DIY Audio Projects Forum

"The Beast" 6336 Dual SET Amplifier
Page 4 of 7

Author:  ToddD [ 08 Jan 2019, 21:28 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

First build question.
How big is too big when it comes to B+ voltage?

Along with the recommended GZ34 rectifier tube I also bought a 5U4GB and a 5Y3GT. I have never had a chance to try different rectifier tubes before and I wanted to try for myself. I measured voltages with the different rectifier tubes.

At the wall, I have 124Vac. Referencing the schematic, B+power voltage measured after the 3rd/4th cap in the PSU. B+ driver voltage measured after the 33uf cap.
With the GZ34, for B+ power tube= 374Vdc, 110mA, for B+driver = 313Vdc, 1.55mA
and for comparison
With the 5Y3GT, for B+power tube = 356Vdc, 104mA, for B+driver = 298Vdc, 1.5mA
With the 5U4GB, for B+ power tube = 346Vdc, 102mA, for B+driver = 289Vdc, 1.5mA

If I use the GZ34, should I get the B+power down to 360Vdc? Spec sheet for the 3663A list 400Vdc Max on the plate voltage. I think I know the answer, but I would like confirmation. I have been using the 5U4GB for now.
I think I am okay with the measured B+driver voltages because of the plate load resisters, but Iet me know if that should be a concern.

Feedback would be appreciated.


Author:  ILoveHiFi [ 09 Jan 2019, 20:01 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

You can have a voltage dropping resistor then short it by a chassis mounted switch this way you can accomodate for multiple types of rectifiers.

If you want over kill absolute reliability solid operation then I would suggest 200v.
"If you are wondering why it appears that the signal voltage can now swing higher than the HT voltage, it is because this is exactly what happens! Inductances abhor changes in current. When current through the transformer increases it stores energy, which is released when the current falls again, allowing up to twice the HT voltage to be developed."

Author:  Suncalc [ 09 Jan 2019, 20:50 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

ToddD wrote:
If I use the GZ34, should I get the B+power down to 360Vdc? Spec sheet for the 3663A list 400Vdc Max on the plate voltage. I think I know the answer, but I would like confirmation. I have been using the 5U4GB for now. I think I am okay with the measured B+driver voltages because of the plate load resisters, but let me know if that should be a concern.
Ok, lets break this down logically.

First, don't worry about the driver stage. All these voltages are just fine.

Now, the design point for the power stage is Vp=261v and Ip=90mA. This puts plate dissipation at about 23.5W (i.e. acceptable). Since this is a self biased amp, the total B+ circuit (transformer, tube, and bias resistor) need to be taken into account for B+ determination. The transformer has a 94Ω primary resistance and therefore a voltage drop of 94Ω*0.090A≈8.5v. The bias voltage is 0.090A*910Ω≈82v. So the required B+ voltage is 261v+8.5V+82v=351.5v. This is about what is on the load line design sheet (1v different). Whatever you do to the component values, this approach remains the same.

Now let's look at your GZ34 measurements. B+=374v and Ip=110mA. You used an 820Ω bias resistor so the bias voltage should be 0.110A*820Ω=90v. The transformer drop is about 94Ω*0.110A=10.3v. This makes the plate voltage 374v-10.3v-90v=237v and the plate dissipation 237v*0.110A=30.1W. This is too hot to run the 6336. In my build, I ended up with about 200Ω of series dropping resistance in the power supply to get the B+ down to where I wanted it with the GZ34.

So you have two choices, stick with the 5U4GB and be happy, or add some series resistance to get the B+ down and use the GZ-34. In the end, it's really about the plate dissipation of the power tube. The design point runs the power stage at about 80% of the rated plate dissipation. Under these conditions the 6336A tubes should last a good long time. At 30.1W you'll burn (literally) through power tubes much more often. Since these are generally pretty expensive tubes, I recommend lowering the plate dissipation (by lowering the B+) if you want to use the GZ-34 rectifier.

Author:  ToddD [ 10 Jan 2019, 19:33 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

Thanks for the response and more importantly thanks for the math.

With the 5U4GB I have a plate dissipation of 25.8W. I will leave that in for now and order some dropping resistors. 470 ohms should get ~ 24.5W plate dissipation with the GZ34.

Follow-up thought,
I also have some 1000 ohm x 10watt resistors. I could increase my cathode resistor and change my bias some which I think will lower my current draw. In this case would there be a preferred direction between higher voltage and lower amps vs. lower voltage and higher amps? Just curious.

Author:  Suncalc [ 10 Jan 2019, 20:33 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

ToddD wrote:
In this case would there be a preferred direction between higher voltage and lower amps vs. lower voltage and higher amps?
In terms of plate dissipation it's really about energy balance. The plates are heated up literally by the bombardment of electrons. The plates cool by means of black body radiation (electromagnetic emission). By design, the plates are manufactured to be able to radiate a certain amount of power at equilibrium. This is set so that long term power flow into and out of the plate structure keeps the plate temperature below an acceptable limit. So at the temperature limit, the plates can radiate a maximum amount of energy per unit time (i.e radiated power flow). If the plate dissipation is below this number you are good. The temperature limit is set based on the allowable outgassing rate of certain coatings; but this outgassing is not linear with temperature. The maximum temperature is set to keep below the point where the outgassing grows to an unacceptable level. Contrary to popular opinion, there is no benefit to the plates or the tubes in limiting the plate dissipation to some low value. Once plate dissipation is below about 90% to 95% of max, the plates will outlast all of us.

So it's really about where you are placing the operating point and load line. At a set load impedance, there is benefit in keeping the current higher at a fixed plate voltage thereby reducing distortion. On the other hand, by reducing the current (again at a fixed load and plate voltage) the bias increases. This allows you to get larger voltage swings in the output transformer primary and hence more power (provided that you don't drive the tube into cutoff). So there is a trade off between distortion and max power.

Power stage design is complicated. :D

Author:  ToddD [ 02 Feb 2019, 10:54 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

Update on my Beast;
As I have said previously, this amp is in my garage system so the updates and observations will come slowly. I mainly only play with it on the weekends and it is January/winter. After getting a good amount of hours on the new build amp, my initial observations are the same. This amp has a lot of grunt and has a full body sound. It has a more pronounced low frequency control than my previous amp (Bottlehead S.E.X amp, a 6DN7 SET with ~2 watts/ch). I use plate amps and sealed bass boxes to supplement the lower octaves and I have needed to lower the gain and crossover frequency on the plate amps with the Beast.

But I have done some changes and have some comments.

- I have stayed with the 5U4GB rectifier for now. I think I like the sound with the lower B+ better, but I like that I can get more B+ with the gZ34 if I change my mind..
- I did try and change my first cap in the power supply from a 10uF to a 22uF The 22uF raised the B+ about 5 volts. I went back to the 10uF for now.
- I changed out the 20K resistor in the driver B+ power supply and replaced it with a 3.3K resistor. More on the choice of value later. This raised the B+ for the driver from ~290 VDC to 339 VDC. It sounded a little cold and sterile at the lower voltage. The higher voltage has a better midrange. It sounds more tube like. So if you choose to build the Beast and have concerns with the initial sound it is worth playing around with the driver B+.
- Replaced the 12AU7 tubes in the driver stage with 12BH7 tubes. This was also a positive improvement. The 12BH7 is a musical tube. Easy change and worth it.
- I replaced the 3.3K resister (which was the 20K) in the driver B+ power supply with a Hammond 156C choke ( 150H, ~3.3K ohms). This is still breaking in, but I am optimistic. I generally have good results when replacing resistors with chokes of the same resistance.

More changes to come. I am enjoying the process and I have learned a lot with this build. Thanks again to Suncalc.

Author:  ToddD [ 02 Feb 2019, 11:11 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

And build question:

I did measure my gain through the driver stages. I used a 1Khz signal at 0.505VAC at my input. I got 6.68VAC out of the first stage and 47.7VAC out of the 2nd stage. That is a gain of 13.23 from the first stage and a gain of 7.14 from the 2nd stage. A gain of 94.5 overall from the two stages.

Would it be better to have the cathode bypass cap on the 2nd stage instead of the first? My understanding is that you can have lower distortion without the bypass cap. If so a higher gain of lower distortion seems better than a lower gain of a higher distortion. Also with a higher input signal into the 2nd stage will it have a greater chance of overdriving the 2nd stage? Not sure if overdrive is the correct term here. I was just wondering. It is an easy enough change and one that I may try.

feedback appreciated.

Author:  Suncalc [ 02 Feb 2019, 15:59 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

Ok, a couple things to address.

First, the distortion. The thing you are missing is that distortion in triodes is primarily 2nd harmonic which is roughly proportional to signal level. The first triode is going to be providing much lower signal swing than the second stage. As such, the natural distortion in the first stage is much lower. The second stage is unbypassed to do two things. First, this lowers distortion. This is important in the second stage because it has such a greater swing than the first. Second, the unbypassed cathode increases the allowed input swing before the tube overdrives. The second stage can take an grid input greater than its DC bias. I think the limit is around 13.6V. This is why the cascaded design can produce over 77v RMS (almost +/-110v peak) with reasonable distortion levels. See my blog post here for performance details on this driver:

Second, about overdrive. When designing a signal chain, it is important to pay attention to when each stage may go into overdrive (grid conduction or compression). Since the goal in a power amplifier is to enable the maximum possible performance from the power stage, you want to design the signal/driver stages to stay clean and reasonably low distortion when the power stage is just at the edge of overdrive. The "Beast" power stage is biased to about 80v. This means that any signal greater than this will drive the power stage into grid conduction and start introducing odd harmonics as the peaks clip. So let's walk back through the signal chain.

You said your second driver stage has a gain of about 7.1. So to get 80v out, you need an input of 80/7.1≈11.3v peak. The second stage will remain relatively clean at this drive level; especially with the cathode degeneration. So the first stage needs produce an output of 11.3v peak. You said that yours had a gain of 13.2 so you need an input of 11.3/13.2≈860mv peak or about 600mV rms to get this output. This is a very easy signal for the first stage to handle.

So using your data we can see that the signal/driver stages will perform just fine up to and beyond the point where the performance of the power stage is maxed out.

Does all this make sense?

Author:  ToddD [ 03 Feb 2019, 16:30 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

If I understand correctly, the distortion is more effected by signal strength than gain. So the distortion of the first stage will still be less than the second stage regardless if the gain of the first stage is more than the of the second.

So I continue to let it play and I did change out the cathode resistors on the output stage from my original 820 ohm to the spec'd 910 ohm. So now I match the signal schematic exactly except I used 0.1 uF coupling capacitors instead of the spec'd 0.068uF caps. I used a 3.3K grid resistor into the output stage. My power supply is very similar to the schematic, but I used smaller value motor run capacitors.

I asked my wife her opinion and she is very good at quick and succinct feedback. She says compared to my other amps, it sounds muffled and is not dynamic. "It has no passion". I think the last comment was directed at the amp. I agree that it does sound muffled and is not as dynamic. No comment on the passion. It sounds like someone lowered the treble on the tone controls, but I don' have tone controls. It does not have the shimmering, twinkling, sparkling treble. I know that your original amp had roll off on the high frequencies and you suggested a lower value grid resistor for the output tube. I used a 3.3K resistor. What else might I consider?

I have to wonder if I am not having a Miller effect going on, but I don't know how to verify. and I am not that clear on the Miller effect or if it would cause this. Not sure if the high gain of the driver stage or the output tube would be more at fault or susceptible to this. I should have enough hours on it by now for all the chokes, OPT, and caps to have broken in and the mids and bass sound fine.

feedback appreciated.


Author:  ILoveHiFi [ 03 Feb 2019, 18:14 ]
Post subject:  Re: "The Beast" 6336 Dual SET Amplifier

If you make the grid resistor too large it acts like a low pass filter and high freunecy gain is reduced.

Page 4 of 7 All times are UTC - 6 hours [ DST ]
Powered by phpBB® Forum Software © phpBB Group