Bottlehead Forum
Bottlehead Kits => Eros Phono => Topic started by: jrhunter62 on February 08, 2021, 12:08:28 PM
-
Just finished my Eros 2 Build and it sounds great, but the left channel is noticeably weaker than the right. I have swapped the L and R tube positions, no change. Swapped cables, no change. I hooked up my Schiit Mani phono pre and the channels were balanced. I have looked over the connections and solder points and nothing looks suspect to my inexperienced eye. What should I check, test, re-solder to fix this?
-
Well, we have seen a fair number of situations like this over the years. The difficulty of diagnosing this with a phono preamp is that the equalizer to correct for the RIAA curve should drag down the response (varying according to frequency), so sometimes people will be debugging the channel with low output when in actuality the EQ isn't connected and it's the loud channel that's causing the problem.
Could you post the voltages on the small center PC board? (Any of those terminals with a wire leaving that say "I" or "O" on the label) Possibly there's something buried in the DC voltages that would give us an idea.
Something else that occasionally happens is that the 100uF caps are installed backwards or not well soldered and that will cause a gain difference, as would a loose 10,000uF cap up on the front terminal strips.
-
Make sure the EQ components are attached at the right points too. I've seen a miswire there cause this problem in the past. This would include the components on the five lug strips inboard of the two front tube sockets.
-
I have gone over the passive RIAA connections and didn't find anything that looked out of place. The voltages checked as follows:
IA - 217v
IB- 217v
OA - 159.8
OB - 166.3
OC - 98.9
OD - 99
OkA - 99.2
OkB - 99.3
OkC - 1.44
OkD - 1.49
-
Yeah, those are dead nuts on. That's sort of a good thing and a bad thing. When you have voltages that are that perfect, you'll be looking at an EQ issue, or a solder joint on the input or output of the preamp. Something like this would have me looking really hard at the solder joints on the RCA jacks and on the coupling caps.
In the absence of something obvious, can you download a signal generator app on your phone, turn it all the way down, then up one click on the level, then us an 1/8" to RCA adapter to send signal into the amp. With your meter set to AC voltage, try measuring the AC voltage across the input jacks with your meter at 20Hz, 1000Hz, and 20,000Hz. This is one of those times where a cheap meter that only has a 200V and 700V AC scale will not quite be up to the task of what you're doing, but the $25 meter at Harbor Freight can resolve low AC voltage levels and I think should at least pick up the 1kHz tone without too much trouble.
Anyway if you can feed a consistent amount of signal into the input jacks and successfully measure it (with the Eros off), then we can move on to additional testing.
-PB
-
I have the tone generator app, the RCA adapter, and a decent multimeter. I don't exactly understand how to conduct the test you suggested. Am I measuring the output jacks?
-
You need an 1/8" TRS to RCA cable to plug into your headphone jack on the TRS end and the input of the Eros on the other. You are feeding the signal from your phone into the Eros.
-
Got it. What am I measuring? The output jacks? What should I see?
-
I have measured both the input and output jacks and I don't read anything on the DMM while sending a 20hz, 1000hz, or 20,000hz tone. I am placing the test leads on the soldering points on each jack (input, output, L, R). am I doing this correctly?
-
AC voltage across each input jack. You can post a picture of your meter if you need help setting it properly.
-
I have attached several pictures of what I am working with and how I am testing.
-
Try 60Hz. The red clamp that you're using probably isn't a good idea. It looks like it's touching the center barrel of the RCA jack and the outside of the RCA jack, which will short out all the signal and give you a 0V reading.
Your meter specifies a 2V scale for AC so it should work properly to take these measurements.
From what I can see of your build I would add a little solder to just about all of the solder joints and a lot more heat.
-
The meter should be connected to an input jack - where you show it - to measure the level you are putting into the Eros from the signal generator. The level should be set to around 5mVrms AC. I would be inclined to test the output from the generator before I even connected it to the Eros just to verify signal is present and near the correct level.
Once you have verified the signal level is present and set where you want, move the test leads to the output jacks and measure that level.
-
At 60hz both input jacks measured .04. I got a reading of 0 on the output jacks with the amp off. With it on I measured 36 on the right and 38 on the left.
-
.04V is 40mV. If possible you want to set the signal level at the input to be about 5mV which is .005V. Once you get that set measure the output level again. Then try 1000Hz and 10000 Hz both set at 5mV, and see what the output voltage reads.
-
I got the input signal as close to 5mV as I could. I'm not sure this meter does well at that low of a reading. With the unit on I got 3.46v on the left output and 6.96v on the right output.
-
Sorry, I forgot to mention it was 3.46v on L and 6.96v at 1,000hz and .035v L and .077v R at 10,000hz
-
OK, now swap EF86 tubes, do those measurements again and see if anything changes from left to right.
-
Swapped EF86 tubes - 3.11v on L and 7.76v at 1,000hz and .033v L and .086v R at 10,000hz
-
Looking at the left output RCA jack in your image. I can't tell for sure because of the overhead angle, but it might be that there is some solder that has run down from the center pin cup and is touching the outer shell.
-
I checked the left output and I can't find any solder touching the outer shell
-
Do you have a spare 6922 to try?
-
I do not.
-
I got the input signal as close to 5mV as I could. I'm not sure this meter does well at that low of a reading. With the unit on I got 3.46v on the left output and 6.96v on the right output.
Gain at 1kHz of the Eros is about 50dB, and 5mV with 50dB of gain applied to it would give 1.5V of output.
It would help to know what you saw on your meter that was as close to 5mV as you could get so we can know whether the lower volume channel is the issue or whether the higher volume channel is the issue. So far it is looking more like the higher volume channel may actually be the issue, but it's not possible to say without concrete data.
-
I have attached pictures of the meter and what I am measuring. Previously I was concerned the meter might not have been up to it because the values were fluctuating, but I am able to get .005v consistently now.
-
more pics
-
I took another picture of the right input measurement at .005v, but it isn't playing nice with the forum, you will have to take my word for it.
-
OK, taking 0.005V in and getting 3.16 out is 56dB. That's a little more than expected, but not impossible. If you are getting 7V out of the other channel under these conditions, I would be looking toward the EQ as the source of your issue. It could be something simple like the two little brown caps being swapped, or it's possible the 10.7K resistor got swapped for a different part by accident.
-
If a visual inspection doesn't turn up some swapped EQ components it might be helpful to measure at 100Hz, 1kHz and 10kHz again and note the voltage difference between channels at each of those frequencies. If the percentage of disparity is larger at one end of the spectrum than the other that might help to narrow down what components to check for a cold solder joint.
-
I can't find any components swapped or mis-wired
100Hz: R- 27.8, L- 18.63
1KHz: R- 7.68, L- 3.14
10KHz: R- .091, L- .026
-
10KHz: R- .091, L- .026
Can you double check that there's not an extra zero here?
I would also be sure that your center C4S board is really well soldered. A loose 75K resistor on that board would do some strange things but the servo might be able to compensate on the DC side enough that it's only noticeable with AC.
Beyond that do you happen to have a spare 0.1uF/400V capacitor around? Or something similar? We can use that to check those levels on the plate of the EF86s, as many meters don't measure AC well in the presence of DC.
-
Sure looks like it's the equalization. Because the 10kHz gain is most affected, my suspicion is that the 0.01uF and 0.03uF are swapped on the left channel.
In any case, the equalization is only four components - those two caps, and the 75K and 10.7K resistors.
-
Found it. I had the both 0.01uF on one side and both 0.03uF on the other. They are now in the correct places. I threw on a record and it sounds much better now. Thanks for sticking with me all week. Even with this hiccup it has been a blast putting this together and troubleshooting it. I have learned a lot and am excited to find my next build. Going to do a lot of listening this weekend and I will check back in and let you know how it goes. Thanks!
-
Fabulous! Glad it's sorted now.
-
This thread reinforces the notion that sometimes it can be helpful to get a second set of eyes on the circuit. Even if those eyes are not experienced at building they may spot something that you have been looking at repeatedly and not seeing. Happens to all of us.
-
I agree completely. Where I work we call it 2nd party verification. Unfortunately my wife had little interest in checking my work. Most days I consider that a good thing!
-
I'm retired now but I worked in IT for 39 years. The classic IBM Mainframe trio of COBOL, CICS and DB2. There were definitely times when we needed another set of eyes to help debug our code. Fortunately, we had tools to allow us to single-step each line of code to see what was happening. Too bad there's nothing like that for kit building. OTOH, with the experience on this forum, problems can be narrowed down to the most likely culprits.
I mentioned this because while I don't really understand any of technical stuff, I find the debugging skills of PB, PJ and Doc fascinating. That's why I enjoy reading all of the threads.
Gerry
-
Gerry - that "line by line" debugging is reflected in our increasing used of multiple pauses in the manuals to test the progress so far. (I didn't make the connection until I saw your post, but the principle is the same. Thanks!)
I honed my debugging using (mostly) FORTRAN from the same era ...
-
What this thread proves if that Bottlehead is one company that follows through in all ways, not just now but for years to come. How many more companies go to the extreme that Bottlehead does, please show them to me so that I may give them my business.