How do I get clean mains ?

[GrimJosef]

Originally Posted by oldeurope

What about a "magnetic voltage stabilizer"?

Hmmm ... I rejected these at the start of the hunt because they're big and heavy, relatively expensive and only regulate to a percent or two (I was looking for better than 0.5% if I could find a way of getting it). But as I've learnt more I've discovered that the horrid mains waveform is a problem too. These look like a pretty good way of sorting that out. If I can find a cheap one I might well consider it ...

Cheers,

GJ

[phase_splitter]

I've just found this (very interesting thread) whilst searching for 'KT66'
GJ, have you made any progress?

[GrimJosef]

Well, yes and no I guess. I thought I had solved the problem. I got hold of a 250 watt CVT which weighed a ton nut which reduced my mains voltage variations by more than a factor of 10. Unfortunately some tests using arrays of light bulbs showed that it had a significant dependence of the output voltage on the load current - roughly 7V/amp. But putting a Variac after the CVT allowed me to correct for this, provided the load current wasn't rapidly changing, so all seemed well.

Unfortunately things went pear-shaped when I tried to use the CVT/Variac arrangement to run a Quad II amp (which was one of the things I wanted to do with it). It turns out that while the average current would have been well within the capacity of the CVT, the peaks that are caused by the Quad power circuit's relatively small conduction angle cause real problems. The CVT can't cope with these and as a consequence the Quad's HT voltage falls well below spec. So I've learnt something, which is always progress I guess, but my problem still isn't fixed.

I'm back to trying to design a mains corrector based on a small phase-shifted AC supply. I'll need to think a bit more about the mains waveform, and in particular about how much I care about its variations, but for the time being this is looking like the best way forward. I will certainly also go with the regulated HT supply for my valve tester. I quickly lashed one of these up, based on the RAT valve tester circuit, and it seemed to work OK. So "all" I need to do now is to build a proper version !

Cheers,

GJ

[Don Collie jnr]

Hi GJ!,
Don`t let me discourage you, but sometimes it is easier to modify something, than build from scratch. That way a lot of the hard work is already done. [chassis, wiring, dial markings, knobs etc] Have you concidered getting a fleamarket valve tester, and modifying it to your specific needs? [Besides, it`s often really satisfying to improve on the experts...!] I`m not clear as to what is the object of your project. What do you want to do, that cannot be done by a commercial valve tester? Even new valves are subject to tolerances on their various parameters of typically + or - 20%. Rarely was it necessary to select valves to tighter tolerances, because the technology of the day usually didn`t require it. Hi -Fi purists who try to balance their valve`s gm to sub 1% values, for example, amuse me, because I don`t believe it is even possible to notice a much larger difference. [...but the subjective is a matter of opinion though, I suppose] You state that you want to make precise measurements of gm under DC conditions, but then say that you are looking for a pure sine wave. It is an easy matter to provide a highly stable Anode, Screen, and Heater supply, but much more difficult, as others have mentioned, to provide regulated amplitude sine waves of low distortion. If you take the Very stable DC approach, you will *still* find that the operating point [ie the gm, etc] of the valve will drift here and there, with time, and temperature. This drift in parameters is the nature of the beast, and one of the main reasons valves have been replaced in precision equipment by transistors under heavy negative feedback or OP Amps. If you want very stable gain, Op Amps are the only way to go, but for audio amplifiers, a change in gain [or gm] of 3dB, or 30%, is *just* noticable by the human ear - lesser values are subjectively not noticable. Does this answer the question you are asking? More gm to you!!.......................................Don.

[GrimJosef]

Hi Don,

Thanks for your thoughts ! As time goes by I'm getting further and further up the learning curve.

First I'm afraid that fleamarkets around here don't seem to be full of valve testers. Instead they all seem to sell on the internet for hundreds of pounds. In fact I've acquired an old Taylor 45A, currently awaiting restoration. But this will certainly be a bigger job, if I'm going to do it properly, than building from scratch a version of the much better RAT tester that Mike A introduced me to (see above). There's a thread entitled "AVO VCM Mk IV tester" over in the vintage test gear forum where RogerWalker says "Good advice from Robin Birch leads me to believe that there is no silver bullet to fix this problem (i.e. getting a commercial tester to tell him the truth) - a lot of components have probably drifted off spec and it'll be a long slow and painstaking process to get it back into spec". I'm afraid that he is right.

You ask "what is the object of your project ?" and "what ... cannot be done by a commercial valve tester ?". You are not the first to ask these questions and my answers are certainly developing with time and with the guidance I'm getting here (for which I'm very grateful !).

My original object was to learn how to get the best out of my Quad II amps. I agree that my ears will not be able to tell the difference between 0.1-0.2% THD, which is where the amps seem to run if I just pick valves at random, and the 0.02-0.03% I can get by carefully selecting the valves. But this is my hobby, so even though I can't hear it I'm nevertheless interested to understand what's going on. To do this I need to know how different valves behave at their real operating point. For the KT66s, for example, this is with about 330V between anode and cathode and about 72mA cathode current. So each valve is dissipating about 24W. By contrast most commercial valve testers check them under very different conditions, with a couple of hundred volts of unregulated AC across them, using the valve itself as a rectifier and dissipating relatively little power. As you say the behaviour of valves is very sensitive to temperature and I'm afraid the commercial testers never get them properly hot.

You also say that the operating point will drift here and there with time and temperature. I agree about temperature but actually, although I don't have a lot of experience yet, I've been surprised about how very reproducible the valves are. However I only discovered this after taking a lot of care to ensure that I was testing them under truly reproducible conditions. It turns out that most of the drift I'd seen before this was not drift in the valves but drift in the test conditions. And most of this, in turn, was due to the miserable quality of my wayward mains ! Which is where this thread started all those months ago...

Cheers,

GJ

[Don Collie jnr]

HI GJ
I posted a previous reply, but it didn`t make it. Oh well... I `m not quite sure what you are doing. Could you explain the purpose of your tests, as this will determine your method, and means. Here`s what I *can* say : Those Amps have a very low output impedance [less than an Ohm], so if you feed one into a Voltage step up transformer that has good Voltage regulation [Toroidal?] you will end up with a well regulated AC voltage source. It might be a good idea to use some VDR`s to prevent back-EMF`s from the transformer damaging the amp`s output transistors, though. You could vary the output voltage by adjusting the output of the low distortion AF oscillator that is feeding the amp. Frequency too. [you`re probably streets ahead of me here]
A 3 terminal regulator of some sort [LM317?] would be an easy way to get a stable heater supply, but off-the-shelf HV regulators are less common - you might be best to "roll your own"... unless you`re lucky enough to have a Heathkit purpose built one.
What sort of accuracy are you wanting to measure? The valves I`ve worked with [I was a Tech looking after (valve) radio, and TV transmitters, etc] are sort of + or - 20% devices. I remember a valve exciter for a TV transmitter, that needed specially selected
valves used in frequency divider circuits. We couldn`t get the darn thing to stay working properly for any length of time, because the characteristics of the valves would change with age, and temperature. It was a critical circuit, requiring close tolerance components, and all the valves we could get, couldn`t be kept to those close tolerances. Perhaps they can now - I don`t know about that. Eventually we solved the problem by installing a solid state exciter. Might have used 7400 series logic for the frequency division, not sure because it never broke down.
I hope this info is of use. Cheers,........................................... ........................Don.

[Don Collie jnr]

I remember an H/P [H/P 606A, I think] manual that says that the best way to test a valve is to put it in the circuit in which it is normally used, and look at the Voltages, and currents. I suppose that eating the pudding is the proof of it. FWIW,............................................. .Don

[GrimJosef]

Hi Don,

I think we may be talking at cross-purposes. My fundamental aim is to understand what it takes to get the best measured audio performance from my Quad IIs. I find that the level of distortion does depend on which valves I pick from my reasonable collection of MOV KT66s. Some pairs are good, others not so good.

So I wanted to measure the properties of all my valves (mainly Ia and gm at 330V anode-cathode and with about -29V on g1) so I could work out what was good about the good pairs and what was bad about the bad ones. I lashed up a basic tester based on unregulated DC power supplies for the HT and grids and a simple AC supply for the heater. Then I found that when my mains voltage changed, so did all the voltages on my tester. It was taking ages to get a clean plot of Ia vs Vg1 which is what I needed to work out gm. This is when I decided to try to stabilise my mains. I was going to use a transistor amp to do this. So, if you like, the valve amps were to be tested and the transistor amp was going to be a part of the test rig. Is this where the confusion is ?

I'm sure the advice about testing valves in situ is good in many ways. I have (at least) three problems with it though. Firstly the Quad II HT supply to the KT66s has a lot of mains ripple on it. This means that precise measurements are hard to make. Secondly while I can quite easily measure voltages inside the Quad IIs it's quite a bit more difficult to measure currents. Baiscally I'd have to disassemble the wiring and insert a meter or a monitor resistor. Doing this is risky in Quad IIs, whose resale value to the audiophiles depends on their level of originality. Repaired wires or terminals may not matter to me or you, but we're not the customers ... Thirdly I sometimes buy second hand valves and I'd like to be able to test them somewhere other than in an expensive amp. I've seen too many blown transformers in the Quad IIs to be entirely happy with just dropping any old unknown valve into them So I do need some sort of robust tester capable of putting a lot of volts onto the valve and surviving if it should arc over.

Cheers,

GJ

[flyingtech55]

Hi GJ

I've come to this thread late....however. Have you considered using a computer UPS. You get two types; on-line and off-line. The on-line type generates 240V AC from the battery pack all the time. On the other hand you could remove the mains supply and it would run from the battery only for the duration of the test.

TimR

[Brian R Pateman]

While the UPS provides a nice spike free AC output it isn't very suitable for this sort of purpose. For one thing unless it's very large (and commensurately expensive) it is only designed to provide a full output for long enough to shut the PC down and save data.

The second reason is that the AC is derived from an inverter driven supply and is rich in harmonics and HF noise.

Regards,

[GrimJosef]

Hi Tim,

I had the same idea and got as far as trying a cheap UPS unit from Maplin. I guess it must have been an off-line one because it just seemed to pass the mains fluctuations through when it was plugged in. I don't recall exactly what happened when I unplugged it (sign of advancing senility - I need to write more and more stuff down these days !) but whatever it was convinced me that this wasn't going to work for me. Brian's probably right and it wqas something to do with the shape of the synthetic waveform.

On a completely different subject though I did use one of these at work to track down the source of 50Hz noise in a complex diagnostic rig involving spectrum analysers. The trick was to watch the 50Hz spur on the LF analyser and work round all the other bits of kit, powering just one of them from the UPS. I knew when I'd got to the one which was making the noise because the spur frequency then shifted to that of the UPS. Bingo !

Cheers,

GJ

[PJL]

You are trying to perfectly match valves for use in a quad under realistic operating conditions? The valve drives a transformer with a known impedance and the anode voltage swings about HT volts. DC testing is OK for small signals but peak current if it stays in class A is 120mA and peak anode voltage is 600V+?

I wonder if putting a small value resistor in the common cathode line might be better as you could monitor the combined current on a scope?

[Don Collie jnr]

Hi GJ,
You have to be very careful about pressing the "Enter" key - another message lost. I hope that you do not feel that I am telling you your own business - I`m not tryng to do that. I have read your replies, and see what you are getting at. I have a technical background, and am not an Audiophial, and approach the same problem from a technical point of view. A constant Voltage transformer, in my experience, produces a "squarish" output waveform, and likes to be operated at full load all the time, but you could clean the waveform with a filter. The same filter would clean the mains input Voltage waveform to a transformer, or the secondary output voltage. You`d be dealing with lethal Voltages, so I`d be very careful. The AF Oscillator feeding an amplifier, and perhaps a step up transformer, would probably work OK, in my opinion.
I`ve built HV regulated power supplies, so I might be able to help you with a design, and know some of the pitfalls - PM me if you like. Everything you ask is possible, but more specifics would be necessary, I feel.
More mA/V to you!!............................................. .........................................Don.

[retro video]

I remember once getting a pro audio magazine. They always put flyers for various equipment suppliers in the magazines... damn things fell out all over the place! This particular flyer caught my eye - it was for a mains cable that allegedly minimized the effects of a 'dirty' mains. I think they were trying to say it was in how the cable was weaved. Anyway, you'll probably find such things in places where you can buy pro (studio) audio equipment, or hi-fi buff shops. It was over £100, though... and this was about five years ago!

[Darren-UK]

This thread's been the subject of yet another Post Report this morning and has also had OT posts either removed or edited.

As a result a disproportionate amount of Moderator time has been consumed and the thread is therefore now closed.