Motorboating could be a result of feedback via B+, that is when the power tubes draw more current this reduces
voltage on the pre-tubes which in turn signals the power tubes and the show is rolling.
Adding resistance in the B+ at an early stage will accentuate the sensitivity to mptorboating.
I suggest you go back to the LC type filtering and add some more caps closer to the preamp tubes
The original amp has RC links to power the pre tube and is quite insensitive to this. You could install the original
components and problems are gone.
voltage on the pre-tubes which in turn signals the power tubes and the show is rolling.
Adding resistance in the B+ at an early stage will accentuate the sensitivity to mptorboating.
I suggest you go back to the LC type filtering and add some more caps closer to the preamp tubes
The original amp has RC links to power the pre tube and is quite insensitive to this. You could install the original
components and problems are gone.
very nostalgic - nearly 50yr back I had fresh MKIII with Speakerlab K-horn clones and a Dyna PAS. The 6AN8 circuit seemed decent sounding. A lot later I had a Stereo 70 with GSI differential cascode 6DJ8 front end which seemed good with a set of 1980 Klipschorns. That ("Schmidt" ?) front end is something like my HF87 and should be better than the 6AN8 Thinking back about flabby bass with some 3-way Dynaco speakers - might Jame's Boak's MK3 regulator fix some of that effect ?
I see what you're saying.
Thing is though, there is an RC network at the preamp tubes - 10K - 34 uF - 39K (top) and 43K (bottom) to the phase splitter followed by 39K - 34 uF - 39K to the input tube.
I can increase the capacitance to the input tube - with the components I have on hand I think I could increase it to 41 uF or 50 uF.
The input tube does have a lot of gain, though. More than the 6AN8.
I did a bit of youtube research followed by some experimentation tonight...
The prevalent theory about motorboating oscillation is that it drives down the B+ and input tube voltages causing a self-propagating oscillation.
One video explained that it's easy enough to tame motorboating by simple lowering the output tube bias point.
Another potential solution was to increase capacitance at the B+ and to increase the value of the driver tubes' power supply decoupling resistors which both lowers the input tube voltage and also mitigates sagging effects on the B+ when the output tubes demand more current from the power supply.
Interesting theory and easy enough to test.
I added 5K of resistance to R10 and alternatively at R11 (schematic is a few posts above) which reduced the problem but it didn't go away.
Then I replaced the 12AT7 with another (vintage) tube I had in my stash. Again, the situation improved but not by much.
Finally I restored the power supply back to where I initially had it at first, with 42 uF before the choke and 34 uF after the choke at the B+ which gave me back the high 525V plate voltage.
But this time I switched the input tube to a 12AU7.
So all the tubes are running higher voltages than nominal, but I could bias the KT88s now to 1.15V (I could go to 1.56) at the bias resistor without any motorboating oscillation which is 51.8 mA per tube and at 525V at the plates means each tube is dissipating 27.2 watts at idle. A bit less than the 33 watts at idle spec-ed by the Dynaco manual but I suppose I could go higher.
The culprit I guess, is excessive gain in the input stage. Or wonky 12AT7s?
The prevalent theory about motorboating oscillation is that it drives down the B+ and input tube voltages causing a self-propagating oscillation.
One video explained that it's easy enough to tame motorboating by simple lowering the output tube bias point.
Another potential solution was to increase capacitance at the B+ and to increase the value of the driver tubes' power supply decoupling resistors which both lowers the input tube voltage and also mitigates sagging effects on the B+ when the output tubes demand more current from the power supply.
Interesting theory and easy enough to test.
I added 5K of resistance to R10 and alternatively at R11 (schematic is a few posts above) which reduced the problem but it didn't go away.
Then I replaced the 12AT7 with another (vintage) tube I had in my stash. Again, the situation improved but not by much.
Finally I restored the power supply back to where I initially had it at first, with 42 uF before the choke and 34 uF after the choke at the B+ which gave me back the high 525V plate voltage.
But this time I switched the input tube to a 12AU7.
So all the tubes are running higher voltages than nominal, but I could bias the KT88s now to 1.15V (I could go to 1.56) at the bias resistor without any motorboating oscillation which is 51.8 mA per tube and at 525V at the plates means each tube is dissipating 27.2 watts at idle. A bit less than the 33 watts at idle spec-ed by the Dynaco manual but I suppose I could go higher.
The culprit I guess, is excessive gain in the input stage. Or wonky 12AT7s?
The culprit sounds like something amiss in the feedback circuit/compensation networks. If you sub in a 12AU7 without changing anything else, you've considerable reduced the feedback. I'd try to track down the oscillation issue with the original circuit, otherwise it may never really be stable or operate anywhere near it's potential. It will always live on the edge, so to speak.
nerdorama wins the troubleshooting sweepstakes!
After prodding and probing and swapping parts I went right back to the beginning.
Removing the feedback circuit from the speaker terminal did very little - I still get motorboating as I raise the bias current.
Then I disconnected the screen grid feedback loop and presto - I could bias way up to 45 watts plate dissipation at idle and it was still stable.
This got me to look a bit more closely and like an idiot, I had connected the output stage out of phase! So the "top" of the phase inverter was connected to the "bottom" output tube as shown on the schematic diagram.
Doh!
Easy to fix, and whammo. All of a sudden this thing is singing properly.
The B+ voltage is still quite high. So rather than bias to the current specified in the manual - 70 mA per tube, which gets 33 watts plate dissipation per tube, I biased to them to 34 watts plate dissipation at idle which at 525V measured at the plates means the tube is conducting more wattage than the book specified. Measured at the Biaset point, I get 1.468V or 66 mA per tube. Close enough that it ought not to make any appreciable difference.
The voltage at to input tube is also high, which concerned me a bit, but the grid voltage at the first phase splitter tube (which is shared with the plate of the input tube) is low enough that it doesn't cause issues with imbalanced gain at each phase splitter triode. So it's outputting more or less even waveforms, top and bottom.
The 1 kHz square wave looks quite good on the 'scope on a dummy load. The 10 kHz square wave looks good too, with two very tiny cycles of ringing on the leading edge. If I dial in too much feedback it goes away but at the expense of the trailing edge sharpness.
It clips at just a hair over 60 watts, so that's great.
With the 12AT7 input / 6CG7 phase splitter it has a canoe-full of gain, though. It can get to a full 60 watts at the output with just 690 mV at the input and it clips at 700 mV input. That's 28 db of gain, before clipping, I think.
And the sound? I think it sounds pretty damn good for an amplifier with brand new output tubes, new rectifier and a NOS phase splitter. The input tube is vintage and very likely has some use on it, though I don't know how much.
I'll let it play music for a few days and report back on the sound.
But at least now I know the circuit boards I designed are sound, have no errors, and most importantly I now have the confidence to get building the second amp.
Last edited:
Glad you got it resolved! Can you take some screen shots of the 10kHz square wave? You might also want to test for stability by dabbing .1uF across the load and seeing if the 10kHz square wave hangs in there or collapses into oscillation. I'd encourage you to try to get the feedback up to normal levels for this amp, it will perform better. If there's too much roll-off that can be adjusted with the phase-lead cap, assuming there is one.
baldrick, perhaps note that the Kara Chaffee modification was done 'by ear' and perhaps on a whim, with no supporting technical assessment as to level of feedback, or alteration of original stability performance, or change to low and high frequency responses. So unless you have come across some others who have taken time to check on response and stability then caveat emptor.
I'm not saying the original was unconditionally stable, but the kit appears to have credible technical design that has not shown up many underlining deficiencies. You may want to review the comments from Dave Gillespie in https://audiokarma.org/forums/index.php?threads/help-understanding-dynaco-mkiii-mods.898002/ as a further aspect of damage prevention.
I'm not saying the original was unconditionally stable, but the kit appears to have credible technical design that has not shown up many underlining deficiencies. You may want to review the comments from Dave Gillespie in https://audiokarma.org/forums/index.php?threads/help-understanding-dynaco-mkiii-mods.898002/ as a further aspect of damage prevention.
That's inspired me to get a 1964 aussie instrument amp out to check its stability. It's main circuitry is a knock-off of the Dynaco Mk3, so pre-dates Sunn 200 use for a bass amp. Given it uses aussie iron, its stability performance won't be the same, but should be somewhat similar as it was a hi-fi category OPT.
trobbins hinted about adding screen resistors in MkIII Good advice, i just ordered some wiring supports for my MkIII to
add these screen stoppers.
add these screen stoppers.
No doubt. But...
This overall topology is pretty much exactly the same as the Mullard 5-10. Same output stage, identical phase inverter and same front end except it's triode rather than pentode. Even the feedback scheme is the same except in Chaffee's case it's got a variable resistor and foregoes the bypassing capacitor.
The Mullard 5-10 is a time-tested design. Sure there could be improvements and tweaks and I'll probably explore those.
Plus I do trust Chaffee's ear. She's designed some nice stuff.
She is an amplifier design engineer. Colloquial writing style aside, one would think that she bench-tested the thing before committing it to a published article.
I have extras. Once this project is finalized, fully tested on both amps and adjustments made I'll make the PCBs available to the community.
UPDATE:
The replacement tubes finally arrived to complete the second amplifier.
I built it over the Christmas holiday downtime and lo and behold it seems to work perfectly. (Pictures to follow)
Interestingly, all the expected voltages are within range, so I went back and re-measured the first amp. Voltages still high on that one.
So probing around with the ol' FLUKE-O-LATOR, I measured the following with the rectifier tube removed:
The line voltage was 121.4V at the time.
So the issue looks like it is in the power transformer of the first amp which is giving me a 20 volts too much secondary voltage and drawing more current in the primary.
The amplifier does sound good, though, but with higher plate voltages it seems to make a bit more gain than the second amp, which unbalances the stereo image unless I twiddle the balance knob on the preamp (I know, that's what it is for).
At this point, I'm interested in any ideas you guys may have to lower the supply voltage of the first amp.
The replacement tubes finally arrived to complete the second amplifier.
I built it over the Christmas holiday downtime and lo and behold it seems to work perfectly. (Pictures to follow)
Interestingly, all the expected voltages are within range, so I went back and re-measured the first amp. Voltages still high on that one.
So probing around with the ol' FLUKE-O-LATOR, I measured the following with the rectifier tube removed:
The line voltage was 121.4V at the time.
| FIRST AMP | SECOND AMP | |||
| Transformer secordary | 434-0-434 V | Transformer secordary | 414-0-414 V | |
| Line amperage | 400 mA | Line amperage | 360 mA | |
| Filaments | 6.65 V | Filaments | 6.66 V | |
| Rectifier filament | 5.42 V | Rectifier filament | 5.43 V | |
| B+ (with rectifier) | 521 V | B+ (with rectifier) | 489 V |
The amplifier does sound good, though, but with higher plate voltages it seems to make a bit more gain than the second amp, which unbalances the stereo image unless I twiddle the balance knob on the preamp (I know, that's what it is for).
At this point, I'm interested in any ideas you guys may have to lower the supply voltage of the first amp.
Alright - I looked through my stash and found a brand new JJ5Y3 I had bought for another project.
That wasn't the answer. That brought the B+ down, but way too much - now 25 volts low and couldn't deliver enough current or plate voltage to get to the correct output tube bias, so that was out.
Then I spotted a dusty old RCA 5U4G that I had marked "tests good".
In it went, but I don't think that it is the right solution either.
With the tubes biased to the recommended current, the B+ voltage was about 10-15 volts lower than the 480 expected by manual, with expected lower voltages at the plates of the input and phase splitter tubes.
It didn't sound bad, per-se, but not as good as with the slightly higher B+, which has better bass control.
I need something in between the 5U4G and the 5AR4.
That wasn't the answer. That brought the B+ down, but way too much - now 25 volts low and couldn't deliver enough current or plate voltage to get to the correct output tube bias, so that was out.
Then I spotted a dusty old RCA 5U4G that I had marked "tests good".
In it went, but I don't think that it is the right solution either.
With the tubes biased to the recommended current, the B+ voltage was about 10-15 volts lower than the 480 expected by manual, with expected lower voltages at the plates of the input and phase splitter tubes.
It didn't sound bad, per-se, but not as good as with the slightly higher B+, which has better bass control.
I need something in between the 5U4G and the 5AR4.
What is your filament voltage in each amp? This can be a telling number more than the B+. You need to shoot for 6.3V when the tubes are heated and operating. I have had high filament voltages and by using an inrush limiter on the primary I could get it under control. This also has a beneficial effect on B+ when it is running high. If the heater voltage is good but B+ is still high then something else might be in order. My recent Mk2 projects have high B+ on the order of 500-510V. This isn't necessarily a problem as long as the bias current is adjusted correctly.