• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

300B Recommendations

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
6A3sUMMER,

Interesting development on your side as I went the other way around...coming from all the indirectly heated stuff to the DHT stuff...

BY the way: I found that a transformer-loaded 6SN7 as a driver beats the hell out of most, if not all noval types...but in RC-coupling it was not engaging enough...but with IT-coupled...completely differetn story...very good resolution, great dynamics (which have not been there with RC) ,musical and smooth.

But back to the subject...

Let me see if I understand you right:

1. You confirm basically that all 300B have a CT-Heater with the exceptions mentioned above (JJ, EML, Sovtek 6A3 ).

2. The EH/Sovtek 300B have a different filament structure than the Svetlanas, you would not rule them out for DC-Heating like you did with the Svetlanas

Correct ?

Well...than we have two different classes of 300B out there:

A. 300B with a 10V filament, which got a CT to get down to 5V

B. 300B with 5V filaments, which are offered with CT as well as 2,5V versions.

I cannot talk about AC-HEating as I did not used it, but assuming DC-Heating, this means that the CT-tubes have a completely different electron-field at there plates than the non-CT-tubes, structurally, let me borrow the drawing of jacmusic/EML to illustrate:

pic1.JPG


This is the case for Type B tubes without CT.

Tubes with CT-Heating approach would actually look like the same picture, but mirrowed so we have two halves from the CT to the two outsides.

Now, when Jac makes the statement that 25ma THERE IS A TOO HOT area with a difference of 25mA compared to the too cold area, hmmm...I try to imagine how the Svetlana guys have build their tube in comparison to the Sovtek as an example:

- Imagine they have only one filament wire and put a CT-connection on it. Now by coincidents this connection goes to your socket pin which you load positive DC to.
- Effectively, you have doubled the trouble as now at the CT-filament wire area you have double heat dissipation as two systems (going to the outside connections) are using the same piece of wire for their +25mA game.
- If you are lucky and your positive connection goes to the two OUTSIDE-wires (physically ends of the filament), the same trouble of 25mA more current than at the CT is at least distributed over a wider area of wire...with the "too cool area in between)"...no ?

So, if this is right, It would be quiet important to understand which Pin on the Svetlana 300 filament goes to the CT of the filament and use this for 0V of the DC-supply.

Overall, I have the impression that the lower the filament-voltage and the more distributed, the better for having an equally heated condition. Plus the distortion coming from the different grid-to-cathode-bias-conditions is smaller.

So, in theory, more reliable and better sounding when you use one of these 300B with 2,5V filaments...can you confirm that ? I believe with the JJ 2a3 40Watt you got one already in your hands and could compare it to the normal ones ?



p.s. Jac insists on the two classical 33 ohm resistors to create an electrical CT...not sure why to be honest...http://www.emissionlabs.com/html/guarantee/HEATER-MODULES/circuits1.JPG and http://www.jacmusic.com/techcorner/...07-How-to-bias-DHT-tubes-without-mistakes.pdf
 
Last edited:
Blitz,

I have never used any EML tubes, have not seen them (don't know about their filaments). I have not seen any 6A3 tubes other than the Sovtek. We can not say All tubes, until we have seen All of them.

I do not know about what 300B tubes other than Svetlana may have similar problems with spacing, and / or filament wires that expand too much, or bow too much, causing them to touch the grid from the bow, or from attraction to the plate. But just from a use-experience of myself and others in the club, I think the Svetlana had the worse reliability of what they used.

I am glad that you took a look at the diagram, you did catch the fact that the positive end of the filament had Less current to the plate, and a Colder end of the plate.

I doubt that there will be a 25mA difference of current to the plate from one end of the 300B to the other end of it. That would mean for 60mA, one end is 30 + 12.5 ma, and the other end was 30 - 12.5mA. That would be 42.5mA and 17.5mA. Well, perhaps it is that much difference. The 300B is 5.5mA/V, but that is integrated across the whole filament (and whole plate). But you can test how much the current unbalance is, if you use a floating 5V DC supply, and: Use a pair of 25 Ohm resistors to create an artificial center tap on a DC filament. Then connect that 'center' tap to ground (for fixed bias), or to the self bias resistor which the other end goes to ground.

Try that and you will find that the ends are not 2.5V away from the 'center' tap. You might find that one is 2.25V, and the other 2.75V, or similar. That would be a 10mA + 10mA difference = 20mA. Now you can calculate how much current is coming from each end to the plate. Do that and report back to us if you do that experiment.

I think that no matter which pin, 1, or 4, that you connect the +5V to, a center tapped filament will have two equal width areas of spread of hot and cold. The fact that the two hot areas are together, or are at ends, would not change the width of each.

DHT: I have not done any real AB testing of the sounds of the various tube manufacturers I have (need 4 identical amps, or 2 for mono). Otherwise, the comparison relies on memory, not what I consider valid. I do not remember any brand that caused me to say, hey, that is to distorted, flabby, rolled off, etc.

I almost always used a 2 resistor artificial center tap for DC filaments. I usually used 25 Ohm resistors. The result is 12.5 Ohms, and at a u = 3.85, the Increased plate resistance, rp = 48 Ohms.
 
Blitz,

I found more differences in sound with other changes to the amp.
5k OPT, 3k OPT, 350V @ 50mA, 300V @ 60 mA, different driver tubes, interstage transformer, RC coupling, etc.
But to make a conclusion about what was best of all that over a couple of decades would be a stretch.
And there were different types of DHT: 2A3, 6A3, 300B, and even 45.
I can say that I enjoyed listening to my DHT amps.
And now I enjoy listening to my 'indirectly heated' amps.
 
Thanks a lot for sharing your experience...I am a bit new in DHT territory...I consider to heat the 300B section now with AC like Kynn Olson did in his Karna...basically using the CT-of the filament transformer as the electrical CT of the 300B, so no additional two resistor to create a CT needed...
 
Blitz,

Some 300B tubes on AC filaments have more hum than others (different manufacturers; and tubes from the same manufacturers). Differential stages and Push Pull stages do tend to cancel that hum.

Lynn Olson has made many very good sounding amplifiers.

AC filaments on DHT, particularly on single ended amps does cause intermodulation of 2 times the line frequency. This was the primary reason I prefer not to use AC filaments on SE DHT.

I did use AC filaments on Push Pull DHT.
And yes, the center tap of a quality transformer eliminates the need for the 2 resistor 'center'.
 
Blitz,

I will list one failure mode of the early Svetlana 300B (I do not believe there were any later Svetlana 300B tubes in production because of the high failure rates of the early ones).

A particular and peculiar failure mode of Svetlana 300B happens with DC powered filaments, and not with AC powered filaments.

On some/many Svetlana 300B tubes, the spacings of the filament to the plate were not equal from one side versus to the other side.
The plates are magnetic.
1.25 Amps DC through the filament has a strong steady magnetic field.
With uneven spacing, the filament is attracted toward the side of the plate that is closest to the filament.
The filament now contacts one side of the grid wires.

That 'bowing' of the hot filament toward the plate causes the filament to contact the grid.
When the filament contacts the grid, the tube is at Zero Bias.

Zero Bias causes high current. The contact is continuous, and the current is continuous.
The output transformer saturates, and the only limitations of plate current is the Zero Bias, the DCR of the output transformer and the power supply current and voltage capability.

Some other manufacturer's 300B may also have uneven spacing of the filament to plate.
But in that case, it is also possible that the filament alloy that Svetlana used has more expansion (and bowing) than other 300B brands. That would also cause the same Zero bias problem and resulting failure.

Does that explain one of the main failure modes of the Svetlana 300B?
(we had 50 or was it 100 of them in an audio club, and we had lots of experience).

You could use AC powered filaments on the remaining Svetlana 300B tubes.
(I would not use the ones that already failed, even if they now seem OK).
But using AC powered filaments on any 300B has the disadvantage of the 2X line frequency intermodulation on all the music frequency/tones. That is discussed in other threads on this forum.

A very interesting set of observations!

But I do not believe that ac-heating will allow safe use of SV300Bs.

To start at the beginning: Consider the cold resistance of the filament: it's about 0.8Ω for an EH/Sovtek, and similar for others, including SV300B. This means that the cold-start filament current is very much higher than the running current.

How much? Remembering that the peak ac voltage is 7.07V (for 5.0V rms) and allowing for a little drop in the transformer, then at startup the repetitive peak currents will be about

7.1V/1.0Ω: 7.1A peak, 5A rms [compared to a normal current of 1.2A dc or rms]

If the ac heat winding is part of a transformer that also feeds the B+, the HV will be unloaded at startup, and the voltage from ac heat winding will be higher still. Increases in voltage give square-law increases in heating power, during the cold metal phase.

This kind of startup current causes rapid temperature changes, sufficient to produce permanent lengthening of the filament. I checked this suspicion with Emissions Labs - EML, and they conformed that it is desirable to limit the startup current to 2x the running value, so 2.4A maximum for standard 300Bs, to avoid "saggy filaments". My current-driven regulators limit the 300B current to 1.2A at start-up.

I see the observation about magnetic materials in the construction, but I think that once the filament gets saggy, we are taking a risk with short-circuits in all events. The shorts do not need to persist for long to put the output transformer at risk.

High startup current greatly increases the risk of outright breakage of the filament, too, which certainly affects the EH, Sovtek and (at times) KR, and I believe that SV300Bs also suffer breakage.

Another service-life problem from ac-heat: The lifetime of the oxide coating is also at risk from the slack tolerances of modern mains supplies: voltage tolerance of UK mains is ±10% now, having been further degraded a few years ago. This may not cause rapid failure, but it certainly shortens the life of our expensive DHTs. Even ±5% is a major threat to long lifetimes: again referring to the express advice of EML, the tolerance of the heating voltage should be ±1%. Excessive voltage is especially bad: this rapidly increases the evolution of barium from the subsurface of the coating, leading to faster exhaustion of the finite supply, and diminishing the emissive capability.
 
You do not have to believe, but just my 2c.
For the last 14-15 years I use the second pair of =C= 300B (not "S" 300B). I listen to music for 3-4 hours each day, average.
I have no problems with NOS Svetlana, but I always use a very slow turn-on with the Siemens medical auto-transformer.
Heating is DC, my own, similar to Rod's; fet gyrator for f+ and ring-of-two CCS BJT for f-
 

Attachments

  • My SET`s.jpg
    My SET`s.jpg
    53.2 KB · Views: 300
  • Variable power transformer.jpg
    Variable power transformer.jpg
    325.5 KB · Views: 290
Oh yes, filament breakage in SV300B's is even a bigger issue than in others! What a pity.

What does it mean hot and cold in reality? It's nothing to worry about when used within the limits. If their tubes can't do that then they are not as good as the original. Period. If the anode current is concentrated in a smaller volume the tube will be a least just as linear or likely better!

I don't like seller's theories about filament supply because they are too simplistic and give a rather wrong picture about real performance and reliability.
It's the usual deceiving stuff repeated thousand times for reasons that have nothing to do with performance. They just want to keep things simple to encourage people to buy expensive stuff even if they aren't experienced. AC heating is worse in every performance aspect for sure. For both SE and PP operations.


I now use DC heating only even with indirectly heated tubes and in amplifiers using GNFB! It's just better.
 
Blitz,

Just more of my personal experience, and some reasoning and theory.

Most of my amplifiers are built on old amplifier chassis (Heath Kit Push Pull 6BQ5/EL84 Mono Integrated amps, and even a Dyna Stereo 70 which originally had four EL34s). I use the chassis and the original power transformer (I know, some people do not like the thought of taking an old amplifier and not restoring it). Instead, on amps that are not in great shape, I build a completely new circuit. To each his own.

Those amplifiers had 115V and 117V primaries. I always use quite a bit less current on the filament and B+ secondary windings than the original amp circuit, to make up for the additional primary heating and lamination core heating at my highest power mains voltage. I do use solid state B+ rectifiers instead of the original vacuum tube rectifiers, but I use Choke input B+ whenever possible, which reduces the peak current on the B+ secondary. If I can not get enough B+ with a choke input, I use a Cap input, but with the input cap that is 1/5 or less than the original input cap (also reduces peak current on the B+ secondary).

My power can be as low as 117V (winter, me and the neighbors with electric room heating, electric water heating, electric stoves, and most everything on at once) but my mains are almost always at 121V to 123V. I design the filament circuit so that filament voltages are dead on with the power mains at 122V.

The 6.3V filament secondaries are usually at from 6.7V to 6.9V. For AC filaments, I use series resistors (to provide equal voltages versus a real or artificial center tap) to drop the voltage to 6.3V, 5V, or 2.5V for the output tubes, and 6.3V for the driver tubes, with the mains at 122V.

For DC filaments, I use a schottkey diode bridge, cap input, series resistor, second cap in parallel with the filament. The resistor is set for 5V or 2.5V on the output tube as needed, with the mains voltage at 122V. Those 5V and 2.5V tubes have 1 or 2mV ripple across the filaments, most of which cancels because of the two 25 Ohm ‘center’ point resistors.

It should be noted that the series resistor(s) in both AC filament supplies and DC supplies does automatically limit the surge current to all of the filament types. The sum of the two resistors are equal to, or more Ohms than the cold filaments.
But as noted, the B+ starts a little higher with cold tubes, and there is a temporary very Slight rise of the filament secondary voltage.

B+ supplies, whether Choke input or Cap input . . . I design so that the highest mains voltage,
and unloaded B+ will not cause the filter caps to be near to their maximum voltage rating.
Choke input supplies that are unloaded will rise to the same peak voltage as cap input supplies.
the 0.9 x rms of choke input does not apply when it is unloaded, instead the Peak (1.414 x rms)
applies, just the same as a cap input supply.

Regardless of the original amplifier’s fuse for the power input, I do something more:
I use a fast blow fuse rated as small as possible that will not blow with the startup inrush current,
and put that in series with a slow blow rated as small as possible with the warmed up current draw, both with no music, and turned up loud.

Sagging problems were known at the very beginning of DHTs. Just look at the old tube data, the tubes are only to be used with the tube vertical, or for some tubes horizontal with the filaments aligned only 1 particular orientation.

Unless you are trying to squeeze nearly the maximum DC plate current, or using A2 grid current that causes extreme plate current, you can live with filament voltage that is little on the low side.
That can help for you to plan how to design for your +/-10% mains voltage.

For incandescent lamps, the life of the filament goes to 1/(fifth power of the voltage).
If you have 1.1 x the rated voltage, that means the life is only 0.62 versus what it would be at 1.0 x the rated voltage.
Unfortunately, I do not know a rule for the life of Thoriated Tungsten and for Barium Coated Tungsten filaments.

If I do use DHT tubes again, I will use DC filament heating. And yes, the inrush current will be limited.
 
Ok, Guys...got the point, will stay with Rod's DC heating...I changed my rectifier from 274B to 866A and got now a bit too much VDC of 457V at the plates with 55mA...I guess I will build some cathode resistors into the game to get Anode-Cathode-Voltage down and have some current-feedback...

I am as well using choke-input only...with the 866a anyhow...and I manually first heat the 300B up before the HV is applied...on top: a soft-start for 7 sec with a 1k5 resistor infront of the CT of HV to nicely treat the rectifiers...after that a relais switchs HV from 380V to 450V...makes sense, I hope.

I have rad some articles about lowering Heating Voltage and TT is behaving a bit more negative when underheated than oxid...I believe there was an Eimac white paper on it...but a bit undeheated typically gives a much extended tube life...why I used 4.9V for the 300B.

And yes, Ac-Voltages changes in the mains are really bad, even here in Germany.
 
Last edited:

PRR

Member
Joined 2003
Paid Member
> Here you go...scroll a bit down

That's for Thoriated Tungsten.

If we are still talking about the 300B, that is Oxide Coated. A very different material with very different wear-out.

However clearly some of you are not getting the life that a big well-made oxide can give. So then you have to ask: what did they do wrong? There's many possible screw-ups. I had never heard the DC magnet idea but I won't giggle at it. It suggests the filament supports are not as stiff as they maybe should be; but the present makers are not building "for themselves" (as WE was doing), but for immediate sales.
 
PRR,

I believe that Western Electric Leased the Theater Amplifiers.
That is like the Leased Telephones (built by WE too) we had at home when I grew up.
It used to save companies money and trouble to make a leased product that will never have to be repaired, or replaced.
Phones or theater amplifiers.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.