Your thd went down for the higher bias, however what I mean is that the % of average is not useful, you have to use single tones, or place the sweeps in the same graph and two harmonics at a same time , it is all bunched up and confusing
I can do the open loop measurements tomorrow morning - I'm not sure if they will help in understanding why certain HD response is preferred but I realize it may be useful information. I previously have done both open and closed loop phase and gain testing to 100's of kHz, along with 10kHz squarewave to tune the gNFB compensation. When I built the amplifier several years ago it had ~4dB gNFB and at some point I started chasing measurements and reworked the input for more gain and gNFB. I've found 15dB seems about right, I initially was targeting 20dB to mimic the designs of the 50's-60's but I think some of that was specmanship.
I run my current 'bad' tube amp at 9 db feedback plus extra 2.5db Cathode feedback.
Ill give you some figures but my initial impression is that your THD is high.
Ill give you some figures but my initial impression is that your THD is high.
Here are the open loop measurements. Same setup, 5W into 8ohm load. The legend is the % at 1kHz, I added the cursor for clarity.
Idle bias 20W (42.5mA) for the 6CA7 and KT77
Idle bias 25W (55mA) for the 6550, KT88 and KT90
I encourage others to post HD responses of setups (amp/output tubes) that they prefer. It would be interesting to see if there is some correlation between HD response and preference.
Idle bias 20W (42.5mA) for the 6CA7 and KT77
Idle bias 25W (55mA) for the 6550, KT88 and KT90
I encourage others to post HD responses of setups (amp/output tubes) that they prefer. It would be interesting to see if there is some correlation between HD response and preference.
Last edited:
Does the input signal pass through a preamp or master volume control before the driver? Did you ground the FB line or let it float?
It is obvious that the Gold lion kt88 are the best, this is no surprise.
Do you realize that the GL 88 performs better without the feedback... this is not right but I would use that without feedback if your output Z is ok with your loudspeakers!
imo, kt77 horrible tube, I use tung sol 6550 which is kt88
Do you realize that the GL 88 performs better without the feedback... this is not right but I would use that without feedback if your output Z is ok with your loudspeakers!
imo, kt77 horrible tube, I use tung sol 6550 which is kt88
Now install a variable resistor for feedback and watch what happens to harmonics.
When I tube rolled my amps it was obvious which ones sound best.
The only one which sounded different was the kt77, I was hearing weird stuff, it had a edgy sound , and what it was was THD, it is a very bad tube for audio, I would never understand why using this and not EL34 or 6CAg
When I tube rolled my amps it was obvious which ones sound best.
The only one which sounded different was the kt77, I was hearing weird stuff, it had a edgy sound , and what it was was THD, it is a very bad tube for audio, I would never understand why using this and not EL34 or 6CAg
Perhaps the first set of benchmark tests should be using at least 5 sets of the same valve, in order to get some appreciation of the spread of performance within one model. I can't quickly find my records, but I recall obtaining THD levels from 0.1 to 0.7% by tube rotating from what I had on hand. Just one spot measurement for a tube, and then comparing that to another spot measurement of another tube is I suggest prone to gross misrepresentation.
The central issue here is the assertion that you could hear a difference in audio quality with tube swaps without having set your amp up scientifically between swaps to give yourself a realistic basis to judge any difference in sound characteristics between the tubes. The distortion measurements by themselves wouldn't lead anyone to believe that the differences are audible at all with the amp's normal GNFB engaged, which was the starting point. It's not so easy to set an amp up for quality comparisons with a music signal input when you have to pass the signal through a preamp or volume control to get a ''similar'' output level which could be very far off on a power level because of the poor sensitivity of our hearing to differences in power that can have real changes that are measurable with test equipment. The results of these tests are only a snapshot of your unique setup, amp, voltage conditions, test equipment and method. I don't believe it's even useful at this point for you to draw any conclusions for yourself going forward with just these graphs.
Last edited:
One of the very basic thinks is to ensure that both the freq response and the absolute level are precisely matched between tubes.
If that is not done, any result is worthless.
It's a well know trick by audio salespersons to play the system they want to sell a bit louder which makes it sound better.
Jan
If that is not done, any result is worthless.
It's a well know trick by audio salespersons to play the system they want to sell a bit louder which makes it sound better.
Jan
The first failure is in not having a control group established for the tubes in the testing. New and just burned in? Old? Matched? Testing one tube at a time under SE conditions is one way but when you put them into a complex PP output what are you able to say about a particular ''type'' compared to another particular ''type''... other than this is what my amp showed for these tubes? And then, you have to interpret what you see in the graphs and say, ''Oh, that's why, the difference of .02% in the 2nd H makes it sound way fuller and with such an open soundstage and such lush mids or such harsh highs.... really?
Last edited: