Speaker matching

[SurreyNick]

Unless someone can recommend a better way I plan to adopt F.J. Camm’s method for matching a speaker to an output valve, this being required for my homebrew TRF.

Essentially Camm’s method involves attaching a millammeter and a wire-wound resistance of 10,000 Ω to the speaker output terminals of the receiver (see pic) and then in 1,000 Ω increments taking a reading of the current in milliamps and from these calculate the valve output in milliwatts. The highest milliwatt output reading thus determining the optimum speaker impedance in ohms. This is all very straightforward, but raises three questions:-

1. Should the receiver be switched on while carrying out the readings? F.J. Camm doesn’t say.
2. How do I produce the 10,000 Ω wire wound resistance? Camm just says to make up the resistance using bare resistance wire wound over a thin former screwed down on to a flat piece of wood covered with white paper onto which a scale is drawn by dividing up the length of resistance into 10 equal parts. No problem with the construction, but he doesn’t specify the material for the former, its diameter and whether flat or round, or the wire gauge, nor how to work out how many turns (help!).
3. Once I have an optimum figure for the speaker impedance what is my chance of finding a speaker to match? I need a small speaker circa. five inches in diameter. I might end up with an optimum impedance of 5,000 Ω, or 6,000 Ω, or 7,000 Ω, or 8,000 Ω…etc. How easy is it going to be to find one to match? Do suitable speakers come in such a wide range of values?

Any and all help would be much appreciated.

Nick

[GrimJosef]

Honestly I think Ed Dinning's advice in your previous thread (post no 10 here https://www.vintage-radio.net/forum/...166#post707166) would be a very much quicker way of solving the problem. But if you want to try the F.J.Camm technique then:

1. Yes, the aim is to measure current and the valve will need to be powered up or it won't supply any current for you to measure. Another point is that this technique relies on measuring AC current through the resistor (or AC voltage across it would do just as well). So the valve will need to have an AC signal fed into its grid or again there won't be any current for you to measure. You want this input signal to be constant, otherwise you can't be sure that changes in the output current are caused by changes in the resistance. So you'll probably need to isolate the output valve's grid from the rest of the circuit (make sure you do this in a way which doesn't upset its bias) and feed in a constant AC signal from an oscillator. Also don't forget that the output from this valve will be coming from its anode. MAKE SURE THAT YOU UNDERSTAND HOW TO AVOID GETTING AN ELECTRIC SHOCK FROM THIS BEFORE YOU POWER THE VALVE. If you're in any doubt whatsoever about this then write your proposed procedure down in detail beforehand and get someone more experienced to check it or, better still, get someone more experienced to come along and actually help you do the measurement.

2. These days making your own resistors isn't really necessary. If instead you buy a 1k, a 1k8, a 3k3 and a 3k9 and connect the four of them in series you will get 10k in total. By shorting out one or more of them you will also be able to obtain 1k, 1k8, 2k8, 3k3, 3k9, 4k3, 4k9, 5k1, 5k7, 6k1, 6k7, 7k2, 8k2 and 9k (have I missed any ?).

3. As Ed said earlier finding a high impedance speaker is likely to be tricky. It will be much easier these days to find a low impedance speaker and use an output transformer. (But if you're going to do that then perhaps you should just eliminate the choke and use the transformer between the anode and the DC HT rail as is far-and-away the more common approach these days.) Whichever speaker you do choose bear in mind that its impedance will vary a good deal with frequency. It might say 3 ohms on it, but it could easily be less than 2 ohms at some frequencies and 10-20 ohms at others. So don't get too hung up on an exact impedance match.

Cheers,

GJ

[Radio Wrangler]

No!

First of all, you do not want to 'Match' a speaker to an output stage. The word match has a very specific meaning: To load something with the exact conjugate of the source impedance of the thing. This comes from the maximum power transform theorem, and will extract from that source the maximum power which can be theoretically extracted. This avoids all practical concerns over trivia like the power a source is capable of giving before damage or other unwanted side effects.

Lets take a reasonably realistic example. Let's take 50Hz mains and neglect capacitance and inductance. Let's say you have a reading lamp plugged into a 13A socket. If we switch the load on and off, while we measure the voltage at the socket we can calculate the source impedance at the 13A socket. A 60W bulb takes 60/240=0.25A. Let's say we find this affects the mains voltage by 0.1 volts so the source impedance is 0.1/0.25= 0.4 Ohms

So to get true matching, we need a bigger light bulb, one which looks like 0.4 Ohms when lit. It's going to drop the voltage at the socket down to 120v, because of the 0.4 Ohm source impedance. So 120v into 0.4 Ohms is 300 Amps and therefore 36kW

The matching calculation takes no account that the cabling won't take it, that there is a 30A breaker feeding the ring and that you don't want that much light. What it has given you is the load which would take the absolute max available power from a 240v source with 0.4 Ohms source impedance. Note also that the wiring will also be dissipating 36kW if you could do it. Matching is very poor for efficiency.

What matching is good for is controlling reflections with transmission lines, with optimising signal to noise ratios.

So let's go back to square 1.

Say you have a 300v power supply. Say your amplifier is single-endedwith a beam tetrode and you want reasonable quality.

Your valve anode is powered through an inductor (actually a transformer, but think of only the primary just now) And the copper resistance of the primary winding is very little, so the average DC voltage across it must be close to 0v. So the valve anode must be at an average 300v Let's say the anode can pull down as far as 50v when the valve is at maximum current.

To maintain the 300v average with a sinewave signal the peak must swing 250v above the supply, to 550v. So the anode swing is 500v p-p = 250v peak =250/root2 RMS = 177v RMS

Nw, say we want to put 5W into a 3 Ohm speaker. Well 5W at 177 V RMS means

5 = (177^2)/Rload so Rload = 31250/5 = 6.25k Ohms.

So we need a transformer with an impedance ratio of 6250/3 = 2083 which means a turns ratio of 45.6 : 1

Now 500v swing p-p into 6250 Ohms is a swing of 500/6250 = 80mA so our valve could sit at an average anode current of 40mA, swinging up to 80mA and down to 40mA, but we do't want to run so close to cutoff, so like the voltage swing earlier, we can assume a minimum current of a similar proportion, so it would go down to 6.6 ma, and up to 86.6mA

The anode transformer must be planned not to saturate with 46.6mA quiescent, and handle 86.6mA peak.

quiescent dissipation in the valve is 300v * 46.6mA = 13.98W at quiescent. So we're being just a bit pushy for a 6V6 yet a 6L6 isn't needed.

This has not matched the speaker to the valve or vice versa. We picked an output power and a power supply voltage, and out of that flowed the O/P transformer ratio and DC current requirement, then we could pick a valve.

Usually this sort of thing is done graphically, by drawing a 'load line' on a set of anode curves for the valve.

David

[SurreyNick]

OK, thank you. It really is good of you to help and I think I understand. But here's the rub...

The set I am building has all the components on show, baseboard mounted and I have taken considerable time to accumulate original, late 1920s / early 1930s components. This homebrew is very much a showpiece, the components are all in pristine condition and invariably in gorgeous Bakelite housings. But I do want to get as good sound reproduction as possible too. I should mention it is a simple three-valve battery powered TRF.

I have the output valve. It's a Triotron SP2 (audio output triode) from 1927. It compliments the two other early Triotrons that I have. I also have a vintage output transformer and the power pentode output choke referred to in my other post. What I don't have is a speaker. To complicate matters the primary and secondary impedance information for the transformer I have is not known, although I know I could work it out with some simple test equipment, I'd just rather not bother if I can avoid it.

So, I want to find a speaker which will match my other components and not find a transformer to pair the output valve and a speaker. Wrong way round I know, but that's what I'm trying to do.

The simplest solution, I assumed, was to employ the output choke plus a 2uF condenser and then, using FJ Camm's method, work out the optimum speaker impedance and try to find one accordingly. (Please note it is FJ Camm who refers to 'matching of the speaker'.) If the output choke proved unsuitable then I was going to repeat the exercise with the transformer and then see if I could find a speaker to match this component. Camm didn't mention the need for an oscillator to employ his method, although I can see the sense in using one, but I don't have an oscillator.

Now I don't know which way to proceed

Nick

[cmjones01]

If you're using an output choke and 2uF coupling capacitor, you'll need a high-impedance speaker, of course. Are there any of those left in working condition? I guess in the late 1920s the speaker might have been moving-iron, so there's some chance its high-impedance coil would still be intact. But I'd say they're quite rare beasts and you're very unlikely to have a choice of different impedances.

In this situation I'd make do with what's available. If you can find one, which will be the challenge, any speaker with an impedance between a few hundred ohms and 10k or so will make a satisfactory noise. If it's not optimal for the output valve, the maximum undistorted volume will be limited, that's all.

If you've already got an output transformer, and can identify the primary and secondary, I'd just hook it up to whatever low-impedance speaker you can find which fits the period. Again, if it's not perfect, the volume will be limited, but nothing should get damaged.

There is a third option: I remember seeing a speaker from the 1930s, branded 'Stentorian' (wonderful name!), which had either rotary switches or multiple terminals which allowed it to present a wide range of impedances. I'm sure it was all done with transformers inside.

Whatever you do, it's likely to be limited by what period components you can find.

Chris

[SurreyNick]

Swell Chris, thanks.
I'll be posting a 'speaker wanted'
Nick

[GrimJosef]

Originally Posted by SurreyNick

... So, I want to find a speaker which will match my other components and not find a transformer to pair the output valve and a speaker. Wrong way round I know, but that's what I'm trying to do ...

That's fine. I absolutely agree that appearances can matter a lot.

In that case the first thing to do is to have a look round and see what you can find in the way of good-looking low-impedance speakers. They should be more common than high impedance ones and you might even find you have a fair choice of nominal impedances (with each speaker's impedance varying with frequency though, as I mentioned above). You could then install your isolating capacitor and output transformer and use the Camm technique on its secondary winding, this time going from, say, 1.8 ohms to 16 ohms or so (put a 1.8 ohm, a 3.3 ohm, a 4.7 ohm and a 6.8 ohm in series and short various ones out to get the test resistances you want).

Cheers,

GJ

[wireful3]

I suspect you will find that most speakers in the late 20s early 30s were all high impedance compared with modern speakers, generally intended to work directly in the anode circuit of the output valve.

If you can see a copy of "Radio Radio" there is a section showing speakers from the very earliest times. Some of them had wooden cabinets with cloth covering behind a fretwork grille. It would be possible to fit a modern speaker and transformer out of sight in the case. However from what you say I think a horn speaker would look more convincing. They do turn up at auctions and BVWS Swapmeets but often have no driver or an open circuit one. If you can find one of these in working order you would not need any transformer to change the impedance.

[kalee20]

The question you ask is not a trivial one!

Radio Wrangler's answer is correct. If you know the DC power supply voltage and either the power you want, or the current taken by the output valve, you can work out the load impedance. But it does assume you have the luxury of as much gain as you want (ie a volume control you can wind up as much as necessary until distortion starts). There are a few assumptions made, and it doesn't work well for triodes.

On the other hand, F J Camm's method is valid if you don't have much gain, and the input signal to the valve is limited at a rather low level. Under these conditions, you might well want to match the speaker for optimum power. It turns out that if the valve is a triode, the result you get is reasonably satisfactory. If it's a pentode or beam tetrode, Radio Wrangler's method will give you less volume than a higher impedance load (which is where FJC's technique would lead you), but you won't be so close to the threshold of distortion.

With the components you are trying to obtain, you won't do better than follow GrimJosef's suggestion when you are first up and running.

[Radio Wrangler]

Thanks, Kalee, I'd posted that before the output valve choice and radio design had been revealed, so I'd headed off down the 6V6/6BW6/EL84 path.

With a vintage triode like that, and being pretty much starved of gain, the set really will be in the regime where true matching of the output will help, not damage. Output power is going to be in the region of 0.25 Watt, maybe a little more.

The speakers used in those days were high impedance to suit valves of the period without transformers, and they were horns to get the best efficiency. You really do need a horn speaker.

Unfortunately high-Z speakers and phones were wound with very fine wire and the enamel wasn't so good. Many have succombed to moisture ingress through cracked enamel and then green-spot corrosion. Something suitable will be hard to find.

David

[mole42uk]

Unless you take a low-impedance speaker that looks right and wind a new voice coil for it. Not trivial, I have done a couple of speaker coil rewinds and they are real piggies to get centered in the gap, but it is do-able.

[G6Tanuki]

I'm wondering if it would be possible to create a suitable high-impedance speaker by attaching a reed-and-horn to an earpiece salvaged from a set of high-impedance headphones?

As mole42 has said, high-impedance speakers are distinctly rarer than Soho virgins!

[SurreyNick]

I had no idea at the outset how intellectually stimulating this hobby would be
Sorry if I sent you down the wrong path David (Radio Wrangler) by omitting an important point about the set design, but please be assured the reply wasn't wasted on me. It really has everything this pastime - mental satisfaction as well as tactile satisfaction.

Do please keep the information coming...

Nick

[AC/HL]

Back in the 20s this was all there was available, there was no confusion with later technology. It looks very simple & basic, but that can be deceptive once the contemporary knowledge has gone. Literary styles were much briefer than the blow by blow hand holding of modern offerings.

[Ed_Dinning]

Hi Nick, I kept my explanation simple, but if you enjoy an intellectual challenge then delve into the theory as the other guys have outlined.
This is Engineering and there are often no absolutes in the way of answers, one of the reasons I chose it as a career. You can often get 75% of what you want for 25% of the cost.
There is also the joy of the "lash-up" and first getting it working, hence my urging to try it, then use an iterative process to get it exactly as you want in terms of the electronics. The cabinet work is something totally different and will no doubt fall into place with your woodworking experience.

Have fun, Ed

[SurreyNick]

Thanks Ed and to everyone else too.

I shall be sure to do as you and others advise. The more I learn the more I want to know I will however endeavour to keep my focus narrow. I'd like to develop really sound knowledge and expertise in pre-war wireless technology, as building sets from circuits and components of that era is what really interests me. I like the relative simplicity and the inherent beauty in components and sets of the 20s and 30s.

I will also be sure to follow your advice in first 'lashing' it together to make sure everything works as intended before I take the irreversible step of screwing everything to the cabinet top.

Regards

Nick

[Radio Wrangler]

A modern speaker could have its voice coil re-wound as an alternative to having a transformer, but either way it will still give that rather low efficiency of modern speakers.

I think you have two mountains to climb... getting the impedance without a transformer, and quite separately, getting the efficiency of a period horn.

The people running single-ended triode hifi amps use modern horns (Lowther for example) to get enough efficiency for a few watts to be useable.

David