| SY |
| I'm curious if anyone who's built an Aikido has done any distortion measurements, especially harmonic distribution. Some circuit particulars, too, please. |
|
|
| SY |
| My goodness, all that interest, all those boards, all those builders, and nobody has put the thing on the bench??? |
|
|
| Andypairo |
One should have a suitable bench ;-)
Me not, at least for the moment.
Cheers
Andrea |
|
|
| Tubes4e4 |
Hi SY,
| quote: | Originally posted by SY
My goodness, all that interest, all those boards, all those builders... |
That was yesterday. Like with all fashions, yesterday means yesteryear. ;)
What might be the next fashion? Maybe using pentodes as pentodes in small signal circuits? Oh, what a truly shocking alta moda that would be :rolleyes:
Yawning, Tom
|
|
|
| Bas Horneman |
| quote: | | My goodness, all that interest, all those boards, all those builders, and nobody has put the thing on the bench??? |
If people knew how to do proper distortion measurements. They would not be ordering boards but would do point to point. The boards just make it that much easier (and neater).
The only distortion measurements I know of are on John Broskie's site.
http://www.tubecad.com/2005/January/blog0032.htm |
|
|
| Gluca |
OK measurements are measurements ... maybe some O/Scope shots ...
Or just listening impressions ... do ya prefer it or the now-classical-outoffashion CF/SRPP/etc... ?
Ciao
Gianluca |
|
|
| maynardg |
Mine is P-to-P, 220V B+ circuit in Blog 30. 6CG7 into 5687. CLCRCRC PS with regulated 12V DC heaters. I built it on a chassis from a AES AE3 preamp using mostly the existing power supply.
Sound is better than the AE3, which was a typical voltage amp-cathode follower 6SN7 affair. Better in every way. Dead quiet, more dynamic and detailed. More true to life for lack of a better term. Piano sounds like piano - nothing soft or muddled, nothing edgy.
No measurements except the one that counts - listening pleasure.
MG |
|
|
| Geek |
I was asked by the creator of the Aikido PCB's to do a test between 1.25mA and 2mA Ia. Sonically irellevent. Because of that, I never 'scoped it.
I agree with maynardg - "No measurements except the one that counts - listening pleasure." :D |
|
|
| SY |
| quote: | Originally posted by Bas Horneman
If people knew how to do proper distortion measurements. They would not be ordering boards but would do point to point. The boards just make it that much easier (and neater).
The only distortion measurements I know of are on John Broskie's site.
http://www.tubecad.com/2005/January/blog0032.htm |
Yeah, I saw those, and that was a good starting point. I'm quite interested in the effects of tubes, operating point, and circuit trimming.
BTW, I'm an exception to your rule- I have it built on a board (one of Broskie's- it is a heroic PCB), yet I know how to do proper distortion measurements... |
|
|
| Bas Horneman |
| quote: | | BTW, I'm an exception to your rule - I have it built on a board (one of Broskie's- it is a heroic PCB), yet I know how to do proper distortion measurements... |
There are always exceptions to the rule. |
|
|
| Bas Horneman |
| Why do you want the figures by the way Stuart? Would you like to compare them to yours? |
|
|
| SY |
| Well, certainly that, but also to probe what the circuit can and can't do, to get an idea of how to really optimize linearity for different applications. I can build about a million variations, but that will take me a million years. So I'm playing around with a couple, but the collective efforts of other guys in their workshops would help map out the design landscape a bit more than it is now. |
|
|
| anatech |
Now if I had a digital 'scope with math functions ...
I generally test everything I build. I find it helps explain what I'm hearing sometimes. My non-technical listening audience are unaware of measurements to remain unbiased. Right now I can only do THD and 'scope the residuals.
-Chris |
|
|
| Geek |
| Ooops, regarding my earlier post... I was informed the preamp PCB I listened to was indeed *not* the Aikido. Sorry for the unintentional mislead :( :o |
|
|
| James D. |
Ummmm... I have done the measurements.... but only on a Pentode Aikido... so not really relevant...
On the pentode Aikido the distortion and spectrum at the output look the same as the distortion and spectrum on the pentode anode i.e. measure the same and has the same spectrum - as far as my measurement setup can tell (say to 5% of measurement).
This using a 6B8-G pentode and driving two PT15s in triode mode PP through and IT phase splitter.
I haven't bothered to measure the all triode Aikido as it sounds so open and transparent that I just listen to it :-) (6N1P and 6N6P)
James |
|
|
| SY |
| How much gain did you get from the pentodes? |
|
|
| James D. |
Hi SY,
I'm runing the pentodes at a gain of about 65 times.
James. |
|
|
| 2wo |
I did post a scope photo on another thread
Look under "improving the Aikido line stage" |
|
|
| SY |
Thanks, guys. With triodes, I find the gain to run about 40-45% of mu, so pentode definitely had me curious.
Distortion results I've gotten so far are quite good, but not blow-away. Distortion reduction requires a LOT of supply voltage, and JB's boards are set up so that the same tube envelope carries both top and bottom triode (to more easily allow voltage amp and follower to be different tubes). This means that the tubes' heater-to-cathode voltage ends up being strained. For the case of ECC88-family, one has to take the plate-to-cathode voltage well above the rated maximum to get the distortion to fall below that of a common-cathode stage.
At output voltages more typical of line stages (2-3VRMS) and a 160V supply, 2nd HD is the only one visible at -55dB for the best-selected tube sample and the resistor ratios tweaked for the best distortion null. |
|
|
| James D. |
As another example Dave Dove posted his pentode aikido into 300BXLS design a little while ago. I designed the aikido stage and that runs the pentode at a gain of about 110 times.
I have run a pentode aikido at 200 times gain on the bench with no problems and a cascode aikido at a gain of 360 times... yes I'm playing with Aikido RIAA cicuits for a DL103 :)
James |
|
|
| SY |
| If the pentode Aikido gives no distortion reduction, why use it? Or by pentode Aikido, do you mean just a single pentode driving the stacked follower? |
|
|
| leadbelly |
| quote: | Originally posted by SY
Distortion reduction requires a LOT of supply voltage, and JB's boards are set up so that the same tube envelope carries both top and bottom triode (to more easily allow voltage amp and follower to be different tubes). This means that the tubes' heater-to-cathode voltage ends up being strained. For the case of ECC88-family, one has to take the plate-to-cathode voltage well above the rated maximum to get the distortion to fall below that of a common-cathode stage. |
Aha! Thanks for that gem :) I had been musing for a while on using 1 dual triode for the top triodes and another of the same for the bottom triodes and posted this in the GB thread. The GB boards do not do this either. I'll have to consider this for a future amp. |
|
|
| leadbelly |
| I just had a look through the datasheets of tubes I have on hand, and the max cathode-heater voltage for a 6N1P-EV is +120V/-250V. Great, I can make a 300V B+ Aikido using the GB boards! Thanks for the info SY :) |
|
|
| SY |
Thanks for the research- sounds like a tube I should try in this.
Here's an interesting comparison I ran during the last week. With 160V B+, 12.5VRMS output, circuit trimmed to a fare-thee-well, the best of the ECC88-family tubes gave a gain of 18, with -40dB 2HD, -50dB 3HD, -68 4HD, and -69dB 5HD. Bouncing the B+ up to 200V dropped the 2nd by 2dB, 3rd by 3 dB. A brief run up to 280V (totally abusing the heater insulation) showed a 2HD of -50dB.
With the same 160V B+, the same tube in grounded cathode (CCS loaded) direct coupled to a cathode follower (also CCS loaded) gave a gain of 41, and distortion of -46dB 2HD, -70dB 3HD, -80dB 4HD, and -85dB 5HD. So the Aikido shows a clear distortion advantage at high voltages, but not at lower voltages. And half the gain.
Interestingly, the relative ranking of linearity of different ECC88 types that I've tested was quite different between Aikido and grounded cathode. For example, the CCa which was the distortion champion in grounded cathode was among the worst in the Aikido.
I'm looking forward to trying the 6N1P- it seems much more optimal. |
|
|
| James D. |
The pentode and cascode aikido are as per the tubecad site i.e as you said, single pentode or single cascode for the cascode aikido driving the stacked pair. The distortion reduction here is by unloading the pentode or cascode i.e. letting the pentode load resistor define the load on the pentode and so enable a very well defined load line to be used that optimises the operation of the pentode (cascode) enabling fine tweaking of the distortion level and spectrum. Then there is the noise cancellation of the circuit that helps open the sound up.
If you used the typical low Gm pentode directly then you have a varying and reactive loadline for the poor beastie to drive - the aikido removes that variability or rather reduces it to low levels. High Gm pentodes do not need quite so much help and, as I said n the 2A3 driver thread, make fine DHT drivers without any help.
James |
|
|
| Sheldon |
| quote: | Originally posted by James D.
As another example Dave Dove posted his pentode aikido into 300BXLS design a little while ago. I designed the aikido stage and that runs the pentode at a gain of about 110 times.
I have run a pentode aikido at 200 times gain on the bench with no problems and a cascode aikido at a gain of 360 times... yes I'm playing with Aikido RIAA cicuits for a DL103 :)
James |
Nice. Please report your findings when you are satisfied. I'd like to use this for an RIAA stage too.
Sheldon |
|
|
| burnedfingers |
While I would really like to comment on this I will stand back and hold by breath. This stems from asking the guys on the MArantz CD63 67 mod thread if they had performed any distortion tests.
It seems that they can hear far better than any distortion tests that could be performed. It is to be understood that all mods concerning the Marantz should be taken at their word.
I consider your thread to be of considerable interest. If the tone of this thread doesn't go sour I will post some Pentode results as well as P-pont and Bas Aikido measurements. Since my personal analyzer took a dump I can still rely on the equipment at the place I work.
It is interesting to see how some little tweeks can change the distortion figures. |
|
|
| SY |
| I would love to see any results you've got. I guarantee it will be received with a "tone" that is grateful.:D |
|
|
| James D. |
Completely agree on wanting to see comparative results and not being rude or abusive about the results :)
I may stand in the corner and cry but I'll do that privately...
James |
|
|
| rdf |
| quote: | Originally posted by burnedfingers
It seems that they can hear far better than any distortion tests that could be performed. |
I don't find that to be completely true, at least not in the usual sense. Single numerical THD figures are nearly worthless when correlated against perceived quality. Wide band harmonic analyzes of distortion components are extremely useful. |
|
|
| burnedfingers |
Quote:
I don't find that to be completely true, at least not in the usual sense. Single numerical THD figures are nearly worthless when correlated against perceived quality.
By the term "single numerical THD figures" do you mean distortion figures in a one place decimal? If so I see no mention of single place decimal test results. Please point me towards them so that I might review them.
Quote:
Wide band harmonic analyzes of distortion components are extremely useful.
Would you expect anything less? |
|
|
| SY |
| I think rdf meant single numbers like "0.1% THD" as opposed to harmonic distribution. And I'd agree with him on that. |
|
|
| burnedfingers |
quote:
Originally posted by burnedfingers
It seems that they can hear far better than any distortion tests that could be performed.
I don't find that to be completely true, at least not in the usual sense. Single numerical THD figures are nearly worthless when correlated against perceived quality. Wide band harmonic analyzes of distortion components are extremely useful.
Point being that I made no mention to distortion tests in a one place decimal so I fail to see the point. The original Marantz specification call out three place distortion figures. They (Marantz CD 63 67 group) claim there is no need for distortion tests because can hear the differences. Poppycock!
rdf
Is there a mention of one place distortion tests?
No, I don't think so. Yes, I would agree they are worthless. Why bring it up? |
|
|
| rdf |
| SY captured the meaning: single numerical, not single digit. I was essentially agreeing with you and disagreeing with those who "...claim there is no need for distortion tests because can hear the differences." Why bring it up? In a thread entitled "Survey: Aikido Distortion" it's relevant. Your post was just a convenient jumping-off point to toss in the notion of in-depth harmonic analysis into the conversation. It's possible I mis-represented those you quoted by presuming they meant a 'one number' approach to THD, but it's my experience those who completely dismiss distortion measurements generally think of it in exactly this way. |
|
|
| SY |
JB was kind enough to send me an octal board with premium components (Teflon caps, bulk foil resistors...). Set it up with 5692s, 10mA, 275V B+. At 100Hz, 1kHz, 10kHz, I got identical distortion readings at 2.5VRMS out. -68dB 2nd HD, no other harmonics detectable.
At 15VRMS out, 1 kHz, -53dB 2nd HD, 3rd nudging the low -90s. Nothing else detectable. Very impressive, much better performance than the ECC88 family.
I'll try to get some pix of JB's boards up; they are massively impressive. |
|
|
| SY |
| A few pix. First, the board in nine pin format. The tubes here are 6KN8. You can see the rather crude job I did with trimming component values... That will be cleaned up shortly. |
|
|
| SY |
I should note that these boards are two sided, PTH, and quite thick (1/8").
The octal board, populated with two different 5692s. The blank spots for coupling caps are reserved for some pretty big Russian Teflon caps. |
|
|
| SY |
| And a closer look at one corner. A nice picture of what genuine 5692s should look like, BTW. |
|
|
| leadbelly |
| Very nice! I could spot some metal films and what looked like thick film resistors, and some other kind that were too small to make out. Would you be so kind as to tell what key component choices JB himself has made :worship: |
|
|
| SY |
Sure. Vishay bulk foils for the cathode resistors. The rest are mostly (I think) Yageo metal films. He uses Wima MKP caps for most positions, but does two output CR coupling networks. In one, I have a cheezy Electrocube. In the other, I will mount John's favorite, some super-heavy Russian Teflon caps. For a volume control, he's designed a nifty stepped attenuator, which I don't have a picture of yet. The boards go together quickly; I think it took me about a half hour to stuff one and check it out.
JB favors a high B+. I reran the spectra at 15VRMS out but with a 400V B+. That dropped 2nd HD down to -57dB, 3rd and higher well below residual. So there's something to be said for that, but I'm not sure the extra dissipation is worth it considering how low the distortion already is at lower voltages.
All measurements were cross-checked between a sound card/spectrum analyzer and an HP 3581A wave analyzer. |
|
|
| Bill*B |
| No measurements, but I like the one I built just fine - it sounds exactly like the recording. Regards, Bill. |
|
|
| leadbelly |
| quote: | Originally posted by SY
JB favors a high B+. I reran the spectra at 15VRMS out but with a 400V B+. |
How so? The datasheet says 5692's have a 100V heather-cathode rating, and I thought JB's boards used upper and lower triodes in the same tube? |
|
|
| SY |
| This is slightly stressing them. At rest, the top cathode sits at about 1/2 the B+ (200V). So if the heaters are biased up 100V, there's no violation. Barely. Signal takes that over peak, so it's not a way I'd recommend that it be run. I'll live with a skootch of 2nd HD to run the tubes more conservatively. |
|
|
| James D. |
In Dave Dove's pentode Aikido driving 300BXLS we are running 430V as B+. The aikido output stage is a 6N6P - so here the heaters are biased up at +120V which is fine. The pentode heaters sit at the same point and are happy so far.
The other reason for a high B+ when using it as a DHT driver is the support for output voltage swing. In the above case that can be +/- 120V. so the hign B+ is used! We have driven it that hard when filling a hall 80ft by 130ft and it stayed very clean sounding all the way through.
Nice to see the exemplory distortion figures and spectrum (such as it is) from the Aikido - it seems to confirm the listening impressions of those who have built it!
ciao
James |
|
|
|
