Fault finding, BCL Clone, help needed.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
A while ago I replaced the OPA627's in my headamp with a "discrete opamp".
I can't share the schematic of it since it is not my design.

Don't know if that has anything at all to do with what I'm looking for help with now.
A while ago I noticed that two caps were "swollen".
I stopped using the headamp until about two days ago, when I finally got around to swapping the row of 6 caps for Pana FC's 1000uF/35V.
Also removed the input caps as my dac has Edcor transformers on the outputs and thus no DC.
Swapped the "discrete opamp" to an opamp.
Sounds great as always, but I noticed within 15min or so that the same capacitors that were swollen were quite a bit more hot then the others.
Turned the headamp off...and that's were I am now.
Removed marked input caps (soldered jumpers in their place) and all of the 1000uF/25V.
Marked the swollen(now hot) ones.
An externally hosted image should be here but it was not working when we last tested it.


Any ideas what could be the problem?
This headamp had worked for years now, and I really like it.

Any suggestions for a headamp that is easy to build with parts I have (so I have something until I get the clone sorted)?
I have a bunch of opamps, tightly matched quad of 2sk170's, four BD140's, four BD139's, a bunch of BC547's and BC557's.
I also have 2 or 3 12-0-12VAC 1500mA transformers, and maybe a 9VAC transformer.
 
Last edited:
Three things that will make electrolytics toasty and swollen are being installed with reverse polarity, more than the rated voltage, or too much AC ripple riding on top of the DC. Plus there is just old age, with the electrolyte drying up, but since your new caps are also getting hot I'm assuming a circuit problem is lurking somewhere. i'm also assuming these electros are being used as power rail filters rather than in the signal path for coupling.

Since your originals were swollen that pretty much rules out reverse polarity, assuming you are sure you got the new ones in with the same polarity as the old ones.

As for voltage, you really need to measure it. Right across the capacitors that are getting hot, from their leads on the bottom of the PC board. See if what you measure is exceeding (or even getting within 20%) of their voltage rating. If you can't get to the underside of the board while the amp is running just temporarily soldering a couple of small wires across the cap to bring the signal out for testing is handy.

As for ripple, put your DMM on the AC setting and do the same measurement across the capacitor leads.

If the voltage is too high and/or your have more than 4 volts or so of ripple then one of your regulator chips may have died (shorted).
 
Last edited:
Three things that will make electrolytics toasty and swollen are being installed with reverse polarity, more than the rated voltage, or too much AC ripple riding on top of the DC. Plus there is just old age, with the electrolyte drying up, but since your new caps are also getting hot I'm assuming a circuit problem is lurking somewhere. i'm also assuming these electros are being used as power rail filters rather than in the signal path for coupling.

Since your originals were swollen that pretty much rules out reverse polarity, assuming you are sure you got the new ones in with the same polarity as the old ones.

As for voltage, you really need to measure it. Right across the capacitors that are getting hot, from their leads on the bottom of the PC board. See if what you measure is exceeding (or even getting within 20%) of their voltage rating. If you can't get to the underside of the board while the amp is running just temporarily soldering a couple of small wires across the cap to bring the signal out for testing is handy.

As for ripple, put your DMM on the AC setting and do the same measurement across the capacitor leads.

If the voltage is too high and/or your have more than 4 volts or so of ripple then one of your regulator chips may have died (shorted).
Thanks for the reply :)
Fiddled around a bit with a pic of the underside of the PCB in paint:
An externally hosted image should be here but it was not working when we last tested it.

Swollen caps are in the black rectangle.
 
Hey that photo helps! Yeah those two caps are power rail filters/bypass on the collectors of the BD139/BD140 push-pull pairs, likely fed by those LM317/337 regulators on the left.

Hmmm. Those two caps may get a little warm normally just from heat conduction via the traces from the regulators and those output transistors, but should not be hot. One thing to check is how hot the BD139/BD140 transistors are getting with the amp on but no headphones attached and no input connected. If something has gone wrong with the bias circuit they may be getting quite hot with excessive idle bias current through them, then that heat making it over to the caps.

Also some voltage readings on the positive and negative rail would be good to see where the regulated voltages sit.

It is interesting though that the two getting hot are on different channels. The top cap in the black box is on the collector of a BD140 on one channel, while the bottom one in the box is on the collector of a BD139 on the other channel. But they are two closest to where the voltage regulator leads attached. May be good to check how hot the voltage regulators are getting, too.
 
Last edited:
All the two caps have in common is, well, common - a.k.a. a ground connection. This should be at the pins not connected to a trace at the bottom.

Looking at the two big 6800µs on the left, see that trace that connects two of their pins? That should be power ground.

There should be continuity between those two points, which would have to be verified. A closer look at the top should reveal where ground is supposed to be be running, and maybe the cause of the interruption as well.

That's a pretty nicely-made board btw. The vias tying input ground to the top ground ground plane indicate that the designer had some experience with RF stuff.
 
All the two caps have in common is, well, common - a.k.a. a ground connection. This should be at the pins not connected to a trace at the bottom.

Looking at the two big 6800µs on the left, see that trace that connects two of their pins? That should be power ground.

There should be continuity between those two points, which would have to be verified. A closer look at the top should reveal where ground is supposed to be be running, and maybe the cause of the interruption as well.

That's a pretty nicely-made board btw. The vias tying input ground to the top ground ground plane indicate that the designer had some experience with RF stuff.
Borrowed this pic from an ebay listing
An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
One thing I consider is to get a good low noise Vreg(dual rail, and that works with my 2x15VAC toroid transformer).
And do this:
An externally hosted image should be here but it was not working when we last tested it.


I don't need the input caps as the DAC feeding the headamp has edcor transformers on the outputs...so zero Vdc to filter.
Since my "discrete opamp" is biased to class A and so is the diamond buffer, I don't know the total current draw of the headamp. That info is useful when choosing the Vreg.

The points I marked for connecting Vreg -> headamp are wrong, just realized that. But you get the idea.
 
AC - VCC+ to GND: 0,012V
AC - VCC- to GND: 0,004V

Whoops, I hadn't noticed that you posted your measurements. :) Well those are not bad at all. 12mVac and 4mVac. The ripple is most likely going to be a sawtooth wave, so this average will be skewed toward the bottom 2/3 or so of the total waveshape, whether a RMS or average reading meter. Expanding that back out a bit to estimate peak to peak values give numbers that are still not that bad. At the very least it says that your regulator chips are regulating (and not shorted).

Hmmm.... One other possibility comes to mind, espeically given your discrete op amp. If the circuit was oscillating at a higher-than-audio frequency the output impandance looking back into those regulator chips will be fairly high. That would leave these rail filter/bypass caps as the lower-impedance path. Your meter may not be responsive at higher frequenies either. Many cut off at 400Hz or 1kHz.

Oscillation can happen with op-amps powering IC buffer chips. I'm not sure how much of an issue that will be with your op-amp plus discrete circuit through. Plus I think you said that you replaced the discrete op amps with an op-amp again (hopefully the same one that was originally in it). I would also expect the probablem to occur with all the rail filter electrolytics too, not just those two near the regulators.

So I'm kind of stumped at this point! Maybe folks on the forum here like jcx and sgrossklass who are better with this stuff than me will have some more ideas for you. :)
 
Last edited:
Whoops, I hadn't noticed that you posted your measurements. :) Well those are not bad at all. 12mVac and 4mVac. The ripple is most likely going to be a sawtooth wave, so this average will be skewed toward the bottom 2/3 or so of the total waveshape, whether a RMS or average reading meter. Expanding that back out a bit to estimate peak to peak values give numbers that are still not that bad. At the very least it says that your regulator chips are regulating (and not shorted).

Hmmm.... One other possibility comes to mind, espeically given your discrete op amp. If the circuit was oscillating at a higher-than-audio frequency the output impandance looking back into those regulator chips will be fairly high. That would leave these rail filter/bypass caps as the lower-impedance path. Your meter may not be responsive at higher frequenies either. Many cut off at 400Hz or 1kHz.

Oscillation can happen with op-amps powering IC buffer chips. I'm not sure how much of an issue that will be with your op-amp plus discrete circuit through. Plus I think you said that you replaced the discrete op amps with an op-amp again (hopefully the same one that was originally in it). I would also expect the probablem to occur with all the rail filter electrolytics too, not just those two near the regulators.

So I'm kind of stumped at this point! Maybe folks on the forum here like jcx and sgrossklass who are better with this stuff than me will have some more ideas for you. :)
I really don't know what to make of it all tbh.
Waiting for a few parts, then I'll veroboard a dual rail LT1083CP Vreg board, de-populate the PSU part of the headamp and hook the Vreg-board up to the headamp. Atleast that should give sonic properties.
The opamp I put in is a LM4562 (headamp came with 5532).
Who knows, the capacitors that got swollen might have been hot before they got swollen and I just didn't notice.
Both DC and AC was measured with nothing on inputs and nothing on outputs.
Worst case I have to get a bare PCB (identical are available on ebay) and populate a new headamp using mostly new parts(I have the BD139´s and 140's, would have to get BC550's and 560's).
 
Lehmann linear clones (and the original too, probably) will oscillate badly with fast op amps, the original is designed around the OPA2134 and so far only jfet ones don't have oscillation signs and acceptable DC offset in my build.

Working opamps:

OPA2132, 2134, 2140, 1642, 1652, 2107, 827

Not working opamps (high DC offset 'till 0.5V and very hot opamps):

OPA627, ADA4627, LME49990, AD797

Borderline (DC Offset under 100mV and hot opamps):

NE5532, LM4562, LME49720, 49960, OPA2209

The best sounding one among working opamps is the OPA1652.

All resistors around transistors should be tightly matched for optimal DC offset, transistors matched for Vbe and HFE also helps a lot.

Also, the power supply can be greatly enhanced on board with 2 zeners and 2 22uF caps.

If you're interested I'll post more about it later.
 
PS mod

attachment.php
 

Attachments

  • IMG_20150807_051027.jpg
    IMG_20150807_051027.jpg
    213 KB · Views: 932
Lehmann linear clones (and the original too, probably) will oscillate badly with fast op amps, the original is designed around the OPA2134 and so far only jfet ones don't have oscillation signs and acceptable DC offset in my build.

Working opamps:

OPA2132, 2134, 2140, 1642, 1652, 2107, 827

Not working opamps (high DC offset 'till 0.5V and very hot opamps):

OPA627, ADA4627, LME49990, AD797

Borderline (DC Offset under 100mV and hot opamps):

NE5532, LM4562, LME49720, 49960, OPA2209

The best sounding one among working opamps is the OPA1652.

All resistors around transistors should be tightly matched for optimal DC offset, transistors matched for Vbe and HFE also helps a lot.

Also, the power supply can be greatly enhanced on board with 2 zeners and 2 22uF caps.

If you're interested I'll post more about it later.

Thanks for your reply :)
I'll remove all PSU components and run the bcl clone of a dual rail LT1083 Vreg board.
When I get it up and running again I'll try it with opa2134.
I've got a small stash of opamps.
The discrete opamp I built (designed by a very knowledgeable guy who designed it with the schematic of the clone) I think is ok to use as well.

Have used NE5532(it came with that), LM4562 and OPA627's on an adapter.
I've had it for at least 5 years (the bcl clone) and it has never been any problems with it until recently.
 
For the BCL Clone, as I wrote, I'll use a LT1083 dual rail reg.
Started removing PSU parts on the PCB.
An externally hosted image should be here but it was not working when we last tested it.


For a JHL headamp kit that is on it's way I was thinking about doing a shunt reg.
This schematic seems simple, but I need higher current and 12VCC out (12-012VAC transformer).
ar.discrete.reg3.gif

From this site

It seems changing R3/R6 adjusts VCC out?
If I'm right 8K2 would give very close to 12V.
Also Q2 and Q3 would need to be replaced to handle more current?
Is D44H11 and D45H11 viable options?

Might be asking in the wrong section, if so, I'm sorry.
 
Last edited:
For the BCL Clone, as I wrote, I'll use a LT1083 dual rail reg.

Since you will have to use a transformer with isolated secondaries anyway to use a positive regulator for both power rails, you might want to take a look at the newer (and quieter and cheaper) TPS7A3301 negative regulator. It comes in TO-220. opc has board with it now, see here

http://www.diyaudio.com/forums/vend...jects-available-here-bal-bal-se-se-lpuhp.html (3rd one down)


If I'm right 8K2 would give very close to 12V.
Also Q2 and Q3 would need to be replaced to handle more current?
Is D44H11 and D45H11 viable options?

That's right, 8K2 for both (be sure to use a red LED and not any other color!), although I would suggest using 7.5K in series with a 2K 10-turn trimpot. That would give you adjustment range from about 9V to around 13V. Even better use a 7.5K in series with a 470R and in series with a 500R 10-turn trimpot. That would give you a range of around 11.4Vdc to 12.6Vdc.

The D4XH11 would work. :) At the currents you will be using their DC current gain is high, around 150 or so.
 
Last edited:
Since you will have to use a transformer with isolated secondaries anyway to use a positive regulator for both power rails, you might want to take a look at the newer (and quieter and cheaper) TPS7A3301 negative regulator. It comes in TO-220. opc has board with it now, see here

http://www.diyaudio.com/forums/vend...jects-available-here-bal-bal-se-se-lpuhp.html (3rd one down)
Thank you for the tip :)
I've got the parts for the LT1083 reg on the way already though. I'll veroboard it as it is simple enough.



That's right, 8K2 for both (be sure to use a red LED and not any other color!), although I would suggest using 7.5K in series with a 2K 10-turn trimpot. That would give you adjustment range from about 9V to around 13V. Even better use a 7.5K in series with a 470R and in series with a 500R 10-turn trimpot. That would give you a range of around 11.4Vdc to 12.6Vdc.

The D4XH11 would work. :) At the currents you will be using their DC current gain is high, around 150 or so.
Thanks again for the advice :)
I'll look at different resistor/trimpot combos, I might even have something suitable at home already.
I'm embarrassingly green at this discrete design thing...
Would you mind explaining the last paragraph of your post?

Just to clarify my idea of the PSU for the JHL headamp:
12-0-12VAC - > rect. bridges (1N4007's) - > capacitor banks - > shunt reg - > JHL.

Writing this on my phone, but hopefully it's understandable in spite of that ;)
 
Would you mind explaining the last paragraph of your post?

Just to clarify my idea of the PSU for the JHL headamp:
12-0-12VAC - > rect. bridges (1N4007's) - > capacitor banks - > shunt reg - > JHL.

A lot of power transistors have fairly low DC current gain, like 10x or 20x, which can suck too much drive current out of whatever is powering them, often requiring another transistor just to drive the power transistor. But the DHX54's are an exception with fairly high DC current gains, right up there with the small signal BC558/BC548 they originally specified.

For your shunt regulators that chain of parts is correct.

For your LT voltage regulators though the big trick there, when using a positive regulator chip on each rail (along with using the isolated secondary transformer) is to tie the grounds together *after* the regulators. So for each power rail you would have isolated secondary -> rectifier -> filter caps -> regulator chip. Then the output of regulator #1 becomes your V+. The ground of regulator #1 and the output of regulator #2 tie together to become your system ground (that is why the xformer secondaries have to be isolated), and then the ground of regulator #2 becomes your V- rail.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.