© COPYRIGHT NOTICE  All Rights Reserved.
Last Update: 18-Jun-2018

Portrait of a tube

full index list

C3g, C3g-S, C3m, C3o.

Author: Jac van de Walle
© Copyright Notice

Perhaps the best small signal pentodes ever made

The "3" I believe is to show they are third generation post tubes. These were low noise audio tubes, made by Valvo, Siemens and Telefunken, exclusively for the German post. SEL (later ITT-SEL) provided their telephone electronics to the German post, and the tubes were private branded for ITT, but still made by either Siemens or TFK. (As they all are). The TFK have a nice logo stamped into the metal. Most have a banderols on it, with a series number, but not all of them have this. These were not for sale for other customers. They were used in repeater amplifiers for long distance telephony.

I have an original 1992 price list (not a copy). Here you can see C3g has cost 257DM for one. That is 130 Euro. The factory price for the standard 50pcs box was 4898 Euro.

Admitted, in the steel can they are not glowing so nice. Believe me, once you feel these tubes get warm, you start to see them as tubes also.

Who can remember the 1970's with a long distance call from Europe to the USA? A call cost you 5$ per minute, and still the voice on the other end was not loud enough, there was a lot of noise on the line, and a one second delay. So you started to speak very loud into the phone, but that didn't help much, and I remember people in the offices would (try to) look important, by shouting into the phone with long distance calls.

Anyway this is not what good signal transmission is about, but it was the limits of what was possible with ocean cables. If there is one enemy of a long distance call,  it's low signal and noise. The problem is, a long distance telephone cable is a noise generator, and an attenuator too. So after a certain length, the noise gets more and signal gets less. This limits the length of a telephone line, and simply amplifying the low signal that comes out, would not help since you can't get rid of the noise.

If this is unclear to you, imagine the following, on a loud market place in the open, somebody shouts a message to you from a distance. You can understand it, regardless the noise. You could repeat the message, and shout it to the next person. So you are a "repeater", and with enough repeaters, you can send a message over great distance, regardless the high noise level. Provided no repeater distorts the message of course. Now suppose you don't want to use repeaters, this is not going to work. The sending person can not shout any louder as maximum, and suppose you go all at the other end of the market place, it will not help to amplify what you receive with amplifying equipment. You would amplify all the random speaking and noises too, and what you try to hear gets lost in the other babble and shouting on that market place. Indeed the only solution is, to repeat the message before it gets lost in the noise. You see, with a long distance transmission cable, it works the same way. So the only solution is, to re-amplify the signal before it gets too noisy. It increases the maximum length of a cable if by putting in repeaters, by amplifying the signal before the signal-to-noise-ratio (SNR) gets really bad. Doing so on our planet, using good cables and finest tube equipment, they could overcome distances from one continent to the other. Though practical situations are extremely difficult, and such projects are higher arts of project management. Imagine the weight of such a cable on a ship, or the trouble with repeaters in salt water, with a 4000 meters column of water pressure on it, inaccessible for ever, somewhere between Honolulu and Paris, on the ocean bottom.

Going further back to the 1960's, this had to be done with tube equipment. Of course they tried to use as little repeaters as possible. The ideal repeater amplifier adds no noise, has no distortion, draws no current, is very small, and needs no service. A partial solution you get from adding an auto transformer (Pupin coils) at certain distances, but there comes a moment where you need active amplification. So the engineers desired the "ideal" tube. However these ideal requirements do not go together well: High Amplification, low noise, long life, no distortion, small size, low filament power. As most of you know, a tube with more heater temperature will have better electrical performance, and more lifetime, but uses more power too. So the solution is a bit more complicated than you might think. Tubes that meet all of the requirements are masterpieces of design, and non-commercial since they are EXPENSIVE. The price of C3g was 295DM for one tube. So that's about 200$ for one tube. In those days you could buy a new car for 5000$. So a standard 50-Pack of those had the price of two new cars. Obviously, cost was no issue with those, and that's logical since with those you could for instance work with a cable that had a bit higher loss, and add repeaters inside to compensate that. Imagine 1000km (1 Million meter) of cable can be made 20% lower cost, who cares if the tube that makes this possible costs 200$. This is the world of C3g, C3m, C3o.

I think most of the secrets of how to make C3-tubes are gone with the wind, but at least we can still buy those miracles of tube art from new old stock. If the days come where we only can buy used one, it's no problem too, since the end-of-life is clearly specified in the data sheets. So you can always tell if the tubes are still good. I found the Funke W19 tester very reliable in picking out used tubes with good lifetime in it.

These German post tubes were first designed in the heads of the amplifier engineers,  that finally wanted to have something without compromise.    The engineers just said,  what must an ideal tube have, and then let the tube factories try to make it.

  • Highest amplification
  • Lowest low noise
  • Lowest microphonics effects
  • Higher lifetime than commercial tubes
  • Stand-by possibility (very few tubes have that)
  • Metal shielding
  • If a pentode is triode connected, it must have curves like a real very good triode
  • Fully complete data sheet, leaving NO DATA out.
  • Optional: Individual series number on each tube

Let me bring to your attention, the C3g at it's introduction in 1952 was the fist Frame grid tube, and it was not available commercially. In short, a frame grid means the grid wire is not self supporting, but is wound around a hard metal frame. Like this, they could use wire which is so exceptionally thin, as it was never possible before. You can not see this wire with the bare eye. This is quite strange to observe, when you take a frame grid tube apart. The grid wire is there of course, and you can "see" there is something there, that you can look through, but you can not see what that is. As if there is a gold colored, transparent layer, that you can see through. Things looks a bit "unsharp" when you look though. Like the low resolution picture below. It is just like this when you look at a frame grid wit the bare eye. Then, if you click the picture it gets enlarged, and you can see the wires.

This technology is more expensive, and was used for very few tube types, perhaps 100 only. Given the 10's of thousands other tube types ever made, this is not much.

Read here about frame grid tubes.

NOTE that in the 1980's when tubes were obsoleted, there was a lifetime-buy option by Siemens for the German post. It stretched over a few years. It is from this period that BIG lots of tubes were made, and just stored for later service of old hardware. Because of the high manufacturing numbers, these were very good quality. Popular tubes were C3g, C3m, and also Siemens ECC801S of remarkable good quality. Probably many other tubes as well, it's just these three Types I ran across myself. Something similar happened in the USA, and from that period many very good 6922, 5687, 12AT7 and 5751 are around. These are those tubes with a barcode label on them.

Then, the whole hardware developed so quickly after the digital multiplexing was used, and analog repeaters were taken out of service everywhere sooner than expected. And so, a six digit number of those tubes were stored and never used. Through the years these were sold, and it seems the German government stocks of C3g and C3m have dried up since 2005 or so. However, these tubes are always somewhere, and find the path to their end user.

Check for a 16 pages (!!)  Data sheet at our website / under Techcorner

I guess they came up with some more nasty things,  that the tube manufacturers all had to comply with.     Just look at how nice the triode connected curves are.     These curves are so linear,  I think there are very few triodes excising with such nice curves!

What to do, if you don't like the metal cap?

You can take off the metal housing, and inside is a very nice glass tube!

C3m inside

C3g with cap


C3g Inside. 

Look at the two round plates above the pins.   The lowest is a metal shield (outside connected) for lower hum.   The other is the mica.   The anodes are open from the sides.   This open construction is the best for finest linearity.   It allowed plate distance adjustments after the tube was assembled.  This noncommercial construction was used already in DHT post tubes from the 1930's. 

Zirconium + Barium Getter

We are lucky to have a lot of the C3g version with the additional zirconium getter. All C3m we sell gave this double getter! So that is additional to the Barium getter ring (or plate).

A Zirconium getter is expensive and it works like this: A Barium getter has most of its function during the short moment (a few seconds) that it is flashed during production, so when it is in the tube in the form of a cloud, while being transferred out of the getter halo, and condensing on the tube glass. This cloud, at the moment it exists inside the tube, absorbs (at that short moment) almost anything whatsoever. Then, after it will be condensed on the glass, the Barium getter is only conditional active. Like during great heat such as with KT88 tubes. However little tubes like C3g can not really use the getter any more after activation. Of course there are some remaining functions left, amongst which is dust catching (yes!) but maintaining extremely high vacuum is not done. Here is where the Zirconium getter comes in. These need no flashing. They start to absorb gasses, whenever they have a sufficient temperature, and all you need to do is, mount it at a warm place. So really top class tubes have both getters. These tubes used to cost 295 DM when new, I have seen an original price list myself.

Zirconium getters can have various appearance. Some can be small square plates as you see here. With other tubes like EL503 they are hidden inside the plates. Tubes like 845 have Zirconium absorbed inside the graphite. With the 845 the Barium flash you see, is only used for initial vacuum during production. Maintenance of the vacuum is done by the Zirconium getter.


Removed cap

You can glue the metal base on the tube.  Put two components glue on the sides. Do not fill glue in the center adjustment hole. That could break the glass pipe in the middle. Note that one  pin hole has a different shape, and is used for positioning. 

Yamamoto Socket

NOS Socket

C3g Siemens
Order Nr: 114-146-50

C3g-S Siemens. From 1975
Order Nr: 114-147-63

C3g Valvo
Order Nr: 114-150-29

C3g-S Valvo
Order Nr: 114-151-93

C3g-S Telefunken
Order Nr: 114-149-97

C3g Lorenz. Some boxes
have Date code 1978.
Order Nr: 114-144-48
C3g-S Lorenz. Some boxes
have Date code 1978.
Order Nr: 114-144-48


C3g, C3m, C3o, what's the difference?

C3g was made after C3m, and C3g has a frame grid and gold pins. Frame grids generally give better quality tubes, though it must be said that the quality of C3m is just as superb, and I make the statement here for C3g and C3m, that there is no better tube available of the same kind. If you think I am wrong, email me with facts from a data sheet, and if you find a better tube, I will add it here.

Generally with C3m and C3g it can be said they have the gain of a pentode, and distortion same as only the finest triodes like E80CC. However C3m will give that low distortion at a gain of 78. (and E80CC only at a gain of 25).

A gain of 78 It means you can drive a 300B with just one C3m, and you have less than 1V input sensitivity. In a few words, this explains why these tubes are so great for HiFi purposes. For applications like pre-amp tubes, or driver tubes for 45 or 2A3 the C3g can be used as well.

They show their extreme low distortion only as pentodes. They can be used triode connected, but then gain comes down a lot, and the distortion increases to a level just below triodes like 6SN7.

  C3g C3o
This is a 6,3V Version of C3m


6,3V 6,3V 20V
Pins Gold Normal Normal
Grid Frame Normal Normal


Gain of the C3g vs. C3m

C3o is a C3m with 6.3V filament. However C3o is exceptionally rare as NOS tube.



C3g-S New

C3g New

C3g after
10.000 hours

C3m New

C3m after
10.000 hours


Heater Volt





Heater mA






Rk-h (ohms)


R Pin-to-Pin (ohms)


Rk (ohms)

Ig1,max (uA)



Gm Pentode (mA/V)
Gm Triode (mA/V

Ri (Ohms)
Triode connected


Ri (kOhms)
Pentode connected


Gain Pentode connected *2

4200 (theoretical maximum)
1625 (theoretical maximum)

Gain Triode connected *1


*1 From TFK Datasheet
*2 Tested with AT1000                              
*3 My personal observations



The "S" Version C3g-S

C3g-S is a hidden treasure. They are EXTREMELY rare. I mean the real ones, not the fakes. Oh yes.... the "S" is faked, but that can be seen so EASILY by tube testing. Please read more about "fake" testing here, because some sellers do so with the U-trace.

So far, I was not able to find the factory specifications for this, but one day, I hope "coincidence" is going to help us out here. I have seen the original test tools for C3g and C3g-S selection myself, in Ulm. This was a dedicated plug in unit, for a universal test bench. The bench was like 1 Meter wide, and the operator had to add plug-in units to it, for C3g or C3m, C3o. etc. The unit had lots of knobs and instruments on it. The tubes were tested one by one, by hand, and many settings had to be made, before all parameters could be tested.

About the S-Version, the following I know for sure, because I was told so first hand by an Ex Telefunken Employee, who was at that time involved in this. The S-Version is selected for a SEVERAL things at THE SAME TIME. One of those the Transconductance exceeds the data sheet value by a minimum percentage. From my personal measurements, transconductance is from 105%...to 110% at normal (so 13mA) plate current. However, make sure, you don't get fooled by auction website sellers, shipping you normal tubes with an "S" printed on them.

fake testingWARNING!


This fake testing trick as as simple as it it easy. Don't be fooled! They begin with showing you official, printed test information.


This is hard to explain in just a few lines. Saying you have tubes with 105% Gm, testing this way, means only you have tubes with 105% Gm tested THIS way, and NOT the Telefunken way! This is really not a good was to test. For this, you need to understand the difference between auto bias testing, and fixed bias testing. If you think now, "...is there a difference? " don't worry, many think so. Now, don't take me wrong, but not knowing doesn't mean it doesn't matter. Much to the contrary! For those tubes, where it DOES matter, you better look twice who sells you what, because if it DOES matter, that is not without a reason. There are some (very few) tubes which have an official data sheet with AUTO BIAS specifications only. Such tubes are always high quality tubes, like C3g, C3m, C3o, or their little sister D3a.

What is the reason for auto bias testing? As we all know, auto bias, biases the tubes "automatically" the same. So an auto bias circuit for 13mA will (and must) produce close to 13mA, for a tube of flawless quality. Any tube from 11...15mA in fixed bias, will perform close to 13mA in auto bias. This means, you must use very tight limits for an auto bias test, much more critical as with a fixed bias test. You will find those limits of course in the data sheet. A new tube, with too much plate current in auto bias, that should better not happen. Such a tube has a tolerance problem, it is simply not build well, or it was dropped, and the inner structure is deformed. A tube with not enough plate current, that is something else. It can have any problem at all. So, whatever it takes, to reduce the plate current, that can be virtual any defect. The idea of an auto bias test is NOT to find all tube parameters. Auto bias is intended for tubes that have good parameters by design anyway, but we want to know if the tube is at the end of life, or perhaps like new.

HOW TO TEST Gm? There is only one way: This has to be done at 100% plate current. There no other way, not for these tubes here, and not for any other tube. And yes, all new made testers (100% of them!) ignore this knowledge, and I would say ignorance that is the best word for this indeed.

The tube tester W20 (Card from above...) makes the same mistake, but at least they have corrected the "good begins at" level to such which (probably) can be expected with fixed bias. So, the 13mA tube is considered to be "good" above 8mA. Whereas the German post, who designed those tubes, never said so, and also never said this is the right way to test them. They are not doing it wrong actually, just not by Post specifications.

This brings us back to auto bias testing, because in auto bias, a good tube is close or at 100% plate current by definition. (and if not, it is bad). Any Gm test at a random plate current, such as you would get with fixed bias, is indeed just something random, and little more than that.

Now for the S-Version. Very simple: These must have >105% Gm at 100% plate current. They must be like this, and nothing else.

A tube which has 105% Gm, but needs 115% plate current for this, is just a normal tube. If these are marked "S-Version" it can not be true.

So at 13mA.

Any tubes with just Gm >105% and the plate current is accordingly lower, these are normally good tubes, but no S-Version. So when Gm is 5% higher and anode current is 5% lower, that a normally good standard version tube.

As a rule of thumb, the average of Gm performance and Ia performance must be 100% for an amuse tube. So Gm = 105% and Ia =95%. Average is 100%, this is a new tube. Or, when Gm=105% and Ia=80%, this is an average of 93%. This is not a BAD tube, but under the assumption this is NOS, this is definitely not an S-Version. Or if we are SURE it is an S-Version, like picked from original equipment, then it must be a used tube.

For NOS tubes, when Gm is at 105% and plate current is 20% lower like 10 or 11mA, this can NEVER be an S-Version tube. When it's printed on the tube boxes and tubes itself, do not be mislead by this, because S-Versions can easily be verified by correct measurements.

Also read (and try...) the below experiment, it will open your eyes.


Some examples:

Gm = 105%, Ia =12.4mA (95%) NOS condition Regular Version

Gm = 95%, Ia =105%: NOS condition Regular Version

Gm = 105%, Ia =13mA (100%): NOS condition C3g-S

Gm = 110%, Ia =13mA (100%): NOS condition C3g-S

Gm = 105%, Ia =16mA (100%): NOS condition C3g-S

Gm = 95%, Ia =105%: NOS condition Regular Version

Gm = 105%, Ia =11mA. NOS condition Regular Version, or USED tube S- Version

Gm = 105%, Ia =10mA. END OF LIFETIME tube, because <10mA is a bad tube.

1st. advise. If the above is not fully clear that is because this is indeed not easy to explain. But, I have tried, and best is, just read it a few l times, until a light begins to shine. Also do the next experiment if you have the equipment:

Experiment: Take a C3g, which is reasonably good, but you sacrifice it for this experiment. Suppose Gm=95% and Ia =12.5mA. So slightly used. Now knock the tube on the table. Such that you deform the inner structure, but not damage the tube totally. Now re-test it, and you will see, Gm=105% and Ia= 11mA. If there was no change, you didn't hit the tube hard enough. You can also try this with an old EL34, because these react to this much more sensitive than C3g. It is crazy, and you wouldn't want to know how many EL34 or C3g have been "improved" this way to sell it on an auction site. This will easily fool a fixed bias test, but it can never fool an auto bias test.

2nd. advise. Any "S" versions are rare and expensive. Makes sure, Gm is tested ONLY 100% plate current. Let nobody fool you.


Build a pentode pre-amplifier stage with C3m

Take a normal Cathode resistor stage, and use this table.
More details in the official Telefunken data sheet

Good applications for these tubes are:

  • Inside active microphones.
  • Very Stabile instruments, with DC coupling, like stabilized power supply.



  • Low microphonics applications. C3g mainly (frame grid)
  • triode connected, or pentode connected
  • driver or preamplifier stages. C3m for high gain.
  • phono amplifiers.  Searching for the holy grail? The C3g or C3m! Finally..  LOW NOISE!

From the fact they both exist, you can see they both have their justification. So don't ask which one is "better". Same as with ECC81, 82, 83, there is c3g and c3m.

Yamamoto A09 amplifier with C3m (Metal cap removed by Yamamoto)


Some small note about low distortion.

In pentode mode, C3m has distortion figures compared with E80CC, but at much higher gain. (at 5Vrms out, E80CC has, with bypassed cathode, a gain of 25 and 0,16% THD. C3m has a gain of 78 and 0,12% THD).

Data sheets. can be found in the tech corner of this website (From the menu)

© COPYRIGHT NOTICE  All Rights Reserved.