because the 50 year old PCBs are crumbling and need replacing. Gives me an opportunity, should I want to rebuild it improving on any aspects that could be improved upon. It could end up that in fact a change to the input stage and a balance pot for the phase splitter is all that is required to make it as good as it can be with the transformers I have. But no harm in exploring other options to see if they are viable enhancements.
Fairy nuff
That covers a good 80% of this forum. People do things because they can. When you don't have to earn a living from it why not?
Fully agree
OK that bit I don't get so will go away and scratch my head until I can agree or disagree.
accept that. The STA-25 as stock was probably that low.
Any references?
Whatever your dismissal of Jones as amateur how can you state that without having even seen the topology. Or are ALL CCS bad in all tube circuits no matter what?
Other than some of my vinyl foobar DRM checker would tend to disagree with you for the levels that I listen at, but we have different speakers, different rooms etc.
No disagreement there.
I see, but I think you pretty much got them covered - lower the gain, and change the PI tube/configuration and of course the standard re-cap and replacement of out of spec resistors. Good luck with your restoration, it's a great looking amplifier, wish I had one...🙂
Fred, an Englishman, decides to make some chicken soup. Cooks have been making chicken soup to a limitted number of known recipes for 200 years. But Fred thinks "Let's try something different, because I can." He further reasons that a) because purer ingredients and fresh chicken makes tastier soup, b) some chap in India said tap water isn't too good, and c) distilled water was not available 200 years ago, he'll try using distilled deionised water instead of tap water. After all, he's not running a restaurant that has to watcgh costs, he can easily pay for the distilled deionised water.
Is he free to do so? Of course he is. Is he being sensible? Nope, not at all.
Are you free to install complicated bias systems? Of course you are.
Would you be sensible in doing so? Nope.
The following is a good overview of sensible tube amplifyier design by a well known master of the subject.
Thiele, N, Some thoughts on the design of valve audio amplifiers 1, Electronics Australia, March 1993, pp110-113
and
Thiele, N, Some thoughts on the design of valve audio amplifiers 2, Electronics Australia, April 1993, pp110-113
A google search will turn up more.
Thiele's various papers should be mandatory reading for all tube audio homebuilders and restorers.
Without having read the book, there is logically a risk that there is something there I haven't thought about.
But it's a bit like using oval wheels on a bike. You don't need to read a book to know its a bad idea. Nor is there any temptation to read a book by some freak who claims oval wheels are good. And there have been such freaks - if I remember where I saw it, I'll givbe you a reference.
It is the case that all CSS tube biasing systems are indeed bad.
I have already explained why: Due to the curvature of tube transfer charaecteristics, tube cathode current rises with signal level. In fighting this rise, CCS circuits overbias the tube to a greater amount than a simple ctahode bias resistor. So overdive behaviour is worse - the tube(s) many even be cut off for a short period after overdrive stops. As the tube(s) on strong signal level will be overbaised with CCS biasing, the tube9s) is/are operated futher into the more curved part of the characteristic near curtoff, there will be increased distortion.
That can really only be the case if you listen at a volume significantly quieter than I do the demonstration at, or if you listen to music that is heavily compressed (such as 1950's rock and roll, or music from AM radio stations). Old vinyl music re-released on CD is still compressed - even when re-mastered.
The thing is, I have speakers than are more efficient than most, and I do the demonstration in a room that is standard bedroom/study size.
Also, I do the demonstation at conversation level, (so we can talk comfortably while it is playing), not at a higher level that I would use for music enjoyment.
Last edited:
Mine is ratty. Had been living in a basement when I got it, so needs a powder coat replacement unless I can start a thing for Rat amps!
But sonically it is most enjoyable which is why I want to restore. And I know mint examples go for £3k+, but its my amp and only cost me £150.
I really must open it up and find what failed.
I have to think. As the CCS is part of the LTP I may have been thinking too transistor. As I did admit I'm coming back in after nearly 2 decades of being dad foisting other technology on an unsuspecting world 🙂
I was talking mainly about CCS cathode baising of output stages - where it is harmful to the sound.
CCS tails for tube LTP's used as phase splitters and push-pull drivers is a bit different. Here its' not actually detremental to sound quality, just a waste of money.
Goodness no. Even I am not daft enough to think that using CCS on output tubes should be considered 🙂
I am however daft enough to thing that LEDs could still be part of the biasing solution.
I am however daft enough to thing that LEDs could still be part of the biasing solution.
Bill, I would recommend the autobias circuit shown in MJ's book. It's very simple and works as advertised. The issue of overload recovery is what drives the idea of DC coupling, and since you have actually read the book, you're aware that this was gone through in exquisite engineering detail. It's further refined in the 4th edition for the Bulwer-Lytton amp, so you might want to check that out as well.
Yes, the biasing scheme I use is only appropriate for pentodes, as you have surmised.
Yes, the biasing scheme I use is only appropriate for pentodes, as you have surmised.
I'll pick up the 4th edition when I have the spare money. Books are good, and opinions are good even if one ends up disagreeing with them 🙂
I still think there is life in the hybrid scheme as it could be set to fail safe and in this arrangement would only need a small positive grid bias. But I have a -ve rail available...
And turning and Ultralinear amp into a pentode just for the biasing would be the height of madness 🙂
I am still trying to fully understand Keit's comment about why the higher impedance of fixed bias is problematic in overload cases in the ultralinear configuration. But I'll get there, just need time and beer.
Proms have just play Belshazzars feast. That is a beast of a piece!
I still think there is life in the hybrid scheme as it could be set to fail safe and in this arrangement would only need a small positive grid bias. But I have a -ve rail available...
And turning and Ultralinear amp into a pentode just for the biasing would be the height of madness 🙂
I am still trying to fully understand Keit's comment about why the higher impedance of fixed bias is problematic in overload cases in the ultralinear configuration. But I'll get there, just need time and beer.
Proms have just play Belshazzars feast. That is a beast of a piece!
You seem to be mixing up terms here. Perhpas that is the problem.
Fixed grid voltage bias - good wrt overload
Fixed cathode current bias (ie CCS bias) - bad wrt overload.
That's a stange comment, given that the Bulwer-Lytton (shown on page 544 of MJ's book) is AC coupled, and has mimimal technical explanation.
The Bulwer-Lytton circuit is pretty ordinary really, except for a fairly pointless CSS for the VAS LTP, and a totally pointless excessively complicated source follower circuit interposed between the VAS and the output stage.
Edward Bulwer-Lytton was a 19th century writer of fiction. A stange name to choose for an amplifier - perhaps it is meant to indicate that the idea that the circuit has any overal merit is fiction.
There seems to be this view that since an LTP works better if the tail impedance is raised well above the tube 1/gm value, it must be better still if we make it as high as technically possible - right up into the megohm range. But a law of diminishing returns sets in quite rapidly. A simple resistor does all that is required.
The same applies with driving output stage grids. The grid driving impedance should not be too high. That doesn't mean you need to make it as low as technically possible. in the region of 100 k to 500 k depending on tube size or number in parallel does all that is required.
"If it's not Australian, it's crrrrrrap!" 😀
The name origin was explained. When the amp was first turned on, it was during dark and stormy weather.
You seem to have missed his explanation of why the high tail impedance was needed. Likewise, the reason for the direct-coupled MOSFET drivers. You might want to actually read the text, then develop a reasoned critique. Morgan is quite a skilled engineer (despite only having a Master's with honours), and like most good engineers, his approaches are logical AND subject to revision when someone points out an error or a better way (this is something I know from personal experience).
The name origin was explained. When the amp was first turned on, it was during dark and stormy weather.
You seem to have missed his explanation of why the high tail impedance was needed. Likewise, the reason for the direct-coupled MOSFET drivers. You might want to actually read the text, then develop a reasoned critique. Morgan is quite a skilled engineer (despite only having a Master's with honours), and like most good engineers, his approaches are logical AND subject to revision when someone points out an error or a better way (this is something I know from personal experience).
When you said 'any form of constant current biasing' Its position in the text did mean that I read that as fixed bias. My bad.
Follow on whilst we are having a healthy discussion and we all agree that overload recovery is vital. Does this imply that, even though you don't design the amp for AB2 operation you have to expect it worst case anyway?
And a really dumb question. If you have regulated bias supplies why do you still put them before the grid stopper? If you accept silicon instead of a simple pot you have the ability to consider different ways of dealing with +ve grid conditions as well as having less of a time constant issue with the coupling cap and the bias resistor. But baby did keep us up most of the night so brain is addled!
The fact is, as I explained, really high tail impedance is not needed. All you need is for the tail impedance to be large compared to 1/gm. If the tail impedance is say 100 times 1/gm, the currents of two sides of the LTP are forced to be within 1%, so clearly the LTP is statically balanced and cancellation of distortion is effective to with 99% of the infinite tail impedance limit.
1/gm is of the order 130 ohms, so a tail impedance of 13 kohm is plenty good enough. Heck, 3 kohm would not be audibly any different. Clearly, all that is needed is a resistor.
You seem to have missed that a law of diminishing returns applies.
Likewise, you have missed that there is a law of diminishing returns. While it is true that driving output stage grids from too high an impedance leads to large signal problems, below a certain value in 100's of kohm range, there will be no audible difference. There is simply no need need for the super low impedance drive that MOSFET followers make possible.
This reasoning that since more is better, so super amounts of more must be extra better is faulty reasoning. It is the same as reasoning that as clean drinking water is needed for cooking, ultra pure water would be better still.
Sure, the MOSFET circuit will give a better audible result than simply connecting the VAS anode coupling caps directly to the grids (with the necessary grid leak to ground). But no better than just a simple resistor in series with the grid leads to limit grid current.
Sy, you might like to develop critical reasoning skills, and apply the common sense test.
You might like to read up journal articles published in the tube era, and tubemaker application notes, and see how real professional engineers avioded issues with grid current effects.
Not so, judging by his ramblings in Wireless/Electronics World and elsewhere. Any twit can eventually get a specified performance by adding more and more complexity. Real professional engineers aren't just guys with Master degrees, they are guys who realy understand the thing and get the same or better performance with simplicity.
Last edited:
Yes, that must be it. 😀
So you still missed the reason for the MOSFET drivers.
So a -200V supply is your idea of simplicity. OK.
Just to make sure we are all on teh same wavelength.....
AB1 means that the tubes are operated such that cutoff occurs during a small part of the cycle (negative peaks), but no grid current flows. The cutoff of one tube at a time in a push pull pair does introduce distortion, which is easily dealt with by negative feedback.
AB2 means that the tubes are operated such that cutoff occurs during a small part of the cycle (negative peaks), and each tube draws grid current during positive peaks. This introduces more distortion less easily controlled by negative feedback, and recovery issues. Even if you don't count that (say by driving from low impedance, since AB2 means operating teh tube over a larger span of it's characteristic curves, ditortion will be quite a bit greater anyway.
The praticalities of making relatively not so costly tube amps mean that AB1 operation must be accepted. In amps of the highest quality it is not accepted - the penalty is more HT power consumption (and perhaps bigger tubes) for the same power output.
In quality audio amplifiers, AB2 operation should never be accepted (ie designed out), but was often the norm in public address systems wher distortion was les a priority than power efficiency and cost.
The grid stopper resistor prevents oscillation in high gm tubes due to wiring inductance and stray capacitance forming a colpits oscillator at HF or VHF frequencies. It is thus essential to keep stray capacitance on the grid pin to an absolute minimum. Hence the resistor should be installed directly on the grid pin with short leads and nothing else should be on the grid pin.
All that is entirely true. But it isn't necessary to add silicon - you just employ a series grid resistor, which you usually need anyway to stop oscillation, and choose sensible component values, including the coupling cap. As I said, the series grid resistor usually needs to be a few kohm to stop oscillation, but if you make up around 50 to 100 kohm it addresses distortion issues as well.
Last edited:
Not to worry. Baby will soon be old enough to eat meat and sleep in till lunchtime after a wild party.
- Status
- Not open for further replies.