Apologies for offence caused
Apologies to everyone I've inadvertently irritated here. Your advice has been useful to me - you've all built your first amplifiers already; I'm working on mine.
It appears my presence has been a little trying on the patience so I shall now withdraw with grateful acknowledgements to everyone who has given me useful advice and pointers. I won't bug you again.
Apologies to everyone I've inadvertently irritated here. Your advice has been useful to me - you've all built your first amplifiers already; I'm working on mine.
It appears my presence has been a little trying on the patience so I shall now withdraw with grateful acknowledgements to everyone who has given me useful advice and pointers. I won't bug you again.
Last edited:
You are attempting something which most experienced people would avoid (4 stages), so problems are almost inevitable. You asked for peer review and received it. This has exposed both errors and gaps in your thinking, which is unsurprising; we all started there!
It seems to me that you have three choices:
1. try a simpler design which has more chance of success
2. soldier on with this one, but taking note of our comments and accepting the risk of eventual failure
3. go off in a huff
It seems to me that you have three choices:
1. try a simpler design which has more chance of success
2. soldier on with this one, but taking note of our comments and accepting the risk of eventual failure
3. go off in a huff
I've seen quite a few Williamsons on this forum, I didn't realise this was something people avoided. So what I'm proposing to build will be something of a rarity then.
In life as in amplifier construction. With both I am happy to troubleshoot.
And in doing so it seems I've annoyed a few people on here, certainly not my intention, it appears I misjudged the purpose of this forum, for which I apologize.
Nobody's going off in a huff anywhere. If, seemingly against all predictions, I end up with a functioning amplifier, I will return and post it in a new thread, with a link to this one. Right now however I will withdraw from here and spend some time reading Morgan Jones and others – which in itself was some advice I was given here a few weeks ago.
Once again with due acknowledgements to those who took the time to look over my drafts and add their comments, which were very useful.
There is a good discussion of the pros and cons of the Williamson design in Morgan Jones' book. In comparing it to the Mullard 5-20 design, Jones points out that where the Williamson achieves very good linearity at the expense of likely problems with stability, the Mullard 5-20 achieves very good stability at the expense of worse linearity. Both are valid approaches. (Once again proving that everything is a compromise.)
I've heard some pretty good sounding Williamson amps, especially a a pair of Heathkit W3 from long ago. However, most will point out that it's not a great idea to wrap a global negative feedback all the way from the secondary of the output transformer to the cathode of the input tube. The combination of the various poles, output transformer phase shifts, and any power supply resonances will inevitably cause stability problems.
One possibility would be to use a local feedback loop around the output stage that does not include the output transformer. I haven't built this, but it should be possible, and might work very well. But this gets into more advanced territory.
You might choose to imitate the Eico HF-87, which is basically the Mullard 5-20 topology simplified by changing the input pentode to a triode, and with differing value load resistors that attempt to enforce balanced output from the 6SN7 LTP.
An externally hosted image should be here but it was not working when we last tested it.
There are several builds on this forum that use that same basic topology, improving it in various ways. For instance... http://www.kta-hifi.net/ and Dynaco ST-70 Modifications (The ST70 Driver Board Upgrade is similar to Eico HF-87 circuit except uses a 12AU7 LTP instead of a 6SN7 in that spot.)
I'd say build your amp and learn from it. If it's a Williamson, then getting the stability and negative feedback right will be the challenge. That sounds like a worthy one!
--
Aros,
I don't think anyone is irritated with your questions. One thing you should keep in mind about diyaudio.com is that users are from all over the world with a huge array of backgrounds. Replies that seem brusque or impatient are not necessarily intended to be so. Sometimes this is a language or cultural issue and other times it's the result of an experience gap.
Deep understanding and experience leads to logical leaps when evaluating a circuit proposal. It is rarely obvious how the dots are connected when you are new to the hobby. This can be really frustrating (I know!) and sometimes makes it seem as though others are trying to discourage you with an off-the-cuff dismissal. The old 'go back and study Jones/Merlin/etc' recommendation feels like the kind of condescending admonishment a teacher would give a schoolboy, but really those authors are the few that have taken the time to connect all the dots. It took hundreds of pages.
You have set your sights very high for a first tube amplifier project and what you are trying to do from scratch (multi-stage push-pull) has been done very elegantly in the past (Williamson, LEAK, Mullard).
Simplicity by itself does not beget elegance. It must also be ingenious.
What appears to be only simple sometimes turns out to be ingenious as well, but knowing the difference requires a deep understanding. By my estimation, your design so far lacks both simplicity and ingeniousness and that is why you are being prodded to continue working on it. But your fundamental understanding is sound (insofar as you have recognized some good suggestions and improved the design over the course of the thread). If you build your design, it will probably work. It just won't be elegant. You have a grasp of the basics and that's where anybody doing anything correctly starts.
Have patience when you get curt criticism and determination as you research the technical whys and wherefores. Experienced and well-meaning designers will point out errors when they see you're on a path to a dead end, but they won't always explain themselves. The irritated and impatient won't say anything at all.
Chin up kiwi classmate.
Soda
I don't think anyone is irritated with your questions. One thing you should keep in mind about diyaudio.com is that users are from all over the world with a huge array of backgrounds. Replies that seem brusque or impatient are not necessarily intended to be so. Sometimes this is a language or cultural issue and other times it's the result of an experience gap.
Deep understanding and experience leads to logical leaps when evaluating a circuit proposal. It is rarely obvious how the dots are connected when you are new to the hobby. This can be really frustrating (I know!) and sometimes makes it seem as though others are trying to discourage you with an off-the-cuff dismissal. The old 'go back and study Jones/Merlin/etc' recommendation feels like the kind of condescending admonishment a teacher would give a schoolboy, but really those authors are the few that have taken the time to connect all the dots. It took hundreds of pages.
You have set your sights very high for a first tube amplifier project and what you are trying to do from scratch (multi-stage push-pull) has been done very elegantly in the past (Williamson, LEAK, Mullard).
Simplicity by itself does not beget elegance. It must also be ingenious.
What appears to be only simple sometimes turns out to be ingenious as well, but knowing the difference requires a deep understanding. By my estimation, your design so far lacks both simplicity and ingeniousness and that is why you are being prodded to continue working on it. But your fundamental understanding is sound (insofar as you have recognized some good suggestions and improved the design over the course of the thread). If you build your design, it will probably work. It just won't be elegant. You have a grasp of the basics and that's where anybody doing anything correctly starts.
Have patience when you get curt criticism and determination as you research the technical whys and wherefores. Experienced and well-meaning designers will point out errors when they see you're on a path to a dead end, but they won't always explain themselves. The irritated and impatient won't say anything at all.
Chin up kiwi classmate.
Soda
Last edited:
I don't know if the original poster was upset by my comments in this thread but if so, I apologize. I don't write very much on this forum as i have limited time, (I am too busy with my own amplifier business) although i have been a member since 2003 and I enjoy other peoples posts every day.
The reason why I posted in this thread was that the original poster made some very basic mistakes and somewhat strange design choices so I want to get an understanding on his reasons behind this and possibly steer him on the right track. If you make a design that is different than anyone else you probably have good reasons for doing that or else you are clueless but as some of the design choices here was a bit advanced, (although without any benefit) I thought that maybe he was advised by some self-appointed "guru".
Maybe my thinking is too simple but I can't understand why a beginner doesn't start with something well known and develop their understanding from there instead of designing something new without first acquiring all the knowledge needed to do that.
By starting with a well known design and develop a deep understanding of the design choices made you can learn a lot, that is how I started in electronics more than 45 years ago and that is how the majority of professional electronic design engineers I have met in my professional life have started to develop their interest in electronics.
PS, The Williamson amplifier can cope with having 4 stages as the first stage is direct coupled to the inverter thereby eliminating one pole, it also have a high frequency phase correction network in the anode of the first stage. The Williamson still have barely acceptable phase margin for low frequencies but it can be improved by an additional low frequency phase correction network. DS
The reason why I posted in this thread was that the original poster made some very basic mistakes and somewhat strange design choices so I want to get an understanding on his reasons behind this and possibly steer him on the right track. If you make a design that is different than anyone else you probably have good reasons for doing that or else you are clueless but as some of the design choices here was a bit advanced, (although without any benefit) I thought that maybe he was advised by some self-appointed "guru".
Maybe my thinking is too simple but I can't understand why a beginner doesn't start with something well known and develop their understanding from there instead of designing something new without first acquiring all the knowledge needed to do that.
By starting with a well known design and develop a deep understanding of the design choices made you can learn a lot, that is how I started in electronics more than 45 years ago and that is how the majority of professional electronic design engineers I have met in my professional life have started to develop their interest in electronics.
PS, The Williamson amplifier can cope with having 4 stages as the first stage is direct coupled to the inverter thereby eliminating one pole, it also have a high frequency phase correction network in the anode of the first stage. The Williamson still have barely acceptable phase margin for low frequencies but it can be improved by an additional low frequency phase correction network. DS
Last edited:
(OP) Where did my ideas come from?
OP's response...
All good. Like I said, good replies and good advice from everyone, greatly appreciated. However I felt like I was treating this as a n00bs forum which I had assumed would be OK but perhaps I misjudged.
Again with grateful thanks for all the advice and information for those who took the time to review my schematics.
Some of you wanted to know where the heck my ideas came from... My original ideas were principally informed by Lenard
Analysing the feedback (no pun intended) it appears the largest problem is the AC coupling of the initial gain stage with the following concertina stage. The DC coupling of those stages appears to be one of the defining features of the Williamson design, and is one I will make a great effort to replicate.
Most Williamson designs I have seen employ the 6SN7 for the initial gain and splitter, DC-coupled, then another 6SN7 as driver. My ideas, again based on Lenard, were to use a 12AX7 / 12AU7 here, but setting up the 12AX7 for low distortion and DC coupling seems unfathomably difficult.
I work on the principle that if something you're trying to do seems "unfathomably difficult", yet thousands have successfully done it before you, you're probably doing it wrong.
Along those lines, I haven't found ANY conclusive evidence of anyone successfully using a 12AX7 in this manner, apart from Lenard in his theoretical design. So, my next step will be to decide whether to persist with the 12AX7 / 12AU7 front end, perhaps using a voltage divider off the anode for the grid of the 2nd stage, or just entirely rework the front end to use the 6SN7 for all stages.
The other alternative is to forget about negative feedback which is also a line I will explore. Implications, pros and cons etc.
But I'll keep work-in-progress schematics away from here until I've built something that works, then I'll post and also with oscilloscope traces etc to substantiate.
Thanks
OP's response...
All good. Like I said, good replies and good advice from everyone, greatly appreciated. However I felt like I was treating this as a n00bs forum which I had assumed would be OK but perhaps I misjudged.
Again with grateful thanks for all the advice and information for those who took the time to review my schematics.
Some of you wanted to know where the heck my ideas came from... My original ideas were principally informed by Lenard
Analysing the feedback (no pun intended) it appears the largest problem is the AC coupling of the initial gain stage with the following concertina stage. The DC coupling of those stages appears to be one of the defining features of the Williamson design, and is one I will make a great effort to replicate.
Most Williamson designs I have seen employ the 6SN7 for the initial gain and splitter, DC-coupled, then another 6SN7 as driver. My ideas, again based on Lenard, were to use a 12AX7 / 12AU7 here, but setting up the 12AX7 for low distortion and DC coupling seems unfathomably difficult.
I work on the principle that if something you're trying to do seems "unfathomably difficult", yet thousands have successfully done it before you, you're probably doing it wrong.
Along those lines, I haven't found ANY conclusive evidence of anyone successfully using a 12AX7 in this manner, apart from Lenard in his theoretical design. So, my next step will be to decide whether to persist with the 12AX7 / 12AU7 front end, perhaps using a voltage divider off the anode for the grid of the 2nd stage, or just entirely rework the front end to use the 6SN7 for all stages.
The other alternative is to forget about negative feedback which is also a line I will explore. Implications, pros and cons etc.
But I'll keep work-in-progress schematics away from here until I've built something that works, then I'll post and also with oscilloscope traces etc to substantiate.
Thanks
We have and welcome people at all skill levels, so no worries. But peer review means that no-one is grading on a curve. 😀 You've gotten some pretty good advice here; consensus among the more advanced folk answering you:
What you want to do for your first project is something at expert level, and that's not going to go well.
If you want to learn, choose a simpler project to start and read some of the better texts (e.g., Jones, Blencoe).
If you need that particular function in your system, build an established design and use it. If you want to learn from it, study the design, characterize it to find out its flaws or tradeoffs, then start modifying and adapting in a directed way.
Designing and building tube stuff is a gas, so to speak. Have fun with it.
Even the 5-20 (or is it the very similar 5-10?) has marginal LF stability; there is a wiggle in the frequency response just next to the LF rolloff.rongon said:
Much of what he says is helpful. Some of it is wrong. You will find that this mix of truth and error is quite common, although some sites veer towards one extreme or the other!aros71 said:
Low distortion is not too difficult: you need a good bias point and a high anode load (which means either a CCS or a high supply rail voltage). Most people manage to get the bias point right but have too low an anode resistor and then complain that the 12AX7 is nonlinear. The 6SN7 (and 6CG7) is much easier to use, which may be partly why it is so popular.
DC coupling can be a problem: the 5-10/5-20 LTP phase splitter can suffer from grid current due to sample variation in the first stage. To make things worse, DC coupling does not always eliminate an LF rolloff - it doesn't in the 5-10, for example.
If your aim is hi-fi then you will need negative feedback.
DC coupled 12AX7 initial gain and cathodyne
So continuing on the DC coupling idea, someone's done it with a 12AX7 but I'm doubtful of the results. This was their setup:
If I plot this on the tube curves I get 0.5mA on the gain stage, 0.6mA on the splitter.
So to my question - this experimenter found this worked, however my concern is with the low anode voltage on the first stage.
here's how it'd look (not my circuit, I just took the liberty of drawing someone else's):
So continuing on the DC coupling idea, someone's done it with a 12AX7 but I'm doubtful of the results. This was their setup:
Using a 12AX7:
Gain stage - 390K Ra, 1.5K Rk, HT = 295V, A = 82V, K = 0.86V, Voltage gain = 76 (@1KHz)
Cathodyne - 100K Ra and Rk, HT = 295V, A = 184V, K = 93V. Max output pre-clip - Anode output = 55Vpp, Cathode output = 80Vpp
The experimenter noted this was based on actual measurements after several resistor substitutions to achieve the optimal results.Gain stage - 390K Ra, 1.5K Rk, HT = 295V, A = 82V, K = 0.86V, Voltage gain = 76 (@1KHz)
Cathodyne - 100K Ra and Rk, HT = 295V, A = 184V, K = 93V. Max output pre-clip - Anode output = 55Vpp, Cathode output = 80Vpp
If I plot this on the tube curves I get 0.5mA on the gain stage, 0.6mA on the splitter.
So to my question - this experimenter found this worked, however my concern is with the low anode voltage on the first stage.
here's how it'd look (not my circuit, I just took the liberty of drawing someone else's):
An externally hosted image should be here but it was not working when we last tested it.
Not sure what your question is, but the first triode has very little bias voltage (-0.8V), so there will be grid current which may increase distortion (although it can also reduce it by harmonic cancallation!). The cathodyne also has little bias, and grid current will impair its balance though perhaps negligibly so.
Last edited:
Yes, the 12AX7 has only a narrow bias window between cutoff and grid current - I guess an unavoidable consequence of the high mu. Both valves could suffer from grid current, although the second stage will be bootstrapped.
Run that circuit from a 400V rail and you have more room. Resistor values may need some adjustment.
Run that circuit from a 400V rail and you have more room. Resistor values may need some adjustment.
I'm doubtful of their setup too. It shows the cathode voltage on the split load inverter (V1b) as 93V, and the anode voltage on the first stage 12AX7 (V1a) as 82V. That difference of -11V forms the grid bias for V1b. There is no way there can be a grid bias of -11V on V1b. That just has to be wrong. Therefore, all other voltages are suspect as well, unless that's just a simple typo, and the cathode voltage on V1b is really 83V. Unfortunately, even that doesn't fix things, because the current across V1b would then be 0.83 mA, not the 0.6 mA reported. (0.6 mA across the 100k resistor would yield 60V at the cathode of V1b.)
--
Last edited:
This is because the grid voltage was dragged down by the voltmeter. In reality it will be closer to 92V.
Now I'm confused.
If the anode voltage on V1a is 82V, and the cathode voltage on V1b is 92V, then that's still a voltage from V1b grid to cathode of -10V.
Maybe you meant the anode voltage on V1a was dragged down by the voltmeter?
And I still don't understand how if V1b is drawing 0.6mA anode current, how does 93V appear at V1b cathode (across R8)?
--
Anode voltage of first section = grid voltage of second section (the grid Merlin was talking about).
I make that 60V...
I may have been wrong with the Anode current though... I took a load resistance of 200K and B+ of 295V then read off the chart where it intersected with the claimed Anode Voltage.
Later on today (it's morning this side) I'll do the full maths on this scenario. As has already been stated, if this is based on actual measurements then the voltmeter could have been pulling voltages around.
Looks like the voltmeter had an impedance of 2Meg, typical of a DMM. Ironically the most accurate and useful voltage to have measured is the one he didn't bother to provide -the cathode voltage of V1a!
I think I'll abandon this design and rather just do what I vowed I wouldn't do... find someone else's schematic and just build that.
I'll be a little constrained by the shopping I've already done, but not disastrously so. I have on order (and paid for) 2 X Hammond 1650R OPTs, and my mains transformer is sitting on my bench, with 400-0-400 secondary, a 60v secondary for bias supply, and 8A of 6.3V centre-tapped. There's not a second 6.3 winding, so this means I can't do an elevated heater supply, but I will select a design that doesn't require it.
I'm abandoning this idea because I can not find a way of using the 12AX7 as the initial gain stage and then DC coupling that to the following Cathodyne stage, which will be a requirement for the amplifier not to be an expensive oscillator.
Bit of actual paid work needs to be caught up on first so this project will have to go on hiatus for a while
I'll be a little constrained by the shopping I've already done, but not disastrously so. I have on order (and paid for) 2 X Hammond 1650R OPTs, and my mains transformer is sitting on my bench, with 400-0-400 secondary, a 60v secondary for bias supply, and 8A of 6.3V centre-tapped. There's not a second 6.3 winding, so this means I can't do an elevated heater supply, but I will select a design that doesn't require it.
I'm abandoning this idea because I can not find a way of using the 12AX7 as the initial gain stage and then DC coupling that to the following Cathodyne stage, which will be a requirement for the amplifier not to be an expensive oscillator.
Bit of actual paid work needs to be caught up on first so this project will have to go on hiatus for a while
- Status
- Not open for further replies.