I think it keeps things tidier and simpler in general by standing the caps up. You might buy yourself some space and allow mounting the caps standing up by using tall feet (f.ex. from Apex Jr -- good enough for creating an air gap between carpet and bottom panel) and then letting the caps drop through holes in the bottom panel. Or you could put a bulge on the top panel...
This issue is one of height. The case has ~150mm inside and the cap is 153mm screw to bottom. I've thought about cutting holes in the top (I've seen other try such things). The easier plan is to use the other, larger heat sinks, giving me the room for the big caps, and better heat sinking for more bias room.
Afaik, no reason not to mount the caps in any position. (up down sideways)
I have seen and done mounting in all positions. Many many years - no failures due to mounting position.
In my book, not a consideration.
Never heard of "polarity reversal" due to running an electrolytic below the rated voltage, or even a fraction of the rated voltage. Maybe, perhaps there is some sort of effect at very very low voltages with a highish voltage electrolytic - but never heard of it if it exists.
The issue I raised has nothing to do with mounting direction.
As far as the cans being at ground - probably not a good idea. On occasion there can be leakage to the can, best not to short things out. There are no electric fields that I am aware of being radiated by filter caps...
_-_-bear
I have seen and done mounting in all positions. Many many years - no failures due to mounting position.
In my book, not a consideration.
Never heard of "polarity reversal" due to running an electrolytic below the rated voltage, or even a fraction of the rated voltage. Maybe, perhaps there is some sort of effect at very very low voltages with a highish voltage electrolytic - but never heard of it if it exists.
The issue I raised has nothing to do with mounting direction.
As far as the cans being at ground - probably not a good idea. On occasion there can be leakage to the can, best not to short things out. There are no electric fields that I am aware of being radiated by filter caps...
_-_-bear
Last edited:
Hum maybe I shuld not but...
The eletrolitic capacitors are well full of electrolite oily like stuff.
went is there to alowd for air and such to get out when they warm up.
Same types are quite tollerant with the mounting positions some are not.
If they are not and the went is in the wrong place instead of letting air and such out the pressure may force the electrolite to came out.
Spec sheet and manufacturer shuld help.
The eletrolitic capacitors are well full of electrolite oily like stuff.
went is there to alowd for air and such to get out when they warm up.
Same types are quite tollerant with the mounting positions some are not.
If they are not and the went is in the wrong place instead of letting air and such out the pressure may force the electrolite to came out.
Spec sheet and manufacturer shuld help.
A long time ago I thought I was told that the external can of a cap is normally at ground potential (for a single rail supply) and to take advantage of this if possible when laying out a PS to block/deflect other electric fields (bridge, transformer) from pcb and inputs. Maybe that's wrong?
Manufacturer's specs say nothing on mounting position, but it's immaterial at this point, I've already soldered in the smaller caps. I'll save the bigger ones for the larger case I'll need to make to fully push the Cascoded XF5. Additionally, I'll try to fit them into another case for the Cascoded Standard F5 (also nearly completed).
I have two Conrad MF35-151.5 but these will only bias so far. I'm hoping to push this amp later as I add the extra outputs.
The Thrill of Victory - and the Agony of (potential) Defeat
I finally finished the Cascoded XF5 tonight.
It was a wonderful build - designing the PCB layout, designing the case layout, selecting a PS voltage, deciding on the diodes, the NTCs, etching boards, collecting unusual parts, putting it all together.
Don't get me wrong, I could never have built such a wonderfully beautiful amp without a lot of hand holding. Nelson Pass was by far the most helpful, but Tea-Bag, Juma, ZhouFang, Patrick, and many others were essential.
I'm not a big believer in hedging one's bets, I'm a flip the switch and look for smoke kind of guy. So far, I'd never had a problem with a self etched amp, until tonight. I flipped the switch, and started to bias. A few seconds later, the voltage at the output for the channel I was biasing jumped from a few millivolts to 15.5V! I made a move for the power switch, but I was too late. I now know what an MPC74 resistor looks like from the inside. I soldered in a new one, checked all the connections and tried again. This time, I got the same result except I turned off the power before I lost another MPC74. I thought I might be getting a bizarre feedback signal overloading the circuit, so I disconnected the inputs. No luck. I removed the board.
I then turned to the remaining board. Did the circuit work? I was novel by DIYAudio standards - while an obvious mix of some common ideas, I'd never seen this exact circuit built and functioning in the forum. NP had given me the clear path to the design. Was the layout flawed? The execution? I soon found out. The other channel biased easily, and is rock solid at 2 amps bias. It's currently playing music in the background as I type.
How does it sound? Wonderful! Just beautiful! Clean, clear, crystalline highs with solid, clean, fast bass. I'm playing the amp with my OB FF85K over Alpha 15s. The amp was well worth the effort, assuming of course I can fix the faulted channel.
About the faulted channel - does anyone have a quick sense of what could cause such readings? I don't have time to run it down tonight, but tomorrow I'll recheck the board and perhaps change a few parts. What could cause a normal starting bias reading then bring the output to nearly the rail voltage, and overload a source resistor? Any thoughts would be appreciated.
I finally finished the Cascoded XF5 tonight.
It was a wonderful build - designing the PCB layout, designing the case layout, selecting a PS voltage, deciding on the diodes, the NTCs, etching boards, collecting unusual parts, putting it all together.
Don't get me wrong, I could never have built such a wonderfully beautiful amp without a lot of hand holding. Nelson Pass was by far the most helpful, but Tea-Bag, Juma, ZhouFang, Patrick, and many others were essential.
I'm not a big believer in hedging one's bets, I'm a flip the switch and look for smoke kind of guy. So far, I'd never had a problem with a self etched amp, until tonight. I flipped the switch, and started to bias. A few seconds later, the voltage at the output for the channel I was biasing jumped from a few millivolts to 15.5V! I made a move for the power switch, but I was too late. I now know what an MPC74 resistor looks like from the inside. I soldered in a new one, checked all the connections and tried again. This time, I got the same result except I turned off the power before I lost another MPC74. I thought I might be getting a bizarre feedback signal overloading the circuit, so I disconnected the inputs. No luck. I removed the board.
I then turned to the remaining board. Did the circuit work? I was novel by DIYAudio standards - while an obvious mix of some common ideas, I'd never seen this exact circuit built and functioning in the forum. NP had given me the clear path to the design. Was the layout flawed? The execution? I soon found out. The other channel biased easily, and is rock solid at 2 amps bias. It's currently playing music in the background as I type.
How does it sound? Wonderful! Just beautiful! Clean, clear, crystalline highs with solid, clean, fast bass. I'm playing the amp with my OB FF85K over Alpha 15s. The amp was well worth the effort, assuming of course I can fix the faulted channel.
About the faulted channel - does anyone have a quick sense of what could cause such readings? I don't have time to run it down tonight, but tomorrow I'll recheck the board and perhaps change a few parts. What could cause a normal starting bias reading then bring the output to nearly the rail voltage, and overload a source resistor? Any thoughts would be appreciated.
I guess I wasn't clear in my post. The voltage changed before I had the chance to turn a screw. No impatience when biasing, I let the amp settle in for hours while I tweak the bias.
No dummy load and shorted too.
I would be interested in other theories.
Last edited:
without exact schematic , known Idss for input Jfets and Ugs for output mosfets, I can't show you a culprit ;
there is no need for theory - one side was biased enough - to try to fill other ( unbiased) side .
but - you didn't use neither bulb trick nor variac .
you were lucky to loose just MPC
there is no need for theory - one side was biased enough - to try to fill other ( unbiased) side .
but - you didn't use neither bulb trick nor variac .
you were lucky to loose just MPC
I think the issue is getting lost in translation. My question is what could yield a normal bias reading then a near rail voltage reading? The schematic is essentially a balanced XF5 with a BJT cascode at the JFETs (liker Cvillers cascode). NP posted the cascode in my other thread.
I'm not sure I understand how a Variac or bulb would answer this issue. Perhaps it would, but I don't understand how.
I would assume the FET was shorted where the resistor burned.
Unsolder it and test it just using the power supply with no load.
Depends on what you have N or P channel way to hook it up. But a 20-50R Power resistor 10 watts or more is good idea to start, measure VGS for a few seconds. Look up vgs testing of fets if you have not already, some good pointers.
Unsolder it and test it just using the power supply with no load.
Depends on what you have N or P channel way to hook it up. But a 20-50R Power resistor 10 watts or more is good idea to start, measure VGS for a few seconds. Look up vgs testing of fets if you have not already, some good pointers.
I'll try to rephrase what I think was written.
The new damaged channel was not yet biased to ClassA 2A. Switching on the power brought up output offset reading that appeared normal with the input shorted and the output open circuit.
This indicates that when cold the current flowing from the +ve supply feeds into the top source resistor on through the Pchannel FET, then through the Nchannel FET, through the lower source resistor to the -ve supply rail.
The low output offset tells you that the effective resistance of the top half is almost identical to the effective resistance of the lower half.
Then something changes. I'll wager that the FET junctions are starting to heat up above their ambient temperature. The cases of the FET could still be at ambient.
How could the change from a cold junction to a partially warmed up junction cause an effective shorting of the Pchannel FET?
Is it conductive debris in the electrical isolator or a busted junction? What else?
The new damaged channel was not yet biased to ClassA 2A. Switching on the power brought up output offset reading that appeared normal with the input shorted and the output open circuit.
This indicates that when cold the current flowing from the +ve supply feeds into the top source resistor on through the Pchannel FET, then through the Nchannel FET, through the lower source resistor to the -ve supply rail.
The low output offset tells you that the effective resistance of the top half is almost identical to the effective resistance of the lower half.
Then something changes. I'll wager that the FET junctions are starting to heat up above their ambient temperature. The cases of the FET could still be at ambient.
How could the change from a cold junction to a partially warmed up junction cause an effective shorting of the Pchannel FET?
Is it conductive debris in the electrical isolator or a busted junction? What else?
Since one channel works, start looking for a wiring error
in the bad one.
Will do tonight. Thanks again for your input - this would not have been possible without your expertise (in more ways than can be expressed)!
It takes a fair bit of current to fry an output stage. If the N FET resistor is ok and you had no load connected, these aren't the exit paths for the current.The path must have been from +V > resistor > FET, out of metal back of FET > insulator > earthed heatsink. A multimeter will tell you if you have a short from the FET to heatsink. Check the FET mounting hole for burrs.
Human error sucks!
I couldn't imagine I'd made a mistake in layout or stuffing the boards (though I still need to run down a remaining problem - see below), I'd checked both a thousand times while waiting for the toroid to arrive. A few of the posts from last night suggested checking the FETs. Of course, there was the error. At the last minute, I'd decided the wiring might be a little easier if I rotated the order of the MOSFETs on the heat sink (I have the MOSFETs mounted to the sink and wire connectors to the PCBs). I forgot to rotate one pair, and consequently connected the 2SK1530 and the 2SJ201 to the wrong sides of the circuit (wrong polarity, only one pair effected).
I corrected the issue and rewired. Now, 3/4 of the amp biases perfectly, but the effected pair is totally screwy. On the positive rail side, the 2SJ201 seems to bias right (voltage across the source resistor seems to be right), but on the negative rail I cannot get a reading across the source resistor. Is it likely I blew out the 2SK1530? I have not had the time to check part by part (I'll do so tomorrow), but I'd be happy to receive suggestions on likely damage.
I hope it's not the MOSFETs, I do not have another pair on hand, and a replacement my be weeks away.
I couldn't imagine I'd made a mistake in layout or stuffing the boards (though I still need to run down a remaining problem - see below), I'd checked both a thousand times while waiting for the toroid to arrive. A few of the posts from last night suggested checking the FETs. Of course, there was the error. At the last minute, I'd decided the wiring might be a little easier if I rotated the order of the MOSFETs on the heat sink (I have the MOSFETs mounted to the sink and wire connectors to the PCBs). I forgot to rotate one pair, and consequently connected the 2SK1530 and the 2SJ201 to the wrong sides of the circuit (wrong polarity, only one pair effected).
I corrected the issue and rewired. Now, 3/4 of the amp biases perfectly, but the effected pair is totally screwy. On the positive rail side, the 2SJ201 seems to bias right (voltage across the source resistor seems to be right), but on the negative rail I cannot get a reading across the source resistor. Is it likely I blew out the 2SK1530? I have not had the time to check part by part (I'll do so tomorrow), but I'd be happy to receive suggestions on likely damage.
I hope it's not the MOSFETs, I do not have another pair on hand, and a replacement my be weeks away.
NYCOne said:
That'd be my guess. I'd also guess that the MOSFETs are okay. I mean the house didn't burn down or anything, so it doesn't sound like you had excessive current going through the output stage. Did the heat sinks on that channel get excessively hot? Pass's version of F5 also had current limiting.
John
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.