Valve Headphone Amp

[Max N]

Think of the lid as a speaker cone and the tx as the motor. I know there is no mechanical coupling, as there would be in a speaker, but because the lid is magnetic it is still being moved by the coil.

So either prevent/damp the movement of the lid or reduce the magnetic coupling. I would try some damping sheet (for car sound treatment) squashed between tx and lid.

[Nick]

Why would this be different than just using the weather stripping and leaving the transformer bolted to the bottom

Just guessing, the bottom of the box may be more rigid than the top as its supported by the sides and corners, also attaching the transformer to the top will increase the moving mass of the top surface and that may make it less resonant at 50Hz.

Damping should do the same thing.

[simon]

The vibration imparted in to the lid is quite something, too much for some damping pads I suspect. Really, it's like it's trying to take off!

Applying damping beneath the tx I suspect might have limited benefit anyway as I think the fixings may well transfer the vibration. Resting the tx on corrugated cardboard made little difference.

The sides of the chassis (and even the top "lip" will have an affect as they stiffen the sides in turn) will make the bottom part much more rigid. But whilst the bottom will have a different natural frequency to the lid I wonder if they would both still vibrate anyway.

Interesting problem.

I'm not sure about mounting the tx to the lid. I wonder again if the much lower stiffness might allow it to vibrate more. Or perhaps it becomes less restrained as it's decoupled. The wires from the secondary to the bridge would have to be quite long though, which wouldn't be ideal.

I think the solution is probably distance of the lams from the chassis. Which I don't have. Or a non-magnetic case. Or a different tx.

[Nick]

May have been already suggested. Make the lid out of Aluminum?

[rowuk]

Why would this be different than just using the weather stripping and leaving the transformer bolted to the bottom

Just guessing, the bottom of the box may be more rigid than the top as its supported by the sides and corners, also attaching the transformer to the top will increase the moving mass of the top surface and that may make it less resonant at 50Hz.

Damping should do the same thing.

What he said!
The mass of the transformer will make the lid a completely different (most likely better behaved) animal. A bit of damping should get the rest. Of course, an aluminium lid would not react to the magnetic field like a piece of steel would.

[simon]

Yes it might make it better, but it might also become worse, or little different. Really, the vibration is quite severe.

An aluminium lid, mdf, or even some perspex that Ant offered me some months ago would do it.

[Ant]

Still got it sat there simon, couple of tap washers or something to space it off so the tx has the necessary clearance and it would solve the problem. Think of the gap as ventilation..

Or i could take it off your hands, just to relieve you of all the associated stress..

[simon]

[simon]

I have a spare pair of ECX10N20 MOSFETs and I've been wondering about building a MoFo type SE buffer speaker amp. Rather than start another thread I thought I'd just continue this one as the circuit would no doubt look rather like ed's original.
.

hpa cct.jpg (46.09 KiB) Viewed 43452 times

.
According to ed's sim it's good for 3W at 12V, and 5W with 20V, but I thought I'd try to actually understand the design a little.

IIRC the gate of the MOSFET is biased at half the 12V on the drain, and there's 265mA through the MOSFET d-s. I presume ed chose the voltage and current to suit a relatively low power requirement for a headphone amp.

Looking at the datasheet the absolute max ratings are:
.

ECX10N20 max ratings.png (133.35 KiB) Viewed 43452 times

.
The gate could be at as much as 14V, compared to the 6V of the headphone amp, and up to 8A through the drain. This would need a fairly beefy 28V power supply. Ignoring the voltage drop across a CCS, this would be a dissipation of c220W, way above the allowable of 125W.

So more realistic could be 20V on the drain, so 10V on the gate, and 1.5A, the maximum current through the LM317 CCS. That's a dissipation of c30W.

I think the output power is Vrms^2 / load, so (10 x root 2 / 2)^2 / 8 = 6.1W, not the same as ed's 5W.

Am I on the right lines so far?

Something I haven't worked out yet is how to choose the amount of drain current. The output power doesn't seem to be linked to current in the power calc above, but surely it must be somehow?

[simon]

I did a bit of reading around the Web last night and watched an interesting YouTube video which stepped through the design. Great, I thought, I'll make a spreadsheet. I failed at the first hurdle though - the video referred to two parameters, Threshold voltage Vth and Conduction parameter Kn, neither of which are specified in the datasheet for the ECX10N20 or IRFP250 used in the MoFo.

I read on one of Rod Elliot's pages that the current through the MOSFET should be (Vdd/2) / load
= 20/2 / 8 = 1.25A. Probably needs to be a bit more for losses in the MOSFET. But I'll go with 1.25A for now so there's a bit of margin in the 317. And it gives me a nice round 1R resistor to set the current through the CCS. It'll need to be minimum 5W.

The 1.5A limit of the 317 limits Vdd to 24V, and the CCS resistor would be 0R84 6W minimum.

I think, anyway.

[Nick]

The following is all IMHO and probably a bit grumpy.

Looks like a lovely device, but the curves makes me want to push the envelope with 20v @ 4A. But where I have to ask the hard question is, if we are going to hand wave a 317 regulator into a CCS, and try and squint and view it in the same way as a choke in a mofo then why not just use an op amp with a pair of medium current bipolar NPN/PNP transistors on the output inside the feedback loop?

OK, yes, maybe its for the fun of it, ok, but somewhere don't we have to accept that we are just using solid state and feedback? Why not just use solid state and feedback?

[simon]

Yeah, it's just for fun. To try to learn a bit about this stuff - I still break out in a cold sweat at the thought of solid state. So there will no doubt be iterations of thoughts.

I'd been reading about the Zenductor 2 on DIYAudio which seems to have a lot of very pleased builds. Essentially it's a buffer with 4 paralleled JFETs with a CCS underneath each in front of a step up transformer for gain followed by 2 paralleled MOSFETs as common drains.

https://www.diyaudio.com/community/atta ... f.1369836/
.

Zenductor2 Schematic.png (129.22 KiB) Viewed 42356 times

.
The step up transformer is suggested as the reason it sounds so good. Is it really any better than a decent SS front end? I don't know but thought it might be interesting to try. Then I thought rather than the paralleled IRFP048s, perhaps use the ECX10N20s? And then I thought is it really any better than a decent MoFo. And round the circle I go again.

The 317 CCS was just an extension of ed's circuit, and a starting point. They are a lot cheaper than a choke, but probably sub optimal in a speaker amp. I don't need a lot of power though.

A 4A PS is reasonably big. Maybe bigger than I need. But I'd like to explore this on the curves. I'm familiar with reading triode anode curves, but pentode (which is what the ECX curves look like?) not so much. If I have the correct graph, is the red dot the OP you meant?
.

ECX10N20 OP1.png (209.59 KiB) Viewed 42356 times

.
If so Vgs is approx. 3.7V, very different from half of Vdd. But the 20V is Vds which is different from Vdd.

Am I anywhere near?

[Nick]

in front of a step up transformer for gain

Except its not a transformer, its an autoformer. That's a very clever use of the autoformer to provide gain almost out of thin air, while also providing the bias voltage for the output stage. With the 25% tap being grounded to AC, but floating at DC.

I like that, its funky.

is the red dot the OP you meant

Yes. Just draw a 8R load line between 50v on the x axis and 6.25 on the y. Replace 8 with whatever load your headphone provides.

[simon]

Yeah it looked like an autoformer to me, but I used Nelson Pass's description. Will draw the 8R load line later.

[simon]

Nuts! The forum's become quite unstable again, logging me out quite frequently, including when I was submitting a reply. Try again.

So I've dobbed a loadline on at the OP, and the existing one for the headphone amp, in green.
.

ECX10N20 OP3.jpg (107.54 KiB) Viewed 39332 times

.
I set off thinking about loadlines in the traditional triode way, then thought hang on, it's a buffer and couldn't quite work out what I should be doing. So I'm trying AI (God help me).

The Triad autoformers are around £20 each, Cinemags more like £60 or £70, Slagle a lot more. But as I understand it there's only 4x gain and I wonder if that might be a bit marginal without some preamp gain, and I use a buffer with none.