Bottlehead Forum
Bottlehead Kits => Eros Phono => Topic started by: maryc27182 on March 14, 2017, 10:26:17 AM
-
Hi folks, how are you?
Short version: 3-month-working-fine Eros now has odd (possibly warm-up related) transient white noise in right channel. Noise starts 3 minutes after powering on and goes away about 3 minutes later. Voltages and resistances are fine, but I cannot guarantee that I measured the voltages during the 3 minutes when white noise is produced. Questions: could it be the 12BH7 doing weird things and should I be concerned?
Long version:
My Eros has been working fine since December. It is currently a stock build. I changed the EF86 tubes from stock to GEC CV4085 in January, and that worked fine as well. I have a modded Rega RP1 into Eros into Stereomour.
Over the past few days, I've noticed something odd. For the first 3 or 4 minutes after powering on, the sound is fine. At 3 or 4 minutes, the right channel sound becomes distorted, volume fades a bit, and then the sound turns to white noise, basically. The left channel is fine. I can control the white noise with the Stereomour volume pot. If I turn the volume down, the white noise is gone.
Usually at this point, I am over by the pre-amp, fiddling with connections, and eventually lightly tapping on the top plate. After 3 minutes of this, the right channel returns, and does not distort or become white noise again for the rest of the playing session, which is usually at least 2 albums long. Light tapping on the top plate after the right channel is restored does not seem to cause the white noise to return.
What I've done so far: I swapped the GEC 4085 between channels. That did not change anything--the right channel is still the one that goes weird. That test was done using a cold system.
I put a different 6922 in. No change. Right channel still goes weird during the first 3-4 minutes, then comes back after another 3 minutes. That test was done using a cold system.
I re-checked resistances and voltages this morning. The only change I saw from original measurements taken in December 2016 was in the kreg readings, and I have changed EF86 since then. Original Soviet EF86 tubes had 1.8VDC back in December 2016; GEC 4085s are measuring 1.4 VDC. (Manual says 1VDC is typical at kreg.) This should be fine. However, I cannot guarantee that I measured right-channel voltages during the time when the pre-amp may have been making white noise.
I do not have another 12BH7 tube. Yet.
Oh: I did the other usual things. Tested another source to make sure it wasn't the Stereomour tubes going bad (CD source is fine); swapped input leads from the phono (no change--it's still the right channel that intermittently goes out in the first 3 minutes.) I'm pretty sure I even disconnected the phono inputs while the pre-amp was making the white noise and saw no change. The sound is such that I really thought it was a tube going bad more than anything like a solder joint issue or a connection.
I have stared grimly and intently at all the solder connections and tried to find any nicked wires. I poked and tugged, but not while the pre-amp was connected to the system.
One test I haven't done, since this is an event that I only seem to be able to duplicate when the pre-amp is cold, is to see if the white noise goes away without me standing over the pre-amp and fiddling/tapping. I think that will be tonight or tomorrow's test, and it will be a really hard one for me hahaha.
Any ideas? Could it be the 12BH7? Does it have any warm-up related 'events'? Thanks, folks.
Mary
-
It's most likely to be a thermal issue. Might also be worth looking for a cap that has been nicked with a hot soldering iron. If you have exhausted the possibilities, the solution that I would use for this in my listening room is:
Turn preamp on.
Wait 7 minutes.
Play music.
-
Yep! As long as it doesn't sound potentially dangerous, and it sounds like it's not, I will probably take one more look, maybe get another 12BH7, and then not worry so much.
Thanks for the response, as always.
Mary
-
Stereomour has three inputs; it's not clear whether you have tried another input pair, or swapping channels. That would check the Stereomour input wiring on that one channel, just up to the selector switch output. I have no reason to suspect that, but just for completeness ...
-
Thank you for the response. I will try another Stereomour input later tonight, or tomorrow.
I just tried the let's-see-what-happens-if-I-don't-tap-the-top-plate test, and although it was hard to just sit there, the same stuff happened: 3-4 minutes in, right-channel music turned into white noise, and then 3 minutes later, the white noise gradually turned back into music.
I could not find any nicked capacitors. I was pretty careful with this build, because I found it fiddly and was the most-difficult one for me so far. That said, because it was difficult, I could have made a mistake.
Mary
-
Might be the next step is to try a different 12BH7.
-
If the 12BH7 doesn't solve the problem, I would recommend going out and buying some clip leads. With the Eros off, clip lead in your meter to measure DC volts. You'll have to run the Eros until the phenomenon happens, then report any deviations. This will have to be done for each of the following terminals:
OA and OB on the C4S board of the offending channel.
IB or IA on the C4S board of the offending channel.
That will take three slow power cycles, as you need to bring the Eros up, wait for the issue, then observe any variations, then shut the Eros down and repeat a few minutes later with the clip leads elsewhere.
FWIW, I don't believe that the 12BH7 will cause this kind of issue, unless there's some kind of internal short, which would become readily apparent when you measure the IA/IB connection.
-PB
-
I have push-on alligator clips. I'll make those measurements, hopefully in the next few days.
I tried a different input on the Stereomour and there was no change in the behavior, so the Stereomour is OK.
Thanks for your help.
-
I was hoping to get all the measurements done and plotted in one posting, but this is going to take longer than I thought (really should grab a DAQ from work...) I have not replaced the 12BH7, as I don't have another one yet.
Here's right-channel OA (VDC) versus time. Summary: At about six minutes after powering on, the voltage drops from 167 VDC to a minimum of about 40 VDC. Then the reading climbs (not linearly), getting back up over 167 VDC about 16 minutes after powering on, and then being stable (I got tired of measuring after 3 minutes of steady-state.) (Note: I'll probably change the time axis to read minutes once I get all measurements done, and put this on a different spreadsheet program so I can fix all the things about this graph I don't like, but that'll have to wait until I have less work to do hahaha.)
-
Just measured OB (VDC). Not worth plotting. Measured 97.7 VDC at 30 seconds after power on. This value slowly drifted down to 96.1 VDC by the end of the test at 20 minutes after power on.
I had the preamp off about an hour between the first test, and this test. (I don't have the preamp connected to my system, so I do not know if I got the white noise.) I'm going to wait a few hours before testing again. Since I do not know what I am expecting at OB during the white noise, I don't know if this measurement aligns with what I measured at OA. I'm considering re-running this measurement in a couple of hours. Might measure IA first.
-
IA (VDC) was 226 VDC. Pretty much only fluctuated about 0.1 VDC. I measured over a period of 10 minutes from powering on.
So: Summary: OA had a transient from 167 VDC down to 40 VDC that occurred about 6 minutes after powering on. OA then rose non-linearly back to 167 VDC over the next 10 minutes of operation, then seemed stable.
OB was ~97 VDC over 20 minutes of operation.
IA was 226 VDC.
-
Can you repeat the test at Kreg on the offending side?
What you're describing could be as simple as a flaky solder joint on an MJE5731A going through thermal cycling.
-
Will do, and thank you for the response, and for all the help.
I was looking at the circuit diagram. Would it be the MJE350 or would it really go back to the MJE5731A? Keep in mind I do not understand how the circuit works. Feel free to simply respond, 'No.'
-
Hmmm... were you expecting to see something here? Because I got 1.45 VDC pretty stable for 10 minutes of operation at Kreg.
Note: for all these tests where I've done measurements, the circuit has not been enclosed. I assume things still warm up enough to get thermal effects on the boards?
-
Sorry, I meant to say MJE 350 on the board up above the EF86 of the offending channel.
The Kreg measurement that interests me is actually Breg on the populated side of the board. I haven't quite worked out how the EF86 plate voltage stays so happy while the 6DJ8 momentarily craps out.
-PB
-
Yeah, I'm skeptical about my third measurement. I was waiting about an hour between tests, but as I mentioned, I can't be sure I got white noise. Ah well. I'm torn between taking another measurement and re-wetting those joints. Scientific curiosity versus engineering pragmatism. Thanks again for the help.
-
Well, since I had the preamp out, I took the Breg measurement. Pretty stable over 10 minutes of operation. 99.3 VDC about 15 seconds after powering on, and this drifted slowly down to 98.3 VDC at 10 minutes.
I am going to put this aside for now--probably put it back in the system this evening and see if the white noise is still there.
-
It turns out that having the preamp powered down for an hour, hour-and-a-half *isn't* long enough to get it back to whatever initial condition that causes the white noise. (I put the preamp back into my system and listened. When powered down for six hours, I got the white noise; when powered down for an hour, I did *not* get the white noise.)
So the OA and Breg measurements are probably OK (they were both done on a system that had been powered down for more than 12 hours.)
The OB and IA measurements were done on a system that had only been powered down for about an hour, so I'll redo those. I also got a new 12BH7, so having that option, along with the option of re-doing solder joints, should keep me busy, since I only like to make one change at a time hahaha.
-
Here's IA (VDC) versus time. The preamp had been powered down for 16 hours. Ambient temperature was probably 5 degrees F colder than when I took the previous measurements.
Summary: IA was ~226 VDC until about 9 minutes after powering on. Then it fell to 160-170 (fluctuated) for about a minute, then dropped below 150 (fluctuating) to a minimum of 128 at about 11 minutes after powering on. There was a cluster of points fluctuating around 140 from 11.5 to 12.5 minutes, then a cluster of 160-170ish VDC from 13-14 minutes, and a linear rise from 180ish to 226 from 14 minutes to 16 minutes (ish). Stable at 226 from 16 minutes to 20 minutes, at which time I powered down.
I'll re-attach the OA vs Time measurement from previously. Difference was that the OA voltage dropped at 6 minutes, but these data were measured previously.
My knowledge of voltage regulators is slim, but shouldn't it give 225 VDC no matter what happens 'downstream'? (Again, I don't know enough about how it works to know if the 'stuff' on the C4S board should be able to pull IA down.)
-
Well, that's very helpful. The IA voltage is indeed the output of the regulator back by the 12BH7. There's a slim chance that this is a 12BH7 acting up (though very doubtful, a failing 12BH7 would actually cause the regulated voltage to go up usually), but more than that, there may be a solder joint acting up thermally back on the PCB above the 12BH7.
I would now recommend measuring Kreg on the PCB over the 12BH7 on the offending side to see if it drops too, but at the same juncture I would reheat all the solder joints on that board as a precautionary step.
-PB
-
Thank you! I was just about to get the preamp out again, as it's been about six hours and should exhibit the behavior again. I will measure Kreg on the board above the 12BH7. Let me take a look to make sure I can figure out which side is the right channel....
I am assuming the "A" Kreg on that board is the right channel?
Would you like the measurement, or would you like me to remove the board and reheat everything first? Thanks again.
-
Re-reading your post, I think you want me to re-do all the solder joints first.
-
I'd reheat everything on that board first, yes.
-
Thanks again. I will be slowed down a bit. While looking at that board, I saw that one of the 220 ohm grid stop resistors was under tension--I remember being concerned about that when I mounted the board, but I didn't know (then) that these resistors are mechanically fragile.
I tried to relieve the mechanical stress yesterday (I had a couple of mm of lead protruding through the hole) but I wonder if the damage is already done. Anyway--I will get some carbon comp 220 ohm 1/4W resistors before I continue.
-
Any old 100-500 Ohm film resistor will do the job down there.
-PB
-
Yeah, all I had in my parts box were 5W metal oxide film resistors, and they are, um, hefty. I can't even remember what I used them for.
p.s. Woah.. so it's 50 posts where you stop being a "Jr. Member" or whatever I was. I don't feel so bad being so verbose now.
-
It wasn't the grid stop resistors being in tension (and perhaps damaged.) I think it was a bad solder joint on a 2N2907A transistor leg. I can't be totally sure, since I re-wet everything on the shunt regulator pcb (as well as the tube socket for the 12BH7.) However, it was the only joint I saw that looked suspect--I had trouble re-wetting it and spent more time on it than anything else.
Although my head knew it's silicon, I built a couple of guitar-related things with germanium transistors recently, and I'm sure my reluctance and anxiety was driven by feeling how warm the 2N2907A was getting as I tried to get a decent joint, as well as my past fear over solder wicking up and bridging to the metal body.
Anyway--I've done two listening trials now, and so far, no white noise (oh--voltage checks were all good.) Thanks for all your help with this.
-
Good job tracking that down!
Yeah, it can be a little tough sometimes to know how much heat parts can take. The little HLMP diodes don't take a lot (they also don't need a lot to be well soldered), then the MJE5731 takes tons and tons of heat to get properly soldered. The PN2907 is somewhere in te middle.
If the issue comes back, we'll be here to help.
-PB