No - the current MUST be limited to about 10mA or less at all times.
Otherwise is risky and dangerous. Capacitors can explode, become
shrapnel, and release dangerous chemicals under pressure.
A 10k series resistor will be very safe in most cases.
Last edited:
I have reformed most of my electrolytics.
I found that reformed very slowly to 100% of rated voltage generally got the capacitor leakage current down to well below the limits set by the manufacturers.
24hours is typical.
Increasing this to 48hours or even 72hours makes very little improvement in reducing leakage current.
I did find that 24hours reforming then a slow discharge through a 1k resistor and then a second slow reform for another 24hours did get a significant improvement in leakage current.
I do not know how permanent that would be, nor do I know what sound quality changes that would bring.
I was just aiming to get the electrolytics to behave as closely as possible to an ideal capacitor.
I found that reformed very slowly to 100% of rated voltage generally got the capacitor leakage current down to well below the limits set by the manufacturers.
24hours is typical.
Increasing this to 48hours or even 72hours makes very little improvement in reducing leakage current.
I did find that 24hours reforming then a slow discharge through a 1k resistor and then a second slow reform for another 24hours did get a significant improvement in leakage current.
I do not know how permanent that would be, nor do I know what sound quality changes that would bring.
I was just aiming to get the electrolytics to behave as closely as possible to an ideal capacitor.
Andrew,
That is excellent news!
Is there anything different from what was described that you did, like resistor value or anything?
The discharge trick it's a great tip too.
There's another thread here about electrolytic capacitors, and the reforming question should be raised there too. In the article I mentioned by Gary Galo, the forming procedure was recommended for new large capacitors, not just for recovering old ones.
I have just seen that this thread was moved to the Parts section. In this case I am not sure it's the right move, because I think this is a matter that affects all amplifiers, as well as receivers and even preamps with large caps.
Perhaps we should ask Cyril Bateman to do some measurements on reformed capacitors and add another part to his series on capacitors.
That is excellent news!
Is there anything different from what was described that you did, like resistor value or anything?
The discharge trick it's a great tip too.
There's another thread here about electrolytic capacitors, and the reforming question should be raised there too. In the article I mentioned by Gary Galo, the forming procedure was recommended for new large capacitors, not just for recovering old ones.
I have just seen that this thread was moved to the Parts section. In this case I am not sure it's the right move, because I think this is a matter that affects all amplifiers, as well as receivers and even preamps with large caps.
Perhaps we should ask Cyril Bateman to do some measurements on reformed capacitors and add another part to his series on capacitors.
It's interesting to monitor the leakage current as you increase the voltage if you can. I did this, when the current stopped reducing I raised the voltage a bit then waited for it to stop reducing and raised it again etc.... When you think about the mechanics of what is happening inside the capacitor (reforming the aluminium oxide layer), this seems a logical way to proceed. https://www.epcos-china.com/downloa...62122e90c/pdf-generaltechnicalinformation.pdf
For the few electrolytic manufacturers that I have looked at, they ALL specify that before testing any parameters they require that the capacitor be freshly reformed.
I have posted this manufacturer recommendation many times on this Forum.
Last edited:
the standard formulae given by the electrolytic manufacturers gives a straight line relationship between leakage and applied voltage.
That is only an approximation.
You will find that the leakage increases exponentially as voltage is increased and becomes quite bad as the rated voltage is exceeded.
You can easily check this yourself.
Reform to 50% of rated voltage. leave it to soak for 24hours and measure the leakage current. The regulated voltage needs to be held VERY accurately otherwise discharge and recharge currents dominate the current flows and obliterate the tiny leakage current. Even the reforming current (the chemical process) can be bigger than the leakage current.
Then repeat the exercise at 100% and 120% of rated voltage.
The leakage current shape behaves a bit like a near saturated iron cored inductor.
Charging, and exceeding, the rated voltage to 120% shouldn't be dangerous and the cap explode?
Reform it to 50% and leave it there for 24 hours should improve the results for reforming the capacitors?
feed a regulated voltage through at least 10k to the capacitor.
At first the Vdop will be high indicating that the current is high.
Let's suppose you have 63Vdc for a 63V rated capacitor.
The initial current flow will be 63V/10k = 6.3mAdc
That current will fall as the capacitor charge builds towards the final voltage.
The current through that resistor will be the sum of { the charging current + the reforming current + the leakage current }
When the capacitor has charged to the final voltage the charging current can be zero uA if the regulated voltage does not change.
That leaves the reforming current + leakage current as the current passing the limiting/measuring resistor.
You need to wait until the reforming current drops to zero uA. That takes a long time. That's why I start with nearer a 100k resistor, so that I can measure the Vdrop at these tiny uA level currents. Just attaching a 10Mohm DMM across the Vdrop resistor changes the voltage applied to the capacitor. Think about how you can compensate for that.
At first the Vdop will be high indicating that the current is high.
Let's suppose you have 63Vdc for a 63V rated capacitor.
The initial current flow will be 63V/10k = 6.3mAdc
That current will fall as the capacitor charge builds towards the final voltage.
The current through that resistor will be the sum of { the charging current + the reforming current + the leakage current }
When the capacitor has charged to the final voltage the charging current can be zero uA if the regulated voltage does not change.
That leaves the reforming current + leakage current as the current passing the limiting/measuring resistor.
You need to wait until the reforming current drops to zero uA. That takes a long time. That's why I start with nearer a 100k resistor, so that I can measure the Vdrop at these tiny uA level currents. Just attaching a 10Mohm DMM across the Vdrop resistor changes the voltage applied to the capacitor. Think about how you can compensate for that.
The current will be down below 10uA. The power needing to be dissipated cannot be more than 10uA times the applied voltage. How hot would that get?
the capacitor continues to reform during the extended soak time. During the 23hours since the cap reached near final voltage the total current into the capacitor is reforming current plus leakage current. You have to continue reforming until the reforming current falls below the leakage current.
I usually reform a bank of capacitors at one time.
Use a star of eight 51k resistors. 8 soldered together allow one 51k to be connected to each of eight capacitors. Then feed the regulated voltage via another 51k to the centre of that star.
You can monitor the total current and you can monitor the difference between the capacitors. They will not be drawing the same current. Expect a ratio of 2:1 across 8 capacitors. Some of that difference is due to different capacitance requiring different charge current. Some is due to different reform current (not sure why?). Some is due to different leakage current. Ocassionally you will find one cap that is much worse than the others. You may want to try to save this cap, but I found the best place was the bin, if leakage was anywhere near the top limit of that manufacturer.
If after a second 24hour reform you then disconnect the voltage supply and short the +ve lead to the -ve lead with a 100k resistor (that puts 151k in series with the individual resistors) you will see very different Vdrops. Some are positive and some are negative.
Capacitors with a high leakage continue to draw current while capacitors with low leakage have become the source of current for the others. Again you will see quite big differences in current through the capacitors. Again attaching a 10Mohm DMM will corrupt every measurement you make.
Last edited:
An ammeter will not resolve the small currents you are trying to differentiate between.
A 10k resistor with a 1.1mVdrop is showing a current of 110nA +DMM effect (~0.11nA).
A 100k resistor allows resolution to lower currents but increases the DMM effect (now you have 1.1mV = 11nA + 1.1nA).
you can reform using a Variac. You cannot compare, nor measure leakage current using a Variac.
a 100mA regulator with an accurate tempco will do very well.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.