Now, I'm more than half convinced that what I have here are not really 12AT7's at all, but in any event, I tried a pair (replacing the original RCA black plate 12AZ7's) in the output buffer stage of my sonic hologram device where they are used as a noninverting long tailed differential amp with feedback and overall I like the result so far - there seems to be some enhancement of micro detail relatively speaking and a better sense of 'air' and naturalness and better separation of subtle details (good), but the overall presentation tends to be somewhat more 'up front' (not sure about that yet) and is more likely to sound 'hard' with some material.
I should mention that, excepting the cross-fed delayed signal (which passes through a total of 4 12at7 sections for each channel), this long tailed diff pair tube stage is the sole active stage in the sonic hologram processor audio signal chain. Each channel's high voltage is fed from its own tube regulated supply, and there are two discrete solid state regulated supplies for the filaments. One floats above ground potential for the tube regulator circuits.
I assume the long tailed differential pair minimizes bias point changes related to different tubes' mu and transconductance (always assuming they are in their linear region), so I'm curious regarding how much of this difference I'm hearing might be related to the harmonic distortion structure of the tube. Could also the larger plates's longer thermal time constant have a subtle effect on the sound detail and stability wise? I should probably mention that I have essentially encased all the tubes horizontally for each channel (including tube regulator circuit) of the sonic hologram device in its own solid block of black anodized aluminum with large holes bored to accept the tubes capped with extruded heat sink sections (grid wires long axis is vertical for all tubes). The chassis is 1U rack mount. It seems not to cause any thermal problems for the tubes and probably does a pretty good job of air-borne acoustical isolation.
I should mention that, excepting the cross-fed delayed signal (which passes through a total of 4 12at7 sections for each channel), this long tailed diff pair tube stage is the sole active stage in the sonic hologram processor audio signal chain. Each channel's high voltage is fed from its own tube regulated supply, and there are two discrete solid state regulated supplies for the filaments. One floats above ground potential for the tube regulator circuits.
I assume the long tailed differential pair minimizes bias point changes related to different tubes' mu and transconductance (always assuming they are in their linear region), so I'm curious regarding how much of this difference I'm hearing might be related to the harmonic distortion structure of the tube. Could also the larger plates's longer thermal time constant have a subtle effect on the sound detail and stability wise? I should probably mention that I have essentially encased all the tubes horizontally for each channel (including tube regulator circuit) of the sonic hologram device in its own solid block of black anodized aluminum with large holes bored to accept the tubes capped with extruded heat sink sections (grid wires long axis is vertical for all tubes). The chassis is 1U rack mount. It seems not to cause any thermal problems for the tubes and probably does a pretty good job of air-borne acoustical isolation.
thoriated said:
I'm still keenly interested in your 12AT7. Ignoring the differences in the bulb and only comparing their plate structure, I see a lot of similarities to the photos I posted back in post #19. It almost appears that your plates are upside down with respect to mine - or maybe mine are upside down with respect to yours. I'm just judging by the apparent length of the plate tabs through the mica. I can't tell by the photo - does yours have the third support mica wafer at the very top? In the limited circuit testing I did with mine, I found the conduction to be much higher than a normal 12AT7. I've never been able to locate any actual curve data, and I don't have the proper equipment to make my own.
Here's two more photos for further comparison:
You're correct - the tab construction between your tube and my '12at7' almost appears to be flipped. Unfortunately, I don't have a competent tube tester, but if someone would be willing to test one of the five tubes, I could loan it for the test because I am interested in what I actually have here.
An interesting (to me) test I did a while ago with a differential input mic preamp circuit having ~50 plus db gain I had breadboarded was the relative level of microphony at the input stage with different tubes. With appropriate heater voltage switching, I could swap in everything from a 12ax7 to a 6dj8 at the input stage without a too drastic effect on performance, except for an expected difference in gain depending on the tubes mu. Listening with headphones, tapping a pencil on the side of a 12ax7 resulted in little sound; a 12at7 had a considerable resonant signature in the midrange (grid wires resonating) and a 6dj8 sounded like shattering glass - quite loud in the headphones connected directly to the mic preamp output. Getting all that transconductance and a reasonably high mu factor at the same time with a low filament consumption seemed to exact a definite price in this circuit; I'm presuming because of the tight spacing geometry needed between the grid and cathode.
The more I heard of the 6dj8/7308 family, the less I liked it; the best I heard in my admittedly limited auditioning was Amperex bugle boys, which were actually pretty good sounding tubes in a lot of ways, of which there are now two in my old Carver CD player which I haven't listened to in years. For the rest, I originally designed them into the differential cross coupled input stage of my DC coupled OTL and as the second gain stage and follower of my phono preamp circuit; fortunately I was able to pretty much replace them directly with 6BK7B's which are still somewhat microphonic but lack the edginess of the 6dj8s in all three applications. I wouldn't go so far as to say a 6dj8 sounded like 'solid state' to me - it was still definitely more transparent and grain free, but the edginess and a lack of ultimate transparency I heard was reminiscent of what is considered one of the characteristic faults of solid state sound.
An interesting (to me) test I did a while ago with a differential input mic preamp circuit having ~50 plus db gain I had breadboarded was the relative level of microphony at the input stage with different tubes. With appropriate heater voltage switching, I could swap in everything from a 12ax7 to a 6dj8 at the input stage without a too drastic effect on performance, except for an expected difference in gain depending on the tubes mu. Listening with headphones, tapping a pencil on the side of a 12ax7 resulted in little sound; a 12at7 had a considerable resonant signature in the midrange (grid wires resonating) and a 6dj8 sounded like shattering glass - quite loud in the headphones connected directly to the mic preamp output. Getting all that transconductance and a reasonably high mu factor at the same time with a low filament consumption seemed to exact a definite price in this circuit; I'm presuming because of the tight spacing geometry needed between the grid and cathode.
The more I heard of the 6dj8/7308 family, the less I liked it; the best I heard in my admittedly limited auditioning was Amperex bugle boys, which were actually pretty good sounding tubes in a lot of ways, of which there are now two in my old Carver CD player which I haven't listened to in years. For the rest, I originally designed them into the differential cross coupled input stage of my DC coupled OTL and as the second gain stage and follower of my phono preamp circuit; fortunately I was able to pretty much replace them directly with 6BK7B's which are still somewhat microphonic but lack the edginess of the 6dj8s in all three applications. I wouldn't go so far as to say a 6dj8 sounded like 'solid state' to me - it was still definitely more transparent and grain free, but the edginess and a lack of ultimate transparency I heard was reminiscent of what is considered one of the characteristic faults of solid state sound.
It's well known that low frequency audio thermal cycling affects the distortion spectra of semiconductor devices, so if some proof existed that vacuum tubes are uniquely immune to any similar effect, then that could be considered further proof of their innate superiority to sand amps, couldn't it?
You would have to excise all ICs from your signal path to be free from this problem, in his opinion. Additionally, I believe it's facile at best to assert that this phenomenon is 'totally absent' within a given device, also, based on an isolated measurement in the same breath that it is claimed to be so considerable due to the thermal + signal coupling between stages within an IC. Self doesn't fall into this particular trap; he claims that heavy feedback nullifies it in typical solid state implementations.
I run well regulated DC on all my SS tube filaments, as does any high end design (DC, at least) I've seen, so that factor would seem to have largely have been accounted for.
I should mention here that I believe tubes in general are less susceptible to this problem than solid state is in almost any implementation due to the continuous heat contribution from the filament which is lacking in sand devices.
I'm more looking for rational explanations why small signal tubes with large plate structures such as 6SN7's and 6SL7's seem to possess their generally acknowledged sonic advantages over flake plate SS tubes, beyond gm vs filament power considerations. Plate (and other structure) size would seem to be one of the next logical areas to consider.
I'd look at the resonance frequency of the bigger structures. If it's in a region where the ear's sensitivity is higher, people will latch on to "different means better."
And that's the RATIONAL explanation. There are a lot of irrational ones... FWIW, although I cheerfully use all sorts of envelopes and bases, the most linear tubes I know of are all 7 and 9 pin miniature types.
And that's the RATIONAL explanation. There are a lot of irrational ones... FWIW, although I cheerfully use all sorts of envelopes and bases, the most linear tubes I know of are all 7 and 9 pin miniature types.
The distortion characteristics are not just related to the design of the tube but to the precision to which the tube is manufactured as well. Tubes with smaller structures would seem to need more precise control over construction than ones with larger elements to meet design goals. This is only one of many explanations as to why a tube with larger plates would sound a little different from the same type of tube with a smaller plate.
You win the Silly Statement of this Thread prize.
John
You win the Silly Statement of this Thread prize.
John
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.