EL33 Headphone Amp OPTs

Design and use of the various types.
Post Reply
Mike C
Posts: 4
Joined: Wed Mar 10, 2021 11:37 am

EL33 Headphone Amp OPTs

Post by Mike C »

I go by the same name on Tube DIY Asylum, where this matter was discussed in June 2020 in a thread that deteriorated. I did appreciate Dave's posts so here I am.
I have Raphaelite 5.5k:32 SET OPTs for 50mA.
The headphone amp was 12SN7 parallelled. No driver stage. Very good 3 stage pi PSU. Stepped attenuator of 10kohm; low value chosen to minimise the 'disconnect' between source and output stage, confirmed OK by listening tests.
Later the valve changed to EL33 trioded - a useful improvement. It works well as gain of 20 is enough not to need a driver stage. It is hard to find a DHT with enough gain for this.

I have tried very many things in developing this amp; modifying professional amps, trying different driver stages. The simplest design of all proves best.

This amp is used in a very high end system, Meze Empyrean phones, Lector or ANK dacs. Sound is very good indeed. The trioded EL33 (EL3N equivalent) is superb as a triode, virtually DHT sound quality.

I am now considering a new OPT.
I do not need to correct anything about the sound and I can't think of any specific area that needs improving, but I'd be interested in any general improvement in sound quality.

Specifics:
3kohm Ri for the EL33 in triode mode
22mA primary current
32 ohm headphones only
Currently using 5.5K : 32 OPTs

Two issues.
Firstly I may well want to use higher current at some time, very likely 30mA, possibly 35mA. Would designing the OPT for 35mA degrade the performance at 22mA?

Next, the existing OPT at 5.5kohm just happened to be that; would a different primary impedance be better? Note that the present setup has enough gain and power, but with not a lot of gain in hand.
A load line of 6kohm looks OK on the curves.

Musical preferences are tonality; vocals; accoustic music, jazz, folk, classical; naturalness.

Any comments welcomed.
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Hey,

Specifics:
3kohm Ri for the EL33 in triode mode
22mA primary current
32 ohm headphones only
Currently using 5.5K : 32 OPTs

The first thing that strikes me is assuming a true 32Ω headphone load you have less than a 2:1 load ratio on your output tube. Since you phones are indeed 32Ω this strikes me a way too heavy of a load which gives you a few options.

The first question is: How is your current gain?

if you never get anywhere near full volume then it might be best to change some of that excess gain into a more favorable load for your triode. Right now you have a turn ratio of 13.1:1 and increasing that to 18.5:1 will lose you 3dB of volume and net your output tube a much more favorable load of 11K. The interesting thing to note here is since you are approaching the situation where Rsource=Rload with your 5.5KΩ load the move to 11K will only lose you about 2dB of gain which isn't really that much.

The other thing that will change sill be the source impedance (damping factor) feeding your phones. In the 5.5K case now you have a 17.5Ω drive impedance for a damping factor of less than 2. Moving to 11000 drops that number to an 8.5Ω source or a DF of just over 3.5. I would expect this change to be as substantial in the overall system sound as the increased load on the tube. I ma not very well versed in the realm of headphones but I would expect the change in DF to be most noticeable in the bass.


The higher current isn't that much of an issue. It will just modify the inductance slightly.

dave
Get Your Fix
www.hifiheroin.com
Mike C
Posts: 4
Joined: Wed Mar 10, 2021 11:37 am

Post by Mike C »

dave slagle wrote:Hey,

Specifics:
3kohm Ri for the EL33 in triode mode
22mA primary current
32 ohm headphones only
Currently using 5.5K : 32 OPTs

The first thing that strikes me is assuming a true 32Ω headphone load you have less than a 2:1 load ratio on your output tube.

The first question is: How is your current gain?

if you never get anywhere near full volume then it might be best to change some of that excess gain into a more favorable load for your triode. Right now you have a turn ratio of 13.1:1 and increasing that to 18.5:1 will lose you 3dB of volume and net your output tube a much more favorable load of 11K. The interesting thing to note here is since you are approaching the situation where Rsource=Rload with your 5.5KΩ load the move to 11K will only lose you about 2dB of gain which isn't really that much.

The other thing that will change sill be the source impedance (damping factor) feeding your phones. In the 5.5K case now you have a 17.5Ω drive impedance for a damping factor of less than 2. Moving to 11000 drops that number to an 8.5Ω source or a DF of just over 3.5. I would expect this change to be as substantial in the overall system sound as the increased load on the tube. I ma not very well versed in the realm of headphones but I would expect the change in DF to be most noticeable in the bass.

dave
Thanks Dave, I appreciate your analysis and agree with it.
The EL33 is known to work well trioded with load 7Kohm, and the sound I get with the 5.5kohm TX is definitely good, the best I've ever heard.
But a higher load could well sound better, the better control with lower Zout is indeed likely to improve the bass.

I can afford to lose 2dB, one step on the attenuator.
Also I'm about to try the EL11 trioded pentode with a bit more gain (25 vs 20), I guess that's about a dB, so I'll only be 1 to 2 dB down. No problem.

I'm happy that I still don't need a driver stage. This gave a substantial improvement to sound; very simple cct, no coupling caps, no interaction via the PSU. I also tried a CF buffer (too many coupling caps) and a TVC; on paper the TVC should have worked well by improving the drive to the valve, but the resistor attenuator sounded better.

I'm happy to go ahead with your design as above.

As tonality, vocals, naturalness are my priorities I assume nickel core would be a good choice?

Please advise how I can place the order for a pair with shipping to the UK.

Many thanks!
Mike
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Hey,

specs of the triode EL11?

sounds like you will be in a good place.... I don't care much what anyone says a 7K load on a tube with a 3KrA is a bit low but I am happy to do whatever you want.

As for the core material.... I am a nickel junkie so 49% would be my suggestion.

the details of getting the deal done should best be handled by email.

dave
Get Your Fix
www.hifiheroin.com
Mike C
Posts: 4
Joined: Wed Mar 10, 2021 11:37 am

Post by Mike C »

Confirmation.

I checked the quietest CD I have (Clair Marlo, Till they take my heart away) and at a good normal level I had 5 clicks on the attenuator in hand, so that is OK.

Also I did some analysis of the load lines. I used EL11 curves as I have them to hand; very similar to the EL33.

I used plus/minus 2V change of grid voltage.
At 5.5kohm load this gave 10mA current change and 56V voltage change.

At 11kohm load it gave 6mA current change and 69V voltage change.

Multiplying I and V gives power 0.56 watt with 5.5k TX and 0.414 watt with 11K TX.
Yes, I know these are not actual power values as values are peak-peak, but they will be in proportion and the ratio will be valid.
The ratio is 0.739, i.e. about 1.3dB loss with the 11kohm TX. Fine!

It was done from a somewhat rough set of curves and may not be exact, but I'm sure it's near enough.

So I'm happy to go ahead.
Mike C
Posts: 4
Joined: Wed Mar 10, 2021 11:37 am

Post by Mike C »

dave slagle wrote:Hey,

specs of the triode EL11?

sounds like you will be in a good place....

dave
Attached are curves.
Triode gain is about 25.
Ri is given as 3.5kohm at 20mA, 2.5kohm at 36mA.
I propose running at roughly 25mA to start with so I estimate Ri at 3kohm.
I think power output as triode can be about 2W, but it then gets a bit non linear at the high current extreme.

This family of EL11, EL12, EL3N,EL33 are known to sound very good as triodes, with pretty much DHT quality of sound.
With my EL33 amp I'm already in a good place, but I'm getting excited at the prospect of being in an even better place. :D
Attachments
EL11 trioded curves
EL11 trioded curves
EL11 curves.jpg (102.19 KiB) Viewed 4727 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

hey

attached the curves with an 11K loadline and you can see how things change. There are a few important points about this situation that need to be taken into consideration before an actual number is chosen.

First.... how much current you run has a profound impact on the Rp to load ratio. I have always used 3X as a minimum but have seen people go as low as 2.5X. In your case the difference between 2.5K and 3.5K for the rA can be substantial when you are playing with the low ratios. Pluws as tubes age the rA typically increases which again robs you of any "headroom" built into the ratio. In your case of the 5.6k output I would bet you were under 2:1 with your ratio.

Next. in your particular application without a driver stage, chances are you will not be swinging more than 4-5V into the grid at which point the "compression" from approaching current cutoff (the lower right corner) may not be as big an issue as if you were trying to eke every last milliwatt out of the circuit. It might be argued that given your limited swing the heavier load may add a bit of "tone" to the sound and juice things up and going to a much higher load may get you into the real of "really clean"

My guess is 11K is a good safe number and if needed the operating point can be adjusted to juice things up if needed.
Attachments
Screen Shot 2021-03-13 at 9.29.30 AM.png
Screen Shot 2021-03-13 at 9.29.30 AM.png (603.17 KiB) Viewed 4722 times
Get Your Fix
www.hifiheroin.com
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Re: EL33 Headphone Amp OPTs

Post by dave slagle »

everything looks right. the only concern is the inductance of the primary might be a bit low. worth a try though.

dave
Get Your Fix
www.hifiheroin.com
Post Reply