Measuring millett?
Aug 10, 2007 at 9:24 AM Thread Starter Post #1 of 12

Twisted85

New Head-Fier
Joined
Jan 28, 2005
Posts
38
Likes
0
Hi,

I've finished my Millett project some time ago and I've been wondering if it's possible to make some measurements of the my Millett performance with my computer. (frequency response, distortion etc)
I've used RMAA to test my M-Audio FireWire 410 soundcard, is it possible to make same type of loop through the Millett? Or should I use totally different program? How can I wire the Millett output to the soundcard inputs? Do I have to use some type of resistors to achieve the correct level or anything like that?

If someone could give me few tips how to get started I'd appreciate it
orphsmile.gif
 
Aug 10, 2007 at 4:04 PM Post #3 of 12
Thanks for your reply

Quote:

Originally Posted by MisterX /img/forum/go_quote.gif
A switchable dummy load is preferable to testing with headphones as a load.
(it's rather pointless to test a headphone amp without a load unless you plan on using it as a preamp)



That is something I had in mind
smily_headphones1.gif


Quote:

Something like this should get you going in the right direction---->
http://www.head-fi.org/forums/showpo...75&postcount=4

IIRC tangent has a tutorial somewhere on his site.


So I just put 30 or 300ohm resistors for both left and right plus wires in series and then wire it to the soundcard inputs?
Can I use same resistors that I use in loudspeaker crossovers if I measure that they are both exactly the same values?
 
Aug 10, 2007 at 7:38 PM Post #4 of 12
Quote:

So I just put 30 or 300ohm resistors for both left and right plus wires in series and then wire it to the soundcard inputs?


Parallel.
wink.gif

The resistors go between the left and right outputs and ground.


Quote:

Can I use same resistors that I use in loudspeaker crossovers if I measure that they are both exactly the same values?



Yes, non inductive resistors are perfect.
 
Aug 11, 2007 at 12:59 AM Post #5 of 12
Regarding the ground lift switch referenced in MisterX's link, when making RMAA measurements it should always be left in the open position even on non-active ground amps. This prevents the output ground return current from finding its way back to the sound card's signal ground, and affecting the accuracy of the measurement.

You should run the tests with the amp under test's volume control turned all the way up. Use the software volume control for your sound card to adjust the RMAA test levels. This will make your results comparable to those that I publish. Running the test with the amp's volume less than fully-open will artificially "improve" the noise and distortion results, at the same time it would make the stereo crosstalk result (in the high frequencies) worse.

Use short (< 1 foot), shielded interconnect cables between the sound card and the amp under test, turn off all sources of potential interference (fluorescent lights, light dimmers, TV, CRT monitor, other equipment with transformers, anything with a motor in it, etc.).
 
Aug 11, 2007 at 8:36 AM Post #6 of 12
Quote:

Originally Posted by MisterX /img/forum/go_quote.gif
Parallel.
wink.gif



Of course
icon10.gif


I tried to use 32ohm resistors in parallel with the output cables with and without the ground cable but I faced exact same problems than I tried before without resistors.

I can't get the levels correct without clipping lights in my M-Audio to light up. When I try to turn up either the Millett volume or M-Audio Gain, and I get to -50dB or above that the clipping lights starts to blink.
And for some reason the left channel level is about 10dB lower than the right channel (same Gain setting) with and without the resistors. (this doesn't occur when listening with headphones)
 
Aug 11, 2007 at 8:44 AM Post #7 of 12
You need to turn down the Output -> out and Mixer-> SW rtn slider settings (assuming that your software control panel is similar to that of the M-Audio Firewire Audiophile) to reduce the amplitude of the signal being sent to the amp's input in order to prevent clipping.
 
Aug 11, 2007 at 10:53 AM Post #8 of 12
Ok, for some reason, I had to invert the millett-soundcard input cables to get proper results. Now I can get the levels to right level with Millett volume all the up.

Here is what I got:

Witout any resistors:

http://koti.mbnet.fi/twisted-/Millett.htm

With 32ohm resistors:

http://koti.mbnet.fi/twisted-/Millett_%2032ohm.htm

Does this look like correct?
In the results it says "Polarity inverted/inverted" (?)

For some reason there is a 1,5dB difference between channels.
 
Aug 11, 2007 at 11:27 AM Post #9 of 12
Quote:

I had to invert the millett-soundcard input cables to get proper results


You had to swap the channels?
(which means they are crosswired?)


Quote:

For some reason there is a 1,5dB difference between channels.


Do you get the same thing when test just the card with and without your dummy load in the chain?

(it looks like poorly matched toobs but without seeing your "baseline results" that is just speculation)
 
Aug 11, 2007 at 11:41 AM Post #10 of 12
Quote:

Originally Posted by Twisted85 /img/forum/go_quote.gif
Ok, for some reason, I had to invert the millett-soundcard input cables to get proper results. Now I can get the levels to right level with Millett volume all the up.


Either your interconnect cables are crossed, or the amp's internal input or output wiring.

Quote:

Here is what I got:
Witout any resistors:
http://koti.mbnet.fi/twisted-/Millett.htm
With 32ohm resistors:
http://koti.mbnet.fi/twisted-/Millett_%2032ohm.htm
Does this look like correct?


You can compare your results against my Millett Hybrid:
http://www.amb.org/ti/audio/millett.html

Quote:

In the results it says "Polarity inverted/inverted" (?)


The Millett Hybrid inverts absolute phase.

Quote:

For some reason there is a 1,5dB difference between channels.


If your amp volume control is set to max, and the software controls are locked between the left and right channels, then the inbalance is most likely due to mismatched tubes. This is not unusual if you just used two artitrary tubes. The mu (gain) tolerance could be quite large from unit to unit.
 
Aug 11, 2007 at 11:47 AM Post #11 of 12
Quote:

Originally Posted by MisterX /img/forum/go_quote.gif
You had to swap the channels?
(which means they are crosswired?)



Yes. Without swapping I can't get the levels correct and inter-channel leakage is red. (only on other channel if I remember correctly).
But the Millett right/left output is the same as with my speakers. (so the channels doesn't swap inside the Millett)


Quote:

Do you get the same thing when test just the card with and without your dummy load in the chain?

(it looks like poorly matched toobs but without seeing your "baseline results" that is just speculation)


Yes I do. I can't remember where did I order my tubes (because it was more than a year ago), but there are markings made by hand in the tube packets. At the moment 12ae6a tubes are connected and the other one is labeled "91" and the other "90". I have also 12fk6 tubes, and both of these are labeled "105".

Here are results without the millett, only the M-Audio:

http://koti.mbnet.fi/twisted-/M-Audio.htm
 
Aug 11, 2007 at 12:21 PM Post #12 of 12
I tried to swap the tubes (left to right and right to left). After swapping the volume difference between channels is also swapped, but the difference is smaller (~0,8dB). But then there was 2,5V difference in bias, and after correcting both bias to 12V, the difference is again 1,3dB.

I haven't tried the 12fk6 tubes.
 

Users who are viewing this thread

Back
Top