[Cloud]

Dont mind if I post my problem here.

The left side of my CKKIII is working fine. Its the right side thats giving me headache. Irregardless of how much i turn the trimpot(R43), the reading across R47 still read 0mv, even if full clockwise. The output transistors are cold, at room temperature.

OPA2134PA is used. Pin 7 reads about 50mv which is stable. Its Pin1 thats irregular. It fluctuates between 50mv to 1.16V.

All solder joints had been reflowed at least 3 times and all resistors are checked.

I thought the transistors are cranky and thus replaced all transistors on board, BC550, BC560, output and 2SK74, 2SK174. Opamp had been replaced too. The problem still exists.

DC offset with OR fluctuates between -20mv to +70mv. It will rise to 'peak' then fall.

Checking with operation voltages supplied by amb shows that voltages before the OPA is fine. after OPA output 1+v, all following voltages are not in line.

Does anyone have suggestion where do i carry on troubleshooting?

 

[Henmyr]

Left channel 2SJ74 measurement between drain and source (soldered to the board): 34.9 ohm

Right channel 2SJ74 measurement between drain and source (soldered to the board): 31.5 ohm

Can this difference cause my problems?

 

[amb]

Henmyr, that measurement is mostly meaningless, so it doesn't help anything.

 

[Henmyr]

I've now gone through the whole board, and every part is the correct one and in the correct location. I've also measured the resistance between the legs of each part, which showed that no part have it's legs shorted.

I have visually inspected every solder point and I can't see anything which looks like a cold solder joint.

I have not measured resistance between the parts yet though. It will require a lot of work to first calculate what the correct value should be and then measure and see if it is what it should.

Any ideas on what the best way to proceed is? Is it possible one part or more can be broken? I've switched the output transistors, and the problem was still there so it can't be them. Maybe the op-amp? Would a broken op-amp make the voltage very unstable across the board?

Problems
* When I measured before between -15v and ground, the voltage first rose to -15.23 and then dropped a lot, so something very strange is happening.
* Voltage vary a lot across the board.
* Voltages measured are often very different from what they should be.
* DC on both left and right output.

EDIT:
The only thing wrong I've ever found was the LED which I soldered the wrong way (and powered on). The led did not light up, so I knew something was wrong and very find the error and resoldered it. The LED is now lit when I turn the power on.

 

[Henmyr]

Some progress:

I bought a new operation amp (same as the part list), and it's more stable now.

1) With both trimports at fully counter-clockwise, the +v and -v are perfectly stable at 14.99V and -15.37V. Both R23 and R47 are stable at around 0-1.2mV, and DC at OG-OL and OG-OR is very low (a few mV)

2) If I measure over R23 and turn the trimpot, I can get it to ~12mV and it is stable. If I increase it past 12.3mV, it's very unstable and varies a lot with spikes around 60mV.

3) If I now measure R47, it also varies a lot (And it's trimpot is still at full counter-clockwise!) with spikes around 60mV.

4) If I now measure +V and -V, they show around 13.xxV and -14.xxV.

5) If I turn off the power and returns the trimpot to zero and turn it on again, +V, -V, R23 and R47 will all be stable like at 1).

6) If I do 2) but turn the trimpot which belong to R47, step 2), 3) and 4) will repeat itself.

Any thoughts?

 

[amb]

Henmyr, when adjusting the bias, you have to do it a little at a time, and wait for it to settle each time. Bipolar junction transistors have a positive temperature coefficient, so when you increase the current, it gets hotter, which will make it increase its own current some more. If you crank up the current too much, it could go into thermal runaway and cause the output transistors to get really hot and flow a large amount of current.

Also, when you make this adjustment, make sure you have the inputs shorted to ground (or if you have a volume pot connected, turn it to minimum). If the inputs are floating, it may pick up noise and mess with your measurements.

Lastly, do the measurements while there is no headphone connected. If it still exhibits unstable behavior, maybe you have a high frequency oscillation. Try adding a zobel network at the output (search the forum if you don't know what a zobel is).

 

[Henmyr]

Quote:

Originally Posted by amb /img/forum/go_quote.gif Henmyr, when adjusting the bias, you have to do it a little at a time, and wait for it to settle each time. Bipolar junction transistors have a positive temperature coefficient, so when you increase the current, it gets hotter, which will make it increase its own current some more. If you crank up the current too much, it could go into thermal runaway and cause the output transistors to get really hot and flow a large amount of current. Also, when you make this adjustment, make sure you have the inputs shorted to ground (or if you have a volume pot connected, turn it to minimum). If the inputs are floating, it may pick up noise and mess with your measurements. Lastly, do the measurements while there is no headphone connected. If it still exhibits unstable behavior, maybe you have a high frequency oscillation. Try adding a zobel network at the output (search the forum if you don't know what a zobel is).

Thanks for your reply

I have the input pins shorted on both channels. I have a bent resistor leg that touches both pins. Thats the way to do it, right?

I have no headphones connected and haven't even connected the headphone output jack.

I tried to get it stable again. I started from zero and allowed it to be on for a while. Both R23 and R47 stabilized around 0-1mV. +-V is stable. OG-OL and OG-OR <1mV.

After approx 10-15 minutes, I increased both sides to 7mV. After 10-15 minutes they both stabilized at around 9mV. +-V is stable. OG-OL and OG-OR <1mV.

Then I increased R23 to 10mV, and then all failed and both channels spiked to ~60mV and varied a lot. +-V has now dropped approx 0.5-1V and varies. R23/R47 does not stabilize even if I turn the trimpots to zero (probably because of the positive temperature coefficient).

If I let the output transistors cool down, I can restart the whole process.

This whole thing is very strange. I might try a zobel network if everything else fails.

My transistorlegs are rather long now since the "repair", and some are attached in two locations, first location is the original location, and the second is at the nearest . The whole thing is very ugly and looks something like:

 

[amb]

Quote:

Originally Posted by Henmyr /img/forum/go_quote.gif I have the input pins shorted on both channels. I have a bent resistor leg that touches both pins. Thats the way to do it, right?

Yes, that's fine.

Quote:

Then I increased R23 to 10mV, and then all failed and both channels spiked to ~60mV and varied a lot. +-V has now dropped approx 0.5-1V and varies. R23/R47 does not stabilize even if I turn the trimpots to zero (probably because of the positive temperature coefficient). If I let the output transistors cool down, I can restart the whole process.

Do you have any heatsinks on the output transistors? If not, try installing some. If there is no heatsink on the transistors, it might be difficult to turn the bias up without getting into thermal runaway.

 

[Henmyr]

Yes, I actually have heatsinks. Se this post for pictures:
http://www.head-fi.org/forums/f6/cav...ml#post5737323

Thanks

 

[Henmyr]

I'm going to make a zobel network for my ck2iii. How important is the 22Ω/1W and 0.047µF/63V?

What I already have at home is 22Ω/0.6W(metal film) and 0.1µF/160V(metallized polypropylen), would that do?

From amb.org, the power output seem to be 1.4Wrms into 33Ω, would I endangered the amp if I used 0.6W resistors?

I will be using the amp for 32Ω Grados a lot.

 

[amb]

The zobel network values are recommendations only -- they can be varied based on need (and availability of parts). Typically zobel resistor values are in the 10-22 ohms range, and capacitors are in the 0.047-0.1uF range. I chose to use the "smallest hammer" because often times it is sufficient. As for resistor wattage, you can go ahead and try the 0.6W. If they don't get hot (when playing music at high volumes) then it would be fine.

 

[Cloud]

Regarding connecting the zobel network

do i connect it like this?


Input-

 

[amb]

No, the zobel is only at the output side, there shouldn't be any connection to the inputs. See the "Wiring &ground" section of the β22 website for info (refer to the 2-channel passive ground figure).

 

[Henmyr]

YAY SUCCESS FINALLY. I'm now sitting here after many months and listening to Blackmore's night with my K501 (have not dared to try my more expensive headphones yet) and I have sound in BOTH channels!

I made a zobel network at the output and after that I could tune in the quiescent current.
DC between OL/OR and OG is about 5mV/3mV, so not perfect.
I have some humming when I use my PX100, but it's not cased yet and it instantly got better when I moved it away from my dac and computer monitor.

Thanks for all the help

EDIT: And when I returned to it after having it on for a while the temperature was hot and the current through the output transistors was way to high. Seems like I have some fine tuning before it works like a charm, but it does work somewhat

Also have some popping sound now and then. I have not soldered the input jacks yet and blame it on them.

EDIT2: Had it on playing for a while now. The quiescent current is about 12.5 and 11. DC offset is about 1mV or less. About once every five to fifteen minutes I hear a popping sound. It's not dangerously loud, so I do not think I'm in any danger. I've soldered the input jacks now.

 

[Henmyr]

I've still got the "pop" sound about once every five minutes. I think it's the left channel, but not sure, it doesn't sound like it's about 80% left and 20% right. (or maybe 100% left, hard to tell).

It's not cased up yet.

What could the problem be?

I've measured every point both left and right channel.