The positive half wave protection transistor needs 4times the base current to kill...
started reforming the elctrolytics yesterday.
Did seven 25V varying from 22uF to 470uF on a slow reform and then left them to soak overnight.
This morning the total leakage current is showing as 0.8µA for the seven caps with a total capacitance of 1446uF
The average leakage comes out at 0.12uA per capacitor.
The Ileak = "X" *C*V comes out at X= 0.000022 (yes, 4 zeros after the decimal place).
Giving Ileak = 0.000022CV
That is equivalent to ~ 212Mohms of parallel leakage resistance across the plates.
With 17 electros per PCB, I have only 27 more to slow reform $days to do one reform @ 24hrs perm, (but I find that leakage improves with a second reform after a slow discharge). That means about 9days of reforming before I can complete the PCBs.
In the meantime I can select all the semiconductors and fit them.
Did seven 25V varying from 22uF to 470uF on a slow reform and then left them to soak overnight.
This morning the total leakage current is showing as 0.8µA for the seven caps with a total capacitance of 1446uF
The average leakage comes out at 0.12uA per capacitor.
The Ileak = "X" *C*V comes out at X= 0.000022 (yes, 4 zeros after the decimal place).
Giving Ileak = 0.000022CV
That is equivalent to ~ 212Mohms of parallel leakage resistance across the plates.
With 17 electros per PCB, I have only 27 more to slow reform $days to do one reform @ 24hrs perm, (but I find that leakage improves with a second reform after a slow discharge). That means about 9days of reforming before I can complete the PCBs.
In the meantime I can select all the semiconductors and fit them.
Last edited:
Reforming the insulating layer on the plate/s to turn the aluminium foil from a resistor into a capacitor.
Aluminium foil electrolytic capacitors degrade in storage after they are manufacturered.
All the manufacturers tell us that their electrolytics must be reformed shortly before any testing.
What's the biggest test we put our electrolytics to?
We listen to the signal passing through them and assess their quality.
The test does not get much more onerous that our listening.
An unreformed capacitor leaks a lot of current. The capacitor plates slowly reform after you apply a DC voltage across the electrolytics. The leakage current decreases as the in circuit reforming is progressing. I have surmised that the many Builders who report "burn in" are actually listening to a gradual reduction in these leakage currents affecting the way their amplifiers behave, or misbehave, when leakage is high.
Higher voltage capacitors can leak so badly at first power on that they blow up due to excessive heating as a result of passing that high leakage current.
Aluminium foil electrolytic capacitors degrade in storage after they are manufacturered.
All the manufacturers tell us that their electrolytics must be reformed shortly before any testing.
What's the biggest test we put our electrolytics to?
We listen to the signal passing through them and assess their quality.
The test does not get much more onerous that our listening.
An unreformed capacitor leaks a lot of current. The capacitor plates slowly reform after you apply a DC voltage across the electrolytics. The leakage current decreases as the in circuit reforming is progressing. I have surmised that the many Builders who report "burn in" are actually listening to a gradual reduction in these leakage currents affecting the way their amplifiers behave, or misbehave, when leakage is high.
Higher voltage capacitors can leak so badly at first power on that they blow up due to excessive heating as a result of passing that high leakage current.
Last edited:
Never done any cap "reforming" and never had any problems.
If a electrolytic is very old it would be good to test it before attaching full voltage via current limited power supply.
All pcb's are tested before mounting in the case using the test power supply with activated current limit. Even power supply pcb's with the big lytics.
If a electrolytic is very old it would be good to test it before attaching full voltage via current limited power supply.
All pcb's are tested before mounting in the case using the test power supply with activated current limit. Even power supply pcb's with the big lytics.
IIRC the oxide layer in the caps is constantly degrading and reforming in a somewhat self-perpetuating cycle, where clearing a defect can lead to other defects. In an array of caps the energy available to clear a defect is multiplied and can cause large cap banks to fail for no obvious reason. I forget the thread where this was uncovered but IIRC reforming the caps was not a solution.
Some background information to "capacitor reforming":
Modern caps mostly do not need reforming as they have already a thick anodized layer. Reforming is only needed:
http://en.tdk.eu/blob/530704/download/4/pdf-generaltechnicalinformation.pdf
BR, Toni
Modern caps mostly do not need reforming as they have already a thick anodized layer. Reforming is only needed:
- if you want to precisely measure the leakage current after long storage cycles.
- if your application can't handle higher leakage currents (then better use MKS or MKP caps)
http://en.tdk.eu/blob/530704/download/4/pdf-generaltechnicalinformation.pdf
BR, Toni
Interesting. I have experienced burn in reducing measured distortion levels at the speaker via microphone measurements. On several brand new - or newly built amps - I could see the levels fall before my eyes over minutes as they amps were swept through the measurements. These were rather high levels of distortion approaching 1% and then falling to less than speaker distortion (-55dB). Always wondered what could have been causing it. So we should apply DC and soak it? Say turn amp on and leave on overnight before testing?
Seems to be a bad design heavy dependant on low leakage lytics or you are using microphones with built in tubes. All my blameless derivated designs have no problems shown so far even on using old lytics...
Best to use the tested drivers. Mouser and me have many on stock ...
Think it may technically work with both variants:
- ksa1220/c2690 have reverse pin numbering, SOA may be ok
- 2sa1535/c3944 looks to be a good replacement, unfortunately also discontinued