I do completely agree, Tom.
In the build guide I've made an indirect referral to it:
"I’ve tried Padis audiophile fuses and found them worse than industrial ceramic ones."
The only audiophile fuses I've tried were those Padis brand and they were no good at all, still sounded different.
Some time earlier I've found that plain simple and cheap ceramic fuses tend to sound better than glass ones.
Somewhere I should have stashed some audiophile fuses (forgot the brand, Isoclean perhaps), 2,5A i think. If you wanna test them, I'll try to find them before we meet.
So I'm having issues with my final assembly. I'll try to get some pictures posted at some point, but for now, here's the rundown. I had tested both boards when I assembled them back in November. They both fired up and played music at the time. Now that I have them in a chassis one channel is working but the other has issues. When I power it up the relay clicks closed then open then closed quickly and the led flickers. My dim bulb also stays dim instead of flashing then turning off like it does with the other channel. When I measure all voltages they are within spec except for the 35v to gnd which only measures 20.6 volts on each regulator. I checked voltages from the transformer, an Antek AS-2222, and they are consistent with the specs on both secondaries. I also gave the board another good cleaning. Any thoughts?
The flashing LED is almost certainly a sign of DC in the signal path. Try to measure the DC between the plus output and a "0V" point on the schematic (the relay is opening and closing so the normal output minus won't work). I have used the bottom of C9 or R13 for the "0V" point. I would disconnect any input, even short the input on the bad channel. In theory, C13 should preclude any DC from the input. That means you have to search for the source of the DC. Also, measure both AC and DC with no input. If you have an oscillation, it will show up as both.
Remind us, are you using the LM318 (standard BOM) as the op-amp?
Jac
Remind us, are you using the LM318 (standard BOM) as the op-amp?
Jac
I would like again point out that I don't see the low impedance heatsink grounding in action..
That is, HSgnd connected to the heatsink, directly. if You have a case, and the heatsink gets connected that way, through the case I mean that is ok too. But in 'standalone' applications like in the photos, pls do it..
It is something that ! sometimes, can burn You.. not always, need to be unlucky as well..
Ciao, George
That is, HSgnd connected to the heatsink, directly. if You have a case, and the heatsink gets connected that way, through the case I mean that is ok too. But in 'standalone' applications like in the photos, pls do it..
It is something that ! sometimes, can burn You.. not always, need to be unlucky as well..
Ciao, George
Hi Jac,
On the Version 1.72 boards that I am using the C9 polarity seems to be reversed to what you are saying. The + is toward C13. See the 10th Run Boards here.
George,
I do have the boards on a 1/8" thick aluminum base plate with a star ground to mains earth ground. I checked for continuity between the heatsink and star ground and it is solid.
On the Version 1.72 boards that I am using the C9 polarity seems to be reversed to what you are saying. The + is toward C13. See the 10th Run Boards here.
George,
I do have the boards on a 1/8" thick aluminum base plate with a star ground to mains earth ground. I checked for continuity between the heatsink and star ground and it is solid.
You are right. All the boards up through v1.5 have had the opposite polarity, so that's a surprise.
Something we can never see in pictures, so I'll ask. Are you using the LM3886TF which is plastic insulated on the heatsink? If you are using the un-insulated version, do you have a non-conductive transistor pad between it and the heatsink?
I am really grabbing at anything because the build looks good. I hope others will join in with ideas.
I keep going back to your power supply voltages. With the 22 VAC secondaries, you should expect about 31 VDC after the filter (where it's marked 35 vDC). 20 VDC after the filter is too low. Again, I wild thought. There are two secondaries from the trafo and two diode bridges. Can you confirm that each secondary is going to only one bridge? If you had one lead of a secondary on one bridge and the other lead on a different bridge, it might do something like you are seeing.
I notice that your inputs and outputs are on the "bottom" of the board. An incorrect polarity at the input would be something I could easily get wrong. Of course, if you get the problem with no input connected, then forget that idea.
Jac
Jac,
Yes, I am using the plastic insulated LM3886TF, no insulated pad or mica between IC and HS.
I was getting about 31 VDC when I first assembled the boards. As for the secondaries, I have checked and double checked that I have the correct secondary wires paired up from the transformer but it couldn't hurt to check again.
For inputs and outputs the problem occurs with nothing connected to input or output. I haven't shorted the inputs during my measurements either.
Here are my current thoughts:
1. Try to track down any DC on the signal path.
2. Re-solder all joints or questionable joints - The iron i used to build these boards wasn't the best. I've since upgraded to a Hakko station that is awesome. Would it hurt to go over all my joints and see if that helps?
3. I had also thought I may have damaged something in the time since I had first assembled the boards to now. I must have moved the boards a million times while doing other things in the garage. Is it possible I could have shorted or damaged a component even thought the board wasn't hooked up to a power source?
4. Can I start testing individual components? Where to start? I have one of these which should be able to tell me quite a bit about most of the parts on the board.
Yes, I am using the plastic insulated LM3886TF, no insulated pad or mica between IC and HS.
I was getting about 31 VDC when I first assembled the boards. As for the secondaries, I have checked and double checked that I have the correct secondary wires paired up from the transformer but it couldn't hurt to check again.
For inputs and outputs the problem occurs with nothing connected to input or output. I haven't shorted the inputs during my measurements either.
Here are my current thoughts:
1. Try to track down any DC on the signal path.
2. Re-solder all joints or questionable joints - The iron i used to build these boards wasn't the best. I've since upgraded to a Hakko station that is awesome. Would it hurt to go over all my joints and see if that helps?
3. I had also thought I may have damaged something in the time since I had first assembled the boards to now. I must have moved the boards a million times while doing other things in the garage. Is it possible I could have shorted or damaged a component even thought the board wasn't hooked up to a power source?
4. Can I start testing individual components? Where to start? I have one of these which should be able to tell me quite a bit about most of the parts on the board.
Hi Koldby.
Nope, I’m just making four channels, maybe two for mono blocks, and I have considered two boards together with an RPI3 streamer, and a phono amp. for a neat integrated one box solution.
That reminds me. I once shorted out the regulated supply (14V) with a probe and an unsteady hand. That resulted in DC in the system. The diodes had burned for certain. You could see it. I replaced the diodes and the transistors, just in case, and everything worked again.
Not There Yet
A few weeks ago, we determined that I had an oscillation because I was stealing power from the regulated supply of the FE. Since then, I have built a completely independent power supply for the balanced line to single ended converters. That part is working fine. The FE boards are v1.2 with the ADA4627 opamp and have a 5 pF mica cap between opamp pins 2 and 6.
I measured both DC and AC at the output with the input shorted and also with the balanced line converter connected. I also measured with C9 in position and with C9 removed and replaced with a short jumper. There was no difference between the input shorted and the balanced line converter connected, so I will only report the input shorted numbers. The voltages will be left/right for each measurement.
C9 - DC = 12mV/42mV - AC = 0.8 mV/1.2mV
jumper - DC = 265 mV/1.36V - AC = 1.0mV/1.3mV
It should be noted that the jumper DC voltages were increasing over time.
Although I haven't tried yet, it should be possible to play music with C9 in place, but clearly a danger to a speaker when the jumper is in place. Of course, even with C9, the 42 mV DC offset on the right channel is troubling.
Jac
A few weeks ago, we determined that I had an oscillation because I was stealing power from the regulated supply of the FE. Since then, I have built a completely independent power supply for the balanced line to single ended converters. That part is working fine. The FE boards are v1.2 with the ADA4627 opamp and have a 5 pF mica cap between opamp pins 2 and 6.
I measured both DC and AC at the output with the input shorted and also with the balanced line converter connected. I also measured with C9 in position and with C9 removed and replaced with a short jumper. There was no difference between the input shorted and the balanced line converter connected, so I will only report the input shorted numbers. The voltages will be left/right for each measurement.
C9 - DC = 12mV/42mV - AC = 0.8 mV/1.2mV
jumper - DC = 265 mV/1.36V - AC = 1.0mV/1.3mV
It should be noted that the jumper DC voltages were increasing over time.
Although I haven't tried yet, it should be possible to play music with C9 in place, but clearly a danger to a speaker when the jumper is in place. Of course, even with C9, the 42 mV DC offset on the right channel is troubling.
Jac
Jac.. I played a bit with math..
First: when you see 12 / 42mV on output, with C9 in place, it straightforwardly means that you have 12 / 42mV offset between input pins.
C9 guarantees unity DC gain. So you have from somewhere a fixed offset voltage on the input constantly present.
Now I forcedly drove out your DC values from my head but left one fact: your DC gain with jumper suddenly becomes 31,77.
12 * 31,77 = 381mV
42* 31,77 = 1334mV
First: when you see 12 / 42mV on output, with C9 in place, it straightforwardly means that you have 12 / 42mV offset between input pins.
C9 guarantees unity DC gain. So you have from somewhere a fixed offset voltage on the input constantly present.
Now I forcedly drove out your DC values from my head but left one fact: your DC gain with jumper suddenly becomes 31,77.
12 * 31,77 = 381mV
42* 31,77 = 1334mV
So going back the first case: 12 /42 mV on the input with these chips is a totally abnormal situation..
There is some 100's microvolts offset from chip and should be practically zero input current contribution.
Should be zero on both chips..
The past is hunting.. it happened already.. could it be residuals under chip, around pins?
Ciao, George
There is some 100's microvolts offset from chip and should be practically zero input current contribution.
Should be zero on both chips..
The past is hunting.. it happened already.. could it be residuals under chip, around pins?
Ciao, George
The small feedback caps? Are they low leakage (cog) types?
Input resistor (100K) ? It is truly 100k?
It is connected? (Your pins..)
Solder flux is painstakingly supercleanly removed after operations..
See how it is easy bombarding you with strange unfunded ideas...������
Input resistor (100K) ? It is truly 100k?
It is connected? (Your pins..)
Solder flux is painstakingly supercleanly removed after operations..
See how it is easy bombarding you with strange unfunded ideas...������
Last edited:
Fascinating! I wish I could think the way you do.
The interesting thing about the DC offset on the input is that I am using a flat jaw alligator clip (some call them duck clips) across the input pins. With the duck clip removed, the DC offset is the same, so the DC offset is coming from somewhere after the input terminals.
I will look a little deeper, but I think you may be correct. Time to look carefully at the connections of the opamp and, if I don't find anything there, remove the opamp, clean up, and start again. I know that you strongly prefer the ADA4627, but I could also try the OPA627 to see if the issue goes away.
Thanks again for your input. When I get to the end of an experiment and can't imagine the next step, you are always seeing further and keeping me on track. Thanks.
Jac
The interesting thing about the DC offset on the input is that I am using a flat jaw alligator clip (some call them duck clips) across the input pins. With the duck clip removed, the DC offset is the same, so the DC offset is coming from somewhere after the input terminals.
I will look a little deeper, but I think you may be correct. Time to look carefully at the connections of the opamp and, if I don't find anything there, remove the opamp, clean up, and start again. I know that you strongly prefer the ADA4627, but I could also try the OPA627 to see if the issue goes away.
Thanks again for your input. When I get to the end of an experiment and can't imagine the next step, you are always seeing further and keeping me on track. Thanks.
Jac