• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

SE Amps with Variable Impedance Loads

Hey guys,

I am in the process of designing a SE 807 amp. In triode mode, I expect approximately 3.4W. I have already ordered Edcor GXSE15-8-5K OTs for this project. I wanted a low power amp for my newly-restored Klipsch Heresy speakers. However, my research tells me that the Heresy speakers vary their impedance from 10 ohms up to 70 ohms or so. Since transformers reflect the impedance of the secondary load to the primary, I can deduce that these speakers would NOT work well with a single-ended amplifier. Is this correct?

If that's the case, how much of a difference would it make if I put a 15 ohm resistor on the outputs? That should keep the impedance within what I would consider an acceptable load. What will this resistor do to the frequency response, and do you think I'd notice an appreciable drop in volume?

Thanks!

Kyle
 
The 15 ohm resitor would certainly help and whilst it would take a bit of power it won't be noticable.
I'd be inclined to include the transformer secondary in the cathode circuit. If you have a cathode bypass cap then instead of connecting it to ground, connect it to the speaker output (with the other end to ground). you will need to experiment as to which way round one of the windings is connected (pri or sec). one way will give you lower gain and smoother respose, and the other will give more gain, and may even oscillate.
 

leadbelly

Disabled Account
2002-12-22 2:13 am
Calgary, Alberta
However, my research tells me that the Heresy speakers vary their impedance from 10 ohms up to 70 ohms or so. Since transformers reflect the impedance of the secondary load to the primary, I can deduce that these speakers would NOT work well with a single-ended amplifier. Is this correct?

Not only is is not correct, but if your impedance dips to no less than 10 ohms, I would consider this a very tube-amp-friendly speaker. If you read up a bit more you will see that you are jousting at windmills, that's why the commonly quoted speaker impedance of 4,6, or 8 ohms is properly referred to as nominal impedance.

If that's the case, how much of a difference would it make if I put a 15 ohm esistor on the outputs? That should keep the impedance within what I would consider an acceptable load. What will this resistor do to the frequency response, and do you think I'd notice an appreciable drop in volume?

Forget the resistor, you don't need it, it's a bad idea.
 
I haven't really heard of putting the secondary in the cathode circuit before. That's definitely an interesting idea.

I got my info from here, and the general consensus was that the Heresy was not SE-friendly. http://community.klipsch.com/forums/p/77869/771253.aspx

Also, many people say never to run a tube amp into no load, which is infinite impedance, technically. Since 70 ohms is much higher than 8 ohms, I would guess that it's not very good for it. That'd be reflecting 43.8KΩ into the primary, I believe.

Kyle
 

artosalo

Member
2010-02-16 9:00 am
However, my research tells me that the Heresy speakers vary their impedance from 10 ohms up to 70 ohms or so.

This is not an unusual impedance behaviour. Rather it is quite typical.
The impedance of a typical speaker is relatively high at bass resonant frequency.
Fortunately there is a cure for this. The proper amount of negative feedback. This reduces the negative effects of varying load impedance.

I can deduce that these speakers would NOT work well with a single-ended amplifier. Is this correct?

It is not correct. Same problem exist with pp-amplifiers.
The worst results will exist with pentode-connected amplifier. With and without feedback.
Best results exist with UL-connected amplifier and little worse with triode-connected. Also with and without feedback.

I have already ordered Edcor GXSE15-8-5K OTs for this project.

If I were doing such amplifier, I would use UL-connection since this transformer makes it possible. Then you will get some 9 watts with same distortion level that 3,4 watts from triode stage.
 
My current design is looking like either a 6SL7 single-ended front end or a 6SL7 SRPP, both possibly with an LED bias, just for fun. Right now I don't have any feedback (that I know of). If I did use negative feedback, what would I need to do and how would I calculate the values?

If the higher impedance really isn't a problem, then why do people say not to run a tube amp into no load? What will happen?

I plan on having a switch to switch between UL and triode modes. I'm working on my schematic(s) right now, so we'll see how it goes.

Kyle
 
A 15R resistor would waste 40% of your output power. Not a good idea.

Triode or UL should be OK. They will help damp the bass resonance. As you are going for SE I assume you don't want a low distortion design with global feedback.

No load is a problem, especially at higher frequencies. A higher resistance is not a problem, especially at lower frequencies. If you want to use feedback then you either need to learn a bit more or copy an existing design, as feedback can create as many problems as it solves if not done carefully.
 
Last edited:
The 70 ohm impedance will not be a problem. That is at a resonant frequency and obviously your amp will not make much power into that load.

But it won't have to, nor would you want it too, as if the amp made the same power, it would be *way* too pronounced at that frequency. If I were you, I'd give it a shot- certainly it will not be hard on the amp :)
 

artosalo

Member
2010-02-16 9:00 am
That is at a resonant frequency and obviously your amp will not make much power into that load.

Without negative feedback the amplifier will obviously be overdriven at this frequency and produce severe harmonics.

I first have to get the amp built, and then get some tweeter diaphragms

If you later would like to add the negative feedback take care that your amplifier has sufficient amount of extra gain. Otherwise this is impossible to do.
 
I'm not so sure that this speaker is typical, the peak that is being talked about isn't necessarily in the bass (at fs) - I think there's a thread on the klipsch forum mentioning a peak at 2.5kHz.

Negative feedback has many consequences, it changes clipping behaviour for the worse and gives the speaker (back emf) and it's cables (rf) an opportunity to send their own signals into the amplifier. It also changes the harmonic distortion profile of the amplifier so you will not end up with a classical SET amplifier.

My advice is to build your amplifier with good quality parts and without negative feedback and try it out with your speakers. It's DIY afterall, so you can experiment with changes to the design if you don't like the results.
 

artosalo

Member
2010-02-16 9:00 am
Negative feedback has many consequences, it changes clipping behaviour for the worse and gives the speaker (back emf) and it's cables (rf) an opportunity to send their own signals into the amplifier. It also changes the harmonic distortion profile of the amplifier so you will not end up with a classical SET amplifier.

Concerning the clipping behaviour:
In HIFI-amplifier there should not happen any clipping during listening with nominal power levels. If such happens the amplifier is too small and the listening result is far from HIFI.
In any case, without negative feedback the amplifier is more prone to clipping - especially at bass resonant frequency of the speaker (see below).

and it's cables (rf) an opportunity to send their own signals into the amplifier.

Please explain me in detail what you mean.


It also changes the harmonic distortion profile of the amplifier so you will not end up with a classical SET amplifier

Yes it changes. It will reduce ALL distortion content of the reproduced signal and improve the linearity of the amplifier.

Once again, if a HIFI-amplifier is concerned, there should be as little distortion added by the amplifier as possible (the quitar amplifiers are whole other story).


My advice is to build your amplifier with good quality parts and without negative feedback and try it out with your speakers. It's DIY afterall, so you can experiment with changes to the design if you don't like the results.

I will give an example what the final result is with and without negative feedback.
At first, the below figure is a typical impedance curve of a loudspeaker:




[IMGDEAD]http://kuva.termiitti.com/image/14583.gif[/IMGDEAD]
As is quite generally known, loudspeakers are designed to be driven from an (almost) constant voltage generator. Then the effect of impedance variations of the speaker is minimized.

Below is a figure when the mentioned speker is driven from three diffrent amplifiers having output impedances of 100 ohms, 10 ohms and 1 ohm.

[IMGDEAD]http://kuva.termiitti.com/image/14584.gif[/IMGDEAD]

Please quess what output power/-voltage curve is from 100 ohms representing a pentode output stage without negative feedback and which one is an UL-stage with "right" amount of negative feedback.

Conclusion: A good classical SET amplifier can be built without negative feedback. But with proper amount of NFB it will be even better.
 
If the higher impedance really isn't a problem, then why do people say not to run a tube amp into no load? What will happen?


Kyle

With no load on the outputs, ie no speakers connected, there very little current flowing, so the various high voltages can go even higher, possibly popping capacitors or worse, frying the output transformers by arcing over from one winding to another.

70 ohms load is well within reasonable operation compared to completely unloaded speaker outputs as far as the voltages go....
 
So it sounds like trying out the Klipsch certainly won't hurt my amplifier. It might be more inefficient at certain frequencies, and it might even sound bad, but it can't hurt to try. Since this is certainly a DIY project, it might take on several versions before I get it the way I like.

The whole part (okay, not the whole part...but a lot of it) part of me building this amp is to learn even more about tube circuits. I think I have a pretty good understanding of what goes on, and I even calculated my own loadline for this project, which I think is a good start. Once I get the basic amp constructed, I think it's only right if I learn how feedback works and is calculated. It will be a good learning opportunity.

Kyle
 
In any case, without negative feedback the amplifier is more prone to clipping - especially at bass resonant frequency of the speaker (see below).

As is quite generally known, loudspeakers are designed to be driven from an (almost) constant voltage generator. Then the effect of impedance variations of the speaker is minimized.

Conclusion: A good classical SET amplifier can be built without negative feedback. But with proper amount of NFB it will be even better.

Clipping has nothing to do with negative feedback, although negative feedback can cause the onset of clipping to be more severe. Clipping is created by distortion as the output devices saturate, which has little to do with feedback.

Speakers are not always "voltage driven". This model is used by a lot of speakers that are designed to be used with transistors. But the voltage model falls apart if you really want the speaker to be tube friendly. Why? Tubes cannot double power as impedance is cut in half. You can add feedback to approach 'voltage source' behavior, but it is done at a price.

The price? Negative feedback results in the distortion of odd ordered harmonics, the 5th, 7th and 9th. Our ears use these harmonics to determine the volume or loudness of a sound so we are quite sensitive to them. If they are distorted the amp might sound bright or might otherwise be fatiguing. In fact this is a coloration- one that our ears are more sensitive to than minor frequency response variation.

So as a result, there is an alternative speaker drive model (it is also much older as a concept) in which the amplifier is more of a 'power source' than a 'voltage source'. And it turns out that speaker designers whether consciously or not have used this model (based on their intention of making the speaker tube-friendly). If you are aware of the design rules, it is possible to get flat frequency response despite the higher output impedance.

This allows for a zero-feedback SET to have fairly flat frequency response on a speaker like the one mentioned. As soon as it is assumed that *all speakers* follow the 'voltage model', you suddenly have an equipment matching conversation that arises, one seems to have no clear boundaries, and is one that many audiophiles have flushed at lot of money into! When you understand that the 'voltage model' is not the only game in town, its easy to see why there are equipment mismatches. Anytime equipment from one camp is used with that from the other, tonal aberrations result.

more:

Competing paradigms in amplifier and speaker design
 

20to20

Member
2010-06-23 9:25 pm
The price? Negative feedback results in the distortion of odd ordered harmonics, the 5th, 7th and 9th.

Our ears use these harmonics to determine the volume or loudness of a sound so we are quite sensitive to them.

If they are distorted the amp might sound bright or might otherwise be fatiguing.

What is the 5th, 7th, 9th harmonic frequency of 1Khz?

Does this mean that if our amps have no harmonic distortion, we won't be able to hear to "Stairway to Heaven?"
 
I have built two SE U.L. tetrode amps (6P1P) that deliver about 3.5W into 8 ohms and tested them with Klipsch Heresy speakers. They sound great and have no problems with driving the speakers (1976 or 77 production). The speakers have never been opened and are all original.

I remember measuring my normal listening levels while in college with these speakers driven by a Sony STR5600SD, and the average power level was only 32mW with peaks reaching 1W. The listening levels were loud enough to enjoy (Dixie Dregs Sky Dive IIRC) but low enough not to require raising ones voice to carry on a conversation.
 
Last edited: