I think it all started off with the LL1660. Now there is a complete family, which is always wound like A+A+B+B : C+C. (See picture on the left). This configuration offers many possibilities, because primary and secondary, each can be connected in series or in parallel, or combinations of series-parallel. Such as put A+B in series, and those two "new" windings are put in parallel. Then, when the total Ratio (N) is changed, it becomes fully another kind of transformer, and many different ratios exist. To offer even more, this transformer family can all be used forwards and backwards, and may be wired for Single Ended of Push-Pull or Parafeed. (Three fully different transformers). Cores can be Amorphous or Steel, static shielding can be added on customer order, and for some types even silver would versions are available. The mother of this family was LL1660, wound as 1+1+1+1 : 2.25 +2.25, but several new types have been added with other windings ratios. |
Windings RatiosI would even say, a large part of the possible applications has not been discovered yet. In any case, in terms of versatility, there is nothing like it from another company, which comes near. LL1692A has the smallest windings ratio, and LL1689 the highest. They all can be used in both directions, but not all of those theoretical options are meaningful. Like indeed LL1689 can be used as a 1:18 step up transformer in theory, but it will be very hard to drive such a load. EE27 board could do it in theory, but it gives quite a capacitive load. Lundahl only published the applications that work well, and here is a short overview of what is in the datasheets. Though Per Lundahl told me, new applications (ALT-...) can be, and will be found.
Generally, the choice is first for output impedance you NEED, and then for the gain of the whole stage you NEED. If you don't know these two numbers. nobody can choose a transformer. Well you can, but the choice may be not good. So not first choose the tube, then choose a transformer to it, and then ask: What is the gain, ignore output impedance, and find out afterwards if you have an issue with microphonics or not. (Don't be confused by this sentence... just read the next part: |
About Noise ReductionPerhaps you recognise this situation: A specific tube seems noisy, and selected replacements are noisy too. But, some other people write, selected versions are not noisy at all. So how come? Do they have different tubes? The answer is: No, they have a different amplifier! It has a different circuit. Microphonics of the SAME tube can very high in one amplifier, and microphonics can be totally gone in another. The MAJOR effect comes from appropriate circuit design. Specially with tube pre-amps the following is the #1 problem: A fully wrong design. Save the stepdown output transformer, adds tube noise, by definition. So my advise is, use an output transformer, if you want low noise. Do not try to "believe" how this works by reading forum talk. Even many professionals try to work without output transformers, and are not fully successful. Which for them is even good, because they sell low-noise, selected tubes, and people are forced to use such. But you see, output transformers are found in every decent piece of studio equipment. That should speak for itself. Please look at the Notes for a few unsorted hints and hopefully good ideas about noise reduction. Tube microphonics. Any kind of tube noise, also hum and white noise, will be reduced SIGNIFICANTLY with a step down transformer. The extra costs are even less than expensive selections of exclusive NOS tubes by selection magicians. The way a step down transformer reduces tube microphonics can not be explained here quickly. We just mention this at some places here. If this is new to you, try to learn more about it. (Read the Notes). |
|
What determines the transformer choice?Well, strangely, sometimes a mistake determines the choice. Much of this text part is inspired by my conversations with my late friend Roger Modjeski, from RAM Labs USA, one of the best circuit designers I had the pleasure to know. He also had a school at Berkeley with technical classes on amplifier design. Roger told me about this major source of mistakes: People fall in love with a particular tube, and then a partner transformer has to be found. He actually wound his own transformers. Users say, the transformer should "fit well" to the tube. Sure this is important, nobody wants a transformer that doesn't fit well. But how can it fit, if we are not even aware of the INTENDED use of the tube? With intended use, we mean, what the tube manufacturer intended, and not what users intend to do with it. Often mainly inspired by the "looks" of the tube. I see it so often, lovely collector's tubes have already been bought on Ebay, and now a transformer must be found, to close the ideal marriage. Which then, may be possible, or just as well not be possible. Like a glass tube is supposed to sound more transparent than a steel tube, and feminine shaped glass sounds more gentle than cylinder glass, which seems to have a harder sound. Then in the end, we are presented a high signal tube, already purchased, which must fit into a low noise application, with a "well-fitting" transformer. So forget for a moment about the optics. Look at the DATA SHEET. Really, if you just read the opening TEXT of the data sheet, you already have the answer! What is the intended use? Are they high gain tubes, or low signal tubes perhaps? That is often not the same. Though for some tubes it is. And here comes already a good negative example. Look at a high gain, low signal tube, declared as such by the manufacturer. This tube will sure have tube curves with large signal swing on it. But these curves are not published to use them as such. Though many people just will do so, saying why can't I use that when they publish it? I will answer that. These curves are only to find the working point needed. If the manufacturer says low "signal tube", you need to accept, that's what they are. Then can, but they will not work well at high signal. Unless the manufacturer was mistaken, but normally they are not. If these are used at large signal swing, that is not impossible, but they will give a lot of distortion. Which you have to deal with, for instance with a high feedback ratio. Such a tube for instance is ECC83. If you see those used as driver tubes, it MUST be with a lot of negative feedback, because otherwise this tube will act a a distortion generator. Their intended use, and you can read from ECC83 (12AX7) data sheet, is for small signal, low noise applications. Another point, is it a low impedance tube, or a high impedance tube? Or a transmitter tube? Or a power supply regulator tube, such as 6C33. Sure, with reduced performance, a tube can be used in another application. Only, the thing is, the more you move away from intended use, the more you will get to learn what those limitations mean for you, and how related problems look like. Another good example is 6C33, which is a series regulator tube for DC power supply, only made for this purpose and no other. This tube is intended to be used in a DC coupled circuit, with very high feedback on the working point. (Because the working point is the output voltage of the power supply). For that reason they are neither specified, or tested for good DC stability. There was no need, because feedback in a power supply circuit is extremely high always. As a result, people say you can not use them unmatched, and they drift a lot, and they are unstable. Well yes in an amplifier, and most specially when foolish designed put them im parallel. They will work fine however in a power supply, unselected, which is how they were intended. To prevent problems, forget about internet forum bla bla, , because risk of listening to fake experts is not a risk there, it is a certainty. There is much better writing available, from the real experts. It is called TUBE DATA SHEET. With writing I mean the text part of it. In the first 10 lines, they say what the tube is intended for. You will find no 300B data sheet, saying this is a low microphonics tube. Same as you will find no 12AX7 data sheet, saying this is a good driver tube. Some things just don't work as well as some whish. So begin with using a tube as intended, and for sure there will be a Lundahl transformer already existing for this. Better is choose a transformer first! So when you choose an inter stage transformer, Lundahl will give some hints about required driver impedance. The all you need to do is, find a driver tube with that impedance. I assume you already know, the output impedance of a driver tube goes up or down drastically at lower or higher bias current? That is good, so we don't have to explain this. Finally, the choice is as simple as that: First choose the transformer (explained below here) , and right after that you will have a signal level, gain and impedance the tube should have, and you can start looking for such a tube. |
How to select a transformer and it's connection scheme?This first question is: Single Ended, Push Pull or Parafeed? Single Ended. This needs only one tube, is easier to build, and easier to design. However there is very high DC current trough the transformer primary, and this pre magnetizes the core. High Current means low voltage in practical situations, like a circuit design.
Push Pull. This eliminates a lot of the SE disadvantages, but it has more complicated electronics. However the transformer is much easier to build. Designing a PP circuit without understanding how to do this from the tube curves, can not be neglected with PP as it is with SE. So beginners do not like PP, and often end up not liking it in general. Also PP has a much better distortion behavior. Add to this the higher efficiency of PP, and you will find, some applications just asking for PP. Then there is Push Pull possible, made of two SE stages, running on one PP transformer, and with a great overlap, you can use PP and SE for many applications. .
Parafeed. This is really the BEST. Period. Also least used, because people do understand quickly how it works, but do not understand as quickly why it works so much better than Single Ended. So the disadvantage of having to use the extra choke is out up front, and they do not look at it any more. This is really a pity, because the choke can be a low quality type, and still the Parafeed transformer will do an much better job than a conventional SE transformer. Read here about Parafeed. The Parafeed connection scheme differs slightly from normal SE connection, because it grounds the driving winding at one end, vs connect it to the V+ with SE or PP. Please let us know if you need a parafeed connection for any of the transformers, we can add it to the selection table at the EE27 main page. |
Simulating LUNDHAL frequency curves, using S-parametersFor this, we can not quickly explain here, how it's done in detail. But... it is not difficult to do, when you understand a little bit about software. If you only want to press buttons quickly, you will still see the example we give here, does the job for you indeed. When you like the results and want to do more with it, you need to work yourself through the basics of this. If doing so, and you get stuck, the schematic we give here for downloading, is always a good anchor, and you can try to modify it. That is easier as beginning from scratch. So what you need for this is, LTSPICE installed. This is freeware from Analog Devices company. Then, download the circuit example (link below here). For this, use the schematic we give you. You can only open it with LTSPICE. After opening, the program can edit the schematic, and it's even a good, free schematics drawing program. Logically the parts in the schematic have all values stored into them. So a coil has not only it's inductance stored, but also it's series resistance and parasitic capacitance, etc. (This was already done by me). Once a schematic is finished, and all components parts data is entered, LTSPICE can now run the schematic as an electrical simulation, and show the test results. That is why you see those two windows. At the analog devices website, there are some tutorials, etc, but this is a lot of work to read. First, to replicate the below picture quickly and live on your own PC, do the following: Do it exactly like this:
After some seconds, an (empty) curve chart window will appear. Now the fun begins. Move the cross hair precisely over the schematic, at any wire, node, or component. At some moments, the cross hair will change into a probe tip. If that happens, give a left mouse click, and a frequency chart will appear of that part you touched. Do this with the output of the circuit, and you see the chart as below here. When you click on the middle of a component, a current probe will appear. So you can measure the voltage and current wherever you like. |
Wow! You made it until here :)
This is about the Western Electric 92A from 1935, driving a pair of 300A with a PARAFEED to Push-Pull transformer.
Search the table above here for "parafeed". There are only some parafeed to output configurations added by me, but when you are interested in a parafeed to PP, let me know, sure the EE27 board can do it.
300A was in production only very short, and replaced by 300B, which is essentially the same tube, but 300B has not this bajonett side pin on the UX4 socket like 300A has.
Note, the transformer schematic suggests a double windings package, same as Lundahl is doing, but I only expect this. Perhaps somebody has both primary windings in series. This reduces capacitance. Here is the full schematic. WE92A has one of the nicest schematics ever made. It has very intelligent schematics, such as C9 and C10 at the 300A, which I think are to increase the bandwidth of the output transformer. Make also good note of the PARAFEED schematic of the 202A tube. Parafeed is normally done with a choke and a zero-DC current output transformer. Now here at WE they did something interesting, they use a very high anode resistor R10, to feed this little 262A tube, from a VERY high voltage. This effectively turns R10 into a current source. Not an ideal one, but still a very good one. Almost like a choke, but not with the problems chokes had in 1935. The return path of the AC signal is not to ground, but look what they are doing: To the cathode of the 202A. Some like to call this "Ultra Path" today. But it is like I always say, there were no more new tube circuits invented ever since Julius Futterman invented the OTL. Some advantage I see here, the power supply hum gets cancelled out with the right balance of C1 and C2. Any power supply hum which finds it's way via R10, into the parafeed transformer, would flow via C1, and generate a signal across the 262 tube's cathode resistor. To make this work, C2 must be in a certain ratio to C1, which is as we can see here: 16x. This is a truly remarkable schematic, almost 100 years old. Here is a full copy.