we later on will record through the preamp with the two different tubes. We will have a burst around 50hz, 200hz, 500hz, 1Khz, 3Khz, 5Khz, 8Khz, 10Khz. These bursts contain sharply cut (64db/octave) white noise samples, which are just 100 milliseconds long.

Now things become very interesting….

If we analyse the recorded waveforms, we can detect a huge difference between the Telefunken ECC83 and the RCA 12AX7 Black Plate. The RCA has a lot more rounded leading edge with deep frequencies and also produces a kind of compression which makes the short burst longer at the time domain, because the decay curve is lifted up, while the peaks are slightly rounded. The Telefunken in comparison handles the leading edge almost without any rounding and applies almost no compression to the signal – Input and output signals look almost identical. If we analyse the other frequencies the outcome is even more astonishing: The RCA reacts on very high frequencies like an electro optical compressor… we can see a slow attack time and after the leading edge has passed, the decay is compressed with a small ratio and a long decay time, which is what we also hear with music – a cymbal has a soft attack and a long decay and is defensive in its character. The Telefunken does nearly the opposite – the attack of a high frequency burst has the same shape as the incoming original signal – there is no compression at all – the frequency burst transmitted through the amplifier is as long as the signal sent to the amplifier. Both tubes act with a tremendous difference! 

And now:

What we have found out by this point is the fact that two ECC83 tubes in the same circuit with exactly the same requirements act differently. But which one is better??? If your room acoustic, the speaker placement and the circuit of your amp, the chosen cartridge and the speakers themselves have a certain acoustic fingerprint one would choose the RCA, while in a different situation the Telefunken might be unbeatable. Do you see how helpful tube rolling advices are…??? 

It all depends on the smallest parameter – one high end enthusiast will tell you a Mullard MC1 will be acoustic nirvana, while another one will tell you the same about a Siemens Halske ECC83 MC2 which sounds totally different. Some people will also tell you that in a good circuit the tube and its sound character does not make any difference which is completely wrong! Each tube has a certain fingerprint and it is not an easy task to find a setting in which it doesn’t act as an equaliser, but instead shows you the beauty of the recreated recording in a neutral way.

A real life example:

If you, like me, own a LS3/5a setup – and in another room with very comparable acoustics a pair of Quad ESL 57, both paired with the same cartridge, mounted in the same tonearm, on the same deck, you will detect that the LS3/5a tweeter is a different pot of tea than the high frequency performance of the ESL57. The KEF tweeter of the LS3/5a is much more unlimited in bandwidth, while the ESL 57 shows a relatively sharp high cut around 18 Khz. A Lyra cart amplified with a high bandwidth phono stage has a tremendous high frequency resolution. With the LS3/5a it could be too much of a good thing – under certain circumstances those tweeters can produce a slight edge on vocals as well as cymbals or high strings and also organ music with high frequency registers. The Quad is a different breed – it needs every single drop of resolution we can squeeze out of our system. The high frequency response of the LS3/5a also acts a little defensive within the presence register around 3 Khz while the Quads behave more linearly in that particular region. So with the LS3/5a we must avoid tubes which create a high frequency sensation, moving away from the path of neutrality, while the Quads need very open sounding tubes. For example the above mentioned Siemens Halske ECC83 MC2 paired with a LS3/5a monitor would sound slightly synthetic in the highest treble section (paired with a Lyra cartridge) while the Quads will show you very, very clearly the HiFi like behaviour of this particular Siemens tube, which has a recessed midrange and some spectacular bass response as well as champagne like treble and air expansion. If I would mount a different cartridge to both systems the outcome will be completely different.

How big are the differences???

If you expect a night and day difference between good quality NOS tubes I must disappoint you – the differences are subtle but clearly audible and in a highly tuned setup they can make the difference between good and amazing! So tube rolling makes sense and can be much fun but also implies the danger of large frustration. A lot of people think that a tubed system can be used to diminish the effect of a bad room acoustic or that ultra expensive super tubes can transform a poorly designed amplifier into something great – both is not the case. The right tube in the right spot chosen with great care is the cherry on the cake. The more resolution your system has, the more important the choice of the right tube in the right spot will be.

Audio is a chain:

Some people do extensive tube rolling with their power amps, they spend a fortune to get rare NOS power tubes and maybe spend the same money on tubes wich have a mega reputation, while the preamp gets less attention. This would be a big, big mistake. As an audio system is a signal flow chain it will all be influenced by the weakest component. The preamplifier and its tubes has a much bigger impact on the whole sound Gestalt than the power tubes – and even more dramatic is the choice of tubes in your tubed phono stage. As a rule of thumb you can say that those tubes with the greatest amount of gain will have the biggest influence on the global sound of your system. But be careful – if you rolled the finest phono as well as superb line stage tubes and you gave everything to get the input, splitter and driver section of your power amp right – BUT you chose some inferior power tubes – a lot of your work will haven been for nothing.

A good advice might be to get all the stages till the power amp as transparent and neutral as possible. With the power amplifier at the end of the signal chain we have the chance to voice the system to our liking (within certain limits). Please take note – no single super NOS tube can repair any mismatches in system assembly – nor can it repair impedance mismatches between the different components or a weak acoustic in your listening room. But within certain limits we can adjust to our taste and maybe ameliorate some special system colourations by rolling specifically chosen tubes. Tube rolling is a little bit comparable with extraordinary cooking…Take your time and be patient – and before you spend crazy money on holy grail tubes – get every other parameter in your system up to the task!!!!

Stay tuned for part two….

E. Strauss

9 Replies to “Tube Rolling – The Big Mystery Part 1”

  1. Hello Ekkehard
    Thanks so much for your writing
    Good thruths to tell
    Very informative blog as I’m Airtight devotee….
    May I ask some advice regarding NOS tube equipment for the AT ATC5 ?
    Kind regard


    1. Dear Philippe,
      as the ATC5 uses ECC81 or 12AT7 in its line section, it is not sooo easy….One of my favourite ECC81 are the Valvo Blue Label 6201 which are available with or without the pinched waist glas design. The pinched waist almost vanished fron circulation and is very, very expensive…but the Blue Valvo with the standard glas shape can be seen sometimes. This one is a very good tube for the ATC5 line section. The phono section is a case for 3 Telefunken ECC83 – in my book it does not get any better!!!! – Hope that helps – happy listening – Ekki


      1. Dear Ekki
        Thanks so much
        have previously followed with great succes your recommendations regarding the NOS tubing of my ATM4
        Obviously that will help a lot


  2. Dear Philippe,
    if you know my article about the 6L6 GC (Air Tight ATM4), than you will also know which Valvo I mean – there is a picture of that particular tube on my blog. Please observe the desired tubes closely bevor buying….because there are also white printed Valvo 6201 available, which are Philips made and not Valvo Hamburg. It must have gold pins and the blue printing on the glas with the special number. And – as always….select only such a rare and expensive tubes if the seller can measure them and those figures looks like a real NOS tube or near NOS specs…because an old tube at the end of its working period is worth nothing. If you need more resolution the Telefunken ECC801S would also be worth to try – but this one is not everyones taste….it depends on your system und speakers. The Telefunken is tremendously open and fast and ultra neutral with an amazing amount of resolution. The Valvo might be a good compromise between outstanding resolution and some warmth…but at the end you are the judge!!!!….happy listening – e. Strauss


    1. Thanks Ekkehard
      Yes I know very well your 6L6 topic
      It’s very difficult to find Valvo BP, a real challenge
      A well known German seller has TFK, …think will take the TFK road even if they
      BTW system is LP12/Ekos SE/ dynavector XV1s/Urika phonostage/ATC5 stock tubes/ATM4 NOS tubes/ Tannoy kensington GR/A23 IC / Kondo copper SC
      so I have the possibility to tune the treble level on the speakers filter if the sound is too sharp or peaky….
      Thanks again
      Kind regards


  3. Dear Philippe,
    if you read my Tube Rolling essay….you will see, that the preamplifier is of the highest importance… is the heart and soul of your system (regarding amplifiers)….The stock tubes are 3X 12AX7 EH gold pins very, very well screened and 2X JJ 12AT7. If you will compare that to 3X Telefunken ECC83 Smooth Plates and 2X Telefunken ECC801S or Valvo Blue 6201 it is not even close….I promise you, you can hardly believe what is going on!!!!! It is STUNNING!!!!
    The ATM4 can only amplify what the preamp sends – so if EH and JJ is delivered to the amplifier – and you did some nice tube rolling there – it might be more obvious that the pre amp ist not up to the task. As the ATC5 has a very, very, very nice phono stage implied…I would drop the Linn phono stage in favour of going all in!!!!! Get a good tranny with could be a 1:20 or 1:30 für the Dynavector – as the ATM5 has 40db MM gain, I would opt for a 1:20 SUT – and then you never will touch the Linn phono again. You payed half of the price – maybe even more for the ATC5 phono stage…so USE IT!!!!! – If you would not use this tubes phono stage – a ATC3 would have been the better choice – because it is a line stage only…But at the end you did it right – but please!!!!! – use that MM phono stage!!!!! It is amazingly good!
    With the Tannoy´s you can use the Telefunken ECC801S easily – without adjusting the treble filter on your speakers…The ECC801S is not sharp, peaky or harsh – it has a tremendous high resolution and is super neutral and amazingly fast in transient response but at the same time super smooth.
    You can compare that with cutting your finger with a scalpel – or using the knive you will cut bread.
    The Telefunken is like a scalpel – it is smooth and very precise at the same time – the EH and JJ stuff is like a breakfast knife – a cheap one. If you select Telefunken ECC83 Smooth Plates for your ATC5 phono section it would be an amazing match with your cart (which is super dynamic with an astonishing bass) and your speakers!!!! As for a not toooo expensive SUT, try to test the ATH3 Step Up – it is possible to get this SUT in 1:20 and 1:40 – there are internal jumper wires which can be configured in either way. It is dead silent (if you ground the SUT at the ATC5 ground terminal – IMPORTANT) and sounds WONDERFUL! There is a new version of this SUT which is fitted with floating ground switch – which is a nice feature – BUT – I would opt for the first ATH3 – the older model – because it has this amazing steel chassis which shields the unit perfectly….the new one is made with an aluminium enclosure, which is not so ideal – Maybe you can get a good price on the older ATH3 SUT…If you have a chance – try to get one for testing during the weekend. Plug you Sondek into the SUT and from there into the RIAA MM Input….use a good shielded cable für the SUT / RIAA connection (if money is not an issue – try the Lyra Phono Pipe – it has an amazing low capacitance – ever if it is triple shielded). It is also a good idea to use a low capacitance tonearm cable….the details about that will come in one of the next essays (Adjusting A Cat – the electrical side of the game… – I am already working on that). Keep me posted if you once tried the ATC5 phono section with a SUT, It would transform the way you will listen to music…..All the best



    1. all my thanks for these advices
      can’t wait to read you
      will start by retubing the line section of the preamp
      Regarding the phono do I need 3 tubes perfectly appaired ?
      Kind regards


      1. Dear Philippe,

        good to hear, that you are working on your preamp….you are in for some BIG surprises!!!!
        As you might have seen in my tube rolling essay – as a rule of thump the importance of the tubes in one system (I know – all are important – but you know what I want to say) goes parallel with the gain one particular stage has to produce. So the phono tubes produce the most gain in a given system – they amplify in a typical MM stage with a figure around 30 – 40db! No other stage comes close! And the phono stage will be the source of an analogue system – so what is lost here is lost!!!!
        Regarding the circuit of the Air Tight Phono section, please read my article about the ATE2 – as all phono sections in the Air Tight product range share the same basic circuit (only the outlay is different!!!) I advice you to read the article. The short version would be:
        You have 3 ECC83 double triodes…each double triode – as the name tells us – is a DOUBLE triode – so it has two triode systems in one tube…that sums up to 6 triode sections in the ATC5 phono stage. If you look at the layout you will see 3 tubes. The one in the middle (with its two triode systems) is the first gain stage! The two triode systems amplify the left and the right channel in this first gain stage…between the first and the second gain stage the RIAA de emphases is done – completely passive!!!! Than there is a second gain stage…
        This second gain stage is implied by one of the triode systems of the left and right (or front or back) of the 3 ECC83….so they have to be very very equal regarding the tubes used in this section. Now – if you have counted all the triode sections we have already used, you end up with 4 used and two left…so what is the job of the two sections we have still on disposal???
        As any Air Tight preamplifier – be it line stage or phono stage – will have an output iimpedance of 600 Ohm – there must be an decoupling stage. There are 3 possibilities (basically) to design such an output stage…:
        1. The usage of output transformers
        2. The usage of a kathode follower stage (wich is very immune to different measured tubes)
        3. The usage of an SRPP stage (wich reacts sensitive to different measured tubes)

        Miura San opted for the last method….he used a SRPP decoupling stage – and that is designed with the last two triode sections left.
        To cut a long story short:
        The 12AX7 (ECC83) in the center could have slightly different measurement figures – but it should have very even triode sections…(now you know why such a tube has always a PAIR of figures published). The two other tubes must be nearly equal!

        Try to get 3 near to NOS or NOS Telefunken ECC83 – get them from a trusted tube dealer and NOT from ebay!!!! – If you want to compare measurement data – keep in mind that any tube tester will do that procedure different – so you cannot compare data collected via a Neuberger tester and the AVO type…for example. So it is a good idea to get those 3 tubes from ONE source! And yes – it will set you back 100,- each – but the effect will be much greater than ANY cable you ever changed in you system – it is not even close! And – maybe – at the end a Linn Urika is up for sale…which is what I would bet on!!!!!

        Hope that helps



      2. Dear Ekkehard
        Very clear as always, understood.
        Work in progress, some TFK already ordered
        Will keep you updated
        Many thanks
        Keep safe


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: