• 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 Load Line and Bias?

mbeards

Member
2012-03-18 10:54 pm
Hi Everyone,

I have been looking at schematics of some typical 300B SE amplifiers after pondering over building one myself.

I have noticed that many of these amplifiers bias the 300B's with fairly large plate current. Typically I have seen 350V supplies with 3.5K loads with a bias of around 60-70mA. This puts the tube very close to grid current. 40mA puts the tube much closer to the load line center.

Why do I not see more 300B's biased more in the center of the load line or with lower load impedances?

Also, any thoughts on a typical gain needed to drive the 300B tube well? I am looking at a calculated preamp gain of 20.
 

tomchr

Member
Paid Member
2009-02-11 12:58 am
Calgary
www.neurochrome.com
Which operating point you use depends on your design goals. If you're optimizing for power in a SE triode, you need to run the tube as close to the max anode dissipation as you're comfortable with. This means higher B+ voltage and higher current. That pushes you towards 400 V, 90~100 mA in case of the 300B.

If you're optimizing for low harmonic distortion, you'll need to run the tube with a higher load impedance. Many people (myself included) use OPTs with 5 kOhm primary impedance. This gives you lower THD but at the expense of slightly less output power. In my experiments, I found very small differences in max output power but big differences in the THD when I varied the load impedance (easily done at mid band by varying the secondary load) from 2.5 kOhm to 6.8 kOhm on my JJ 300B running at 90 mA, 400 V. So I settled on a 5 kOhm load. I get 10~11 W out at 400 V, 85 mA, 5 kOhm, 1 kHz, 3 % THD using JJ 300B tubes.

There are operating points between the "current starved" 350 V, 40 mA and the "max anode dissipation" 400 V, 100 mA that are used because of personal preferences. If you prefer a certain distortion signature, biasing the tube at some intermediate point between these two extremes may be a good option.

Note that some 300B tubes have issues with high B+ voltages. They tend to go into some sort of thermal runaway above 380~400 V.

~Tom
 
Last edited:

mbeards

Member
2012-03-18 10:54 pm
Thanks for the help, I had to refresh my SE load line knowledge. I have been looking at preamp stages for so long I drew it as an anode resistor which a transformer impedance is clearly not.

I have never worked with the 300B before, or any tube where the max anode voltage intersects the load line. 450V in the case of the 300B. To avoid destroying the tube, I assume you have to make sure not to drive the tube beyond this point? Thus, the intersection of this limitation in anode voltage determines the max output; rather than using the entire load line?

Thanks again for the help!
 

45

Member
2008-12-18 2:29 am
UK
I have never worked with the 300B before, or any tube where the max anode voltage intersects the load line. 450V in the case of the 300B. To avoid destroying the tube, I assume you have to make sure not to drive the tube beyond this point? Thus, the intersection of this limitation in anode voltage determines the max output; rather than using the entire load line?

Thanks again for the help!

No. The maximum rating for anode voltage is only referred to quiescient plate voltage. The tube can swing to higher voltage. No problem at all! This is true in general.
Best bias I found with (good chinese replica) 300B was 420V/75 mA and 5K load. It didn't sound good at low plate currents and/or low plate volges. In theory you could use it to replace a 2A3 at 250V/60 mA but I found that it didn't sound good, the 2A3 was quite better!
 

kmaier

Member
2007-07-12 1:42 am
Take a look here: Western Electric

I also would use a 5K load with 450V and lower current. You can get a solid 10-watts output and still keep plate dissipation at a reasonable level. Distortion will also be lower.

Driving the 300B is not a simple task... after some initial bread-boarding with a 300B, I've opted to use an interstage driver transformer.

Regards, KM
 

tomchr

Member
Paid Member
2009-02-11 12:58 am
Calgary
www.neurochrome.com
I also would use a 5K load with 450V and lower current. You can get a solid 10-watts output and still keep plate dissipation at a reasonable level. Distortion will also be lower.

That's my philosophy as well. However, not all 300Bs are created equal. Some have been reported to perform some sort of thermal runaway stunt at higher B+ voltages (>380 V). During the prototyping stage, adjust Vgk on the 300B to a value that's supposed to shut off the tube (say, -200 V) and verify that the anode current does indeed reduce to the uA range.

I use some JJ 300B tubes that are quite excellent. They're relatively inexpensive. Relatively... I've run those at over 400 V without issues. I recently tried some Fullmusic TJ 300B/n mesh plate. Nice tubes! But I think one of them is defective as it redplated and tried to run away at 400 V B+. At 380 V, 65 mA it seems happy... I'm working with the vendor to get a replacement tube.

My point is: Be a bit careful. When plugging in different tubes or testing your circuit for the first time, have the bias dialed to minimum anode current.

Driving the 300B is not a simple task... after some initial bread-boarding with a 300B, I've opted to use an interstage driver transformer.

I wholeheartedly agree. To drive a 300B, you need a driver with low output impedance that can deliver about 180 Vpp swing at low THD. This is no simple task. I have experimented with source followers (see Tubelab's PowerDrive) and cathode followers. After three years of experimenting and prototyping, I'm finally happy with my amp. You can read about it here:
http://www.diyaudio.com/forums/tubes-valves/218217-deathtrap400-pretty-damn-good-300b-amp.html

~Tom
 

kmaier

Member
2007-07-12 1:42 am
Hi Tom,

Some interesting points.... my 300B matched pair is WE. I've seen some thermal runaway (as you called it) on some RCA 2A3 tubes. Appears to be high sensitivity to grid circuit resistance, lowered it and the problem went away. Still, I would consider this a defective tube for practical purposes as the grid circuit resistance was less than half of the spec sheet's maximum. It was also isolated to two tubes (both RCA) out of more than 50 tubes tested in the breadboard prototype.

Driving the 300B... I consider a minimum Pk-to-Pk voltage closer to 250... as my bias point has a grid voltage closer to -100V (I also use self-biasing). Using a stout driver (non-transformer), I was not happy with the dynamic compression as the Pk-to-Pk signal gets closer to zero grid volts, the linearity suffered far too much (in comparison to a similar topology used for a 2A3 or 45 output stage).

My stock of tubes are all NOS... the only new production anything I have are EML solid plate 45s, which, IMHO, are not really a 45, they bias and perform close to a real one, but alas... you need to push them closer to 40ma of idle current for better performance, they're also butt-ugly and very sensitive to any mechanical vibration.

Saw your amp some time ago... :D I did a low-cost 45/2A3 design a few years back which used a cathode follower driver. Works very well for the 45/2A3, just not for the 300B (again, IMHO).

Regards, KM
 

tomchr

Member
Paid Member
2009-02-11 12:58 am
Calgary
www.neurochrome.com
my 300B matched pair is WE.

Dang. Nice tubes.

I've seen some thermal runaway (as you called it)

Yeah, I don't like the term either, but what seems to happen is that the anode gets rather hot (cherry red in some cases). This seems to cause more current to flow. Whether it's because of emission or impact ionization, I don't know. The increased current flow causes the anode to get hotter and the tube races for self-destruction.

So far I've been able to catch runaway tubes before they developed issues, but a warning about this seems in order.

Driving the 300B... I consider a minimum Pk-to-Pk voltage closer to 250...

Which only increases the driver design challenge.... :)

I did a low-cost 45/2A3 design a few years back which used a cathode follower driver. Works very well for the 45/2A3, just not for the 300B (again, IMHO).

In my experience, that depends on the cathode follower. I tried a 6SN7 and/or 6J5. Mushy sound. That tube doesn't have a high enough gm to work well as a cathode follower in a 300B driver (IMO). That's one of the reasons I ended up in ECC99 and 6N6P land.

How about a parallel 6SN7 triode driver for a low impedance source?

I take you mean a standard grounded cathode amplifying stage. The output impedance of that circuit will be fairly high (roughly (rp/2)||Rload) and the circuit cannot source current.

You need a driver with an output impedance of a few hundred ohm or lower. Lower is better. This implies that you're using a tube with a gm over 10k (Rout ~= 1/gm) at the operating point of choice for the cathode follower.

~Tom
 

mbeards

Member
2012-03-18 10:54 pm
I take you mean a standard grounded cathode amplifying stage. The output impedance of that circuit will be fairly high (roughly (rp/2)||Rload) and the circuit cannot source current.

You need a driver with an output impedance of a few hundred ohm or lower. Lower is better. This implies that you're using a tube with a gm over 10k (Rout ~= 1/gm) at the operating point of choice for the cathode follower.

~Tom


Yes. I was plaining to use an 6J5GT input (essentially 1/2 of a 6SN7) directly coupled to a zener biased 6SN7 with both plates in parallel. This gives an output resistance of around 3.3K.

I am surprised if this output impedance is to high to effectively drive the 300B. Where is all the the current draw coming from to load down the previous stage? The inter-electrode capacitance's of the 300B roll off the output beyond the audible frequencies and the gird leak resistor is somewhere around 300K.

or... am I way off?


Also, dare I say.. why use the 300B at all if it is such a dog of a tube drive?
(yet I want to build one :D)


Matt
 

kmaier

Member
2007-07-12 1:42 am
Tom,

Good post... re the thermal runaway. I don't know what else to call it either, but it seems appropriate. It could likely be some residual gas in the tube, but the getter flashing is fully intact on my 2A3 tubes.

The 300B bias point I'm using is 455V cathode-to-plate and requires 100 volts of negative grid bias with an idle current ~67ma. At a calculated minimum, 200V Pk-Pk is required for drive. Having decent headroom makes sense, as the driver stage does not reach it's limits when fully driving the output stage. You also have some class A2 capability with a well designed transformer-coupled driver.

EC99 - good tube. I also like the E182CC (7119) :D For the 45/2A3 amp, I used a 12AT7. First stage is a voltage gain amp direct-coupled to the second stage as a cathode follower. It has good linearity, wide bandwidth and low distortion and can easily swing over 100V Pk-Pk driving the 45 or 2A3. Again, the idea was low-cost and simple.

Mbeards - 6SN7 is a good tube, and many circuits about using one as an input/driver for the 300B. For under 8-watts output, it's fair, but limited. I don't like to parallel tubes, so I've taken a different path. Search the net for "bugle45 amp". There's a PDF article from Gordon Rankin for his Bugle 45 SE amp. Good article and has some good formulae in the sidebar for calculating slew rates and bias currents for driving the grid, bypass values, etc.

Regards, KM
 

tomchr

Member
Paid Member
2009-02-11 12:58 am
Calgary
www.neurochrome.com
Yes. I was plaining to use an 6J5GT input (essentially 1/2 of a 6SN7) directly coupled to a zener biased 6SN7 with both plates in parallel. This gives an output resistance of around 3.3K.

I like the 6J5GT as well. I have a handful of Pinnacle 6J5 tubes.

To avoid blocking distortion, DC coupling to the 300B is needed. This either means that you're using a plate choke on the driver or using a cathode/source follower.

I tried using a d3A in triode with a 70 H plate choke. That works really well, but to minimize the THD of the driver, a variable cathode supply is now needed. Each tube needs it's own supply and the supply voltages are rather different due to tube-to-tube variation. Sounds good, though.

Due to the supply complexities, I opted to use a source follower. Now the challenge becomes finding a good tube for this. One that can handle the signal swing and has high gm. That means d3A, ECC99, 6N6P, and 12BH7A.

My goal is to get as clean of a sound as possible. If you're more into "sound shaping" than I am, then the 6SN7 or 6J5 are good candidates. It's also possible that they'd work well in a configuration where the 6J5 provides the gain up front and a high gm tube is used as a cathode follower. I know they work well with a source follower. BTDT.

The 300B is drawing a little bit of grid current as Vgk approaches 0 V. If your driver can't deliver this current - or it results in a signal dependent voltage sag across the output impedance of the driver - high THD will follow. With a 6J5 grounded cathode driver, you'll probably get 5-ish W at 3 % THD. I seem to recall measuring 0.6 % THD @ 1 W in such an amp. With a driver with lower output impedance, you'll be able to get more power out. That's how I'm getting 10~11 W @ 3 % THD and 0.1~0.2 % THD @ 1 W.

Also, dare I say.. why use the 300B at all if it is such a dog of a tube drive?
(yet I want to build one :D)

Yes. Some of us are masochists... :devilr:
Some of us get into 300B designs because of the sound quality of such designs. We later realize that it's a royal pain to drive that tube. Once I came to that realization and wrote down the requirements for the driver circuit, the circuit topology became rather obvious and only a small handful of tubes fell out as possible candidates.

I actually rather enjoyed the challenge and design process. But then I'm used to the multivariable optimization problems bound by the laws of physics from my day job. :)

I suspect some like the 300B amps because with a non-ideal driver, you can get a sweet tone from it. That's the "sound shaping" aspect. Each to their own I guess.

~Tom
 
I seem to recall measuring 0.6 % THD @ 1 W in such an amp. With a driver with lower output impedance, you'll be able to get more power out. That's how I'm getting 10~11 W @ 3 % THD and 0.1~0.2 % THD @ 1 W.

~Tom

Hi Tom,

The 0.6% you mention is that at 1kHz?

I am busy with my first built, a Herb Reichert "Flesh and Blood" derived 300b so 6SN7 GTB cascade. The power supply is of my own design, a low DCR design with four 5V3 tubes and a CLCLC with 150mH and 2.6H chokes optimized for low impedance and fast impulse recovery.

It is currently on auto bias but I will change that to fixed bias once I have the EML 300b-XLS (500VDC B+).

What do you think of the distortion plot taken at 1 W and 84mA plate current? Is it normal that the second harmonic is curved like it is?

Regards,
Peter
 

Attachments

  • Distortion-Frequency plot Auto bias 84mA @ 1 watt 482V anode-50%.jpg
    Distortion-Frequency plot Auto bias 84mA @ 1 watt 482V anode-50%.jpg
    180.8 KB · Views: 674

tomchr

Member
Paid Member
2009-02-11 12:58 am
Calgary
www.neurochrome.com
The 0.6% you mention is that at 1kHz?

Sorry for not being specific. Yes. All distortion numbers are THD+N at 1 kHz measured using an HP 8903A.

What do you think of the distortion plot taken at 1 W and 84mA plate current? Is it normal that the second harmonic is curved like it is?

I haven't looked at the individual harmonics in detail in my amp. But from your plot, the distortion is clearly dominated by 2nd and 3rd harmonics. That's the case in my amp as well. I would expect the THD to go up at lower frequencies as the OPT gets closer to saturation and any magnetic nonlinearities would tend to show here. I seem to recall my design with 6J5 having higher THD at higher frequencies as well, but I'd have to check my notes to make sure. Bottom line: Your 2nd and 3rd harmonics look normal to me.

In my current design, the THD is highest at low frequency, flat (at 0.15~0.2 %) from 200-ish Hz through 10-ish kHz. then it drops slightly at higher frequencies. The bandwidth of the amp is 32 kHz.

In your plot, the fourth and certainly fifth harmonics seem to be below the noise floor of your equipment. I'm guessing, but that's what the graph looks like.

~Tom
 
Bottom line: Your 2nd and 3rd harmonics look normal to me.

Thanks for confirming that I was a bit worried about the high frequency distortion level.


In your plot, the fourth and certainly fifth harmonics seem to be below the noise floor of your equipment. I'm guessing, but that's what the graph looks like.

Yes, my 192kHz soundcard broke down so measurements were done with a 96kHz soundcard sampling rate. I just bought a second hand EMU 1616 as replacement but I am currently sailing in the Indian Ocean. I will be back home in three weeks time then I can do some better measurements and install the 300b-XLS and connect the fixed bias.

I have a Tentlabs bias module but did all my simulations with autobias not realizing that the Tentlabs is of course a fixed bias. Stupid mistake, now I have a anode voltage of 500VDC in fixed bias, to much for a standard 300B. The EML 300B-XLS should work well on that voltage.

The low 2nd harmonic at the mid freq. band could be the result of cancellation, the 2nd stage 6SN7 actually has quite high 2nd harmonics distortion.

I can start a new thread if anybody is interested, as said the circuit is reasonable standard except for the low DCR power supply.

Peter
 

Attachments

  • FFT plot Auto bias 84mA @ 1 watt 482V anode-50%.jpg
    FFT plot Auto bias 84mA @ 1 watt 482V anode-50%.jpg
    129.5 KB · Views: 665
  • Frequency plot Auto bias 84mA @ 1 watt 482V anode.jpg
    Frequency plot Auto bias 84mA @ 1 watt 482V anode.jpg
    120 KB · Views: 656
  • 2nd stage Distortion-Frequency plot Auto bias 84mA @ 1 watt 482V anode.jpg
    2nd stage Distortion-Frequency plot Auto bias 84mA @ 1 watt 482V anode.jpg
    161.7 KB · Views: 649