I’ve been rebuilding a few tube radios that I own. I realize that our line voltage is a bit higher than it was back in the day, so I am installing a thermistor on each of them to bring it down a few volts and I’m even open to adding a large resistor right before the primary winding to bring it down a few more.
I know that many of the capacitors from back in the day had both their working voltage and a peak voltage stamped on the can. I replaced the caps in this particular radio were rated for 400v and 450v. I replaced them all with 450v. I measured them on power up and the voltage shots up to 480-490v on all three caps and then settle at about 380v on the original 450v cap and 350v on the two caps what were 400v rated caps. So that means the two 400v caps were seeing nearly 20-25% higher than rated voltage, granted only for a few seconds. These originals don’t have a peak voltage printed on them.
I looked on the spec sheet of my replacements and couldn’t find anything on peak voltage. I’m using Nichicon CA, and another very good quality Nichicon along with good quality Panasonic, all 105 c of course. I haven’t run into an issue yet, but I was just curious as to how high these newer caps can go. If I were to swap the 400v rated caps with 400v rated, could the new 400v caps see that 480-490v for that several second duration like the vintage caps could?
Dan
I know that many of the capacitors from back in the day had both their working voltage and a peak voltage stamped on the can. I replaced the caps in this particular radio were rated for 400v and 450v. I replaced them all with 450v. I measured them on power up and the voltage shots up to 480-490v on all three caps and then settle at about 380v on the original 450v cap and 350v on the two caps what were 400v rated caps. So that means the two 400v caps were seeing nearly 20-25% higher than rated voltage, granted only for a few seconds. These originals don’t have a peak voltage printed on them.
I looked on the spec sheet of my replacements and couldn’t find anything on peak voltage. I’m using Nichicon CA, and another very good quality Nichicon along with good quality Panasonic, all 105 c of course. I haven’t run into an issue yet, but I was just curious as to how high these newer caps can go. If I were to swap the 400v rated caps with 400v rated, could the new 400v caps see that 480-490v for that several second duration like the vintage caps could?
Dan
No, not hot, but warm and all of the power transformers run warm. But this is the same with all of the tube radios and amps I’ve been doing. The Fisher 100 amp that had 500v rated caps. On power up the voltage went to 550-560v I believe and then one the tubes started their thing the voltage dropped to like 480v. Every single one of my radios high voltage goes high and then settles, same with every tube device I’ve seen worked on on YouTube (d lab, blue glow electronics, etc). I was told that this behavior is normal.
Dan
Did you change the value on the caps? I like to keep a relatively low value on the first cap in the pi-filter. It seems the inrush voltage peek can be tamed that way. Also, replacing the resistor with an inexpensive 1-2H inductor in the pi-filter can do wonders (Triad Magnetics C-24X and C-17X are good candidates). I always model my power supplies in PSUDII and try and eliminate the inrush peak.
The surge voltage is standardized in IEC/EN 60384-1. For aluminum electrolytic capacitors with a rated voltage of up to 315 V, the surge voltage is 1.15 times the rated voltage, and for capacitors with a rated voltage exceeding 315 V, the surge voltage is 1.10 times the rated voltage.
The surge voltage is the maximum voltage which may be applied to the capacitor for short periods
of time, i.e. up to 5 times for 1 minute each per hour. Surge voltage testing is conducted according
to IEC 60384-4.
some useful information about al elcos from TDK (EPCOS) can be found here: https://www.tdk-electronics.tdk.com...b0cff75f8/pdf-generaltechnicalinformation.pdf
The surge voltage is the maximum voltage which may be applied to the capacitor for short periods
of time, i.e. up to 5 times for 1 minute each per hour. Surge voltage testing is conducted according
to IEC 60384-4.
some useful information about al elcos from TDK (EPCOS) can be found here: https://www.tdk-electronics.tdk.com...b0cff75f8/pdf-generaltechnicalinformation.pdf
Solid state rectifiers conduct instantly, directly heated rectifiers conduct very quickly, much faster than the circuit begins to draw curent, so voltage is much higher until warmup. Inrush limiters, resistors, chokes, reducing capacitance won't change this a bit, as no current is being drawn until power tube(s) warm up. Indirectly heated rectifiers have a warmup time similar to the other tubes, so will reduce voltage overshoot. Otherwise, stay within surge ratings as listed above, or take your chances.
With enough series resistance, electrolytics further down the filter chain will limit the the overvoltage because of their leakage (limited by the series resistance). This occurs even with higher-voltage rated caps, since they don't see rated voltage for a long enough period to completely re-form. Can't count on this with film caps - they'll charge to the full upstream voltage (no leakage), so they should be rated same as input caps.
With enough series resistance, electrolytics further down the filter chain will limit the the overvoltage because of their leakage (limited by the series resistance). This occurs even with higher-voltage rated caps, since they don't see rated voltage for a long enough period to completely re-form. Can't count on this with film caps - they'll charge to the full upstream voltage (no leakage), so they should be rated same as input caps.
Without a schematic of a particular radio, and its original line voltage rating, there is not much to determine what voltage rating to use.
Of course that is too much work, and is not necessary.
Instead, use the following generalization, and following rule:
Generalization: Most modern electrolytic capacitors are smaller than the old electrolytic capacitors of the same voltage and capacitance ratings.
A higher voltage rated capacitor may fit in.
Rule: Be conservative and reliability minded; increase the replacement capacitor's voltage rating, versus the voltage rating of the old capacitor that you are replacing.
Example: You might be able to fit a modern 500V electrolytic where the old 400V electrolytic was.
Just an idea for consideration.
Of course that is too much work, and is not necessary.
Instead, use the following generalization, and following rule:
Generalization: Most modern electrolytic capacitors are smaller than the old electrolytic capacitors of the same voltage and capacitance ratings.
A higher voltage rated capacitor may fit in.
Rule: Be conservative and reliability minded; increase the replacement capacitor's voltage rating, versus the voltage rating of the old capacitor that you are replacing.
Example: You might be able to fit a modern 500V electrolytic where the old 400V electrolytic was.
Just an idea for consideration.
I often buy "mystery brand" caps and always measure them.
I am talking SS amp values, so think 2200/4700uF 35-40-50-63V typically, scale up for tube voltages.
I have a 100V supply, and charge them through a 10k resistor, so injecting around 4-6 mA, and let them charge as long as needed to get a stable value, a "Zener capacitor" value if you wish, what actually means losses balance injected current.
It does not take long, just a few minutes.
I take note of that value and call it "absolute maximum surge voltage", which in any case is derated by 10-15%
Then I disconnect cap and let it self discharge through its own internal losses; it will do it faster at the beginning and slower after a minute or two which means losses are negligible.
This second value is my "working voltage" which again I derate by 10-15%, not becuse of the capacitor itself (I trust measured values over datasheets any day of the week) but to cover against mains line variations.
It has worked very well for me, allowing use of "second source" suppliers, wider availabiity, huge savings ... and still I can trust values since they are measured.
Since I make Guitar amps commercially, I can NOT "just buy Nichicon - Panasonic - etc. and call it a day" which some might suggest to play it safe; that is fully acceptable and desirable for home/Hi Fi/boutique builds of course, bot not in a cutthroat competitive market.
Sample measurements?
Here:
EPCOS on top, Suntan on bottom, both nominal 4700uF x 63V
4700/63 TRec on top, 2200/63 Epcos bottom.
Are Suntan unusable junk?
Not at all, they simply can be trusted up to 55V Working Voltage, period.
Again, talking about competitive Commercial products.
I am talking SS amp values, so think 2200/4700uF 35-40-50-63V typically, scale up for tube voltages.
I have a 100V supply, and charge them through a 10k resistor, so injecting around 4-6 mA, and let them charge as long as needed to get a stable value, a "Zener capacitor" value if you wish, what actually means losses balance injected current.
It does not take long, just a few minutes.
I take note of that value and call it "absolute maximum surge voltage", which in any case is derated by 10-15%
Then I disconnect cap and let it self discharge through its own internal losses; it will do it faster at the beginning and slower after a minute or two which means losses are negligible.
This second value is my "working voltage" which again I derate by 10-15%, not becuse of the capacitor itself (I trust measured values over datasheets any day of the week) but to cover against mains line variations.
It has worked very well for me, allowing use of "second source" suppliers, wider availabiity, huge savings ... and still I can trust values since they are measured.
Since I make Guitar amps commercially, I can NOT "just buy Nichicon - Panasonic - etc. and call it a day" which some might suggest to play it safe; that is fully acceptable and desirable for home/Hi Fi/boutique builds of course, bot not in a cutthroat competitive market.
Sample measurements?
Here:
EPCOS on top, Suntan on bottom, both nominal 4700uF x 63V
4700/63 TRec on top, 2200/63 Epcos bottom.
Are Suntan unusable junk?
Not at all, they simply can be trusted up to 55V Working Voltage, period.
Again, talking about competitive Commercial products.
Last edited:
On SS stuff it’s easy to over spec the voltage on caps but on tube stuff it’s hard and expensive to go over 450v on most electrolytic caps. On name brand quality caps the surge rating is usually listed in the specs and can be trusted. On unknown junk brands I don’t trust them at rated voltage. I’ve had too many bad experiences of caps well within ratings fail spectacularly. If you have room for film caps, especially the first after the rectifier, use one there. Also if it gets hot under the chassis the cap will eventually fail on over voltage even if it took it at first. Buy name brands and use film if possible, it will save you grief later.
If the radios are old enough, the electrolytic capacitors might need replacing.
If you have the equipment, you might test the radio's original capacitors using JMFahey's method.
But Ouch, you have to test them at 400V or 450V.
Let us know if any of the radio's original capacitors pass JMFahey's tests.
I bet that most do not pass.
If you have the equipment, you might test the radio's original capacitors using JMFahey's method.
But Ouch, you have to test them at 400V or 450V.
Let us know if any of the radio's original capacitors pass JMFahey's tests.
I bet that most do not pass.
No not really, just by modern standards I guess. The originals were 30uf @ 450v and the other two 20 uF @ 400v, so I went with a 33uF and two 22uF. I’d I run into a 50uF I’ll use 47uF.
That’s seems intriguing, can you show me with an example of a schematic, the change you make. What value of inductor?
Dan
Yeah, that’s what I’ve read over the years of restoring these radios, I just had no idea how much higher over the rated voltage of the caps they went. I know this has partly to do with elevated line voltages which is why I’m trying to lower it a bit by using the thermistor and of course still open to adding a resistor to the primary winding. Doing the math I figured on maybe a 10w 4-5 ohm resistor. With an input of 121 Vac the thermistor drops it to about 118, then figure with the resistor I could drop it to about 114 or so with the resistor.
Awesome info, thank you for supplying that. That’s what I was looking for. Well I’m definitely exceeding the voltage by 1.15 the rated voltage, especially on the 400v rated caps since all three caps all peak at the same voltage. Though nowhere near for 1 minute durations and 5 times an hour. We’re talking 5-6 seconds total maybe before it declines to the safe voltages. I know they don’t spec it, but I wonder if they could handle 1.2 to 1.25 times the spec’d voltage for under 10 seconds at a time. I don’t imagine any of these radios would be power cycled more than 5 times in an hour.
I have a few values on hand rated at 500v, they’re all Nichicon and like you say they weren’t cheap. I guess my options are to either add some 500v rated caps to my next order or the less expensive route is to reduce the line voltage even more by using resistors. If I’m remembering my math correctly I want to say by using a 4-5 ohm resistor that it would be dissipating about 5 watts of power. So I could get away with a 10w, but was thinking of maybe using a 15w or a pair of 10w.
Here is the schematic, though it is slightly different. This one looks to be for the 101.619 and 101.619A and mine is the 101.619B, which there is no schematic.
Here is a close up of the power supply and the three caps in question, it’s not a great scan…
You can see that the schematic calls for 450v rated caps for all three positions, I’m guessing that’s because they all peak at the same voltage, but for example C49 settles pretty low, the lowest at like 320v.
Here are the original cap
I need to get everything 100% functional first so that everything that can pull the voltage down is able to. There are two tubes that have bad heaters if I’m remembering right and none of the lamps are working. So get the circuit right and then measure the peak voltage again.
Film is a good idea, but 33 uF 500v-630v rated polypropylene would be mighty big.
Yes, already did and still having these peaks.
I would imagine that they wouldn’t pass, considering that all three caps were spewing their guts out of the bottoms. Also I couldn’t even get the high voltage to come up with the 30uF cap in circuit. The voltage wouldn’t go about about 5v, I’m guessing it was reforming as I was bringing it up on the variac. So I just cut them out of circuit. But as I stated in my first post, these caps have already been replaced. I snipped the leads from the original caps so that they could remain in the radio and look original, but not sure I’d want to try that at 400v-500v anyways lol. I have a power supply capable of 4000v, but don’t like to touch it haha.
Dan
F&T makes 550Vdc capacitors, Askjanfirst has a good choice - there should be someone in the USA providing needs for radio restoration community...?
P.S. In my last build transformer had a little higher voltage then expected so the first cap is 500Vdc unit.
Standard 450Vdc cap was tried, worked but for safety sake I have installed 500Vdc cap.
P.S. In my last build transformer had a little higher voltage then expected so the first cap is 500Vdc unit.
Standard 450Vdc cap was tried, worked but for safety sake I have installed 500Vdc cap.
Try 2 caps in series, I do this quite often in my old radio restorations. While many consider balancing resistors are absolutely necessary to avoid over voltage on one of the 2 caps, they self balance quite nicely. This way you can achieve substantial voltage de-rating. I will put resistors across the capacitors just for safety discharging.
I believe that if a value between 450 and 550v is required, the procedure described by JMFahey is effective and does probably ensure the same reliability result than most 500-550V capacitors. They probably are nothing more than selected/tested specimens anyway. It was used by at least one small but well respected Italian tube guitar amplifier manufacturer. On those amps, I sometimes see 450V Nichicons with a sticker bearing the manufacturer logo and the words "tested at 500V" or "550V". They have been in use at 500V or more for a decade and they usually are still fine. By the way, the cost of a small thermal label printer is negigible today. The professional looking sticker is a useful hint for me that the supposedly under specified capacitor has been indeed tested by the factory, and it is not some makeshift repair that need to be replaced straight away.
The only shortcoming I see of the procedure described by mr. Fahey at post #9 is that it does not factor in the increased leakage current at high temperature. This is not a concern for a traditional power supply inside a well built amplifier, but it may be a issue inside a crammed high-power switched mode power supply or inside a product with subpar component placement.
A reputable capacitor manufacturer is supposed to test over the full temperature range, and this may explain why expensive brands seems to specify a working voltage that is far lower than the procedure results at room temperature.
The only shortcoming I see of the procedure described by mr. Fahey at post #9 is that it does not factor in the increased leakage current at high temperature. This is not a concern for a traditional power supply inside a well built amplifier, but it may be a issue inside a crammed high-power switched mode power supply or inside a product with subpar component placement.
A reputable capacitor manufacturer is supposed to test over the full temperature range, and this may explain why expensive brands seems to specify a working voltage that is far lower than the procedure results at room temperature.
If the thermistor works for you, just add another one or two in series.
Allow for air circulation near each of them.
The balancing resistors are indeed necessary, even if (at first) the leakage currents are similar. They won't be later on.
You are right, of course, I should have gone a little further in my description.
As I said at the beginning, not preaching or arguing with anybody, just describing a practical test which works for me.
What I should have added is that I use tons of those caps in my bread and butter 300W amps, both a Bass amp and an 8 channel box mixer.
I mount a fan at the end of head cabinet, and first thing in the airflow path is the power transformer and rest of the power supply, and only then airflow goes out cooling the thick aluminum back panel which is the actual heat sink, trough full width 10 mm slots above and below it.
Meaning besides transistors which are the "expected" components in the path, I also fan cool power transformer, main caps, diode bridge and drivers.
Transformers were included early, when I found much increased winding DCR (copper has an important PTC thermal coefficient) caused unacceptable +/-V rail voltage drop after , say, 1 hour use.
Contrary to Home Hi Fi use, MI amps are regularly used full tilt, even continuously clipping for hours on end.
Full EIA/FTC testing was dropped from most Consumer electronics for being "unrealistic" .......
Well, not here
Will post some pictures later showing fan mounting and airflow path.
Last edited: