SY,
The second half of my post wasn't directed at you, but more a general reply to the thread. You weren't trying to tell me what I could or could not hear, but several other people were making that claim. As far as I'm concerned, they can take a hike.
Regarding the main function of grid stoppers, I disagree. There are very few circuits that require these to stop oscillation. I did a quick web search of some of the classic amps, including Dyna PAS-3x, Dyna Stereo 70, Marantz 8, Marantz 9, and McIntosh MC-30.
The McIntosh didn't have any grid stopper resistors anywhere in the circuit
The Marantz and Dyna amps use 1k resistors at the grid of the power tubes. I don't think they are to prevent oscillation, but it's been so long since I read a tube manual that I can't say for sure.
There were only two instances of what we might now refer to as "grid stoppers". The first was at the phono input of the PAS-3x. The second was at the input of the Marantz 8 (but oddly enough, not the 9), where an R-C feedback network produces a -3 dB point of around 130 kHz. But these are not used a tools to "tune" the sound of the amplifier. Instead they are to prevent hearing your local radio station when you are playing LPs.
In contrast, knowledgeable modern designers use grid stoppers at each tube. These are chosen by ear to provide the best sound quality. If you have ever played around with these, you will know how significantly they can affect the sound quality. I would assume that they do so by altering the product's ultrasonic response, but this is different than just trying to avoid hearing the local radio station coming through your stereo when playing LPs.
The second half of my post wasn't directed at you, but more a general reply to the thread. You weren't trying to tell me what I could or could not hear, but several other people were making that claim. As far as I'm concerned, they can take a hike.
Regarding the main function of grid stoppers, I disagree. There are very few circuits that require these to stop oscillation. I did a quick web search of some of the classic amps, including Dyna PAS-3x, Dyna Stereo 70, Marantz 8, Marantz 9, and McIntosh MC-30.
The McIntosh didn't have any grid stopper resistors anywhere in the circuit
The Marantz and Dyna amps use 1k resistors at the grid of the power tubes. I don't think they are to prevent oscillation, but it's been so long since I read a tube manual that I can't say for sure.
There were only two instances of what we might now refer to as "grid stoppers". The first was at the phono input of the PAS-3x. The second was at the input of the Marantz 8 (but oddly enough, not the 9), where an R-C feedback network produces a -3 dB point of around 130 kHz. But these are not used a tools to "tune" the sound of the amplifier. Instead they are to prevent hearing your local radio station when you are playing LPs.
In contrast, knowledgeable modern designers use grid stoppers at each tube. These are chosen by ear to provide the best sound quality. If you have ever played around with these, you will know how significantly they can affect the sound quality. I would assume that they do so by altering the product's ultrasonic response, but this is different than just trying to avoid hearing the local radio station coming through your stereo when playing LPs.
Charles, I would disagree for two reasons:
1. My own experience. I've designed and built quite a few circuits with high gm triodes characteristic of modern designs (the ones you cite use stuff like 12AX7, 6CG7/6FQ7, and the like). You WILL see all sorts of squeggy oscillations in these circuits if grid stoppers are not used. For the case of cathode followers where high gm tubes are a must for performance, cathode stoppers are often needed, too. No doubt that people will tune for sound, but the best sound comes from stable circuits. For example, in my Heretical preamp, the grid stopper calms down the ECC88, but doesn't significantly affect the ultrasonic response- but for the input transformer, the circuit kernel has a bandwidth greater than a megahertz. Without it, I see bits of raggedness at roughly 50MHz.
2. Textbooks. Although I'm not one to blindly follow texts, I always find the good ones to be worth considering. The best modern text on audio design with tubes is (IMO) Morgan Jones's "Valve Amplifiers." He specifically points out that the combination of the inductance of grid wiring and the Miller capacitance causes high gm tubes to have a tendency toward oscillation. His advice is reminiscent of the stuff found on MOSFET data sheets- grid stopper as close as possible to the pin, sized experimentally to calm things down. He suggests that the body of the resistor be placed right at the grid pin. Sound familiar?
As a side comment, the grid resistors in power amp output stages have several functions. They do act as grid stoppers (many output tubes have high gm compared to 12AX7, 6SN7...). But if sized properly, they also act to limit grid current when the tube is overdriven.
edit: A reasonable question might be, "Since the Marantz 9 used 6DJ8s as input and driver tubes, how did they get away without grid stoppers?" The answer is that they ran the tubes at very low current where the transconductance was similarly low, and relied on circuit balance to reduce even-order HD.
1. My own experience. I've designed and built quite a few circuits with high gm triodes characteristic of modern designs (the ones you cite use stuff like 12AX7, 6CG7/6FQ7, and the like). You WILL see all sorts of squeggy oscillations in these circuits if grid stoppers are not used. For the case of cathode followers where high gm tubes are a must for performance, cathode stoppers are often needed, too. No doubt that people will tune for sound, but the best sound comes from stable circuits. For example, in my Heretical preamp, the grid stopper calms down the ECC88, but doesn't significantly affect the ultrasonic response- but for the input transformer, the circuit kernel has a bandwidth greater than a megahertz. Without it, I see bits of raggedness at roughly 50MHz.
2. Textbooks. Although I'm not one to blindly follow texts, I always find the good ones to be worth considering. The best modern text on audio design with tubes is (IMO) Morgan Jones's "Valve Amplifiers." He specifically points out that the combination of the inductance of grid wiring and the Miller capacitance causes high gm tubes to have a tendency toward oscillation. His advice is reminiscent of the stuff found on MOSFET data sheets- grid stopper as close as possible to the pin, sized experimentally to calm things down. He suggests that the body of the resistor be placed right at the grid pin. Sound familiar?
As a side comment, the grid resistors in power amp output stages have several functions. They do act as grid stoppers (many output tubes have high gm compared to 12AX7, 6SN7...). But if sized properly, they also act to limit grid current when the tube is overdriven.
edit: A reasonable question might be, "Since the Marantz 9 used 6DJ8s as input and driver tubes, how did they get away without grid stoppers?" The answer is that they ran the tubes at very low current where the transconductance was similarly low, and relied on circuit balance to reduce even-order HD.
PM,
You are making two incorrect assumptions. What makes you think the Ayre designs are more sensitive to these effects than other products? From there you further assume that I "don't care about preventing RF pickup". As our fearless idiot leader would say, "Nothing could be more further from the truth."
You are making two incorrect assumptions. What makes you think the Ayre designs are more sensitive to these effects than other products? From there you further assume that I "don't care about preventing RF pickup". As our fearless idiot leader would say, "Nothing could be more further from the truth."
Charles Hansen said:
(JP move this one along with the rest)
Allen Wright tells an interesting story about his 1st use of high GM tubes (ECC88). I know he has no hair now, but if he had hair before the ECC88 he didn't have after.
This was before ECC88 became really popular, he was designing based on his 12A*7 circuits. For 2 weeks he tried to sort the "hum". Then a O-scope probe touched to the grid set a light off. The whole story is in his pre-amp cookbook (besides Jones another highly recommended -- this book it more about attitude than circuits).
You tell Allen grid stoppers aren't important for RF and you'll be in for a couple hours of lively (and wonderful) discussion. (Note: all of Allen'd stuff have response at least 2 decades beyond the usual "audio spectrum)
dave
Hi all, what frequencies and magnitude of oscillation are you finding? I'll admit to much confusion regarding the necessity of gridstoppers since I've yet to see anything on a 100 MHz digital scope at 20 mv/div. My experience isn't broad but includes tubes with reputations such as the triode-connected E180f and EL84, as well as 6C45-PE. The latter is operating at 24 ma and 10 ma in various circuits without stoppers and with no visual indication of oscillation. Could construction methods account for some it? I stick with Pete Millet-style ground plane construction, shortest possible loop paths and the use of bypasses - especially on the cathode - aimed for low impedances to ground at RF frequencies. In the latest one, and CCS/LED-baised 6C45 cap coupled to a triode EL84, even unshielded Litz wire interconnects work passably. A hint of radio with an ear pressed to the speaker but no more.
I've seen anything from full blown oscillation to intermittent stuff. More often, I see a fuzziness riding with waveforms, especially square waves, which goes away when the resistor is added. Allen Wright's anecdote is not atypical. FWIW, I use an analog 100 MHz scope.
I've used a variety of construction methods in the past and haven't seen any good correlation. IIRC, pinkmouse did up some groundplane boards for my preamp design and had to do some reworking to get the grid stopper closer after seeing some oscillation. My own version was done point to point with a star ground, and no problem- but the grid stoppers were soldered directly to the tube socket pin.
I've used a variety of construction methods in the past and haven't seen any good correlation. IIRC, pinkmouse did up some groundplane boards for my preamp design and had to do some reworking to get the grid stopper closer after seeing some oscillation. My own version was done point to point with a star ground, and no problem- but the grid stoppers were soldered directly to the tube socket pin.
Thanks SY, that makes a bit more sense. I never considered checking with an excitation signal. This digital scope also has a residual noise level in the 25 mv p-p range. Fuzziness is its nature at full sensitivity. The few times a potential oscillation appeared the waveform remained with probes removed, shorted and held in mid-air. Sounds like a better scope is required. Fortunately this one's on loan from work. Oh darn, how I so hate buying new toys.
Have you noticed any correlation between the need for stoppers and interstage negative feedback? So far mine have been without.
Have you noticed any correlation between the need for stoppers and interstage negative feedback? So far mine have been without.
I'll have my amp on the bench, scope at the output measuring noise or distortion, and once the pot goes past a certain point I get oscillation. Turning down often kills it, but i've had it bad before.
When this happens, I usually clean up my wiring and change devices and eventually make it disappear. Can't say I've nailed it though. e.g.
I used to run 6550's and I found I needed 22k grid stoppers to stop an oscillation at (20kHz i think or something high).
I ran a 33 as a driver once and it was microphonic. It oscillated at 5kHz.
The 12at7 I used in my phono amp kicked off an oscillation at 4kHz using the volume control, replacing it with 12au7 fixed the problem, but trying to duplicate it today, it wasn't happening.
My sig gen has show up trouble at 10kHz, possibly by noise injection.
I am reluctant to increase the grid stoppers too far for f response reasons. BTW, heavy buss and star gnd, decoupled stage rails. Phono, stage1, pot, driver,o/p.
When this happens, I usually clean up my wiring and change devices and eventually make it disappear. Can't say I've nailed it though. e.g.
I used to run 6550's and I found I needed 22k grid stoppers to stop an oscillation at (20kHz i think or something high).
I ran a 33 as a driver once and it was microphonic. It oscillated at 5kHz.
The 12at7 I used in my phono amp kicked off an oscillation at 4kHz using the volume control, replacing it with 12au7 fixed the problem, but trying to duplicate it today, it wasn't happening.
My sig gen has show up trouble at 10kHz, possibly by noise injection.
I am reluctant to increase the grid stoppers too far for f response reasons. BTW, heavy buss and star gnd, decoupled stage rails. Phono, stage1, pot, driver,o/p.
SY said:
I'd like to LOL but I once had a computer monitor to fix. It wouldn't turn on and my wife said it must be the power switch. I was floored to find out it was the switch.
The pot is an Alps. Despite using it often, I seem to have forgotten it existed. Maybe I'll change it and see what happens. (Maybe it has a tiny flat spot)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.