I'm a bit of an idiot with biasing, so please be kind. My amp has a very simple biasing system and excellent instructions available on line, so I did it myself.
I have a Audio Research Dual 75 stereo amp that came to me with three of the four original GE 6550 tubes. The fourth had exploded due to a power supply meltdown, but that's another story. Anyway, the old tubes are in fine condition and I didn't want to replace them for financial and audio quality reasons.
I've managed to find another GE 6550, but it's not matched to the other in its channel. I have to split the difference on biasing and come as close as I could. What am I risking here? It seems to run fine, the channels are balanced as far as volume goes. I understand that the audio performance may be compromised, but is there a chance of a tube failure?
I have a Audio Research Dual 75 stereo amp that came to me with three of the four original GE 6550 tubes. The fourth had exploded due to a power supply meltdown, but that's another story. Anyway, the old tubes are in fine condition and I didn't want to replace them for financial and audio quality reasons.
I've managed to find another GE 6550, but it's not matched to the other in its channel. I have to split the difference on biasing and come as close as I could. What am I risking here? It seems to run fine, the channels are balanced as far as volume goes. I understand that the audio performance may be compromised, but is there a chance of a tube failure?
How did you measure the biasing?
DId you measure the plate current and adjust the bias to set the complimentary tubes near equal current with no input?
Very low probability of tube failure due to minor miss-match in tube current.
From the experiments I've done, if they are within 5% it's "good enough for jaz". Worst case I've seen with bias off over 10% was increased second order harmonics and uneven clipping.
DId you measure the plate current and adjust the bias to set the complimentary tubes near equal current with no input?
Very low probability of tube failure due to minor miss-match in tube current.
From the experiments I've done, if they are within 5% it's "good enough for jaz". Worst case I've seen with bias off over 10% was increased second order harmonics and uneven clipping.
Quiescent current checks are fine for low level class A (in the AB part) but as the amp handles highish signal levels into the B part, then the transconductance, the AC signal gain of each tube will show current discrepancies and none are the same. One can measure cathode current on near full sine signal to get some idea how far out they are. If you find the full signal condition is way out compared to the others, despite the quiescents are set the same, there's not anything one can do except find another tube.
In reality, the odd tube could have a higher transconductance and be a signal current hog; or the opposite, be a slouch, in which case the other becomes the hog. The result of a serious mismatching is power output problems and increased distortion and one is quite amazed how the global nfb loop can clean up alot of problems. If you find another GE6550 version, then things may be easier finding a match.
richy
In reality, the odd tube could have a higher transconductance and be a signal current hog; or the opposite, be a slouch, in which case the other becomes the hog. The result of a serious mismatching is power output problems and increased distortion and one is quite amazed how the global nfb loop can clean up alot of problems. If you find another GE6550 version, then things may be easier finding a match.
richy
Here's where my inexperience hits
Thanks for the replies gentlemen. As far as how I biased, I followed the "underequipped" option on the schematic:
"5. If without distortion measuring equipment, set all 4 output tubes to 50 MA each after 30 minute warmup"
Each pair of tubes has only one adjustment pot and since I couldn't get both tubes to 50 MA, I had to set it so one was over and one was under. The fourth tube is currently an old Mesa Boogie -- if I'm not mistaken they were rebranding NOS US 6550s at the time.
I'm asking now because I just bought a GE 6550 that looks identical to the ones already installed. I wanted to make sure I wasn't going to blow anything up. With any luck the new tube will match the old tube better than the Mesa tube.
Thanks for the replies gentlemen. As far as how I biased, I followed the "underequipped" option on the schematic:
"5. If without distortion measuring equipment, set all 4 output tubes to 50 MA each after 30 minute warmup"
Each pair of tubes has only one adjustment pot and since I couldn't get both tubes to 50 MA, I had to set it so one was over and one was under. The fourth tube is currently an old Mesa Boogie -- if I'm not mistaken they were rebranding NOS US 6550s at the time.
I'm asking now because I just bought a GE 6550 that looks identical to the ones already installed. I wanted to make sure I wasn't going to blow anything up. With any luck the new tube will match the old tube better than the Mesa tube.
Each pair of tubes has only one adjustment pot and since I couldn't get both tubes to 50 MA, I had to set it so one was over and one was under. The fourth tube is currently an old Mesa Boogie -- if I'm not mistaken they were rebranding NOS US 6550s at the time.
[/QUOTE]
The single setup preset/pot was common in the early days and obviously the matched pairs worked fine. As Gimp mentioned, one can get away with quite a mismatch,esp with E&I cores, but not with toroids. In fixed bias, the reality to acheive perfect anode currents is impossible as miniscule variations in heater voltage, B+ and temperature will effect emission drift even after hours of use. However, there are many audiophiles who are obsessed about stabilised DC heater supplies in power stages which is unnecessary unless the AC line has wide variations. Front ends fine.
Generally, nowadays I design/modify tube amps with separate bias controls for each tube and some kind of interlock/watchdog to protect the o/p stage if the neg bias fails. This does offer a bit more flexibility. However, Ye bigger the amp, the more of everything is required, esp as I live in a wooden house.
Richy
[/QUOTE]
The single setup preset/pot was common in the early days and obviously the matched pairs worked fine. As Gimp mentioned, one can get away with quite a mismatch,esp with E&I cores, but not with toroids. In fixed bias, the reality to acheive perfect anode currents is impossible as miniscule variations in heater voltage, B+ and temperature will effect emission drift even after hours of use. However, there are many audiophiles who are obsessed about stabilised DC heater supplies in power stages which is unnecessary unless the AC line has wide variations. Front ends fine.
Generally, nowadays I design/modify tube amps with separate bias controls for each tube and some kind of interlock/watchdog to protect the o/p stage if the neg bias fails. This does offer a bit more flexibility. However, Ye bigger the amp, the more of everything is required, esp as I live in a wooden house.
Richy
I was looking for a place to put this and this is the first thread I found about biasing, so I apologize if this is a bit of a hijack.
I just finished measuring 26 6P1P-EVs with various manufacture date codes to try to figure out how to match them for 6P1P PP amps (Like the Meng Yue Mini, Class A). I tested the tubes in a push pull amplifier with fixed bias (510Ohm bypassed with a 330uF cap). I measured the bias voltage (Cathode to ground), anode current, and ac signal gain across a 1 ohm resistor in the anode circuit, with a 10Vrms input to the tube grid. The Bias voltage and anode current were measured without an input signal as that shifted the bias.
Tube Vgk Ia Va (Gm*10)
1 16.91 31.00 28.70
2 16.67 30.80 30.00
3 16.27 29.10 30.20
4 16.67 29.10 28.60
5 15.65 29.10 29.00
6 16.59 31.20 29.10
7 16.15 29.70 29.30
8 15.63 29.00 27.90
9 15.43 28.60 29.00
10 15.98 29.80 27.90
11 16.80 31.40 27.80
12 17.80 27.50 26.20
13 16.67 25.90 26.50
14 15.33 28.50 29.00
15 15.61 29.00 29.40
16 16.48 25.80 24.50
17 15.87 29.40 29.30
18 16.79 26.40 26.20
19 15.65 28.30 27.60
20 17.34 26.60 27.00
21 16.70 30.80 30.70
22 17.15 27.00 27.50
23 16.16 29.80 28.40
24 18.75 29.00 25.50
25 17.63 32.30 27.90
26 16.74 26.10 26.40
What I found by doing various sorts was that there is no correlation between Ia and Gm. That is to say that a tube that biases up with a higher anode current does not correlate either to higher OR lower Gm. It is random.
For Class A this won't matter. Match the tubes for Ia to minimize shifting the hysteresis curve of the output transformer. This will maximize output and help minimize distortion, but it won't be a major factor. The difference in Gm of the two output tubes will result in one tube driving harder than the other, but it will do so on both positive and negative swings so there will be a minimum of additional odd and even harmonics.
For Class AB1, this will matter. If both tubes are biased the same, but one tube has a higher Gm, it will amplify more. For low level signals it will behave like class A and have a minimal effect, but when it moves into class B operation the tube with the higher Gm will result in that phase having a greater amplification than the opposite phase when the weaker tube is driving. This will cause an offset in the saturation of the transformer, shift the bias point and result in greater harmonic distortion due to unequal phases.
This might be desirable to some degree in a guitar amp, but not in an audiophile amp.
I just finished measuring 26 6P1P-EVs with various manufacture date codes to try to figure out how to match them for 6P1P PP amps (Like the Meng Yue Mini, Class A). I tested the tubes in a push pull amplifier with fixed bias (510Ohm bypassed with a 330uF cap). I measured the bias voltage (Cathode to ground), anode current, and ac signal gain across a 1 ohm resistor in the anode circuit, with a 10Vrms input to the tube grid. The Bias voltage and anode current were measured without an input signal as that shifted the bias.
Tube Vgk Ia Va (Gm*10)
1 16.91 31.00 28.70
2 16.67 30.80 30.00
3 16.27 29.10 30.20
4 16.67 29.10 28.60
5 15.65 29.10 29.00
6 16.59 31.20 29.10
7 16.15 29.70 29.30
8 15.63 29.00 27.90
9 15.43 28.60 29.00
10 15.98 29.80 27.90
11 16.80 31.40 27.80
12 17.80 27.50 26.20
13 16.67 25.90 26.50
14 15.33 28.50 29.00
15 15.61 29.00 29.40
16 16.48 25.80 24.50
17 15.87 29.40 29.30
18 16.79 26.40 26.20
19 15.65 28.30 27.60
20 17.34 26.60 27.00
21 16.70 30.80 30.70
22 17.15 27.00 27.50
23 16.16 29.80 28.40
24 18.75 29.00 25.50
25 17.63 32.30 27.90
26 16.74 26.10 26.40
What I found by doing various sorts was that there is no correlation between Ia and Gm. That is to say that a tube that biases up with a higher anode current does not correlate either to higher OR lower Gm. It is random.
For Class A this won't matter. Match the tubes for Ia to minimize shifting the hysteresis curve of the output transformer. This will maximize output and help minimize distortion, but it won't be a major factor. The difference in Gm of the two output tubes will result in one tube driving harder than the other, but it will do so on both positive and negative swings so there will be a minimum of additional odd and even harmonics.
For Class AB1, this will matter. If both tubes are biased the same, but one tube has a higher Gm, it will amplify more. For low level signals it will behave like class A and have a minimal effect, but when it moves into class B operation the tube with the higher Gm will result in that phase having a greater amplification than the opposite phase when the weaker tube is driving. This will cause an offset in the saturation of the transformer, shift the bias point and result in greater harmonic distortion due to unequal phases.
This might be desirable to some degree in a guitar amp, but not in an audiophile amp.
Last edited:
Using unmatched tubes you loose inaudible fraction of loudness. Higher THD does not matter since you don't need to cancel even order harmonics in favor of odd order harmonics. Also, I don't think that somebody who is sane would ever use class AB amp without NFB that linearizes it.
Quite right .
Another "Audiophiles" myth bites the dust !
(I have try over the years a lot of testing in PP with gross different tubes and Wavebourn , I say again , you are right )
Global nfb.. The fantastic reduction of output tranny secondary impedance, a bonus not to be missed !
And 20db is the magic number?
I've seen a graph is a couple of articles that plotted haromonic distortion vs feedback(in db). Second order harmonics decreased with GNFB, but higher harmonics actually increased up to around 12-15db nfb, then started decreasing and reached the initial level around 20db.
So, less than 20db would decrease second order hamoonics at the possible expense of higher order harmonics.
I've seen a graph is a couple of articles that plotted haromonic distortion vs feedback(in db). Second order harmonics decreased with GNFB, but higher harmonics actually increased up to around 12-15db nfb, then started decreasing and reached the initial level around 20db.
So, less than 20db would decrease second order hamoonics at the possible expense of higher order harmonics.
Last edited:
Getting a bit late.
The Gimp. Those figures in the chart. I'm curious. They look too close. You obtained these in a p-p amp with global nfb ? Big difference. The feedback loop (if enough of it) will fight to correct tube tolerances on a cycle to cycle basis, in an attempt to correct the output tranny primary currents within the available loop gain and phase shift available. Try without global nfb.
Really somewhat akin to the error current ramp behaviour in a current mode SMPS.
Everyone agree to the analogy ?
richy
The Gimp. Those figures in the chart. I'm curious. They look too close. You obtained these in a p-p amp with global nfb ? Big difference. The feedback loop (if enough of it) will fight to correct tube tolerances on a cycle to cycle basis, in an attempt to correct the output tranny primary currents within the available loop gain and phase shift available. Try without global nfb.
Really somewhat akin to the error current ramp behaviour in a current mode SMPS.
Everyone agree to the analogy ?
richy
The numbers were without GNFB. I was trying to see how the tubes compared and did not have a SE amp to test in, so I used the PP amp. Two tubes at a time make for quicker testing as well.
I'm going to re-run the test with higher bias next time (35mA, 40mA, 45mA, 50mA) to see how it effects the tube operation. I've got another 50 tubes I can test if the sample size is too small, but it gets rather time consuming waiting for the dark-room timer to ding so I take measurements at the same time relative to power on.
I'm going to re-run the test with higher bias next time (35mA, 40mA, 45mA, 50mA) to see how it effects the tube operation. I've got another 50 tubes I can test if the sample size is too small, but it gets rather time consuming waiting for the dark-room timer to ding so I take measurements at the same time relative to power on.
- Status
- Not open for further replies.