Little Dot MK8SE / MK6 Super Mods (All verified mods are on first page)

[baronbeehive]

What I meant is remove the impedance mod (which I have installed based on calculations for 6AS7 values) from the two power rails that feed the 6N5P tubes, as the default resistor has the correct value for them and leave the mod on for the power rails that feed the 6080 tubes, so those two will have the added resistors delivering more power.

Here the idea would be to reduce the imbalance between the 6080 and 6N5P on the same channel. It is also an imbalance that I've had for most of the time for the last six months as I've been using the combination of the Tung-Sol 6080WA and the Shuguang 6N5P with the same power resistor values, be it with or without the impedance mod.

As for the possible switches, the amplifier already does have individual switches (for low/high gain) leading to each of the four power tubes. Maybe it wouldn't be exceedingly difficult to also have individual 6080/6N5P/5998 switches for them as well




Since starting with the new PCB in spring, the meters have been steadily at ~60 ma values on all kinds of tubes, only hopping upwards on bass hits or having some imbalance when the impedance matcher is attached. The impedance mod had no effect on the values.



Well, as stated above, I have already had that kind of imbalance for the last six months. Now that I think of it, there has been this effect which could occur if I left the amp running for a longer time - above 90 minutes - where it sounded like the sound characteristics of one of the two mixed power tube types would somehow become more distant or hollow sounding. I think I had thought it is normal for tube amps to have their sound vary over time with tubes warming up and capacitors charging up, but it could perhaps possibly be better attributed to phase or impedance fluctuations in the WCF. Rebooting the amp would change it to varying degrees, not always fixing it. But putting four 6N5Ps on instead of the mix always did.

Anyways, I now have the amp back up running after removing the impedance mod resistors from the 6N5P's part of the circuit. I'll keep an eye on if there is some change on that temporal stability thing over a few weeks.

Yes, you're quite right members have been doing this tube mixing for years with no adverse effects, I just don't feel qualified without detailed knowledge of how the WCF circuitry works with this kind of alteration to it to advise. As I said one member even mixed 2 completely different types of tube, and I think he survived!

Hopefully Maxx is available to comment, if not there is probably someone who will.

That power output from the meters is quite normal so no problem from those tubes, the 5998's operate at half that output so there is scope for variation, so long as it doesn't go too high to burn out any internal components.

It does seem that mixing has some effect if that problem goes when you substitute 4 of the same tubes.

Edit: Thinking about this, you have one signal from the driver stage per channel. As it goes to the output stage it gets split into 2 signals which are then recombined to cancel each other out eliminating noise. But before that, within each of the signals it is further split within the WCF into 2 in phase/antiphase manner, this is where my head starts to hurt. What happens when one of those signals outputs a different impedance than the other due to a change of Ra and then recombines with the other with a different impedance? How can the impedance be constantly changing, and if it is what effects does that have on sound and on the circuit function?

I don't know if I'm understanding it correctly, it may be that when the signal halves in the WCF are recombined any differences are cancelled out in the same way that noise would be so differences such as caused by different resistors in each half of the signal within the WCF may well not be there any more at output so impedance changes might not in fact be there. I imagine something like this must happen as it would if one of the resistors went off also.

Answers on a postcard please .

 

[baronbeehive]

Sorry ... ... all that is baloney. You are only proposing to change one resistor per output tube in which case I should have realised that the impedance won't be affected because it will be the same for this tube, even though it will be different for the other tube. And that is probably what is causing the sonic differences when mixing in this way.

See what happens when you make me think .

Edit: Did you get to see WRC Finland btw? I was disappointed that Elfin Evans lost out.. again.. to Seb Ogier!

Edit: When I said baloney I meant what I said, not what you said LOL, just to avoid confusion!

 

[Maxx134]

It is actually one of the reasons I like this amp. With a combination of two different power tubes on each channel it is possible to get characteristics from two different power tubes at the same time and imo the combinations also sometimes do things that the respective single tubes can't do on their own as a set of four

Yes this is good idea.
I was able to simulate an expensive tube, by mixing 2 lesser tubes together..
As long as your anode resistors have higher wattage (like 5W), you should be ok with differences.

Im changing barons quote, ha

Anyways, I now have the amp back up running after removing the impedance mod resistors from the 6N5P's part of the circuit. I'll keep an eye on if there is some change on that temporal stability thing over a few weeks

Should definitely change the tube influence levels.

Answers on a postcard please

You are correct in your assumptions for changing the resistor according to the tube, for the impedance setting from the carveli formula.

The reason is that no one ever did this before, to set impedance for different tubes.

What you essentially are doing, is optimizing the output level of that tube, for that impedance setting (for the headphones).

So your tubes are laid out :
L~Positive + L~Negative, then other side is:
R~Positive + R~Negative

(LP)+(LN), (RP)+(RN)
So your signal should have more equal output levels per each half (positive half & negative half) of the signal.

A "skewed" level ouput would be achieved, by NOT changing the resistor according to the tube.

Part of the sonic difference would be, which tube takes over the resulting sound.

IMO, the positive signal tube would lead the attack (wether it sounds slow or fast), while the negative signal tube would lead the decay( wether it sounds thick or not).
But that is only my general observations.
Your tubes I haven't checked which one is naturally more gain (mu) so maybe the same resistors work in lowering gain of louder one, or lower gain of lesser one.
Striking a nice balance is trial and error.

Maybe you like the greater influence of one tube over another.
I need to try this mixing tubes, on my amp as well...

 

[CopperFox]

Anyways, I now have the amp back up running after removing the impedance mod resistors from the 6N5P's part of the circuit. I'll keep an eye on if there is some change on that temporal stability thing over a few weeks.

It didn't require weeks to notice the difference - with the resistors for the 6080 and 6N5P tubes at the correct values, the sound was noticeably better than before right at starting the amp up... but then degraded a lot more and a lot faster than before, in less than 30 minutes.

After some initial confusion I began to suspect the reason for this would be capacitor failure of the WCF caps, which were 0,33 uf 600V Miflex KCPU-01s.

To investigate further, I removed the Tung-Sol 6080WBs from the amp and put the second bassiest of my 6080/6AS7 tubes in their place - which were GE 6AS7s from the eighties. With these in place instead of the Tung-Sols, there was noticeably less of the sound degradation over time, but there was still some of it.

So here my thought was that the Tung-Sols perhaps put out an amount low frequency current that the 0,33uf WCF caps weren't able to handle, and this amount has also been increasing with the various mods and now also with the power resistors at correct values.

The next logical step was to replace the WCF caps. I first put there some 1uf power delivery caps that were leftovers from other moddings. Those sounded stable but were bright and low on bass. Then I remembered that I have 0,47uf Miflex KPCU-01s on my shelf that I had ordered accidentally while trying to order something else at some point.

So I installed those in the WCF position and with these in place the sound seems to stay very stable over several hours and sound quality is better than before. I will now keep those there for at least some days to see if that changes or not with further burn-in. In any case, it does look like the 0,33uf capacitance perhaps is not high enough here in my use at least, unless the caps had been somehow severely damaged from being soldered on and off several times.

I could try putting the 0,68uf Miflexes in the WCF position and the 0,33uf or 0,47uf as the coupling capacitors. AFAIK there is far less current going through the coupling capacitors than the WCF caps. I do also have the 0,56uf Duelunds that I haven't given a proper burn-in yet and they might burn in faster in the WCF position because of the higher current. I have been planning to get some Miflex KFPM-01 caps to try in either position as I'm interested to see how a copper foil propylene cap would sound without the paper/oil part. There have been at least some people who said they preferred those over the KPCUs even if they are overall less hyped.

The 0,47uf KPCUs are quite large and with those installed I could only put my chassis extension back on without the side plates and if I keep the 0,47uf caps there I will need to carve those a bit to make them fit again.

Is there some kind of limit to how high a capacitance would be good in the WCF position, would the maximum be 1uf as with the decoupling caps or something else? If it is better to have high capacitance caps in the WCF than as decoupling caps, I could perhaps fit in some brand or model with 1uf capacitance.

 

[CopperFox]

Edit: Did you get to see WRC Finland btw? I was disappointed that Elfin Evans lost out.. again.. to Seb Ogier!

Didn't have enough experience for the marshall job, missed out on the deadline for other volunteer stuff by a couple of days and no spectators were allowed due to covid. So I watched the Arctic rally's stream from the internet. It's coming up again in a few weeks and I'll likely go watch whichever series will be ran. There's been some rumours that Sweden might get cancelld again due to legal issues with the event's relocation.

Yeah people losing narrowly to someone named Sebastian has been a recurring theme for a couple of decades. That might change next year without Ogier.

 

[baronbeehive]

Didn't have enough experience for the marshall job, missed out on the deadline for other volunteer stuff by a couple of days and no spectators were allowed due to covid. So I watched the Arctic rally's stream from the internet. It's coming up again in a few weeks and I'll likely go watch whichever series will be ran. There's been some rumours that Sweden might get cancelld again due to legal issues with the event's relocation.

Yeah people losing narrowly to someone named Sebastian has been a recurring theme for a couple of decades. That might change next year without Ogier.

Hey, if you do get any pics in August PM me, I would be interested!

Those rally cars from last year were so fast it's getting quite scary .

 

[baronbeehive]

It didn't require weeks to notice the difference - with the resistors for the 6080 and 6N5P tubes at the correct values, the sound was noticeably better than before right at starting the amp up... but then degraded a lot more and a lot faster than before, in less than 30 minutes.

After some initial confusion I began to suspect the reason for this would be capacitor failure of the WCF caps, which were 0,33 uf 600V Miflex KCPU-01s.

I would have thought unlikely. And the .33uF is the right value for the cutoff. Having said that I think it does require experimentation as we haven't been there ourselves. The power handling you mention should not be a problem, they should handle it and anyway the current output from the power tubes is small, 60mA at the meters which you can see. If the output was too high it would need an OT to handle it which is why we have a low power OTL design on this amp. Also your components are operating well within parameters. The tube could be driven by much more power, and you have uprated components to a higher wattage to avoid reliability problems.

The effect you mention is like that of a buffer filling up and then unable to disperse, or similar to a cap charging and unable to discharge properly over time.

I will have to have a think about this but the obvious things that happen due to fluctuations over time are mainly heat related. Solder, joints, wires losing, or partly losing contact when heat expands the joints. This is something that would not be noticed normally if you check the joints because they would appear OK until they heat up. I don't however think it is that because the sound of a cold solder joint for example would be more of an intermittent noise rather than a gradual decay probably.

Another obvious thought is the tubes themselves, a defective tube for example might be losing control of the electron flow and the anode might be flooded with electrons and unable to get rid of them, so the first thing is try other tubes to see if the same happens.

Another thought is interference of some sort possible due to close proximity of components such as your large PSU caps. If the problem persists I would maybe try to separate them. In some amps you will see a metal shield in between PSU components.

My initial thoughts over tube mixing a few posts back were wrong as Maxx said you are just optimising power for each tube, I got hold of the wrong end of the stick with my comments as I misunderstood what you were planning, so I don't think this would affect anything. As I said that's where my head started to hurt. I know that for class A push pull amps the signal first gets split into push and pull halves at input by the NPN and PNP transistors. Then the two halves are obviously able to retain the different signals produced by tube mixing in each of the 2 signals. But at some point they get recombined by the balanced part of the amp and this is where the noise cancellation takes place, and I'm a bit fuzzy on the exact details. But it's interesting you said the effect you found disappeared when you put in 4 of the same tubes. So I'm still not quite sure how the circuit handles this.

So I installed those in the WCF position and with these in place the sound seems to stay very stable over several hours and sound quality is better than before. I will now keep those there for at least some days to see if that changes or not with further burn-in. In any case, it does look like the 0,33uf capacitance perhaps is not high enough here in my use at least, unless the caps had been somehow severely damaged from being soldered on and off several times.

I could try putting the 0,68uf Miflexes in the WCF position and the 0,33uf or 0,47uf as the coupling capacitors. AFAIK there is far less current going through the coupling capacitors than the WCF caps. I do also have the 0,56uf Duelunds that I haven't given a proper burn-in yet and they might burn in faster in the WCF position because of the higher current. I have been planning to get some Miflex KFPM-01 caps to try in either position as I'm interested to see how a copper foil propylene cap would sound without the paper/oil part. There have been at least some people who said they preferred those over the KPCUs even if they are overall less hyped.

The 0,47uf KPCUs are quite large and with those installed I could only put my chassis extension back on without the side plates and if I keep the 0,47uf caps there I will need to carve those a bit to make them fit again.

Is there some kind of limit to how high a capacitance would be good in the WCF position, would the maximum be 1uf as with the decoupling caps or something else? If it is better to have high capacitance caps in the WCF than as decoupling caps, I could perhaps fit in some brand or model with 1uf capacitance.

I would think fast acting WCF caps would be more important than high capacity there. The cathode bypass caps are already doing the power handling so what we want for WCF position is something to act fast within the feedback loop in each power tube. Those film Miflexes you mention look to be a good bet.

Anyway it looks like you might have the problem sorted. You could also double up the .47 Miflexes, ie connect in parallel to give a combined value of .94uF and see what effect that has on power recovery after bass dips for example. That would avoid having to buy any at the 1uF value.

I have Audyn True Copper in my other amp as bypasses and I like the copper foil sound.

I would try this first before going higher but I don't think higher than 1uF would work as a coupling or WCF cap due to oscillations. Incidently I was wondering if that was what you are hearing.

Finally you need to remember what you did before the problem became apparent to narrow it down, did it happen after you put in those 200uF Wimas for example, if so maybe they may be the culprit? I think that was the last thing you did?

That's just a few thoughts thrown out there. I don't know really what the problem is, AFAIK nobody here has experienced that.

 

[CopperFox]

I would think fast acting WCF caps would be more important than high capacity there. The cathode bypass caps are already doing the power handling so what we want for WCF position is something to act fast within the feedback loop in each power tube. Those film Miflexes you mention look to be a good bet.

Anyway it looks like you might have the problem sorted. You could also double up the .47 Miflexes, ie connect in parallel to give a combined value of .94uF and see what effect that has on power recovery after bass dips for example. That would avoid having to buy any at the 1uF value.

I have Audyn True Copper in my other amp as bypasses and I like the copper foil sound.

Did you try the Audyns somewhere in the mkVI at some point, how did they sound there?

I ordered 0,68uf KFPMs today so will see soon about those. Order number 666 from audiomiflex.pl !
Did some resistor upgrades today too. Put in Mills MRAs for 56k and 2k2 resistors near the psu, Audio Note niobium for the 150K resistor and finally put 1K Mills in place of the stock 3K3 "last r" resistors. Those will probably need some burn in now.

One thing I noticed while doing that was the WCF caps were the fourth warmest caps inside the amp after turning it off. Warmest were PSU electrolytics, PSU output films and power tube decoupling caps. The WCF caps had more heat in them than the coupling caps despite being right next to each other and having fans blowing directly at them. So that heat most likely comes from the caps themselves and not their surroundings.

 

[baronbeehive]

Did you try the Audyns somewhere in the mkVI at some point, how did they sound there?

No, I actually ordered a full set because I liked the sound of them in the reviews but sadly they didn't fit for the compact version and I didn't want to extend the chassis at that point so I changed in favour of the Jupiters.

I ordered 0,68uf KFPMs today so will see soon about those. Order number 666 from audiomiflex.pl !
Did some resistor upgrades today too. Put in Mills MRAs for 56k and 2k2 resistors near the psu, Audio Note niobium for the 150K resistor and finally put 1K Mills in place of the stock 3K3 "last r" resistors. Those will probably need some burn in now.

Oh dear .

When you mentioned your resistor findings I assume they were for anodes? Good job putting in Mills at those locations, they shouldn't cause trouble in future.

 

[Maxx134]

Did some resistor upgrades today too. Put in Mills MRAs for 56k and 2k2 resistors near the psu, Audio Note niobium for the 150K resistor and finally put 1K Mills in place of the stock 3K3 "last r" resistors. Those will probably need some burn in now.

That's excellent. That's what I did because wirewound has superior capabilities.

Also, I am glad you are addressing the WCF cap effects in the circuit.
It will be interesting to know.
I am also convinced that a larger size would be beneficial here.
In beginning of the thread, I did first notice the importance, and then upgrade them from their meager .22uf stock values, to a .33uf for a noticeable improvement. I think the reason I didn't go bigger was that it was early on before I rolled the output tubes, and my tubes were smaller than the ones in your MK6.

But it's interesting you said the effect you found disappeared when you put in 4 of the same tubes

Yeah the first thought for me was a tube issue. A Tube that can perform normal, can also change under stress.

 

[Maxx134]

. So that heat most likely comes from the caps themselves and not their surroundings

Good observation.
I too did notice all the PSU section gave off alot of heat.
I was especially surprised how the caps took on heat, especially the last two after the power transistors.

For those WCF caps to be warm is very good observation that I overlooked. This tells us how much more important it is than I thought.

 

[CopperFox]

While looking at the 6080 tubes' data sheets for clues about temporal stability in operation, there actually were some to be found on the second page in the Tung-sol 6080WA's data sheet.

There in the section "Additional tests to insure reliability", it is stated that the tube's transconductance can change up to 10% from the specified 7000 micromhos in the first hour of use, may drop to 5800 micromhos in the first 100 hours of use and to 5500 micromhos in the first 1000 hours of use.

(I haven't found any data on transconductance variation by hours of use in other manufacturers' data sheets(!!))

Those changes would throw off the calculations of the Cavalli formula for the impedance mod quite a bit. If the tube spends most of its life near the 5500 micromhos transconductance rate, then the correct resistance resulting from the formula would be 280 ohms and not 220 ohms, which is the result with the specified 7000 micromhos rate.

And that 280 ohms is closer to the stock 330 ohms value than the 220 ohms value... so the best resistance value would likely be something other than the result from the spec transconductance.

I don't know how many hours 6080 tubes are expected to last either, but from what I know it would be a lot more than 1000 hours, so the transconductance rate (and average transconductance rate across the tube's life span) may likely drop lower still.

It should be also noted that the 7000 micromhos specified transconductance only applies with specific operating conditions which are listed at the beginning of page 3 of the data sheet, which are Ef = 6.3V, Eb = 135V, Ec = 0, Rk/k = 250 Ohm. (What does that Eb voltage refer to?)

And on the last page of the data sheet there are curves of how the transconductance varies by DC grid voltage and some other variables. How should those be taken into account here?

(I've now removed the impedance mod completely myself but might put it back later on... with some values).

 

[Maxx134]

(I've now removed the impedance mod completely myself but might put it back later on... with some values).

 

[Maxx134]

When changing my values for impedance mod, I tried different by adding paralleled resistors, and noted that the specific tubes I used performed better when I selected my new value, so I stopped at that point, bit the improvement was of both clarity and holography on the tube, so I would suggest testing as the formula (as you have pointed out) is not perfect due to variables.

 

[baronbeehive]

Also, I am glad you are addressing the WCF cap effects in the circuit.
It will be interesting to know.
I am also convinced that a larger size would be beneficial here.

Yeah, also a cap like the Miflex film cap type would also be good bearing in mind those .22uF's were Wimas originally.