> They don´t tell me the tolerance about thier selection.
Then you have NO specification and no Warranty about what you will get. I don't see why they would take any of these tubes back. If they were *dead*, or maybe half the current of others in the carton, they might take them back.
If you need Specific Specs, and the seller is not bragging on their specs, YOU need to write that into the contract. This is often done on big jobs. I have at hand a spec for buying sacks of concrete with sample tests of weight and full-lot tests if any sample looks 5% shy of nominal.
I believe the Military and Fender Amps were getting custom-spec tubes. Of course this costs more. For a run of 100, probably far more than you would want to pay.
I have always worked to an assumption that the amplifier should "work" with tubes varying +/-20% even +/-30%. In self-bias this may be +/-15% different gain. In mono this was acceptable (you have a knob for that). In stereo it could be a bit much unbalance so stereo rigs often had a bit of NFB even in minor stages. In cross-country telephone lines with hundreds of repeaters cascaded and a +/-4dB total gain tolerance, every repeater had 100X the required gain midband and threw that away with two 1% parts. A single tube could age to less than 10% _-90%) nominal gain and you'd never know it. (There were internal check-points to find and fix such problems before gain dropped enough to notice.)
Raining, raining ----
Mu as percent of nominal "100"
Mu 94%-104% = +/-5.02%
Ia has interestingly large spread. Self-bias would reduce this.
Ia 53-173 (62-140 over 98 units)
Normalize Gm for Ia (assume you force specific current NOT specific voltage):
Gm/sqrt(Ia) 1.40-1.52 +/-4%
You essentially have two hundred 5% parts. 180 2% parts. I call that pretty darn good.
What I do not see is NOISE measurements. Build a zero-feedback stage, shield well, short the grid, listen to the output. An averaging meter would put numbers on that. But what you probably find is 90%-98% all very nearly equal hiss, and few-percent with excess hiss (dirty cathode, grid, glass, gas). Even the mild hissers can be used in high-level stages. Heavy hiss suggests real contamination and possible short life. As an added test, ground grid with extra-shielded 100K film resistor to check for grid current hiss.
I also don't see grid leakage numbers. The Mu tester probably has low grid resistance. 1uA in typical amp 1Meg grid resistor is 1V of unexpected grid bias, which on a 12AX7 is a LOT. A part-volt in 1Meg may be normal because there are several sources of grid current.
Then you have NO specification and no Warranty about what you will get. I don't see why they would take any of these tubes back. If they were *dead*, or maybe half the current of others in the carton, they might take them back.
If you need Specific Specs, and the seller is not bragging on their specs, YOU need to write that into the contract. This is often done on big jobs. I have at hand a spec for buying sacks of concrete with sample tests of weight and full-lot tests if any sample looks 5% shy of nominal.
I believe the Military and Fender Amps were getting custom-spec tubes. Of course this costs more. For a run of 100, probably far more than you would want to pay.
I have always worked to an assumption that the amplifier should "work" with tubes varying +/-20% even +/-30%. In self-bias this may be +/-15% different gain. In mono this was acceptable (you have a knob for that). In stereo it could be a bit much unbalance so stereo rigs often had a bit of NFB even in minor stages. In cross-country telephone lines with hundreds of repeaters cascaded and a +/-4dB total gain tolerance, every repeater had 100X the required gain midband and threw that away with two 1% parts. A single tube could age to less than 10% _-90%) nominal gain and you'd never know it. (There were internal check-points to find and fix such problems before gain dropped enough to notice.)
Raining, raining ----
Mu as percent of nominal "100"
Mu 94%-104% = +/-5.02%
Ia has interestingly large spread. Self-bias would reduce this.
Ia 53-173 (62-140 over 98 units)
Normalize Gm for Ia (assume you force specific current NOT specific voltage):
Gm/sqrt(Ia) 1.40-1.52 +/-4%
You essentially have two hundred 5% parts. 180 2% parts. I call that pretty darn good.
What I do not see is NOISE measurements. Build a zero-feedback stage, shield well, short the grid, listen to the output. An averaging meter would put numbers on that. But what you probably find is 90%-98% all very nearly equal hiss, and few-percent with excess hiss (dirty cathode, grid, glass, gas). Even the mild hissers can be used in high-level stages. Heavy hiss suggests real contamination and possible short life. As an added test, ground grid with extra-shielded 100K film resistor to check for grid current hiss.
I also don't see grid leakage numbers. The Mu tester probably has low grid resistance. 1uA in typical amp 1Meg grid resistor is 1V of unexpected grid bias, which on a 12AX7 is a LOT. A part-volt in 1Meg may be normal because there are several sources of grid current.
Hi Karsten,
I did some calculations on your data with minitab. So be aware before doing any statistics, your data ( for any measured parameter) shows no normal distribution.
This may have many different reasons. First of all you have to do a Measurement System Analyses to be sure the problem is not due to your measurement tool.
Separate the measurement of triode 1 and triode 2 of the tube..
If you have done this without getting a normal distribution it is due to manufacturing process ( ie two or more different punching tools) or may be JJ does a pre selection.
I did some calculations on your data with minitab. So be aware before doing any statistics, your data ( for any measured parameter) shows no normal distribution.
This may have many different reasons. First of all you have to do a Measurement System Analyses to be sure the problem is not due to your measurement tool.
Separate the measurement of triode 1 and triode 2 of the tube..
If you have done this without getting a normal distribution it is due to manufacturing process ( ie two or more different punching tools) or may be JJ does a pre selection.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.