Single ended Class A OP stages.
 

After mucking about for far too many days trying to build a class A SE amp, I was left scratching my head trying to figure out why I could get more than 1.5w out. Tried altering every parameter, but always got asymmetrical clipping at about 3v RMS OP.

So,in search of answers I read this article - https://www.tubecad.com/2018/01/blog0409.htm and this passage "The peak symmetrical voltage swing across the primary is equal to the idle current against the primary impedance; thus, with an idle current of 0.1A and a primary impedance of 2500 ohms, the peak voltage swing equals 250V, which would leave zero volts for the output tube. " What I think he's saying is that a lot of power, or is that voltage is lost in the OPT,but how?

Could anyone elucidate a bit further on this please? Andy.

Re: Single ended Class a Op stages.
 

Class A output stages run with a standing current.

When you apply a signal, the valve swings up and down in current, and the anode voltage goes up and down, too.

As you drive the valve harder, you run into clipping. The valve can't pull its anode down any further than the 'bottoming' voltage, which is generally about 25V for pentodes and beam tetrodes, so you get clipping on this half-cycle at this point. And on the other half-cycle the current swings to zero, and can't reduce any further, so you get clipping on that half-cycle too.

For symmetrical clipping, you want the clipping on each half-cycle to kick in at the same input level, so nothing is 'wasted'. This is the condition for getting maximum undistorted output power out of the valve.

So if you have 290V HT, and a standing current of 0.1A, then the downward swing is going to be (approx) 250V, because 25V is lost across the valve, and 15V is lost across cathode bias resistor. And the upward voltage swing is, symmetrically, 250V too, as the current reduces from 0.1A to zero.

This means your optimum load is 250V / 0.1A = 2,500Ω.

And maximum output power is Vrms x Irms = Vpk x Ipk / 2 = 250 x 0.1 / 2 = 12.5W.

The input power to the stage is 290V x 0.1A = 29W so efficiency is 43%. Of course, there is also the screen-grid current (and heater power) so the total power draw is probably nearer 40W. And the 12.5W output is also going to be a bit reduced because the output transformer isn't ideal - it will have a bit of winding resistance, so when the output transformer has instantaneous 250V across its primary, about 7% of this in a typical transformer will be absorbed by winding resistance so the load thinks it's only 233V. But, hopefully, you get the idea.

Re: Single ended Class a Op stages.
 

Thanks. Ok, and that means Va shoots up because our OPT is an inductor, until that is we hit zero current. I suppose what's limiting OP power is available HT and turns ratio of the OPT and the losses you outline.

Andy.

Re: Single ended Class a Op stages.
 

Yes, Va shoots both up and down. Ideally, when you reduce current to zero, Va shoots up to Iq x Rload and stays there. In practice, because of finite inductance, it will decay exponentially back to Vb. But, you've got the idea.

What limits power is HT volts, and quiescent current - assuming you have access to an infinite array of transformers with turns ratios.

If you have a 'good' transformer and want to use that, then the turns ratio hence actual load is fixed, so you choose quiescent current to suit. Then if you have chosen your valve, you determine if it's under-run, or over-run (either on current, or anode dissipation).

If under-run, then fine, it will have a long life (unless excessively under-run - when you'd choose a less gutsy valve). If over-run, then just put two in parallel, or use a bigger bottle.

Re: Single ended Class a Op stages.
 

In practice, most of what are called "class A single-ended" outputs [the sort of 6V6/EL84 things you find in broadcast radios or record-players] are actually working in Class-AB1 - which lets you get the most out of a single-ended stage without too much noticeable or objectionable distortion.

In general, the higher the anode-voltage [within the valve's ratings] the greater the efficiency it will have at converting DC voltage/current to usable audio-power: choose a HT supply that's at the upper end of your valve's anode-supply rating and then scale your transformer to cope.

It's worth noting that the 'maximum anode voltage' shown in valve spec-sheets is actually the maximum anode-cathode voltage - if you're using a cathode-resistor to supply the grid-bias then you can add the static DC-bias developed across this resistor to the 'permitted' maximum HT-voltage!

[I've run a 6V6 with 340V HT without issue...]

Re: Single ended Class a Op stages.
 

Thanks for those pointers Kalee20. What I found interesting whilst experimenting was that secondary load had less effect on Pout than I'd have thought. I'd assume that the higher the reflected primary Z the higher Vout, the same as using a bigger anode resistor. This was the case to a point,that is until Imax was reached.

Another thing I noticed was how the load line for all the Class A stages I drew fell well below the knee, which for a PP OP stage is a big no no.

Interesting. I found I couldn't get the full voltage swing on the grid, EG a 6AQ6A was biased at -12v, but Vg1 was only 2v ish RMS, the OP stage clipped way before the grid could swing more negative, that and the load lines I drew only reached as far as Vg1 -2 or -4v grid lines.

I also found the OP of the triode gain stage started to get distorted, as seen from g1 on the pentode/BT. Couldn't really determine whether this was down to grid current of cutoff.

I still haven't nailed all the nuances of Class SE design, there's a lot going on, it seems to me this OP stage and Class is very touchy,very tricky to get right. In comparison Class AB1 PP is a doddle.

Thanks again all for your IP, Andy.

Re: Single ended Class A OP stages.
 

I'm sure this will be a silly question, but is the OPT suitable for SE class A? SE transformers will usually be gapped to reduce magnetic saturation issues.

I have fond memories of building many SE class A amps using mainly EL84's and the odd 6V6 plus one of about 10W IIRC using an EL34 back when I was a teenager (it was retro then... I'm talking about 1979 ish) and they always seemed to work well without being particularly sensitive to anything that I can recall... other than getting the feedback the right way round from the OPT secondary!

Re: Single ended Class a Op stages.
 

Hmmm! I've been musing on this...

...Class AB1 actually drives the valve into cut-off. That's OK in push-pull: the other valve then does all the work rather than half the work. Grid drive waveform needs to peak-up, but negative feedback takes care of that by predistorting the grid waveform to compensate.

But single-end? Once the valve is cut-off, it completely loses control of what happens to the output load. So, any drive greater than that to just make the anode current kiss the 0mA line for a brief instant each cycle, will distort horribly.

The higher the reflected load, the higher the anode voltage swing will be, for a given grid drive. But that higher anode voltage swing of course gets transformed down by a greater amount.

What you need to remember is, you generally have the luxury of being able to adjust the grid drive level according to your needs, and if you change your load, you can readjust your input. The load for maximum power assumes you can wind up or down your volume control with each change of load, to the point where either the top or the bottom goes into clipping, and when you reach the point where they both go into clipping at the same level, you have your optimum load.

It's quite possible that the stage runs with plenty of bias, and the signal swing is relatively small. With a high HT voltage, current will be small for a given output power, so you will have a high -ve bias. And the signal swing will be a smaller number of milliamps, so the grid swing to achieve this will be small too, superimposed on a big bias.

This is of course, why valves work best with high voltages: if you try and get the same power with 50V HT, you'd need a high standing current (so maybe only -2V bias) and it would need a large drive (10V signal on that 2V bias). Very difficult.

Easy way to tell, pull out the OP valve! Or 'scope g1, if the grid goes more positive than the cathode, it's bad news unless you have a really gutsy driver stage.

Re: Single ended Class A OP stages.
 

I thought AB1 only applied to push-pull amps with class A being single ended only, unless I'm missing something here.
If the OP is expecting much to happen with a single valve, it will need a ridiculously high anode voltage, a very efficient TX and biased rather hot and a short lifespan.
Just re-read the bit regarding grid volts, if it's zero volts or higher (positive), it's time to worry, next stage is red plating!

Re: Single ended Class A OP stages.
 

With no signal, the valve sits at its chosen quiescent anode current for class-A.

In the inductor, there is energy stored = 0.5 times L times (I squared) joules.

It is this which powers the output when the anode swings above the supply rail. The inductor re-charges its field while the anode is at or below the supply voltage.

If the inductor has a gapped core to control flux and keep it out of saturation, this will reduce the inductance and the peak stored energy because it reduces the inductance. So you need to redesign the inductor to still have the wanted amount of inductance AND to have low enough flux density to not saturate, so you need a new inductor on an appreciably larger core and with a gap. You lose in two ways.

David

Re: Single ended Class A OP stages.
 

Maybe it is worth showing the amp schematic, as well as the EL90/6AQ6A plate curves that show the operating idle point (V,I) and a simplistic straight-line showing the nominal loadline for the output transformer being used with a nominal resistive speaker load. It may make the discussions easier to appreciate and hone in on relevant issues.

Re: Single ended Class A OP stages.
 

That's all true of course, but it's beyond the scope of the OP's initial question. For sake of argument, and for analysis of how the valve is operating, it is simple to assume the inductor is sufficiently big not to matter - assume infinite inductance, and later on, calculate the effects of it not being infinite. As frequency rises, the circuit moves closer to the 'ideal' condition.

Re: Single ended Class A OP stages.
 

So long as Andy is working at a high enough frequency to neglect having finite transformer inductance, he can play the game backwards. Take the impedance of his dummy load, multiply by the OP transformer turns ratio squared and draw that load line across the valve's anode characteristic curves. Find the operating Va and Ia quiescent point of the valve and plot it 3V RMS is +/- 4.24V at the dummy load, so multiply that by the transfotmer turns ratio, and mark out +/- this voltage either side of the operating point. He should then be able to see where the valve is swinging with respect to compression or cutoff.

As you said, running to the region where the curves relate to grid above 0c Vgk will guarantee the onset of grid current and a dramatic drop in the Z presented to the previous stage.

You can tell a lot from seeing a load line plot. If you do start running with inadequate inductance, or with an inductive load resistor, the load line shows hysteresis and opens out from a line into an ellipse which makes life a lot more complicated. You have to do it for transmitters to see the effect of mismatch on output power capability. Audio is so much easier!

If the stage hasn't been carefully designed and a load-line fitted to the valve's HT voltage and current capabilities - and then created in realiy by the right transformer turns ratio, then the available output power is reduced from what it could have been either by cutoff at one end or saturation at the other. The right turns ratio and load meets the onset of both limitations at once, and is a best-fit into the available operating area.

Once a real-world loudspeaker is used on the output, with its impedance varying all over the shop, and a few resonances thrown in for good measure, the operating load seen by the anode goes all over the place too and the available power gets severely curtailed as the load line swings around.

Negative feedback doesn't just flatten the gain of the amplifier, it modifies the amplifier's output impedance and it acts to curtail the influence of speaker impedance on the amplifier.

Loudspeaker designers have noticed the trend to transistor amps, with relatively huge amounts of negative feedback and now design their speakers and crossovers assuming that they are driven from an almost perfect voltage source. Drive them with a valve amp, and they are significantly away from their design conditions. The analysis of this becomes very complicated, but it's needed in order to know what's going on when the two things are connected together.

Alternatively you just connect the two things together and measure voltages/currents over a sweep across the audio band.

It's why I laugh when I come across the attitudes that transistors and feedback are evil :-)

In the background of all this, speaker people changed their definition of speaker sensitivity from dB SPL at 1 Watt, to dB SPL at a defined drive VOLTAGE. It got everyone out of the quandry of having to provide a non-flat drive to put a defined power into a very non-flat impedance, and it recognised the dominance of constant-voltage drive from feedback-dominated amplifiers.

Too long, and too deep, maybe, but it's what's going on. Ironically, the simpler-looking the circuit diagram, the more complicated it gets :-) The universe has a sense of humour.

David

Re: Single ended Class A OP stages.
 

Yes,the OPT's is designed for SE operation, came out of a Lowe radiogram driven by an ELL80 which was kaput.

Class AB1 has puzzled me too. When drawing a load line for a PP stage the Class B line is 1/4 of the load,your draw that then the Class A line which is 1/2 the load, then slide the Class A line up a bit until the two lines intersect,that's your bias point. But in a SE stage what load does Class B see? Here on my graph I've assumed 1/2 of the load, not sure that's right. Can't see how you can run a SE stage in Class B without heavy distortion.

With these OPT's I tried a PCL86 first, then tried an ECC88 and 6AQ6A,my reason being the ECL/PLC 86 is designed for amplifying low voltage sources like radio or ceramic PU. In my application the signal source will be an Iphone/MP3, an ECC83 type triode has way too much gain for this ap.

I also tried different OP valves and loads to try and find out why I could only get about 1.5w out this despite different biasing & different loads. Also tried a different OPT, still could only get about 3.5v out before clipping. 3.5^2/5 = 2.45w 3.5^2/8 = 1.5w THD at these levels was over 12%. Figures are approximate.

Looking at my load lines you'll see we get about a 350v V swing, so with a 44:1 T ratio - 350/44 = 7.9v P-P, which is 2.79v RMS. Therefore the limiting factor in all this is the HT & high OPT T ratio, but HT most of all.

I'll have another read of your comments, it'll take awhile to digest em but have include a schematic- valve is PCL86 and load lines. Load line graph is on the Vg2 200v, actual is Vg2 230v but this was too messy to show.

Andy.

Re: Single ended Class A OP stages.
 

Are classes B and AB1 relevant to single-ended audio stages?

Discontinuous modes are OK in resonated RF power stages, or in push-pull stages where one device takes over from another. But in single ended audio what passes the signal to the output while the active device is cut-off?

Room, elephant.... elephant, room..... fez.

David

Re: Single ended Class A OP stages.
 

I've had a look.

First, the cathode voltage at 9V is effectively 'lost' from HT, so you need to pretend that HT is 241V.

Second, you have 240V on g2, less 9V on cathode, so really you need a set of curves for Vg2=231V not 200V. However, they won't be markedly different, save that you'll need a bit more bias.

Forgetting all that, if your operating point is at 250V and 25mA (so 6.25W standing anode dissipation) then the load-line must pass through this point. You can swivel around here, to get different slopes, and see what happens to your limit points of Ia=0 and Va=50V (approx-the 'bottomed' voltage where the characteristic goes around the knee).

Without using a ruler or whatever, at sight it does look to me as though the optimum load is about 8kΩ.

Re: Single ended Class A OP stages.
 

So is the Mullard 3-3 class A, or something else?

Cheers

Aub

Re: Single ended Class A OP stages.
 

If you look at the right-hand side of your load line, where it goes down to low anode currents, look how the curves for different g1 voltages get closer and closer together. If you can see the non-linearity in this picture, you'll see the non-linearity in the output with a good sinewave applied to g1. It really needs a lot more -ve g1 swing (w.r.t. bias point) than it needs positive swing (w.r.t. bias point)

This would need either pre-distortion or negative feedback to combat.

For these simple amplifiers you get accelerating increase in distortion until you decide enough is enough, and that's your rated power.

David

Re: Single ended Class A OP stages.
 

It looks like Class A to me, the operating point is approx. half way along the "straight" part of the Ia-Vg curve so far as I can make out.

Lawrence.

Re: Single ended Class A OP stages.
 

It's class A yes. All "conventional" single ended amps for audio use are class A.

Re: Single ended Class A OP stages.
 

Yes, thanks both. That's what I thought, but post 5 put doubt in my mind.

Aub

Re: Single ended Class A OP stages.
 

Yes, all single-ended linear audio amplifiers have to be.

The largest audio componets tend to be in the bass. So a class B, C or AB stage will be cut off over part of the cycle (with big signals around for class AB) so over part of the cycle of the double basses or drums, the sound of the piccolos and triangles and cymbals will be cut-off as well.

If that isn't intermodulation distortion on steroids, I don't know what is!

Push pull stages get around this by always having one device in linear mode, conveying signal (all signals!) at all times. Class B has the difficulty of getting them to meet up exactly without a glitch. Class AB gets around this by having some overlap. It gives you a place where the gain gets doubled, so care over having a smooth segue from one into the other helps a lot. Many people take the class A everywhere escape route.

David

Re: Single ended Class A OP stages.
 

"Many people take the class A everywhere escape route" Including my good self for my best amplifiers where cost is of little object. Class A has many benefits beyond an absence of crossover distortion.

Re: Single ended Class A OP stages.
 

Douglas Self did an interesting examination of B vs. AB and demonstrates that a properly set up B is much better than AB. From the book "Self on Audio", a fascinating read.

Re: Single ended Class A OP stages.
 

Analysis of the criticality and stability of setup tends to point the other way. But if you do set it up right and maintain it, then there is a potentially smaller zone of perturbed gain.

There is the Blomley design with low-level commutation that could be developed to give very carefully controlled class B crossover, but its current-mode output lets the speaker impedance peregrinations get into the amplifier loop stability.

David

Re: Single ended Class A OP stages.
 

Self's definition of class B and class A/B differs from most....

The Blomley amp, in the versions I know of anyway, was not current output. AFAIK it was never implemented commercially. It kind of moved the need for precision in set up to an earlier stage, which pretty much defeated the object. Walker and Albinson's current dumping concept nailed it on that one I reckon!

IMHO class A push pull is still the non plus ultra for sound quality, though admittedly the vast extra cost, weight and heat are (literally!) a huge price to pay for the degree of improvement compared to the very best A/B models. If you really want the best though...

Of course most of the legendary <30W valve hi fi amps from the likes of Leak, Radford, The Williamson Amplifier etc etc were class A or class A for the majority of their output!
Valve sound quality? Or class A sound quality?

Re: Single ended Class A OP stages.
 

Andy, have you measured the turns ratio of the OPT you are using? Was it 44:1 ? For a 6k primary that would be a 3 ohm secondary.

You show a 6k primary loadline, with a nominal 200V anode swing available to saturation end, and 150V swing available to cut-off end using the 250V 25mA idle point.

As indicated your idle operating point should be moved to 250V minus 9V (for cathode), and minus 17V for DCR (=700 ohm - did you measure that?) of OPT primary, so circa 220V.

Using an EL90 datasheet pentode curve with a 6kohm loadline, would suggest you need to have an idle up toward 50mA, given a 12W rating for anode. If you chose 40mA, and could maintain a 250V B+, then anode voltage idle point would be down around 210V. Anode voltage swing might then have limits of about 40V to 440V, so a swing limit of 170V and 230V. That may be a better starting point to see if you hit saturation clipping about the same time as cutoff distortion/rounding gets too noticeable. That's about 2.5W output capability.

To get a higher output power you likely need to increase B+, and push your idel operating point closer to the 12W max rated limit, or increase your speaker impedance to give a higher primary side impedance.

Re: Single ended Class A OP stages.
 

There's a lot needs unpacking here in your replies, I'm going to have to go through and write it all down. I thought I had a reasonable grasp on the subject but see i was only skating the surface. As Francis Rossi sang, I need to go down, down, deeper and down.

Andy.

Re: Single ended Class A OP stages.
 

What's the value of the cathode bias resistor in the schematic you posted?

Lawrence.

Re: Single ended Class A OP stages.
 

I reckon "Radio Wrangler" has it in his post #18. I'd try some NFB.

Re: Single ended Class A OP stages.
 

Hello,

Just checking a pair of single ended EL84 output transformers salvaged from a Trio W-10 stereo amplifier to hand onto fellow forum member for his ongoing 3-3 projects and I’ve measured a primary impedance of 4.3K with a 8 or 16 Ohm load (separate taps) – now after reading this thread I wonder if Trio were using a slightly lower primary impedance of 4.3K, compared to the more common values of 5K, as used in 3-3, to add their take on the bottoming out and loading of the EL84 in a single ended configuration?

Just a thought.

Terry

Re: Single ended Class A OP stages.
 

They may have been running the EL84 at a slightly higher standing current than the 3-3, in which case a lower load would be appropriate.