"Like" AX84 on facebook

'load lines for dissimilar output tubes?'
Author:tapehead ted (registered user: 6 posts )
Date: Wed, Mar 15th, 2017 @ 16:45 ( . )

Hello all. I'm working on a design with a pair of 6L6 and a pair of 6v6 in parallel, in p-p amp, and trying to optimize the load lines. I will probably have a custom OT made, so there's the possibility of having a separate taps for each type of output tube, but this may not be necessary.

It seems to me that this is not as simple a case as paralleling two like tubes- Let's say I'm using a 4K OT. With four 6v6's, I would have a nice clear cut class B load line (Class AB overall) of 2K. But using dissimilar tubes, unless the current is the same in both the 6v6 and 6L6, I am thinking I will end up with a different Class B line for each tube type. This, I am thinking, will apply regardless of whether or not there is a separate OT tap for each tube type.

So how would one determine what the class B loadline would be with dissimilar currents, which I think is really all that is dissimilar about these "dissimilar" tubes when it comes to plotting loadlines?

(I'm using separate screen supplies to let me find more compatible operating points.)

I'm thinking this can be resolved using V=IR:

Say the 6L6 is biased to 60 ma, and the 6v6 to 40 ma. My thinking: At these currents, adding the 6v6 to the 6L6 is like adding 2/3 of another 6L6, not a whole other 6L6, and from the perspective of the 6v6, adding the 6L6 is like adding a 6v6 and a half, when it comes to determining the resulting loadlines.

Doubling the number of tubes, with the same currents, from the perspective of two 6v6s (40 ma each) adding two 6L6s (60 ma each) would be like adding three 6v6s. This I found a little easier to contemplate, and as far as I can tell the result is the same.

Is this how this works? Or is it at least close enough to how it works to make useful guesstimations?

My results show me that at these currents, a 4k OT looks like 10k to the 6v6s (for a 2.5k class B loadline) and the same 4k OT looks like 6.67 to the 6L6s (for a 1.67 class B loadline). This actually makes the tubes more compatible, offering more appropriate load lines to each, if in fact this thinking resembles reality.

Does it?

It seems like with dissimilar currents, the 6v6 and 6L6 couldn't be both on 2k class B loadlines with a 4k OT.

I'm focusing on the class B loadline as it is the one that crosses the 0 vg1 line, where screen dissipation becomes a serious issue, especially for the 6v6 with it's very small screen dissipation limit.

Any thoughts?

Thanks!




-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:CChurchill (registered user: 163 posts )
Date: Wed, Mar 15th, 2017 @ 23:19 ( . )

On 03/15/2017 @ 16:45, tapehead ted wrote :
Hello all. I'm working on a design with a pair of 6L6 and a pair of 6v6 in parallel, in p-p amp, and trying to optimize the load lines. I will probably have a custom OT made, so there's the possibility of having a separate taps for each type of output tube, but this may not be necessary.
:
: It seems to me that this is not as simple a case as paralleling two like tubes- Let's say I'm using a 4K OT. With four 6v6's, I would have a nice clear cut class B load line (Class AB overall) of 2K. But using dissimilar tubes, unless the current is the same in both the 6v6 and 6L6, I am thinking I will end up with a different Class B line for each tube type. This, I am thinking, will apply regardless of whether or not there is a separate OT tap for each tube type.
:
: So how would one determine what the class B loadline would be with dissimilar currents, which I think is really all that is dissimilar about these "dissimilar" tubes when it comes to plotting loadlines?



:
: (I'm using separate screen supplies to let me find more compatible operating points.)
:
: I'm thinking this can be resolved using V=IR:
:
: Say the 6L6 is biased to 60 ma, and the 6v6 to 40 ma. My thinking: At these currents, adding the 6v6 to the 6L6 is like adding 2/3 of another 6L6, not a whole other 6L6, and from the perspective of the 6v6, adding the 6L6 is like adding a 6v6 and a half, when it comes to determining the resulting loadlines.
:
: Doubling the number of tubes, with the same currents, from the perspective of two 6v6s (40 ma each) adding two 6L6s (60 ma each) would be like adding three 6v6s. This I found a little easier to contemplate, and as far as I can tell the result is the same.
:
: Is this how this works? Or is it at least close enough to how it works to make useful guesstimations?
:
: My results show me that at these currents, a 4k OT looks like 10k to the 6v6s (for a 2.5k class B loadline) and the same 4k OT looks like 6.67 to the 6L6s (for a 1.67 class B loadline). This actually makes the tubes more compatible, offering more appropriate load lines to each, if in fact this thinking resembles reality.
:
: Does it?
:
: It seems like with dissimilar currents, the 6v6 and 6L6 couldn't be both on 2k class B loadlines with a 4k OT.
:
: I'm focusing on the class B loadline as it is the one that crosses the 0 vg1 line, where screen dissipation becomes a serious issue, especially for the 6v6 with it's very small screen dissipation limit.
:
: Any thoughts?
:
: Thanks!
:
:
:
:
--



So, this is an interesting question. I've though about doing this in passing (but not researched it). I wanted to touch on a couple of things you mentioned, but I'm sure you will get some more informed responses from those with expertise in magnetic coupling.
I'm not sure if I misunderstood you, but it sounds like you might misunderstand load impedance in the output stage somewhat. (please correct me if I'm wrong)
To oversimplify this, apart from the inherent tube internal resistance, an output transformer "reflects" the load impedance from the load - ie. speaker/resistor - to the tube. Often, when you look at the tube datasheets, you will find several different load impedance examples based on different operating conditions.
To your point, a 4K primary impedance with the same load would present a 4K load to any tube. Unless you were using different taps and maybe that's what you were talking about. An output tube doesn't "have" fixed output impedance under all conditions. However, there are popular choices which we often see used with different tubes.

However, I would think that the different operating characteristics of the dissimilar in a single primary winding (even with different taps) the tubes would cause instability and maybe damage to the transformer. I don't know for sure, but here is my guess and how I would approach it. I would use a separate push pull primary winding for each type of tube on the same core, as long as the transformer was designed to handle the stress and power of the two different tube types. This is an intuitive guess, there may be a reason it might not work because I've never seen it done this way, but this is how I would proceed.

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:tapehead ted (registered user: 6 posts )
Date: Thu, Mar 16th, 2017 @ 16:41 ( . )

Thanks for your reply.

The whole scenario I relate in my initial post is, to clarify, envisioning the same tap and load for both tube types. I do, I believe, understand the basics of the reflected load idea. This is assuming that the speaker load remains the same, and thus the reflected load remains the same from anode to anode.

Each tube is looking at the same 4k load, but in the same way it looks like an 8k load to each of two pair of similar tubes, at assumedly the same current, it looks like 8k, possibly plus or minus some amount, to each of the two dissimilar tubes. I just have been nagged by the idea that some difference between the two dissimilar tubes, notably with the 6v6 and 6L6 that one is quite a bit larger than the other, would influence the load lines of the two dissimilar tubes so the load lines were not identical. I was struck by the idea that perhaps the different currents the two tubes were handling would be the crucial difference. Perhaps in fact both load lines are identical, just like I was using two tubes of the same type and at the same current. Hopefully someone has an answer to that- I look forward to hearing more about it.

Is there really a possibility that this could damage the output transformer, assuming that it is rated to handle the total current? I can't see how that would happen- I would think the two dissimilar tubes on each side of the output transformer would appear to the transformer just like some other composite single tube.

FWIW I have heard and used this combination in an amp a friend built that was supposed to be able to handle one or two pair of 6v6 or a single pair of 6L6. (This with a 4.2K transformer switchable to 2,4, or 8 ohms.) One pair of each sounded really good in that amp, adding a whole different dimension to the 6L6 tone while still having the clear lows of the 6L6, which I am quite addicted to, and breaking up much more smoothly than the 6L6 alone. Unfortunately, that amp eats tubes for a variety of reasons including that the rectifier tube is really too small for anything more than a single pair of 6v6's, the power amp grid leaks are too high a value, the screen stoppers are tiny, among various others.

So I have been inspired to design an amp addresses those and several other problems and makes life as safe as possible for 6v6's and 6L6's at the same time.

So I've been trying to determine what was going on as far as loadlines when I was hearing those lovely sounds through that other amp. I suppose I could just conclude that the combination works nicely with that anode to anode load, and move ahead with it, but I'm hoping to learn some more about the dynamics of it.

Thanks again.

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:Malcolm Irving (registered user: 29 posts )
Date: Fri, Mar 17th, 2017 @ 09:23 ( . )

Yes. The total plate current of the two dissimilar tubes would correspond to a load of 8k. But I expect the 6L6 will take more of that current than the 6V6.
I think you could decide on the B+ and the screen voltage you are going to set for both tubes and then look at the plate characteristics. The two load lines both start at the B+ value on the horizontal axis. You then need to juggle around with the slopes of the two load lines, so that total plate current stays the same (as for two 4k loads). If you move one line up by 5mA on the vertical axis the other line has to come down by 5mA. Keep juggling until you find a pair of load lines so that at a given grid voltage on the load lines (say -10V) the corresponding plate voltages agree.
Assuming the 6V6 takes less current, its load line could go ‘below the knee’ – so watch out for screen current on that one – a good size screen stopper may be needed.
Would you set the same bias voltage for both tubes, or with some added complication you could have different biases?
The output transformer would just see the sum of the two currents – so no problem there, that I can see.
My own gut feeling is that there won’t be any more stability problem than usual.


-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:CChurchill (registered user: 163 posts )
Date: Fri, Mar 17th, 2017 @ 14:17 ( . )

Okay, yes, I see what you were saying.
I still think your problem would be with instability, even if not necessarily damage, I haven't tested it, so I really dont know.
I guess my questions would be, first, are you trying to match up the point within the curves so that both types of tubes begin to transition, basically, into class B?
If so, would it be equally as important to have both types of tubes enter cut off as well as grid clamping? 6V6s and 6L6s have very different and put sensitivities, so I think it would be difficult to maintain load balance using the same primary.
But it's an interesting challenge, and I love the idea of seeing if different parallel output tubes. So let's try and make it happen

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:CChurchill (registered user: 163 posts )
Date: Fri, Mar 17th, 2017 @ 14:20 ( . )

Sorry that was supposed to be *6v6s and 6L6s have different "input" sensitivities.
The folly of speech to text on my phone

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:Malcolm Irving (registered user: 29 posts )
Date: Fri, Mar 17th, 2017 @ 14:54 ( . )

On 03/17/2017 @ 14:17, CChurchill wrote :
...
: I guess my questions would be, first, are you trying to match up the point within the curves so that both types of tubes begin to transition, basically, into class B?
: If so, would it be equally as important to have both types of tubes enter cut off as well as grid clamping? 6V6s and 6L6s have very different and put sensitivities, so I think it would be difficult to maintain load balance using the same primary.
...


Yes, I think there could be various changes in slope in the load lines, as different tubes go into cut off or grid clamping (including those on the 'other side' of the push-pull).
Biasing the 6V6 and 6L6 independently might give some control over this.
It's tricky, but I think whichever of the tubes grid clamps first, might also clamp the grid signal to the other, depending on the design of the input circuits to the grids.

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:CChurchill (registered user: 163 posts )
Date: Fri, Mar 17th, 2017 @ 19:51 ( . )


:
: Yes, I think there could be various changes in slope in the load lines, as different tubes go into cut off or grid clamping (including those on the 'other side' of the push-pull).
: Biasing the 6V6 and 6L6 independently might give some control over this.
: It's tricky, but I think whichever of the tubes grid clamps first, might also clamp the grid signal to the other, depending on the design of the input circuits to the grids.
--



Yes! you make an interesting point. I suppose if you used a common coupling capacitor feeding both of the grids, once the cap begins to charge from the 6V6 clamping, It would begin to shift the bias closer class B...
...no, wait a minute. Let me think about this for a second. ;)
I'm not sure that scenario would clamp the 6L6 at all. I guess the first thing you have to figure out, is the bias configuration for this stage. Because no matter how you couple the output tubes, the effect will probably be heavily dependent upon each other. I want to think about this more. Look forward to hearing your thoughts . Though

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:tapehead ted (registered user: 6 posts )
Date: Sat, Mar 18th, 2017 @ 13:49 ( . )

Malcolm- thank you, I'm going to have to copy down your first post and study it for a while. Perhaps that's what I've already been doing... I can't think so well while looking at a computer screen- I'll have to hole up with my charts and look into that approach.

I've spent some time (hours and hours) trying to find achievable screen voltages and bias points where I6v6 and I6L6 are as similar as possible, while keeping the overall point to point swings similar and the bias voltages similar as well. I've decided that having the 6v6 clip a little sooner might be really nice, as it clips much more smoothly in my experience, so I've set the bias voltages so the 6v6 runs maybe -17.5 while the 6L6 runs -19. The closest I could get while meeting all the above criteria was I6L6/I6v6=1.25 at quiescence.

A stiff fixed bias barking overdrive is not really the best for me, so I'm making life complicated by allowing some bias shift. I tracked the ratio of the currents as the bias shifted, which would be the relevant ratio at vg1=0, and found that they diverge more and more as the voltage goes more negative. To keep the ratio from diverging more than about I6L6/I6v6 as suggested in the original post, I'm using about 7-8 volts of fixed bias, 1/3 to 1/2 of the overall bias voltage. I had an elaborate scheme with separate "fixed" bias controls for each tube derived from a back bias, so everything would shift, more moderately than cathode bias, together, but that was way more complicated than I have a real appetite for. The combination of fixed and cathode bias should keep things reasonably well balanced and allow some bias shift.

I'm definitely aware of a tendency for the 6v6 to tilt towards below the knee, hence the above precaution with a mixed bias and definitely healthy screen stoppers.

Screen voltage at design target are 285v for 6v6s and 255v for 6L6's. Allowing for some mains variations, this is the most advantageous situation I could devise.

Who knows but I may rethink all of this once I've applied Malcolm's post.

My own intuition- a very well developed and reliable design resource, in my experience- tells me that loadlines will not tilt so very far as I calculated. As long as I'm having a custom transformer built, I'm envisioning two taps at 3.3k and 4.4k for the 6L6s and 6v6s respectively, although I believe it would work fine with both with the one 4k tap, pretty much as I heard it on by buddy's amp, but a little gentler on the 6v6 screens.

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:Malcolm Irving (registered user: 29 posts )
Date: Sat, Mar 18th, 2017 @ 15:21 ( . )

The suggestion (in my first post) of how to find the load lines only works if there is just one tap for both tubes, as I was searching for a condition where both plate voltages are the same.
Good luck with your design.

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:tapehead ted (registered user: 6 posts )
Date: Sat, Mar 18th, 2017 @ 15:35 ( . )

Thanks Malcolm, I thought that was the case.
I can now envision your approach clearly- I'll be curious to see how similar the results are to those using the method I came up with deriving some equations from V=IR.

How to apply any of this to multiple transformer taps is really stretching out my head just now... It looks from here like there's no real advantage to be gained that way, especially without any clear way to calculate it.

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:stringbend (registered user: 1245 posts )
Date: Fri, Mar 17th, 2017 @ 16:30 ( . )

All I can say is I wasn't willing to put this much thought into it when I did mine. I just used cathode bias, with separate resistor values for each pair.
It sounded great, but I play clean so I can speak to the distortion qualities.

I hope you'll cobble together a schematic once you've built this thing.

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:tapehead ted (registered user: 6 posts )
Date: Sat, Mar 18th, 2017 @ 13:59 ( . )

FWIW the amp I initially heard and played this combination of tubes with had an unusual bias scheme that guaranteed that at all times the bias voltage was identical with all four tubes. That all broke up very nicely together.

With my loadlines as best I can figure them, the two tubes transition into class B at a similar place, if not exactly at the same moment. I've figured this out at 108, 120 and 125 VAC mains voltages, optimizing for 120 VAC but keeping things workable at the extremes.

Thanks everyone very much for a very rewarding discussion!

-- REPLY: [With No Quote] --- [With Quoted Text]

'load lines for dissimilar output tubes?'
Author:tapehead ted (registered user: 6 posts )
Date: Sun, Mar 19th, 2017 @ 19:31 ( . )

Malcolm's suggestion was very easily done- I thought at first suspiciously so. I just tried it on the only charts I had with -10 lines, just drew a load line through -10 on each. They were equidistant in current from the 2k class b line, and right through the knee on both graphs! The results tallied exactly with my equations, but the graphic approach made it so much easier to see. I'm no longer concerned about bias shift. it's the ratio of currents at vg1 eequals zero that counts, and that is constant on the linear part of the characteristic. You could do this with only the vg1 equals zero chart.

This design just got a whole lot simpler! Thank you!

-- REPLY: [With No Quote] --- [With Quoted Text]


--- 51 Users Online --- 41 Recent Unique Posters

Q160=1490256540 - Threads: / 1490256540