Megatron Electrostatic Headphone Amplifier

[kevin gilmore]

somewhere there is a mini board that is a stn9360 to to126 pinout.  but bending the pins is something i have also done.

[G600]

Wow have I ever told you that you are a bold man, Joachim?

Impressive, again!

[Shawn]

Finally, I have time to hand in the Megatron XL. Having some questions here.

What`s a decent static plate current and voltage for 300B?

For the tube rectified power supplies, can we do a solid-state rectifier instead of the GZ34/EZ81?  Assumed we have other soft-start high-voltage circuits populated.

I'm wondering if it's possible to use the Golden Reference HV supply to provide ±450for XL. Does anyone know if the current capability is sufficient?  I'm concerned the total load might be too much for this PSU.

Edited May 6 by Shawn

[MLA]

Regarding you last question, I'm using GRHV in that role so I would say it's doable, but in addition to adjusting the current limiting resistor you need to put some protective measures in place; otherwise there is a good chance the SicFet will blow up during startup (I killed 4 before understanding the problem  ). 

See SimmConn's post for great advice: 

 

[Shawn]

MLA, Thank you for the reference.

Since the C2M1000170d is ending the life cycle, I don`t want to trash them.😅

I may just go with the universal power supply which was used on Megatron before.

2n3904 current limiters needed to remove for the universal supply. I`m wondering if the 2n3904s in GRHV do the same purpose? if yes, can we remove them to increase current?

The single universal power supply worked fine on my previous Megatron build(Each El34 draw about 27ma). And I assumed each 300B draw 33mA as Kevin mentioned it is the sweet point for 300B. Don`t know if I calculate it correctly.

 

 

Edited May 7 by Shawn

[JoaMat]

2n3904 does the same thing. Can be removed. Anyhow it won’t protect you from sudden shorts.

[Shawn]

On 5/6/2025 at 10:24 PM, JoaMat said:

2n3904 does the same thing. Can be removed. Anyhow it won’t protect you from sudden shorts.

JoaMat always helps. I appreciate the input regarding the 2N3904. I've been considering a modification to handle higher output current more safely. Would it be feasible to add an IXTP08N100D2 (Depletion Mode MOSFET) before the C2M1000170D to serve as a hardware current limiter?

Drain of IXTP08N100D2 connects to the Source of C2M1000170D (main current path).

Source of IXTP08N100D2 connects to a sense resistor Rs, then to GND.

Gate of IXTP08N100D2 is tied to GND (0V) or through a simple RC filter to mitigate noise.

The current limit would then be:

I = Vgs(th)/Rs.

Would this be a viable approach to implement a basic current limiter before the C2M1000170D, or would there be better alternatives? 

 

Datasheet attached below:

https://www.mouser.com/datasheet/2/240/media-3320804.pdf

Edited May 10 by Shawn
Fixed. Lower the current limit resistor to increase current.

[JoaMat]

Did this today today – in full dayligt. Replaced EML 20B-V4 with 2A3 tubes in a Megatron with solid state constant current sources. Changes are filament voltage and cathode resistors (from 640 ohm to 3600 ohm. Only light is from blue and red LEDs and some light from the tubes themself.

[Shawn]

The latest version of Megatron XL comes with 4 pots each 300B.

What are those pots adjusted for(cycled in the pic)? I'm not familiar with the DHT tube.☹️

I assumed they are for Grid voltage adjustment.

[MLA]

Yes; gives roughly +/-4V of adjustment range in my build. Can be upped if larger range is needed.

[Shawn]

7 hours ago, MLA said:

Yes; gives roughly +/-4V of adjustment range in my build. Can be upped if larger range is needed.

Thanks for your help, MLA.

Another thing I’d like to confirm: based on the 300B datasheet (see chart below), I’m planning to use a 400V B+ supply. According to the curve for Eb=400V, it looks like a plate current of 70mA corresponds to a grid bias around -85V.

Would you consider this a good operating point?

[MLA]

I'd say that's a high operating point for this amplifier; the math of push-pull output stages is different from single ended and as long as the transition is smooth you can pick much lower operating points. See for example:

https://www.valvewizard.co.uk/pp.html

Sorry, I should have remembered; Megatron is NOT push-pull. 

Megatron (XL) has about 22-23mA through the EL34 current sources. I've tried up to ca 30mA but not beyond that.

Edited May 21 by MLA

[Shawn]

24 minutes ago, MLA said:

Megatron (XL) has about 22-23mA through the EL34 current sources. I've tried up to ca 30mA but not beyond that.

Thanks for the feedback! I wasn’t aware of the differences in operating point strategy between push-pull and single-ended, so I’ll definitely take the time to study the article you linked—much appreciated.

Also, if it’s not too much trouble, could you share what grid voltage you typically use for the 23mA setting? And when you went up to around 30mA, did you notice any significant sonic differences?

Thanks again for the quick and helpful reply!

[simmconn]

The Megatron final stage is CCS-loaded single-ended output working in class-A. There is no problem selecting operating point that way. However the final stage is self-biased. The grid bias eats up part of the B-, in other words B- is not equal to Eb in the tube datasheet. 20 to 30mA is more than enough for an estat amp.
 

Since 300B has a low mu (2.85), which is about 1/3 to 1/4 of a triode-strapped EL34, the undistorted output voltage will be less than with EL34. But the DHT fame and the aesthetics probably more than compensate for that.

[Shawn]

9 minutes ago, simmconn said:

However the final stage is self-biased. The grid bias eats up part of the B-, in other words B- is not equal to Eb in the tube datasheet. 20 to 30mA is more than enough for an estat amp.

If the output tube (like 300B) is self-biased, should I measure the voltage between the cathode (K) and the control grid (G1) to determine the actual grid bias (Vgk), and then adjust the trimmer resistor accordingly?

I understand that in a fixed bias setup, you’d measure between GND and G1, but in a self-bias configuration, G1 might be at 0V and the cathode elevated — so the true bias would be relative to the cathode, right?

Just want to make sure I'm adjusting based on the correct reference.

Thanks in advance!

[simmconn]

Ok, I stand corrected. The Megatron XL uses a 'hybrid' biasing scheme for the 300B. The grid is adjustable between 0 and 0.99% of B- ('fixed'-biasing), which cancels out less than 10% of the bias generated by the cathode resistor (self-biasing).

The voltage between the cathode (K) and the grid (G) is the actual grid bias (Vgk), regardless of self-bias or fixed-bias. In a fixed bias setup, you’d measure between GND and G when K is grounded.

[Shawn]

This post is a bit long, but if you're interested in alternative filament supply ideas for DHTs, feel free to read on.

I know @kevin gilmorehas already mentioned and implemented a solid filament supply solution for 300B tubes using five modular switch-mode power supplies from Traco. One 12V unit is used for controlling soft start, and the others supply 5V for the 300B filaments. It’s a good and proven design. However, I believe it may take up more space than I can afford in my chassis, so I’ve been working on an alternative.

Summary of My Design Concept:

1. Inspired by @JoaMat approach using rectification + DC-DC regulation, aiming for a similar or smaller footprint.

2. Native soft-start behavior with no need for external resistors to set ramp-up current or output voltage.

3.Up to 5V @ 3A output per module.

4. Molded inductors for reduced EMI

5. Adjustable switching frequency (to avoid audio band).

6. Dual-random spectrum modulation for further EMI suppression.

7. No heatsink required.

 

I’m currently working on the PCB layout based on my schematic (slow progress – I’m not a professional). Any suggestions, corrections, or feedback would be greatly appreciated!

1.Design Rationale (skip if you’re only interested in the result):

There are many DC-DC converter chips available, but my criteria were:

Very small footprint

Adjustable switching frequency (to avoid 20Hz–20kHz band)

Integrated soft-start, over-voltage, and over-current protection (to safeguard DHT filaments)

Output current ≥2A @ 5V

Low noise and ripple

Adjustable output voltage (to support both 2.5V for 2A3 and 5V for 300B)

 

After some research, I settled on the TI LMQ664x0 series, which offers 1A/2A/3A variants. To leave some headroom, I chose the LMQ66430, rated for up to 3A output and 36V input.

Datasheet:https://www.ti.com/lit/ds/symlink/lmq66430.pdf

This chip is extremely compact – only 2.6mm x 2.6mm – though I’m still unsure if I can hand-solder something that small.☹️

2. Circuit Topology (text version for now):

Although I’ve drawn a test schematic, the files are on another computer. I’ll post them later if needed.

In general, the topology is quite typical: 

Use a transformer with a secondary in the 9–16V AC range, Rectify the output, Feed it to the LMQ66430 module.

Key connection notes (based on the datasheet):

PIN 1 and PIN 3: Rectified DC input

PIN 6: GND

PIN 7: Output through an inductor (preferably molded)

PIN 8: NC

PIN 10: 1μF to GND

PIN 11: Feedback resistor network for setting Vout (see page 29)

PIN 13: Set switching frequency (tie to GND for 2.2MHz; see page 14)

PIN 14: NC

PIN 15: GND

For layout reference, check Page 39.

Also, combinations and example circuits are shown in Page 9.2

 

Disclaimer:
I haven’t verified every detail yet. In this way, there may be changes before I finalize the layout and perform live testing.

I have some concerns if two modules can work in parallel. In Megatron XL, the two 300Bs per channel share the same filament supply. If I use two separate LMQ66430s in parallel, I’m unsure whether uneven current draw might become a problem, since these chips don’t have a master/slave current-sharing mechanism.

I have another separate question, please bear with me for bringing this up here:

If I substitute 300B with EML 20B-V4, I understand that I no longer need the pair of 3300Ω balance resistors and should instead use a cathode resistor connected to the center tap (CT) of the filament.

But I’m unsure about how to treat the G-stage resistor R4 (680kΩ). Should it connect from the grid to the cathode, or from the grid to -450V?
My guess is that the G-stage connects to -450V through the 680k, and the cathode bias is set independently via the CT, not the filament ends. I admit I’m not very confident when working with DHTs – I often get confused by how CT-based cathode bias works when combined with DC filament supplies.

Any insight would be really appreciated!😊

[Zhang]

my friend see your post at head-fi planing chasis for megatron and look very nice

look same as chasis you build for kggg

look like good chasis many use
can you tell where to buy from?
can you share the pcb files for the custom pcb in kggg?
 

[kevin gilmore]

so that chassis in various sizes was available on ebay. i do not know whether it is still available. will have to look later.

i know that some people think the tent labs filament supply is gods gift to dht tubes. real fact is that it does not really like hanging off -450v. as it was never designed to do so. there are better, cheaper and smaller ways of doing this.

Edited Wednesday at 09:41 AM by kevin gilmore

[Shawn]

2 hours ago, Zhang said:

my friend see your post at head-fi planing chasis for megatron and look very nice

look like good chasis many use
can you tell where to buy from?
can you share the pcb files for the custom pcb in kggg?
 

You should easily find it on eBay or somewhere else.

Regarding the file, it is not mandatory for GG. You should complete the build by populating the GG boards only. Otherwise, you need to figure out the drilling distance to perfectly make the center board fit well, which is unnecessary.

[Zhang]

13 minutes ago, Shawn said:

You should easily find it on eBay or somewhere else.

Please can you share the link? my friend canot find the right one

13 minutes ago, Shawn said:

Regarding the file, it is not mandatory for GG. You should complete the build by populating the GG boards only. Otherwise, you need to figure out the drilling distance to perfectly make the center board fit well, which is unnecessary.

is center board unnecessary?? is the same true for the board mounted on the sides?

[kevin gilmore]

quotting shawn

"I have some concerns if two modules can work in parallel. In Megatron XL, the two 300Bs per channel share the same filament supply"

no they do not. each output tube (all 4 of them) has its own filament supply. they have to be seperate supplies for the 300b because they are in the audio circuit.

in the d&g i use 2 switching modules in series to get the equivalent for 10v.

Edited Wednesday at 07:28 PM by kevin gilmore

[Shawn]

6 hours ago, kevin gilmore said:

no they do not. each output tube (all 4 of them) has its own filament supply. they have to be seperate supplies for the 300b because they are in the audio circuit.

in the d&g i use 2 switching modules in series to get the equivalent for 10v.

Thanks Kevin. I went back and double-checked the schematic. You're right, each 300B does have its own separate filament supply.

In this way, would it be possible for me to use four individual 2A/3A DC-DC modules to power them separately?

BTW, how’s the D&G project going? I remember it requires something like 10V at 8A or even 10A for the filament supply. That’s already more current than a typical household socket can handle for just the tubes! I’d love to hear more about the D&G if you’re open to sharing.

14 hours ago, Zhang said:

Please can you share the link? my friend canot find the right one

I got the chassis at the end of 2024, the link was expired for a while but i do believe you can find it somewhere else.

14 hours ago, Zhang said:

is center board unnecessary?? is the same true for the board mounted on the sides?

For GG building guide, you can browse the GG thread, it has all the info and issues you may like to know.

[Zhang]

6 hours ago, Shawn said:

Thanks Kevin. I went back and double-checked the schematic. You're right, each 300B does have its own separate filament supply.

Where did you found the schematic?
We canot find anywhere in the thread.

6 hours ago, Shawn said:

I got the chassis at the end of 2024, the link was expired for a while but i do believe you can find it somewhere else.

maybe you can share the link to the classified?
sometimes one can find new sellers searching the pictures.

6 hours ago, Shawn said:

For GG building guide, you can browse the GG thread, it has all the info and issues you may like to know.

I am deeply sorry I meant the to ask if the center board unnecessary?? is the same true for the board mounted on the sides? if unneccessary why are they there??

 

[JoaMat]

On 6/4/2025 at 9:06 AM, Shawn said:

Adjustable output voltage (to support both 2.5V for 2A3 and 5V for 300B)

Nice idea. If you can implement an auto detect function as well – like Elekit has done with TU-8900. 

Power module…
 
filament supply in black square,

and the tiny - 2mm x 2mm - step-down converter.