• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Y A A B - Yet Another Aikido Build

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I did some testing with AudioTester and the Audiophile 192 card this weekend.

Here is the FFT noise spectrum of the card for reference, running at 192 khz/24 bit input:
 

Attachments

  • aikido loopback test for audiophile noise.png
    aikido loopback test for audiophile noise.png
    22.1 KB · Views: 747
I'm calling the current Dackido setup "Configuration A".

This is a 207v B+, GZ34 tube rectifier, 6n23p-ev/5687 tubes, 10 ma/19mA, LCLCLC power supply as described previously.

The following chart shows the noise spectra of the Aikido boards without any input, one color for each channel:
 

Attachments

  • aikido config a self noise.png
    aikido config a self noise.png
    23.7 KB · Views: 742
I haven't been able to figure out how to calibrate the Audiophile 192, or maybe calibration is broken, so the 1 khz signal is >20db down.

It would be better to be able to get a 0db tone so as to be able to measure THD and distortion accurately.

AudioTester has a nice feature where the FFT can be continuously recreated and displayed. I set this mode and twiddled the noise injection pots on the Aikido boards, fully expecting the noise spectra to change.

Nope, nada, nothing. Or the changes were too subtle to notice. What?

One thing I found out was with my 100k pots the minimum range of adjustment is about 200 ohm. So if the PSRR nulling requires a resolution less than 200 ohms, I'm missing it. But if the null is that sharp, probably everyone is missing it without test equipment and good resolution pots/resistor combos. This issue needs more investigation.

Next I did some frequency sweeps. I found a sweep tone WAV file and played it via the Squeezebox->Pace Car->PCM63 DAC->Aikido.

The AudioTester results appear to work fine above 50 hz or so. I'd like to see levels below that at some point. Here is one view of the frequency response:
 

Attachments

  • aikido config a freq response.png
    aikido config a freq response.png
    23.7 KB · Views: 713
AudioTester can also do a sweep on the Audiophile 192 output, but I wasn't able to get this working with SPDIF into the DAC. It's possible that the sweep tones I picked up on the Internet aren't completely accurate in the low frequencies.

Here is another view, one color for each channel, zoomed in:
 

Attachments

  • aikido config a freq response zoomed in.png
    aikido config a freq response zoomed in.png
    27 KB · Views: 712
Thinking about it more, the reason the test tone is down ~25db is because the DAC is low gain, given the low I/V resistor value. The test tone is probably at 0 db. However, calibration is still an issue because even with loopback testing of the Audiophile card, I'm still under 0 db on test tones.
 
Thanks for the info. I am interested in your THD level with the Aikido on Autio Tester. I get 0.15% THD with my DAC-60 modded with a 50 ohm I/V, Jenson PS caps, Hammond choke, 5k resistor on the output, and Amperex tubes.

Here is the 0dB 1khz sine tone tested with an EMU0404 soundcard.

Dac-601khz0db.jpg
 
And this I took down -20 dB to get the peak at the same height
as yours. It looked like your peak was at -30dB? You seem to have a 60hz hum problem that is carried foward thru all the even and odd harmonics, 60-120-180-240, etc. The Aikido is supposed to have good PSRR so I think the problem is before the Aikido. Your IV resistor may be too low.

I would like to email you a sibalance test file I have and mybe you could give me an impression with your setup.

I am very close to turning my D1V3 into a PCM63K -> Aikido, but want to make sure I am doing the right thing.


DAC-601khz-20db.jpg
 
Nice charts, Regal -- really appreciating your contributions.

Regarding the 60 hz bump -- my next post was going to be entitled "Who you jiving with that harmonik debris? ", with apologies to Frank, but you beat me to it. :)

It isn't from the DAC, it's coming via the power supply, I believe. I posted a chart that shows the self-noise of the Aikido. That is without the DAC connected. Same 60hz bump and harmonik debris.

In fact, I'm more confident now after this testing that the I/V is working well since the noise levels are similar with and without DAC connected. Frequency response also looks fine. I do want to try the same testing with the Buffalo because the bass doesn't seem right -- it's possible the Sabre chip doesn't like the low I/V resistor, or there is too much noise not running differential I/V.

The Aikido is supposed to have good PSRR but how come changing the supply noise injection ratio via the pots had no effect that I could detect? Did I build it wrong? Or is the null very narrow? Or something else I haven't considered?

Also, no fair! You have a regulator! :) Even the crappiest regulator after my LC cells would knock down noise dramatically. But will it sound better? That is one of the goals of this project, to see if that is the case.

I'm definitely interested in your sibilance test file. You don't have email activated, so if you click on my name in the posting header and send me an email with your email, I'll send you my email.

I actually think your THD is even lower than you measured. It looks like you might be running into the noise floor of your sound card. Either that or you are running at 96khz? Also the bass response looks suspect. It would be interesting to see a loopback test chart for reference.

What is your D1v3, your modified Lite DAC?
 
the D1V3, is a Dac modeled after the Pass D1.


Also I just changed the Dac-60 SRPP stage to the same total pole config as Broskie's Aikido (the input.) The distortion jumped from 0.15 to .6 THD.

I am guessing this is because I am using an LED on the bottom cathode or something else. I need the cathode bypassed for the gain, or I may look for a higher mu tube.
 
I couldn't understand why the Aikido totum pole was giving poor distortion results but sounds better than the SRPP. TI found that the input impedance of my soundcard is only 3khz. The output impedance of the totem pole is 2.6K, so its not a fair test. I need to go to Radioshack to test with my headphone amp in bewteen.
 
The Lundahl LL1673 20 Henry choke I am using has an interesting feature.

The choke is composed of two 10 Henry chokes. Currently I am running the chokes serially to get the 20 henries. The chokes can also be run in parallel, for large currents, at 5 henries.

But there also a third mode -- as a combination common mode choke, "for improved common mode rejection", and regular choke. One choke is used in the out current path and one in the return path. Ground is not applied until after the choke. Like in this picture from the Emission Labs site:

274-2.gif


I'm going to try this mode and see how it goes. I'll measure ripple with my oscilliscope and compare to normal mode. I'll also rerun the spectrum analyzer.

I've moved temporarily from the Lux 300b amp to a slightly modified Trends TA10.1 Class-D amp. I want this Dackido project to sound good on a variety of amps.
 
I wasn't able to detect any difference, either in sound or measurements, between the Lundahl LL1673 serial and common mode configurations.

Ripple stayed the same at 3-4 mV.

The FFT looks very similar, too:
 

Attachments

  • dackido config a self noise lundahl common mode choke.png
    dackido config a self noise lundahl common mode choke.png
    23.7 KB · Views: 592
Power seems reasonably clean here in Western Oregon so maybe I'm not getting common mode noise in quantities that need filtering. What is more of a problem are voltage swings up to 127 vac. That is one reason why I have a Power Plant Premier, to regulate voltage. I tested the LL1673 common mode with power directly from the wall and also through the PPP.

This year the power may be particularly clean, what with strong flowing rivers due to a twice normal snowpack and strong winds along the Columbia River Gorge. Earlier this week the power operators had to tell the wind farms to shut down equipment as there was too much juice getting delivered.

There was another river drowning today, an 18 year old guy, this time on the South Umpqua. People don't realize that deeper levels of the river flow faster than at the top, pulling them down. That can be deadly in combination with the ice cold water.

*****

I've moved on to Configuration "B". This is the "at least my cold war tax dollars were good for something, sort of, indirectly" special.

This Soviet Special is using the same 6n23p-evs as Configuration "A", but with 6n6ps replacing the 5687s. Kevin Carter of K&K turned me on to both of these tube types, by the way.

Voltage is 204 volts with a Mullard GZ32 rectifier. I guess this voluptuous British babe is a bit of Glasnost in the circuit. Cathode resistors remain 161/132 ohms. 6n23p-ev current is 9 mA while the 6n6p is running at 15 mA.
 
I finally did the CCS Grounded Cathode into Cathode Follower mod. I used a cascaded INX on the plate and just a single on the cathode. Basically Broskie's compound amplifier circuit. Sy recommends it in a few threads

The improvement on the measurements is amazing. THD+N went from 0.22% with the SRPP to 0.03%. Incredible for 40 year old tubes. All traces of power supply noise are gone, I can hear details in Joni Mitchell's voice I never knew were there.

Here is the 1khz FFT, 0dB's.


Dac-60afterCCS.jpg
 
There was another river death yesterday, this time a University of Oregon football player on the beautiful McKenzie.

On a happier subject, the Soviet Special sounds pretty good. My initial thoughts are that the 6n6p sounds bigger than the 5687. It is energetic with a silky midrange. There's a touch of coarseness and glassiness in the treble, although perhaps it is exposing something elsewhere in the system. Bass perhaps is not as nuanced as the 5687. Maybe. I'll need to go back and forth a few times to see the differences, sometimes they are not as big as one thought -- tricked!

But at the moment, I'm moderately surprised the second Aikido current stage tube would make such a difference in sound.

The treble edge was removed by replacing the first PSU cap with a paper-in-oil variety, and subbing in a Mullard GZ34 fat base. Call this configuration "C". This is the best the Trends TA10.1 has sounded in my system. The Trends amp can't believe he's actually dating a real Manhattan supermodel.
 
Here is a police lineup of the current suspects:

An externally hosted image should be here but it was not working when we last tested it.


That is a Tung-Sol "D" Getter 1950s production 5687 on the left. In the middle is a 6n6pi, which is a "pulse version" of the 6n6p. A regular 6n6p is on the right.

One audio company sells the 6n6pi as superior in it's higher spec gear. I couldn't detect a sound difference in the Aikido circuit, however. Note: K&K isn't the audio company.

The two 6n6ps shown look identical to me except for the height of the glass envelope. Same year: 1978. Both made in Novosibirsk factory,denoted by the pentagon symbol. If I am not mistaken, all 6n6ps were made at Novosibirsk.

It's interesting the plates and disc supports are almost the same size between the 5687 and 6n6p. But the space above the plate structure is quite differently sized. If anyone has any knowledge of what a larger outer glass envelope does in terms of pros and cons I'd like to hear about it. My wildass guess is that it allows for more getter flashing, which could extend the tube life.

The internal structure is much beefier in the 6n6p versus the 5687. There is a large middle screen between the triodes and the supports are made out of finally machined... springs?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.