[jacobmj]

Chime rectifier voltages, gain, and reducing noise

Hi.

I'd appreciate some clarification about acceptable voltages in my
Chime.  I've been experimenting with a Mullard 5AR4 in lieu of a Sovtek
5Y3.

First, about the sound: By ear, I notice that bass notes with the 5AR4
are more prominent, and perhaps a bit 'faster'.  I question whether
this is simply a wholesale increase loudness at bass frequencies, but
I can't convince myself so, and haven't measured carefully enough to
differentiate.  The midrange appears identical to my ear.  Highs
appear to have the same tone, but are more spacious.

I would like to clarify a few points about the greater voltages that
result from the 5AR4.  Here are my numbers:
     

                             5Y3     5AR4
                             -----------------
H+                        8.13    8.22
6.2V, after R423   6.33    6.41
B+                        369     378
305V (R438L/R)    304     313


It makes sense that I can adjust R423 to bring down the heater
voltage.  Is 6.2V, as on the schematic what I should aim for, meaning
I should adjust it for the 5Y3 as well?  Is there any reason to
similarly adjust the B+ voltage?  If so, how could such an adjustment
be made?

----------

On another note, I find the level of noise at silence (no input,
volume at nil) to be rather high, and quite noticeable from my
speakers at a distance.  I am running direct from the Chime into a
Bryston 3B-ST, and a pair of Sound Dynamics 300Ti speakers.  I correct
the noise by attenuating with an L-pad between the Chime and Bryston
by about 10dB.  This allows me to use much more of the volume control
range: about 1/2 for classical, 1/3 for typical music.

Is it possible to reduce the gain of the Chime output stage?

I've considered the aspect that this noise is brought upon by my
tubes, so I have made a few measurements.  The following are
measurements taken with a radio shack meter against the waveguide of
the tweeter on one speaker, always using the same amplifier channel.
All measurements are A-weighted.


12AX7             12AU7    Red Output(left)   White Output(right)
----------------------------------------------------------------
Telefunken      Brimar   75db                      69db
Electro-Harm. Sovtek   64db                      64db
Electro-Harm. Brimar   75.5db                   69db
Telefunken      Sovtek   65.5db                   64db


I had previously noticed, and measured by oscilloscope, that the noise
on the left channel (right on this reversed board revision) is of
approximately twice the voltage amplitude.  This appears to be solely
due to the Brimar 12AU7.  Even if I corrected this, presumably with a
new tube (suggestions welcome), I don't consider the 64db
measurement level quiet enough for listening at a couple of meters; it's
noticeable.

Incidentally, the Telefunken and Brimar make a striking improvement to
the onset of all notes, imaging, and smoothness of vocals.  The change
is far greater than that between the 5Y3 and 5AR4.

So, aside from attenuating after the Chime, can anything be done about
this noise?


Thanks much.
~Jacob

[poty]

I would like to clarify a few points about the greater voltages that
result from the 5AR4.  Here are my numbers:
                             5Y3     5AR4
                             -----------------
H+                        8.13    8.22
6.2V, after R423   6.33    6.41
B+                        369     378
305V (R438L/R)    304     313

Are you shure about the measurements? They have been puzzled me. 5Y3 has more resistance to the hi-voltage part, so the load on the transformer should be eased, so the H+ voltage (not involved the tube for rectifying) should be higher than in case of 5AR4, but not otherwise.
According the values itself.
H+ IMHO should be in the range of 6.15-6.35 V. Less than 6.15V greatly deteriorates tubes characteristics, more than 6.35V poses a risk of short tube life.
Such high B+ should be the result of wrong values of R424-R425 (if you are using specified transformer of course). The V401 stage on one channel should draw near 6.5mA of current, the V400A +0.5mA - so about 7mA total, which should substract about 7.8*7=54.6V from the raw rectified voltage. Your measurements shows the comparable difference with the values marked on the schematic. You have all the measurement rather high - whan mains outlet voltage do you have?

I find the level of noise at silence (no input, volume at nil) to be rather high... I am running direct from the Chime into a Bryston 3B-ST, and a pair of Sound Dynamics 300Ti speakers. I correct the noise by attenuating with an L-pad between the Chime and Bryston by about 10dB.

Hmmm... It seems you have a classic case of impedance mismatching. While "in specifications" all seems correct (1k - output impedance of the Chime, 50k - input impedance of the Bryston on the unbalanced input) in reality something is out of order. That is why adding the L-pad has the effect.

Is it possible to reduce the gain of the Chime output stage?

The output stage has gain=1. I'm far from thinking that almost classic first stage would add such noise while the grid is effectively at ground. So the problem is somewhere else. You can do classic trick with adding a capacitor ~1uF between the grid and the ground of the output tube. It effectively eliminates everything AC from the first stage while preserve the DC point. With such test you can check where the noise is from?

So, aside from attenuating after the Chime, can anything be done about this noise?

In my opinion the problem is in the output tube. It's cathode has rather high potential (around 150V) relating the heater, for some tubes (especially old) it can be out of specs or noisy.

[jacobmj]

Thanks for writing.

I'm rather confident in these measurements, and do not believe I
have, for example, reversed the labels.  To clarify, the Sovtek
is a 5Y3GT.  Voltage at the wall during my measurements was right
about 121V.  Throughout a typical day, it ranges 119.5-123.

I bought this Chime second-hand.  I don't know whether it was build by
the fellow I purchased it from, or bought complete.

R424 and R425 are 3.9K.  R424 does have a diode in series.  For
sanity of all components, a couple of photos of the board are
here: http://jjoseph.org/photos/projects/hagerman_chime

Hmmm... It seems you have a classic case of impedance mismatching. While "in specifications" all seems correct (1k - output impedance of the Chime, 50k - input impedance of the Bryston on the unbalanced input) in reality something is out of order. That is why adding the L-pad has the effect.

Forgive my ignorance, though I'd appreciate a bit more explanation of this.  My intention with the L-pad is simply to attenuate, not better match impedance.  The Chime has at least 10dB more gain than I need with my speakers.  I drop the absolute amount of noise by 10dB while retaining plenty of signal.

The output stage has gain=1. I'm far from thinking that almost classic first stage would add such noise while the grid is effectively at ground. So the problem is somewhere else. You can do classic trick with adding a capacitor ~1uF between the grid and the ground of the output tube. It effectively eliminates everything AC from the first stage while preserve the DC point. With such test you can check where the noise is from? In my opinion the problem is in the output tube. It's cathode has rather high potential (around 150V) relating the heater, for some tubes (especially old) it can be out of specs or noisy.

I'll give a capacitor a try.

~Jacob

[poty]

I'm rather confident in these measurements, and do not believe I have, for example, reversed the labels. 

OK, It may be coincidence with something else.

Voltage at the wall during my measurements was right about 121V.  Throughout a typical day, it ranges 119.5-123.

Difficult to see the transformer wiring clear, but I think the primary connected right for 120V.

R424 and R425 are 3.9K.  R424 does have a diode in series.

Could you measure the voltages on
- (R424-R425)s (the "+" side - to the positive of C103, the "-" side - to the positive of C406);
- the cathodes of V401 to the ground;
- R438 (between leds);
- anodes of V400 to the ground?

Forgive my ignorance, though I'd appreciate a bit more explanation of this.  My intention with the L-pad is simply to attenuate, not better match impedance.  The Chime has at least 10dB more gain than I need with my speakers.  I drop the absolute amount of noise by 10dB while retaining plenty of signal.

The output of the Chime has limited ability to drive the following stage input. According to specs for both interconnected devices - they should be coupled nicely, but something around that is wrong. It may be the side effect of high amplitude signal from hagdac too (then it should be checked too). The output voltage of the Chime is 1.2Vrms, the input voltage of your power amplifier - 1V (all numbers - according to their specs), it is only 20*log1.2 = 1.58dB more than the input of your power amp! But! You've said you have the noise even if the volume control is at zero position! Then the noise is coming from the analog tube part itself. Then - main goal is to find the part that adds to the noise.

[jacobmj]

I've made some more measurements, and double checked their
veracity.  The only mistake was the branding of the signal tubes:
It's a Brimar 12AX7 and a Telefunken 12AU7.

Adding a capacitor from V401 grid to ground does eliminate nearly
all of the noise, particularly on the right channel (L on board).
Presumably due to my rather long jumper, it does result in some
60 and 120hz hum.  The noise is effectively gone though.
Curiously, the left channel retains some noise--I believe the
difference is due to the tube itself.

Further grounding the grid on V400 effects no reduction in noise.

Using a simple tone generator on my laptop, and terminating the
chime into 22kohm, I notice that I achieve 4.44V peak-peak,
whether noise or a pure sine.  The noise is approximately 2mV
p-p.  That 4.44V p-p strikes me as somewhat higher than the
expected 1.2Vrms * 2.8 = 3.36 V p-p.  Am I mistaken here?


As for voltages, here's a new table, including a few more than
you'd asked, if only for my own records.


                        5Y3GT   5AR4
B+, C103             365      375
C407  L                302      310
          R                302      311
btw R438,R429 L 294      301
                         R 294      302
drop R438 L         7.74     7.83    Some drift
                    R      7.55     7.63    Some drift
drop R424 L         32.2     32.0
                    R      31.9     31.7
drop R429 L         159.5   161.1
                  R        153.1   154.8
V400 plate L        136      138.9
                   R       142      145.3
V400 cathode L   0.969    0.988
                       R   0.926   0.943
V400 grid L          0.2mV  0.2mV
                R          0.2mV  0.2mV
V401 plate L        300      309
                  R        301      309
V401 cathode L   141      144.9
                       R   147      150.1
V401 grid L          136.0   138.7
                R          142.5   145.3   

(For voltage drops, I did not include the series diode.)

Thanks.
~Jacob

[dnewman]

FWIW,

Another source of noise in Chime's can come about when using the S/PDIF inputs (as opposed to the USB input).  Since the Chime lacks isolation pulse transformers on those inputs, you can inject problems into your Chime via the S/PDIF inputs.  I had such a problem and fixed it by introducing an isolation transformer (Digikey 470-1006) and 75R termination resistor.  Eliminated the noise making for nice, dead silence on my very sensitive speakers (98 dB).  Eventually, I fixed the problem with the source thereby eliminating the need for the isolation tranny.

Dan

[poty]

Hello Jacob,
Dan in the previous message gave good advise. I'll even broaden the thought: to be completely shure about the place to cure we should eliminate all the possibilities that the noise is induced from the digital side. I'd completely disconnect Chime from all sources and probably remove the hagdac board for the time of testing. Let's analyse your findings:
Begin with measurements:
I think instead of "btw R438,R425" should be "btw R437,R425"? Then all readings are correct and within the right margins. You mentioned some "drift" in measures across the R438s. It's not good. It may be the Brimar (main suspect) or a solder problem of R437, or leakage in C400, C411 (in doubt, because you have the noise in both channels - very small possibility that you made the same mistake in both channels or have bad parts in both channels; the fact I'm mentioning this is that the noise from one channel may affect the other).

... grounding the grid on V400 effects no reduction in noise.

... means we do not pollute the input from the previous stages. But there is some possibility to pollute through ground though.

Adding a capacitor from V401 grid to ground does eliminate nearly
all of the noise, particularly on the right channel (L on board). Presumably due to my rather long jumper, it does result in some 60 and 120hz hum.  The noise is effectively gone though. Curiously, the left channel retains some noise--I believe the difference is due to the tube itself.

Yes, the hum is from long wires. The fact, that in one channel the capacitor was more effective than in another could mean that the source of the problem lies in one channel and pollute the other through the common circuits (for example - ground). But it also could mean different spectrum of noise in channels or amplitudes.... I'd begin with replacing the AX7 tube. It is the main suspect up to now, common to both channels, can add noise, easily replaceable, the only active part around the supposed noise source stage.

Using a simple tone generator on my laptop, and terminating the
chime into 22kohm, I notice that I achieve 4.44V peak-peak, whether noise or a pure sine. The noise is approximately 2mV p-p.  That 4.44V p-p strikes me as somewhat higher than the expected 1.2Vrms * 2.8 = 3.36 V p-p.  Am I mistaken here?

It is definitely higher, but I'm far from certainty that you have supposed 1.2V - 30dB = 38mV of signal on the grid of V400 to match the specifications.

----
For heating voltage - I'm really enjoying after replace the R423 with the Sharp LDO regulator (I got it from mouser - PQ6RD083J00H). You don't need any additional parts and would forget about heating voltage instability. The onlu thing that may prevent you from doing that - the space inside the box. The regulator should have a heatsink.

[jacobmj]

Dan, Poty, thank you both for your assistance.

I only use the usb input of the Chime in typical usage.  All of the
tests here are with a minimal number of connections -- power only, and
either a scope (22k termination) or the amp on the output.  For
considering the noise, I've kept the volume at nil.

I have now removed the HAGDAC; the noise persists, and has not changed
in in level (still 64db) or character by ear.  Interestingly, I did
notice that two of the capacitors closest to the front edge of the DAC
were of questionable connection.  But, they're not the immediate cause
of trouble.

To be sure, I've changed both (singly and together) the AU7 and AX7
tubes.  This doesn't affect the noise beyond the db values in my first
post.

Poty, you're right about the resistors, I transcribed my notes
incorrectly.  I meant the portion labeled 295V on the schematic.  I've
corrected the post.

You mentioned some "drift" in measures across the R438s. It's not
good. It may be the Brimar (main suspect) or a solder problem of R437,
or leakage in C400, C411 (in doubt, because you have the noise in both
channels - very small possibility that you made the same mistake in
both channels or have bad parts in both channels; the fact I'm
mentioning this is that the noise from one channel may affect the
other).

By 'drift', I mean that the measurement was somewhat unstable over
second timescales.  They varied approximately 0.06V, possibly with
some larger peaks.  I didn't use a scope for that measurement.  I'll
retouch the solder on R437L/R to be certain of them.

With respect to drift, I also notice that the output of both channels
exhibits a slowly changing DC offset, ranging as much as 10mV over
several seconds.  I'll need to arrange for a long exposure to share
the scope trace, should you be interested in seeing it.

The LDO regulator seems like a good idea.  I'll pick one up with my
next order.

Thanks again.
~Jacob

[jacobmj]

To follow up on a bit more testing:

I've retouched the solder on most of the the components above the tubes on the schematic, on one channel.  This made no difference.

~Jacob

[poty]

All of the tests here are with ... power only, and either a scope (22k termination) or the amp on the output.  For considering the noise, I've kept the volume at nil.
I have now removed the HAGDAC; the noise persists, and has not changed in in level (still 64db) or character by ear.

Good to mention conditions of tests. Found: we do not have external electrical influence and should look into the circuit itself.
Then, to eliminate the last possibility of internal pollution - could you disconnect the Chime from the outlet and listen to what happens? The heaters will supply emission for several seconds, the B+ capacitors - allow to work the same seconds. We eliminate switching noise from heaters and B+ rectifiers, capacitive coupling to mains and several other things.

To be sure, I've changed both (singly and together) the AU7 and AX7
tubes.  This doesn't affect the noise beyond the db values in my first post.

OK! It should prove that the tubes are not the source of the noise, but I'm not convinced fully (oxidized pins? bad batch of tubes? bad socket? bad socket's solder joints? broken PCB contact pad? ... ). There should not be such noticeable noise in the circuit!

By 'drift', I mean that the measurement was somewhat unstable over
second timescales.  They varied approximately 0.06V, possibly with some larger peaks.

Such small deviations may be from EMI from nearby devices or the noise inside the multimeter you use. It's difficult to understand at this point if they are important or not.

With respect to drift, I also notice that the output of both channels
exhibits a slowly changing DC offset, ranging as much as 10mV over several seconds.  I'll need to arrange for a long exposure to share the scope trace, should you be interested in seeing it.

The analog part circuits are DC coupled, so the output drift can be result of anything, beginning from the power supply infralow-frequency oscillations and ending with positive feedback somewhere. It can be of course the source of the noise. If you can - do the trick with o'scope.

[jacobmj]

Good to mention conditions of tests. Found: we do not have external electrical influence and should look into the circuit itself.
Then, to eliminate the last possibility of internal pollution - could you disconnect the Chime from the outlet and listen to what happens? The heaters will supply emission for several seconds, the B+ capacitors - allow to work the same seconds. We eliminate switching noise from heaters and B+ rectifiers, capacitive coupling to mains and several other things.

I hadn't thought of doing this.  I'm testing by pulling the plug.  Alas, for the approximately 2 seconds that the power light remains on on the Chime, the noise remains.  There is no change in any quality of it, so far as I can hear.

OK! It should prove that the tubes are not the source of the noise, but I'm not convinced fully (oxidized pins? bad batch of tubes? bad socket? bad socket's solder joints? broken PCB contact pad? ... ). There should not be such noticeable noise in the circuit!

Given the number of times I've swapped tubes lately, I'll bet all of the contacts are squeaky clean!

For isolation, would it be acceptable to disconnect B+ from one channel, say by pulling R424, and possibly grounding it instead?

Thanks again.
~Jacob

[poty]

I'm testing by pulling the plug.  Alas, for the approximately 2 seconds that the power light remains on on the Chime, the noise remains.  There is no change in any quality of it, so far as I can hear.

Then the power supply is not the source.

Given the number of times I've swapped tubes lately, I'll bet all of the contacts are squeaky clean!

I hope so. Have you tried different brand tubes sourced from different places or all the tubes are from the same source?

For isolation, would it be acceptable to disconnect B+ from one channel, say by pulling R424, and possibly grounding it instead?

Unfortunately it raises the B+ voltage if you just pull R424 or cooks the R424-425 if you ground them. I'm afraid If you are absolutely sure about good tubes and there is no signs of broken traces or colder joints around the tubes the next steps are going to be rather difficult.

[jacobmj]

Thank you for your continued help.

The tubes are of very different origins.  Tested 12AU7s are an old Telefunken and a new Sovtek.  12AX7s are an old Brimar and a new Electro-Harmonix.  I've tried a variety of rectifiers, old and new.

Unfortunately it raises the B+ voltage if you just pull R424 or cooks the R424-425 if you ground them. I'm afraid If you are absolutely sure about good tubes and there is no signs of broken traces or colder joints around the tubes the next steps are going to be rather difficult.

I wasn't proposing simply grounding them!  If it's worth bothering, I could presumably correct B+ back down with an equivalent resistance for that channel.  I'm asking if it is indeed worth doing so.  And, whether it is appropriate to run the tubes with one channel grounded or disconnected.

I'll work through more of the solder joints, but it strikes me as unlikely that this would affect both channels equally.  (As you point out, one may 'cause' the other.)  I tried to go through most of what was in common between them, and your suggestion of removing power further suggests that these weren't likely the cause.

So, what's next?

~Jacob

[jacobmj]

Hello again.

A complete resoldering of the tube sockets, and all resistors on the L channel yielded no change.  There was a reasonable amount of flux remaining at some joints, and a thorough cleaning produced no change either.

Since I had previously accepted this level of noise as inevitable, I'd like to confirm that this thing should be dead silent with speakers such as mine.  As a means of calibration, I can say that touching the 12AX7 (Sovtek or Brimar) during operation increases the noise by about 6db, about 70db measured 0 distance from the tweeter, and adds a bit more high frequency content to it.  Out of curiosity, I put a grounded metal can over the signal tubes--no change with it either.

With more careful oscilloscope measurements, I believe the p-p amplitude of the noise is around 1mV plus a 120hz component of about 5mV.  I'd never noticed the hum on my speakers, so I think this 120hz may have something to do with running sans case.  I don't have the DAC installed.  I can't really do better than this on my oscilloscope.  I haven't yet moved the amplifier to my test bench, or I'd scope the noise on its speaker terminals.

~Jacob

[dnewman]

FWIW,

When I throw one of my Chimes on a spectrum analyzer, the 120 Hz peak is around -73 dBV and 60 Hz around -80 dBV.  I have to put my ears right up to my speakers to hear that at all.  I'm not a pro at doing such testing; those numbers may be incorrectly measured.

Dan

[poty]

Hello Jacob,
If you are going to experiment further, I'd recommend to begin with less aggressive ways first. For example, you can try to connect a comparable value capacitor (47uF) in parallel with C400, C406 (better to connect the positive lead directly to the top leads of the anode resistors according to schematic).
Of course it is possible to go through the circuit with system approach. In that case I'd cut the first stage from the second. The main problem is stay with DC work point of the output tube and changes in the voltages due to non-regulated power supply. I recommend to unsolder the R423 and solder a pair of wire in its place. It allows to change the R424 without harming the PCB. Another possibility - solder wires to a spare tube socket and put the other ends to the socket on the PCB (sort of extender). The idea is to pull out the V400 and insert into the tube socket a pair of resistors to maintain the right anode voltage. I estimate the value of the resistors in the range 140-150k and they should be put in 1-3 and 6-8 holes respectively. To maintain the heating voltage in the appropriate range you should either tweak R423 or insert the heating wires from the extender into the corresponding holes of the PCB socket. After that - if the noise stops - the problem is in the tube itself or heating circuit, if persists you have to go further - eliminate the R437, C400, C411 from the circuit (you can do that by unsolder the R438 and add a resistor between the 305V point to the left side of R430) and eliminate the socket (solder the resistors from the socket directly between the left side of R430 and the ground). ... Circuit by circuit - you'll find the problem. All abovementioned you could do in one channel of course, just realise your idea of switching off the second.
\\\
Vladislav

[jacobmj]

Hi.

Before going too much farther, I thought I'd try to establish a means and baseline of measurements by scope on my bench.  I'd previously measured at the outputs of the chime, but this was at the limits of my (min 2mV/div) scope.  I'm somewhat surprised that measuring at the amp's speaker outputs still presents a challenge.

Let me just get straight to the plots.  The following traces are measured at the speaker terminals, with a speaker connected.  They look the same if the speaker is disconnected.  I'm using 10x probes that don't update the readout, so all of these plots are 20mV/div.  All are with the Brimar 12AX7 and Telefunken 12AU7 tubes, measuring the right channel (L on board).


Grounded amplifier inputs.  No audible noise with ear on the tweeter.




Full output.  ~69dbA with the meter upon the tweeter.  It's clearly audible at 1meter.










-10db attenuation using L-pad between the Chime and amp.  Much less noise by ear.  Almost below background at 1 meter.  This is what I've been using for music.






I'll order a few components for any more tests, so there'll be some delay on my end...

I remain mildly suspicious of whether all of the noise is coming from the 12AX7 tube itself, and whether it should be considered "normal".  I don't believe I can hear it at all when I connect a different 8ohm speaker (the Sound Dynamics are 4ohm).  That the Brimar 12AX7 tube is louder than the Sovtek, and that simply touching the Sovtek increases the noise as much as the difference between them, leads me to think that the 12AX7 is the primary cause.

To get back to my original question, is it possible to reduce the gain in the 12AX7 stage, whether by adjusting the resistors around it, or by changing to a lower-gain tube in the 12A*7 series?

Another option might be to purchase the lowest-noise 12AX7 I can obtain.  Presumably this would help.  Curiously, I greatly prefer the imaging and sharpness of notes with the Brimar rather than the Sovtek.  It's so much better that I'm still using it, despite the noise.  Might you have a suggestion for which new tube I should buy, or from where I should get it?

One other thing I thought I'd note is that the noise decreases a little if I leave the Chime on for a few hours.  I tend to turn it off when I'm not using it, and have read mixed opinions of whether this is a good thing to do, particularly with respect to tube life.  What's the consensus here?  Especially in the winter, when I could use a little more heat, I'm not concerned about the power consumption of leaving it on.

I did look at a spectrum of the output by recording from the Chime output using a computer.  The 120hz and 180hz peaks dominate the signal, by far.  I need to try with a better sound card, which I have but didn't use, and possibly measure at the speaker terminals.  I'll report back if I get this working...

Thanks again.
~Jacob

[poty]

Hello Jacob,
You've done a grandiose work! I looked through your pictures with great interest and have some thoughts about them. It's a shame, but I didn't understand them fully.
The first pic - from the Bryston output while the input is short-circuited?
The second and the third - the same as 1st, but the Chime connected and sensitivity switched from 2mV for 2nd to 5mV for 3rd? What the words "full output" in the legend before the pictures means then?
Also what are the 4th and the 5th pictures for? And what the symbols H and O in the right bottom side of the 5th picture means?
If you can still bear my comments I'll try to comment on your message.
12AX7 as a class of tubes should not be the source of noise itself if they are good. For example the tube is used in the first stage of Cornet 2 phono amp of JH, which I built myself and prove that the level of noise is very small despite the fact that the input signal voltage is almost 10 times less. That is why I asked you to try different tubes. I also tried Brimars in my Cornet, but changed them - can't bear more noise and instability too. For long time I used cheap Philips NOS tube, then changed them to even cheaper Russian tubes (with very small tweaks) - for my ears they are most analytical and truthfull. My friend - also DIYer - doesn't like them, he says they are not so "rich" in presentation comparing to the Philips. To each his own... sort to say... In the circle there are many suggestions for the AX7. Since the Cornet 2 and the Chime circuits are essentially the same in the analog part you can use Cornet 2 suggestions in the Chime too.
According the gain. Yes, you can reduce the gain. You can get a hint from the http://www.audiocircle.com/index.php?topic=74326.40 topic. At the 3rd page there is a schematic incorporating the volume control between the DC coupled stages in the Cornet 2, which are identical those in the Chime (minus RIAA eq circuits of course). I'd put a capacitor with more value in place of C12 though.
My last words would be rather wild I think. I'm looking at the 5th picture in your post and a similar pattern come to my head. The infralow frequencies are most common in LC PS filters, not RC. The sharp peaks are common for SS rectifiers, not tube. JH designed the filter very carefully, but IMHO in minimalistic way, still the behaviour cannot be explained just by not enough capacitance for example (still it is possible, that the electrolytic caps in the PS filter are lost their capacitance for some reason). I think about the serial SS diodes in the PS filter (serial with R425s, R438s). They can "delay" the propagation of the voltage (inductive behaviour) from the PS and cause sharp rising of current (SS diodes behaviour). I don't know why the diodes are here, maybe you can try to short circuit them and listen?

[jacobmj]

Vladislav,
Your comments have been very helpful.  I only hope you'll be patient enough to keep them coming.

My labels could have been clearer, but I think you've deciphered them as I intended.  The first plot is the amp with input short circuited.
"full output" means unattenuated input from the Chime to Bryston, and applies to photos 2-5.  2 and 3 differ only in timescale.  4 is identical to 3, but with the intensity turned way down.  5 has more vertical sensitivity.  The breakage in traces at the left is due to movement of the trace during the exposure.  6 and 7 are after attenuating by 10db with my L-pad between the Chime and amp.  They differ only in timescale.

 "H_O" stands for hold-off of the trigger, which is a tunable delay of how frequently the trigger can re-fire.  It helps to get a more stable trace in some situations.

I had previously shorted those diodes; it made no difference.  I don't know why they were added either.  I do have a plot of B+ with the sovtek 5y3gt.  The following is 500mV/div.  The diodes were not shorted at this time.



One other thing...  I had previously mentioned a slowly changing (with some faster jumps) DC offset at the Chime output.  This appears to be entirely correlated with appliances in the house switching on and off, and probably my neighbors' houses too, who share a transformer on the pole.

~Jacob

[poty]

I had previously shorted those diodes; it made no difference.  I don't know why they were added either.  I do have a plot of B+ with the sovtek 5y3gt.  The following is 500mV/div.  The diodes were not shorted at this time.

OK, I hope you have shorted all the diodes at once. Relating B+ plot. More interesting would be the plots in the 305V and 295V points. In my opinion the p-p pulses should not exceed 10 mV in 305V point and much less than 1 mV in 295V point.
Vladislav