Need help with an unusual problem

Larpy · 48883

0 Members and 1 Guest are viewing this topic.

Offline Larpy

  • Full Member
  • ***
    • Posts: 230
Reply #30 on: April 05, 2024, 11:27:06 AM
Would hearing an audio file help?  Is there a way for me to share a brief (40 second) audio file of me switching back and forth between the same music on LP and CD?  The difference in sound quality is dramatic.  I could also attach the waveform of the recording as captured on Vinyl Studio so you can how LP waveform is asymmetrical whereas the CD one is not. 

I'm technologically backwards enough to have no idea how to share a brief MP3 file with others on a forum.

 

Larry


Offline Larpy

  • Full Member
  • ***
    • Posts: 230
Reply #31 on: April 06, 2024, 01:03:05 PM
OK, I'll drop the audio file idea.

Today I tried bypassing the Sowter SUTs and listening to (and recording the waveforms of) an LP.  Had to turn the preamp volume up all the way and there was tube rush, but the LP seemed to sound distortion-free.  What’s more, the resulting waveform, though tiny and hard to magnify clearly, looked symmetrical.

Is that a clue?  Could the asymmetrical waveforms I’ve been seeing and the distortion I’ve been hearing be caused by the SUTs stepping up the signal from my cartridge too much?

If so, this wouldn’t make sense:  I have a cartridge (Hana ML) that puts out 0.4mV into Sowter 1990 SUTs wired 1:10.  That should put the input to the Eros circuitry at the low end of its 4-6mV target range.  And yet when I’ve recorded music from LPs in the past few days, the signal into my digital interface has clipped a few times.  That means the Eros is putting out a signal somewhat hotter than line level, yes  The badly distorted mass strings I heard the other day looked to measure 9v p-p on the oscilloscope.  From what I understand, that shouldn't be too hot an output signal from the Eros.

As far as I can tell, my Eros is wired correctly.  The oscilloscope measurements of my RIAA network now read as expected.  I’ve checked and rechecked my wiring of the Sowters (and I double-checked that they are indeed 1990s).  And yet could the AC signal into/out of the EF86s be too high?  Is there a wiring mistake associated with the 6922 tube that would explain my symptoms?

I’ve swapped out all of the tubes and that makes no difference.  Changed interconnects.  Surely a bad AC cable couldn't cause this.  Right?

I’ve been reading up on asymmetrical waveforms and, from what I’ve gleaned, DC offset can be the source but when it is the mean amplitude will be displaced from the central axis.  That doesn’t seem to be the case with my asymmetrical waveforms, so I’m thinking DC offset is not behind my problem.  From what I remember observing my oscilloscope while I play music through the Eros, my mean amplitude is typically –1 volt or so.  And the waveforms from music I’ve recorded on Vinyl Studio look on-axis.

If I correctly understand what I’ve been reading, an asymmetrical AC waveform where the mean amplitude is on-axis (or close, as in my case), the source of the asymmetry can be high pass filters or “aggressive low-end processing.”

Is it conceivable that I’ve somehow inadvertently created a high pass filter in both channels of my Eros?  What kind of miswiring would create such a high pass filter?

This is the most puzzling and frustrating problem I’ve faced in 20-25 years of tube DIY builds.  And I’ve done dozens.

Larry


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5833
Reply #32 on: April 06, 2024, 02:16:38 PM
Somewhere in the thread so far there is a statement that the voltages are all within spec. But some of the other observations make me wonder if the 6922 is not operating correctly. If you can measure the DC voltages at the pins of the 6922 (pins 1-3 and 6-8) that would help.  The correct voltages are only a little different from incorrect ones, so it's easy to mis-interpret them.

Paul Joppa


Offline Larpy

  • Full Member
  • ***
    • Posts: 230
Reply #33 on: April 07, 2024, 07:39:17 AM
Measurements on the 6922 tube:
pin 1:  160vDC
pin 2:  97vDC
pin 3:  98vDC
 
pin 6:  160vDC
pin 7:  97vDC
pin 8:  98vDC


Larry


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5833
Reply #34 on: April 07, 2024, 09:33:15 AM
That's pretty close to the design target, though I'd rather see 170v at the plates. The difference means your 6922 has a slightly higher transconductance than nominal, which is generally a good thing but can upset the operating point a bit in this direct-coupled circuit..

If you have another 6922 on hand, preferably a well-used one, it might be worthwhile trying it. If not, you can measure the voltage difference between cathode and grid (pins 2-3 and 7-8), which should be about 1.50 volts. I don't expect that to show a problem, but our expectations have been a poor guide so far!

Paul Joppa


Offline Larpy

  • Full Member
  • ***
    • Posts: 230
Reply #35 on: April 07, 2024, 11:20:36 AM
I don't have any well used 6922s other than the Bugle Boy 7308 I've been using for a couple years.

Difference between pins 2 and 3 on it is 1.5 vDC.  The difference between pins 7 and 9 is a mere 1 volt.

I have a reputedly NOS Siemens 6922 that measures 160 and 155vDC on their plates, and
the voltage between pins 2 and 3 and 7 and 8 is a bit below 1.3 vDC in both cases.

The Russian tube that came with the kit measures 162 and 156vDC on their plates.
The difference between pins 2 and 3 is only .7 volts.  The difference between pins 7 and 8 is a paltry 0.55 volts!

I'm consistently low with all three tubes.  Might this be an important clue?

Larry


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5833
Reply #36 on: April 07, 2024, 02:06:06 PM
It is indeed - those grid-cathode voltages are too low for proper performance.

The first thing to check is the 47K/2W resistors - are they the right resistance? Are they properly connected? You should be able to measure the resistance from the plates (pins 1 and 6) to ground, with the amp power off.

While you are at it, also check the resistance from the cathodes (pins 3 and eight) to ground, which should be 27K - this will take some time, as the 100uF/160v caps will have to charge up. Also confirm that those caps are properly oriented.
« Last Edit: April 07, 2024, 02:13:02 PM by Paul Joppa »

Paul Joppa


Offline Larpy

  • Full Member
  • ***
    • Posts: 230
Reply #37 on: April 08, 2024, 05:11:44 AM
The resistance from 6922 plates to ground is 44K in both instances.  With the power off.  I checked the 2 watt 47K resistors on the board above the 6922 socket and they are the correct 47K resistors.  Just to make sure, I desoldered one leg of one and measured its resistance out of circuit: 47K Ω.

I reflowed the solder connecting those two resistors but no change in resistance.

Am I right in guessing this suggests something is miswired elsewhere and pulling the 47K Ω resistance down about 10%?

Measuring the cathodes to ground proved easier if I disconnected the bypass cap (which is not the stock electrolytic but a Panasonic DC link cap and so not directional).  Resistance from 6922 cathodes to ground is 27K Ω on the nose.

Larry


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5833
Reply #38 on: April 08, 2024, 08:24:58 AM
I hate to say this, but things are pointing to that 100uF bypass cap. The circuit is behaving as if something is draining current in parallel with the 27K resistor, and the only candidates I can see are that cap or the EF86 screen grid.

To test this, with the cap disconnected, repeat the voltage measurements, including the grid-cathode difference as well as the plate, grid, and cathode voltages.

Paul Joppa


Offline Larpy

  • Full Member
  • ***
    • Posts: 230
Reply #39 on: April 08, 2024, 12:17:25 PM
Don't hate to say anything!  Any advice you offer is immensely appreciated, and I'll be ecstatic when you find my problem.  I haven't been able to listen to LPs in 5 months, so at this point I'd happily replace any component to get my Eros happy again.

Measurements with 100uF caps disengaged (they actually measure 110uF):
Pin 1: 162v
Pin 2:  98.8v and then drifted down.  After a minute or so it was down to to 97.7v
Pin 3:  99v

Pin 2 to Pin 3:  1.1v

Pin 6:  157v
Pin 7:  101v
Pin 8:  101.3v

Pin 7 to pin 8:  started at .4v and slowly rose (over a few minutes) to .85v and then began drifting back down.  After a couple of minutes it was .76v.  Now that is weird.

Larry


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5833
Reply #40 on: April 08, 2024, 02:22:05 PM
This is a real puzzler, all right.

At least we've found a symptom that explains the one-sided clipping problem. The 6922 bias is too low and it is going into grid current, clipping one side of the waveform. The problem is, the voltages and currents don't make sense

The only idea I have left is a possible radio-frequency oscillation problem. Those are usually dependent on layout and lead dress, so any non-standard component or change in wire routing becomes a suspect. Can you locate the original 100uF caps and install them on the PC board as shown on page 70 of the manual? Leave the film caps detached from the cathodes. Are there any other deviations from the stock configuration? They may have to go, too.

Paul Joppa


Offline Larpy

  • Full Member
  • ***
    • Posts: 230
Reply #41 on: April 09, 2024, 07:12:04 AM
I don't seem to have any 100uF electrolytic caps in my cap drawer.  I'll order a couple from Mouser.  If I remember right, they were Lelons?  This is probably a question for Doc: have Eros kits been shipping with Lelon 100uF/160v caps recently?  I'll play it safe and order a pair of Lelons and a pair of Nichicon UCY caps--they're rated for 12,000 hours.

I'll report back once the caps have arrived and I install them.

Here is the complete list of the other non-stock components in the Eros:
• SBE 716P "orange drop" caps in the RIAA network
• Auricap caps as the output caps
• 0.1uF Russian K40Y-9 caps on the voltage regulator board
• Dale CPF 2 watt 47K Ω plate resistors for the 6922 tube
• Vampire RCA input jacks

The big Panasonic caps are the only component that required non-stock lead dress.

The interesting thing is that most of these components were imported from my first Eros.  I used the Panasonic, Auricap, and Russian caps in my first Eros for a year and a half with no problem.  Quick recap:  I happily used an Eros for a year and a half when, last fall, it suddenly started sounding distorted.  My attempt to identify the problem turned into a nearly comical creation of new problems, but I'm 99% sure the original problem was a flaky transistor solder connection.  But I got so frustrated I ordered a new Eros kit during last Thanksgiving's BH sale.  I put this new kit together with all new parts except for the Panasonic and Auricap caps--they came off my old Eros and were installed in my new Eros.

I gave my first Eros to an audio friend, but he's out of the country until later this month.  Otherwise, I'd have him come over with my old Eros to compare.

Larry


Offline Larpy

  • Full Member
  • ***
    • Posts: 230
Reply #42 on: April 11, 2024, 11:19:22 AM
100uF caps arrived today and I soldered them in to the board above the 6922 tube socket as shown in the instructions.  I put the Bugle Boy 7308 tube back in (it seems the happiest of the ones I have for duty in that socket) and the measurements looked better:

pin 1:  167v        pin 6:  161v
pin 2:  94v          pin 7:  95v
pin 3:  96v          pin 8: 97v
pin 2 to 3:  1.6v  pin 7 to 8: 1.3v

I hooked the Eros back into my system but unfortunately the distortion is still there.  And the asymmetrical waveforms.

I guess the next step is to remove all the other non-stock components I've used.  Since I was ordering the 100uF electrolytics, I also ordered replacement Dale resistors that get soldered to pins on sockets A-C (the same ones that came with the kit).  I think I'll rebuild all three sockets and their connections and go really, really slowly.

« Last Edit: April 11, 2024, 11:21:22 AM by Larpy »

Larry


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5833
Reply #43 on: April 11, 2024, 02:47:48 PM
I don't have a better suggestion at this point. I wish I did!

Paul Joppa