A direct question!
See Fig. 1 in the paper. Merlin samples at the cathode.
The paper does not state if or how the EIN is corrected for the unbypassed variable cathode resistor.
DT
It is an inverted gain stage, not an unbypassed cathode resistor or cathode follower. It's gain is the same as any ordinary gain stage. It was just more convenient for me to arrange it that way so the load resistor and CCS could be ground-referenced along with the control circuitry.
Hi Everyone,
I updated the measurement list on my website to include additional samples of the JJ E88CC and Tung Sol Reissue 6SL7GT.
Downloads – Tavish Design
I modified the test fixtures for the E88CC / 6DJ8 / 6922, and also the D3a, in response to questions in this forum about possible “excess” (1/f) noise in the plate load resistors. I replaced the metal oxide power resistors (which have very low excess noise) with the same value 5W precision wirewound resistors, which have essentially no excess noise (most references have a noise index of 0 for wirewound resistors). I then re-measured the lowest noise tube samples with the new test fixtures. Within the repeatability of my measurements (±3%), there is no change. I included an updated photo and schematics of the revised test fixtures.
The latest samples of JJ E88CC are better than the previous ones, but I still discarded one of the 5 samples as high noise. The JJ E88CC is puzzling due to its high variability. Some samples show input-referred noise <0.55 uV rms, while 5 of the 13 had such high noise that they had to be discarded. Maybe it's just bad luck.
Scott
I updated the measurement list on my website to include additional samples of the JJ E88CC and Tung Sol Reissue 6SL7GT.
Downloads – Tavish Design
I modified the test fixtures for the E88CC / 6DJ8 / 6922, and also the D3a, in response to questions in this forum about possible “excess” (1/f) noise in the plate load resistors. I replaced the metal oxide power resistors (which have very low excess noise) with the same value 5W precision wirewound resistors, which have essentially no excess noise (most references have a noise index of 0 for wirewound resistors). I then re-measured the lowest noise tube samples with the new test fixtures. Within the repeatability of my measurements (±3%), there is no change. I included an updated photo and schematics of the revised test fixtures.
The latest samples of JJ E88CC are better than the previous ones, but I still discarded one of the 5 samples as high noise. The JJ E88CC is puzzling due to its high variability. Some samples show input-referred noise <0.55 uV rms, while 5 of the 13 had such high noise that they had to be discarded. Maybe it's just bad luck.
Scott
I suspect this is just the effect of flicker noise in tubes. It was common in ancient times to select tubes for low flicker noise. RCA tube mixers always used selected tubes in their first stages. I suspect your low noise samples exactly these.
Cheers
Ian
Yes, I suspect the same. But I am still puzzled by the difference in variability between, say, a JJ 5751 and JJ E88CC. The tubes must have similar construction and cathode materials. And the best samples of both tubes have input-referred noise ~0.55 uV rms. But the 5751 has sample standard deviation of 0.06 uV vs. 0.29 uV for the E88CC. I've measured enough samples that this doesn't seem to be random chance. Or is it??
Scott
it could quite easily be process variations. There is no guarantee the two tube types are made in the same factory let alone on the same production line. Not a lot seems to be known about the exact causes of flicker noise in tubes and before anyone got close to the answer, tubes went out of general use so there was no incentive to pursue it further.
Cheers
ian
is there any difference in the life use of the two types? If one set has been sitting on a shelf for donkey's years and the other is fresh outta the factory then you might observe differences due to outgassing. Just a thought.
I would expect a 5751 and E88CC to have somewhat different cathode surfaces. The latter needs a very smooth flat surface, as the frame grid is very close. The former does not - the cathode may be cylindrical and a bit more variable. Since they are made (or, at least, finished) on different lines I would expect some difference in noise statistics.
Wild guess: the E88CC cathode has to be machined flat, and this leaves some surface damage which creates excess noise?
Wild guess: the E88CC cathode has to be machined flat, and this leaves some surface damage which creates excess noise?
Oh I didn't realise they're completely different valves! I assumed Tavish was comparing equivalent types. Makes sense now; standard and frame grid constructions are quite different.
The 5751 is a 12AX7 variant, I didn't make that clear.
The consensus that I'm reading here is that 12AX7 types could have different noise statistics than E88CC / 6922 types because of construction differences, possible cathode-to-grid spacing, which seems reasonable.
Looking at my table, the 12AX7 types seem to have lower variability than other types I've measured, as well as lower average noise than other currently-produced tubes. At least so far. I could get a noisy batch of some 12AX7 type which could change the statistics.
Scott
The consensus that I'm reading here is that 12AX7 types could have different noise statistics than E88CC / 6922 types because of construction differences, possible cathode-to-grid spacing, which seems reasonable.
Looking at my table, the 12AX7 types seem to have lower variability than other types I've measured, as well as lower average noise than other currently-produced tubes. At least so far. I could get a noisy batch of some 12AX7 type which could change the statistics.
Scott
It is likely that almost all differences in excess noise arise from cathode details: mechanical, chemical etc.
What current are the 12AX7 types running at when you measure their noise?
Cheers
ian
It’s all about the Bass
Hello All,
You know the pop song ‘It’s all about the Bass’; in this case it is all about the cathode and some other things.
In the 1920’s a couple of grey beards by the names of Johnson and Schottky, a guy by the name of Fry was in the fray as well. In 1925 Johnson wrote a paper ‘The Schottky Effect in Low Frequency Circuits’ discussing Schottky’s earlier work. Johnson noted that the effect was critically dependent on the temperature of a tungsten cathode and he also added that the effect appears to be due to transient and local changes to the cathode surface.
One year later, 1926, Schottky published a discussion of Johnson’s paper. Schottky said that the low frequency noise was independent of shot effect and likely caused by fluctuations in the properties in the cathode surface. Schottky called this ‘flicker effect’.
Subsequent, mid to late 1930’s, to all this discussion and naming of the “flicker effect” researchers at University of Oxford experimentally found that the flicker effect due to tungsten cathodes has a corner frequency octaves lower than coated filaments.
Point is, first it is all about the cathode.
Once the cathode is accounted for, (experimentally controlled) we can look at other things;
1. Current
2. Differential voltage between the cathode and anode. Remember, no voltage no current.
3. Other structures between the cathode and anode. The geometry may make a difference.
4. The voltages applied to the structures.
5. The space charge, the relative velocities of the electrons and the chaotic, convoluted, paths the electrons traverse between the cathode and anode.
6. Add your own thoughts / variables here.
All these variables comingle, one affects the others. Experimentally if a single variable is held constant we do not know what modulating values of this variable will have on the noise performance of the valve.
DT
https://www.youtube.com/watch?v=7PCkvCPvDXk
Hello All,
You know the pop song ‘It’s all about the Bass’; in this case it is all about the cathode and some other things.
In the 1920’s a couple of grey beards by the names of Johnson and Schottky, a guy by the name of Fry was in the fray as well. In 1925 Johnson wrote a paper ‘The Schottky Effect in Low Frequency Circuits’ discussing Schottky’s earlier work. Johnson noted that the effect was critically dependent on the temperature of a tungsten cathode and he also added that the effect appears to be due to transient and local changes to the cathode surface.
One year later, 1926, Schottky published a discussion of Johnson’s paper. Schottky said that the low frequency noise was independent of shot effect and likely caused by fluctuations in the properties in the cathode surface. Schottky called this ‘flicker effect’.
Subsequent, mid to late 1930’s, to all this discussion and naming of the “flicker effect” researchers at University of Oxford experimentally found that the flicker effect due to tungsten cathodes has a corner frequency octaves lower than coated filaments.
Point is, first it is all about the cathode.
Once the cathode is accounted for, (experimentally controlled) we can look at other things;
1. Current
2. Differential voltage between the cathode and anode. Remember, no voltage no current.
3. Other structures between the cathode and anode. The geometry may make a difference.
4. The voltages applied to the structures.
5. The space charge, the relative velocities of the electrons and the chaotic, convoluted, paths the electrons traverse between the cathode and anode.
6. Add your own thoughts / variables here.
All these variables comingle, one affects the others. Experimentally if a single variable is held constant we do not know what modulating values of this variable will have on the noise performance of the valve.
DT
https://www.youtube.com/watch?v=7PCkvCPvDXk
Last edited:
MarcelvdG
Thank you for posting the name of the text ‘Noise’ by van der Ziel, 1954, it was delivered to my home this past week. As you suggested I have been perusing chapter 8.
Interesting, the author says concerning flicker effect there are multiple models operating all at the same time. He says something like (my words) flicker sort of follows 1/f some of the time. He also says that flicker effect is a function of current for some tubes and for others flicker effect is independent of current.
One example tube was the 6SJ7. The author tested a sample (the number of tubes was not stated) of factory fresh tubes and a sample of old tubes. At high current the old tubes had noise 3 orders of magnitude greater than the factory fresh tubes. The author reported that the old tubes had a layer buildup of barium orththosilicate at the cathode. The author did not state the age of the old tubes in donkey’s years. He did report that the old 6SJ7’s precisely followed the I^2 rule, the new tubes did not.
The author reported that 6AG5, 6CB6 and EF40 are low flicker effect tubes and should be preferentially selected. Something else he said is that tube noise varies widely among tube types and between tubes of the same type he added that tubes should be “selected”. My guess at what “selected” means is; tested and the good ones selected.
Fun stuff!
DT
Hi Everyone,
I updated the measurement list on my website to include additional samples of the JJ 5751, and also an additional tube type, the JJ ECC99.
Downloads – Tavish Design
The JJ 5751 continues to be one of the most consistent and low-noise tubes in my study. But I’ve noticed that as more tubes are measured, the average input-referred noise tends to rise. This is true for all tube types, and it is because as more tubes are measured, it is more likely to get a high noise tube that raises the average. The noise distribution is actually not Gaussian – while there seems to be a lower limit for the noise of a given type, there is no upper limit, so there is a tighter distribution on the low side of the average and a wider distribution on the high side of the average. In the future, I may begin reporting the median as well as the average for each type.
The ECC99 is a high transconductance tube that I believe was originally intended for RF applications. I’ve experimented with it as a driver tube for my hybrid amplifier, but had not considered it as a low noise tube until Lewm on the Audiogon forum suggested it to me. The ECC99 has transconductance comparable to an ECC88, but without the extremely tight electrode spacing or wide bandwidth.
I’ve only measured 4 samples of the ECC99 so far, but it seems promising. The average input-referred noise of 0.711 µV RMS puts it in the middle of my list, and I intend to measure more. One of the 4 samples had “popcorn” noise and was discarded.
Scott
I updated the measurement list on my website to include additional samples of the JJ 5751, and also an additional tube type, the JJ ECC99.
Downloads – Tavish Design
The JJ 5751 continues to be one of the most consistent and low-noise tubes in my study. But I’ve noticed that as more tubes are measured, the average input-referred noise tends to rise. This is true for all tube types, and it is because as more tubes are measured, it is more likely to get a high noise tube that raises the average. The noise distribution is actually not Gaussian – while there seems to be a lower limit for the noise of a given type, there is no upper limit, so there is a tighter distribution on the low side of the average and a wider distribution on the high side of the average. In the future, I may begin reporting the median as well as the average for each type.
The ECC99 is a high transconductance tube that I believe was originally intended for RF applications. I’ve experimented with it as a driver tube for my hybrid amplifier, but had not considered it as a low noise tube until Lewm on the Audiogon forum suggested it to me. The ECC99 has transconductance comparable to an ECC88, but without the extremely tight electrode spacing or wide bandwidth.
I’ve only measured 4 samples of the ECC99 so far, but it seems promising. The average input-referred noise of 0.711 µV RMS puts it in the middle of my list, and I intend to measure more. One of the 4 samples had “popcorn” noise and was discarded.
Scott
As the ECC99 was only recently introduced by JJ it is unlikely to be intended for RF - unless it is based on an existing design with a different (naming-convention-compliant) name. Modern valve datasheets can be works of imagination, or mere copies of earlier datasheets for a vaguely similar valve.TavishDad said:
