Replacements for an EL33

[McIntyre]

What would be the most suitable replacement for an EL33? I have been working on a Pye 16G and have tested and found the output EL33 valve past its usefulness. My Avo valve data manual gives the 6AG6 as a direct equivalent. The problem is that the only output valves I have, by the dozen, are EL42, UL41, 6V6, 6F6 and 6K6’s – obviously only the last three being suitable. I tend to think that a 6F6 would be best as the output transformer required, in the data sheet, would be the same at optimum 7K. Output would be slightly lower (at a maximum 3,2W) and anode current about the same. Is using the 6F6 the best option as I have read elsewhere on this forum of using 6V6’s??

Andrew

Andrew, There was a thread discussing the possibility of using a 6V6 in place of the EL33. The main difficulty is that apart from the 6AG6 (rare), very few American beam tetrodes have the slope (gm) to match the European valves. If your Pye is a 3 valve set as most in this series were, there will be insufficient audio gain.

The octal near equivalents (some need a bias adjustment) are the 6P25, 6P1 (Mazda) and the KT61. Your scrap EL33 could form the basis of an adaptor, enabling you to fit an EL41 (almost identical characteristics) or an EL84. The problems with the EL84 are that it is not the most reliable device, being prone to becoming gassy and that the audio nuts have inflated the prices.

Leon.

Written offline - Paul S beat me to the other thread reference. Can't agree about the 6V6 though!

[XTC]

My 1966 copy of Radio Valve Data (Iliffe) gives a string of equivalents for EL33 - 6AG6G,KT61,6M6G,N147,BVA264/5/6/7, all of which are rare and expensive. NOS EL33s are available at a price (£15 or more). EBL31 and EL3 look like direct electrical equivalents, with rewiring/rebasing. They're damned expensive - but if you had one to hand.........

Reading the thread Paul referred to, it looks as if Ferranti, at least, regarded 6V6 and EL33 as near equivalents, with circuit adjustments.

If the radio was of particular value to me, I'd be inclined to splash the 15+ notes, otherwise cautiously try the 6V6 sub with mods as described in the other thread.

6V6 vs 6F6? Dunno, both have far lower gm than the EL33. EL41 looks pretty close. Yes, there'd be a nasty looking adapter to build with an octal base and B8A valveholder.

Pete.

[paulsherwin]

Can't agree about the 6V6 though!

I think it's fair to say that Leon attaches more importance to the slope of the various output valves than the other respondents. It's perfectly true that short superhets need all the gain they can get, and I can quite believe that using a 6V6 instead of an EL33 will reduce the maximum volume available, especially from weaker stations. All I would say is that it's a very easy sub to try. Just plug in the 6V6 and see how it sounds. If it's clearly not loud enough, you need to find an EL33 or close equivalent. If it sounds OK, you can change the cathode resistor to the correct value and use a 6V6.

In the case of a normal 4 valve superhet with an audio preamp stage, there's usually enough gain to overdrive the output valve anyway, so the reduced slope shouldn't have much impact (though, as Leon said in the other thread, it'll reduce the negative feedback applied which may reduce sound quality).

HTH, Paul

[McIntyre]

You state that a KT61 is a direct equivalent. I think I should have a couple of these will check tonight. My Avo manual also gives a CV2938 which I think I can get my hands. It sounds asif you guys have quite an organised network of valve dealers. I don't think there is one here in RSA. This style vintage equpiment has not caught on - I picked up a crate of NOS valves (aprox 100) and a couple of radios - a Bush EU13 - Similar to your Dac90, I think, a Zenith, and the Pye 16G, for R200 = aprox 19 Pound. It's more a case of knowing someone here.
Andrew

[Sam]

It was just what I read on the National Valve Museum Website!

http://www.valve-museum.org/

They have equivilants listed with a picture of valves they have. Another useful source of info is the TDSL on the Duncan Amps site:

http://tdsl.duncanamps.com/

You can download it to your computer, and it is a useful little thing to have-it gives ratings, short data, and links for quite a few types of valves. It isn't exhaustive though!

Sam

*added* After I properly read the whole thread, XTC mentioned the KT61 here. Sorry!

[XTC]

The KT61 and other subs were quoted as subs in Radio Valve Data , 1966 edition. It was a widely used reference.

The KT61 is referenced working with a cathode resistor of 90 ohms and grid bias of -4.4, a gm of 10mA/V, Ia 40mA, Is 7.5mA and Rl 6K with 4.3 W output.

EL33 is referenced with cathode resistor of 150 ohms and grid bias -6, gm 9mA/V Ia 36mA, Is 4mA and Rk 7K with 4.0 W output.

Putting a KT61 into a circuit biased for an EL33 should make its bias more -ve than the standard conditions and drop the cathode current.

Whatever you substitute, it doesn't matter too much if the power is too low, or there's some distortion, because you can take it out again. What you don't want to do is make a sub which draws much more cathode or heater current than the EL33, risking damaging a transformer.

I noticed the EL41 sub because it's listed just below the EL33 in the Brimar section of the data book. It's pretty close. Then I posted and noticed that Leon had posted the same thing.

If you try the EL41 bodge/sub, watch the cathode current carefully. The cathode resistor is quoted as 170 ohms, rather than 150 ohms, giving a grid bias of -7 rather than -6.

Pete.

Paul, I think you might be missing a trick in suggesting that I attach unreasonable importance to valve slope.

For an open loop configuration used in many radios, with a given (fixed) input, if you halve the valve slope, the power output goes down by a factor of 4.

Suppose an output valve has a slope of 10mA/V, and sees 1V rms input at the grid. The anode signal current (superimposed on the static anode current) will hence be 10mA rms. If the load is 5k Ohms, the signal voltage will be 50V rms and the power output consequently 500mW.

Keep everything the same, and let the slope be 5mA/V. The anode signal current becomes 5 mA rms which gives 25V rms across our load. The power output thus becomes 5 x 25 = 125mW.

500/125 = 4

As the human response to an audible energy source is logarithmic, this is a simplistic approach in terms of perceived result, but you would notice the difference.

Regards, Leon.

[paulsherwin]

Paul, I think you might be missing a trick in suggesting that I attach unreasonable importance to valve slope.

Leon, I didn't say that, and didn't mean to suggest it. I simply meant that there are a range of opinions regarding the practical significance of subbing output valves with different mu characteristincs, and that your view is at one end of the spectrum. You're perfectly entitled to your opinion which is clearly based on an understanding of the principles involved. My point is that you can often substitute different valves despite these differences in slope - given the variablity in performance of old valve radios, the difference may be completely undetectable (or it may be *very* detectable - I only suggest that people experiment).

Best regards, Paul

Paul, Fair enough; I'm a lazy soul, and prefer to experiment wherever possible with my pencil!

Regards, Leon.