EL34 schematic confusion

[SwedishWings]

It's always "right" with a smail

I ususally spend a lot of time with SPICE for (god forgive me) solid state designs (i design all sort of electronics for a living).

However, I have less good experience with tube simulations. The poor modeling of the tubes is the main reason, another that SPICE transformer models are not so good. So far, the best models I've found come from Duncan's site, that is realative close with some adjustments. However, i frequently run into convergence problems with the models and ususally give it up and build the damn thing and meassure it up, which is faster

If you have better models of ECC83, EL34 and "real" transformer models, I would greatly appreciate it!

Btw, we just had a talk about age here in the ofice. Me and my partner concluded that a mans age is determined from the balls, and they usually dong get much older than 17 years

 

[Johan Potgieter]

I have exactly the same problem with tube Spice models. I do not even use them. Especially an UL output transformer is not as simple as any "simple" Spice model. So many variations between makes. I am fortunate to have my own wound locally, which comes cheaper than importing, all costs included (something like US $130 for a good C-core 100W output transformer).

I also usually hook up and measure in practice. One can usually get a rough idea even with a scope only when measuring without feedback to begin with. I also have an old THD analyser (but broken at present), which can give a ball park figure. Otherwise experience.

Regards

 

[TonyTecson]

1) The extreme bandwith is hard to achive in other ways, and will allow for better error correction in the feedback chain, hence the net distortion product of the amplifier will propably be better than the common anode resistor solution that lacks bandwith and will introduce more phase errors.

you can try also SY's mosfet followers.....i do not find EL34's hard to drive as drive requirements are not stiff imho..

 

[JojoD818]

now you guys are confusing me again!

don't be. let's continue this project.

I'll probably be starting before the end of this year.

 

[navin]

don't be. let's continue this project.

I'll probably be starting before the end of this year.

I'll follow on your coat tails!

 

[SwedishWings]

you can try also SY's mosfet followers.....

...or maybe just abandon VT entirely and go for MOSFETs No, i should not be like that. I have used that solution in some places, but it somehow feels the amp get contaminated...

Solid State: A component that has been specically designed to make a (guitar) amplifier sound bad. Compared to tubes, these devices can have a very long lifespan, which guarantees that your amplifier will retain its thin, lifeless, and buzzy sound for a long time to come.

btw, anyone who knows how to set this thing to show the last posting first?

 

[Dominique_free]

A montage that can be interessant in that case is the ultra-linear cascode. CAD: Ultra-linear cascode But maybe at your SRPP will be better because of a lower output impedance.

The advantage of the ultra-linear cascode will be in the bass frequency, it will go down to DC. With the SRPP and a fc at 5.7 Hz, you will have phase shift up to around 57 Hz. Simulation are good to see this, but they will fail to show the high level swing of a vacuum tubes montage, because all the models we have today are too poor.

And even with better models, a simulation will never show the sound of a montage.

The only good enough spice model we have today is the Ferugini diode model. All the free triode models are good enough at pure class A, that mean at low level output swing. At high level, the main problem with the triode model is at the parameters of the real component are changing, and the model don't modelize that very well.

In class B, it is even worse, because the models that have a G1 model don't show the very big increase of G1 at low Ua.

For a pentode, it is worse, because it have the same limitations as the triode, and when Ua get lower as Ug2 (possible even in class A), the model don't show the increase of Ig2 in that case. And if you change Ug2, as in an ultra-linear PP, the models don't change the parameters of the tubes accordingly.

I find recently a very good pentode-tetrode model Tetrode simulation , but I have never see this model in use with audio tubes. I begin to try to do a 807-6L6 model based on this work, but I am not done with it yet. It is very time consuming because this model use the constant current curves to calculate the parameters, such curves don't exist for audio tubes, and I don't have matlab to calculate them.

For those that don't have a spice simulator, LTSPICE is free, is one of the best for analog electronics, and it doesn't have limitation as the free version of Pspice. It even run very well and fast on linux with wine. For linux, I recommand the gEDA and ltpsice or ngspice (ltspice is better at that time but gschem (part of the gEDA) is maybe the best schematic program I know).

 

[dragos_g]

EL34 Schematic (Ray_Moth)

Hi Ray,

Which version is your latest and gratest?
I would like to build it (UL instead of triode) since it looks to me one of the best designs (I built something similar on transistors)
Do you have any kind of document describing the design principles?
I went through some your past replies and I have an idea, but I dont know if you did update the schematic since the original post.
Can you share also your PSU schematic?

I think that a PDF document similar with the one Claus Byrith has on Lundahl, will be a great help for a lot of guys.
I am thinking to use Lundahl, at least for OPT if not for the PSU as well since they are not much expensive than Hammond.

Regards,
Dragos

 

[pred]

primary impedance

Hi everyone,

I am "hijacking" this thread for a small question, since I did not think another EL34 amp should deserve a new one

In all schematics I have seen there is a huge difference for a pp EL34 amp transformer primary impedance. It goes from 3k up to 7-10k. I was wondering how one chooses. If I read right the Mullard sheets on EL34 it should be around 6k (or 6.6k as some note) for the lowest distortion.

So what you gain what you loose with the choice of primary impedance?

Another thing is transformer choice. Here in Switzerland I cann't find a thing that is not way out of my budget so I think I would go for a Hammond 1650p. It is a 60W trannie but by the frequency specs I do not think I will suffer high frequency for increased power rating(or maybe I am wrong?). I hesitate to take the potted version 1650pp but I am not sure if it deserves double the price?

So if someone has a thought on these things the answer is more then welcome.

Predrag

 

[PeteN]

I've wondered the same as the previous poster, there seems to be a fairly wide range of impedances used with the EL34. Even more if you consider parallel valves. Any insight or recommendations our resident benevolent masterminds could provide would be appreciated.
More variation seems to be on price though - anywhere from 'cheap' as you will see below to several hundreds of pounds.
(as a poor student I object to such silly prices for something which has been produced for what... 50 years? *wink*)

Regarding price, the cheapest output trafo I have seen (quite possibly ever) is THIS model (warning: Maplin). Here is the description:
Chassis mounting 25W push-pull output transformer suitable for 2 x EL34 valves. Secondary tapped at 4/8/16Ω. Primary, 5k6Ω anode to anode with ultra-linear taps. Grain oriented silicon steel interleaved laminations.
No futher information available.
The manufacturer is OEP, who seem to make mainly non-audio trafos, but I cannot find *any* information about this item on anywhere but Maplin's website.
If using 2xEL34 in push pull as they say it is designed for you'd have a 1.4k loadline (pretty steep, I've drawn a few and with ~400V HT this over-powers the valve for [I think] about 40% of the loadline that exists between the 0mA axis and the 0V grid voltage curve), so you'd need to use lower (300V?) HT to prevent making the OPT go poof.
Or just oil cool it hehe.

Ummm anyway, uneducated conclusion: More primary impedance => shallower loadline => higher HT needed to make loadline stray into the nether regions above the max anode dissipation curve => more gain (more anode swing given only-a-bit bigger input) => more power.
Correct me if I'm being dumb here.

I'd be loathe to use an OPT that cost 25 quid on anything particularly spectacular, but if I had a bit more money to play with I might give it a whirl for a lower powered or mostly "junk-box" amp just to see if it is in fact a bargain.
Not to mention that you'd need to hide it away inside the chassis to stop the cat getting a numb nose

 

[pred]

If I understood well the higher the resistance less power you have but also less distortion from output tubes. But this is a still mistery to me. Since for me the investment of time and money for housing is the biggest then I would like to start from the good basis, choose a nice transformer (not too expensive lets say less than 200$ a piece) and then play with the driver and tubes. So still in the dark what to do...

I also found an audio note transformers with a dissent price (there is a 25W one and 50W one) but they are like open, everything sticks out and almost a completely closed enclosure is a must (or somebody) has some other idea. I thought about some mounting with the cage for tubes (a small curious daughter ) but not to enclose the transformers as well...

Pred

 

[SY]

The Mullard datasheet is a perfect lesson in load tradeoffs. There is no one right answer- there are a lot of good ways to run the tube. After all, would you expect that class A triode and AB pentode should have the same optimal operating conditions? To make things more complicated, optimum load varies depending on bias type (fixed versus cathode).

 

[pred]

I really secretly hoped you will answer

No not really I did not hoped that the performance will be the same for triode pentode connection. But I always thought that there is a sweet spot or a good compromise for the operating point and the transformer output value...

If not in your opininon for the triode mode and arround 15-20W of power is a 6k a good option?

Thanks in advance,
Predrag

 

[SY]

The data sheet shows 5k as the recommended load for triode at that power level. You'll need 430V, it will not be class A, and you'll want a driver than can handle the Miller capacitance. But absolutely do-able; that's close to the operating point and load of the classic Marantz 9.

At the cost of a few dB of power, you can really make a distortion champ with a 10k load, 430V, and an unbypassed cathode resistor.

 

[pred]

In this point according to the datasheet I will put like 70mA into the tube that for Va=400V leads to 28W of disipation, a mighty lot for EL34 tube. I just repeat this in order to be sure that I understood your point and "and it will not be the class A". This means I would have to bias it, to prolong the life of the tube with lets say 50mA and have like first 10W in class A and the other half in class B?

Pred

 

[SY]

You're right, you have to flog them pretty hard to get the 15-20W you want if you use triode and bias for minimum distortion. Give them some breathing room and a moderate amount of airflow and the lifetime will be acceptable.

In UL, things are easier, but the output transformer quality and the actual impedance of the load become more critical. Trade-offs.

 

[pred]

Problem comes now that if I want to bias them in a class AB UL according to the datasheet the impedance falls down to 3.5k which is almoust double. Now I understand why the Dynacos and similar designs choose 4.3k so somewhere in the middle between 5k and 3.3k. To have something of both of the worlds.

Since I would like to try not to use feedback, at least try I will have to see for my self. Maybe I will by a Tubecad soft and calculate to see what suites me best...

While we are there if I look at Hammond transformers they are all with the same specs that means for me that the bigger Iron the better, more low frequency. Am I right?

 

[Johan Potgieter]

In this point according to the datasheet I will put like 70mA into the tube that for Va=400V leads to 28W of disipation, a mighty lot for EL34 tube.
Pred

Hi Pred,

By this time SY had just about explained all. I would just like to add that the 70mA you mention probably includes screen current, so the dissipation will be for both anode and screen. Also, if you use cathode bias remember to take that off the Va. But that is rather academic - one feels nervous for a dissipation of 28W and sighs for relief as soon as it drops below 25W. These are of course points on a line; nothing catastrophic is going to happen after exactly 25W. It is a question of heat, which then brings in ambient temperature, etc.

From the Mullard data one finds a "preferred" operation with cathode bias, at round about 450V(B+), distributed load at 43% taps, 6,6K primary load, for an output of about 25W. The distortion curves are quite low.

But also (linking in with SY) you would probably know that a loudspeaker load is anything but constant. Your amplifier will probably see loads from 3K all the way up to 10K (at higher frequencies). So its a tube's life out there.

In your post #589 you also ask about loss of h.f with a bigger transformer. Not necessarily. That will depend on the leakage reactance - h.f. response usually goes beyond 20KHz. It is only that compensating capacitors with NFB would need adjustment.

I am not trying to make this into a tutorial (you may already know all this), but one must be careful to simply argue that "class A as far as possible" gives reduced distortion. Too high a cross-over point and things get worse as one has something called Gm-doubling. This is especially prevalent with transistors, but not absent with tubes. There is roughly a best quiescent current for p.p. tubes, although here the zone is pretty broad, especially with UL.

I may overstay my welcome with a last point but must always ask: why no NFB? Just hoping you are not avoiding it for all the wrong reasons .....

Regards

 

[pred]

thanks

Thanks a lot for the answers I have a lot of experience in electronics and very low with tubes. I know how triode works but am not sure about the operation of pentodes So any comment is welcome.

No NFB is just a trial. When I started to do the hobby I first thought about the distortion and making as small as possible, but then why bother I desided I will do by my ear, for me this more making and instrument I like. Anyway my friends do not care about the HiFi so even if I wan't except my vife I do not have a person to show it to So I will try if I like it I will use it if not no. No real reason behind it

 

[ashok]

"......So I will try if I like it I will use it if not no....."

Great ! One of the CORE reasons why one does DIY !