Max Input Cap

[trem]

I did a search...and there is some info...but I am still not sure.
Tube Diodes in general, but specifically the 6X5GT... says to use a 10 Mic cap at most.
I know we frequently run plate voltages over spec, but how critical is the max cap input spec.?

If I use a cap that is 22 Mic and the book says 10 Mic (that is double) how "bad" is that.?

Is it the Rectifier Tube that is at risk of Failing/Shorting...and then maybe exposing the rest of the amp to High Voltage AC.?
How close do You Guys adhere to the Max Cap Spec.?
Thank You

[Enzo]

You can always count on someone chiming in that they used a 200uf cap and it "worked just fine" and so there you are. Unfortunately, like driving drunk, "getting away with it" is not the same thing as "good idea."


yes, the risk is to the rectifier tube. The larger the cap, the larger the charging current. How we treat power tubes like 6V6s really has nothing to do with cap ratings. If you use a 22uf instead of a 10uf, likely nothing bad will come of it. But you can use much larger caps farther downstream, so in my mind there is not a lot of reason to max the first cap. You want me to draw a line not to cross? Sorry. I'd suggest you stay within sight of the limit on that spec. This is the internet, so you will find someone preaching that if 10uf is the limit then their personal choice would be OK, and that might be 20,30,40,50,god knows what.

[trem]

Thanks Enzo -
I am getting ready to build a little filter cap board for this amp (2x 6K6) and I realized I had automatically grabbed a 22/450 Cap. I have 10/450, so I will use one of those.
After it hits that 10-450 and the choke, I am not sure there is much "ripple" left. I guess I will just use 10-450 throughout and see how it goes.
Thank You

[Chuck H]

I used a 200uf cap and it worked just fine!

[soundguruman]

Fender had 35 with no problem. (two 70 in series)
Seymour Duncan used 100 (two 200 in series) and it still keeps working, year after year.
I'm still waiting for it to fail....and waiting...and waiting...
so despite this doom and gloom, a lot of amps work just fine with bigger filter caps.

I think you are too paranoid about the rectifier shorting. It's pretty rare.
I have been around a while, and it's a pretty UN-common occurrence.

BUT I would use two caps in series, and bypass resistors.
Seems like the amps with caps in series have the least problems.

AND you gotta consider, that older components had more leakage.
Newer capacitors with improved construction probably don't suck as much current as the older style ones did.

In an old style build, there would be a 10 on the rectifier output.
Then a big honk'in 10 ohm resistor in series, then a bigger cap.
So they had a smaller cap on the rectifier output, a small voltage drop, then a bigger capacitor for better ripple rejection.
It was a compromise that worked acceptably. But they were also fighting caps that leaked more than the newer ones
we use now days.

[Chuck H]

Fender had 35 with no problem. (two 70 in series)
Seymour Duncan used 100 (two 200 in series) and it still keeps working, year after year.

True. But the OP is using a tube with a particularly small spec. It seems irresponsible to post these large values since they would be exponentially over spec for the OP's rectifier.

I think you are too paranoid about the rectifier shorting. It's pretty rare.
I have been around a while, and it's a pretty UN-common occurrence.

Agreed. But this is an uncommon rectifier tube. Saying this isn't a critical consideration because other, more familiar tubes seem to tolerate it seems irresponsible.

BUT I would use two caps in series, and bypass resistors.
Seems like the amps with caps in series have the least problems.

Interesting. I wonder what it is about the series caps and additional load of resistors that REDUCES current draw on the rectifier?

AND you gotta consider, that older components had more leakage.
Newer capacitors with improved construction probably don't suck as much current as the older style ones did.

Older amps also didn't commonly use voltage bleeder circuits except incidentally as balance resistors for series caps. Which was actually rare. I don't understand how a new cap with less leakage and a bleeder circuit is better than an old cap with more leakage.

In an old style build, there would be a 10 on the rectifier output.
Then a big honk'in 10 ohm resistor in series, then a bigger cap.
So they had a smaller cap on the rectifier output, a small voltage drop, then a bigger capacitor for better ripple rejection.

I actually haven't seen this as standard practice. It's a worthy idea though with the only caveat being that designers were also battling power supply sag and adding more resistance behind the amplifiers would counter this objective. Maybe that's why I haven't seen it.

It was a compromise that worked acceptably. But they were also fighting caps that leaked more than the newer ones
we use now days.

I don't know that the leakage was so horrible as to be noteworthy but you keep bringing it up. It seems to me that if a cap were leaking as much as five or ten percent that wouldn't tax the loading any worse than a higher value vs. a lower value within the typical +/- 20% spec for caps at the time. And they didn't leak that much.

[loudthud]

In the 1959 RCA tube manual (RC19) the data sheet for the 6X4 (you are directed there from the 6X5GT data sheet) says higher than 10uF can be used but the effective plate-supply impedance should be increased to prevent exceeding the maximum rating for peak plate current. The rating for plate current (per plate) is 245mA peak.

[R.G.]

Fender had 35 with no problem. (two 70 in series)
Seymour Duncan used 100 (two 200 in series) and it still keeps working, year after year.
I'm still waiting for it to fail....and waiting...and waiting...
so despite this doom and gloom, a lot of amps work just fine with bigger filter caps.

The amps do work fine - and guitar amp makers historically did abuse the specs of the parts they used.

It's not doom and gloom - the tube makers flatly said not to go beyond X because they offered warranties with their parts and did not want to get caught with a bad rep. The situation with tube makers today is quite different. Largely, I doubt they give a rip about reputation because they can sell all they make, and can just change company names if there's an issue.

The issue with tube rectifiers and maximum cap values is one of peak currents. In rectifier circuits, the current into the cap happens only on short pulses. The cap gets charged to near the peak of the AC wave applied, and when the AC starts back down, the diode shuts off, leaving the charge in the cap. The diode keeps the AC out of it and the cap supplies charge to the amp circuit with no further help from the AC and runs down until the AC wave gets high enough to turn on the diode(s) again the next half-cycle. The diodes conduct for the time the ripple voltage rises, and do not conduct during the time the ripple voltage sags. The higher the cap, the smaller the ripple voltage from peak to peak, but the shorter the time it has to charge when the AC half-wave nears its peak again. That means that if the diodes only get 5% of the total time in an AC cycle to charge the caps, then they have to stick in 100% of the charge to be used in the total cycle in 1/20 of the cycle time. So the current in that conduction time has to be 20 times as large as the DC current going out on average.

Peak currents in rectifiers can get well and truly out of hand with big caps. If you want small ripple, the diodes conduct for a very short time - and in that time pulse in enough current to get the cap through the next discharge. The current during that pulse is big. The bigger the cap, the bigger the pulse because the shorter the time the diodes get to conduct to charge the caps.

Vacuum diodes can't conduct large currents well. Even though plate resistance limits the peak current a bit, the space charge at the cathode gets exhausted on current peaks, and the cathode coating gets eroded by big current peaks. This is fundamentally why tubes have current limits - bigger currents kills the cathodes on oxide coated tubes.

So sure they'll work - but it wears out the rectifiers faster.

I think you are too paranoid about the rectifier shorting. It's pretty rare.
I have been around a while, and it's a pretty UN-common occurrence.

I think you're confusing things. Too-big capacitors do not make rectifiers short - it makes them less efficient and their rectification efficiency declines much before it should. It's completely unlikely to make a tube short at all, so I guess the anecdote is right, but it reminds me of the joke about trying to scare away elephants.

BUT I would use two caps in series, and bypass resistors.
Seems like the amps with caps in series have the least problems.

Just off hand, I'd guess that two caps in series doubles up on the ESR and other minor losses and helps keep the peak currents down. It's as good a theory as not shorting rectifier tubes.

AND you gotta consider, that older components had more leakage.
Newer capacitors with improved construction probably don't suck as much current as the older style ones did.

Which is neither here nor there as regards maximum cap values. It makes no difference to the max cap value at all what leakage happens, within any kind of reasonable limit. Leakage is like added load current.

In an old style build, there would be a 10 on the rectifier output.
Then a big honk'in 10 ohm resistor in series, then a bigger cap.
So they had a smaller cap on the rectifier output, a small voltage drop, then a bigger capacitor for better ripple rejection.
It was a compromise that worked acceptably.

... which makes sense from the perspective that the first cap is what experiences the current peaks, and the 10 ohm keeps the second cap from simply adding to the current peaks from the diodes.

But they were also fighting caps that leaked more than the newer ones we use now days.

And that is as critically important to the issue of what maximum first filter cap to use with a tube rectifier as whether or not to use Pine-Sol to wash tube pins before putting them in circuit.

Note: That means it doesn't matter at all. I got concerned that I was being too subtle.

[Enzo]

I am now interested to know WHICH Fender and Seymour Duncan amps used those cap values with 6X5 rectifier tubes? It really doesn't matter what we use with a 5AR4, GZ34, 5U4 when we are discussing 6X5.

[Chuck H]

I am now interested to know WHICH Fender and Seymour Duncan amps used those cap values with 6X5 rectifier tubes? It really doesn't matter what we use with a 5AR4, GZ34, 5U4 when we are discussing 6X5.

I mentioned that too. SGM didn't iterate clearly enough but I think his point was that going higher, even doubling the max uf spec, is common. And it is. But I don't know much about the 6x5 so I didn't dare say it myself. I will say that they use the $h!t outta them over on the AX84 site and there may be some info about what kind of abuse they'll take and how much fudging on uf's you can get away with. I'd look there for sure.

[R.G.]

The problem with figuring out how much capacitance a rectifier can stand is that it's dependent on things that are not on the tube datasheet, like the exact size and conformation of the cathodes, the heating efficiency, the residual gas in the tube, the cathode material substrate and plating (if any) and the composition and treatment of the oxide surfaces. At one time these were predictable, as the big tube suppliers were very consistent. But today, we have over 50 years more knowledge of what materials do and how to do them, and a few tube suppliers that may do funny things with cathodes that are not like the originals. A tube supplier may have decided that an existing cathode from another tube would make a fine 6X5 cathode, ignoring those funny grid things and leaving them out. Or that an oxide formulation is OK if it puts out less electrons but is "tougher" to ion bombardment. It likely varies with the "6X5", whether it's a 60's relic or a new manufacture.

I personally would either stay with the datasheet recommendations for "vintage" tubes, or go to SS diodes plus resistor approximation if I felt I had to have big input filter caps. But then my opinions only matter (and cost!) the same as everyone else's, which may be different.

[trem]

Well...I suppose I will just go with the 10 Mic listed in the book.
I would gladly real a link, but I am not sure where to look....Merlins book...Randall Aiken sight...RG's pages.....how do you determine if a certain value filter cap is big enough to do the job.?
This amp will use a Pi Filter with 10 Mic on each side, then another 10 Mic each for the Phase Invert and the preamp tube.
Is their a way to figure the ability of a 10 Mic cap to keep up with a pair of Cat Biased 6K6 tubes.?
Thank You

[R.G.]

Yes, there is a way. The only thing a filter cap does for you is to make the ripple voltage smaller. If the cap is nearly zero, then the ripple voltage is equal to the peak AC voltage, dropping all the way to zero as the AC half-wave heads for the zero crossing. Ugh!

But putting a cap in lets the diodes charge the cap to the peak value of the incoming AC half-wave, and then the cap holds up the load voltage until the next AC voltage half-wave comes along.

If the load is not so heavy as to run the cap down a whole lot between chargings, then the ripple voltage runs down at an almost linear rate between AC half-wave peaks. So you can calculate the ripple as
Vripple = Iload* (time between AC peaks) /Capacitance (in farads).

The first step is to estimate the current for a pair of cathode biased 6K6 tubes. The tube databook should give you that, to a first order. My Sylvania tube book says a pair of 6K6s in Class A1 self-bias has a max signal plate current of 61ma for a plate supply of 285Vdc. So the ripple voltage you get for this situation will be

Vripple = (0.061A)*(0.0086Sec)/C

If C = 10uF, Vripple = 0.061*0.0086/10E-6 = 52V

UGH! 52V ripple! That's way too big. Ripple should be under 10%, and under 5% if you can get it.

Using 20uF gives half that, 26V ripple.

Looking at the 6X5 datasheet, Sylvania did not list a maximum capacitance, but did note that "additional impedance [in the supply to the plates] may be needed for filters with more than 40uF". I don't have other datasheets for other brands of 6X5, but from that you ought to be able to get to 33uF without much worry on the life of the rectifier if I interpret that right. So

using 30uF, you get
Vripple = 0.061*0.0086/33E-6 = 15.9V ripple, which is 5.5% ripple at max load, and should be fine.

[Rhodesplyr]

Hammond Organs in the 2-series, (B2, C2, RT2) made from 1950 to 1954, used 6X5s in their AO10 preamps and used 50uF on the first stage. However, the overall continuous current draw of the preamp is small, about 14mA. I'm not sure how much series resistance there is in the power transformer, but the tubes tend to last a long time, unless they suffer heater-cathode shorts, which are the Achilles heel of the 6X5 and 6X4 rectifier tubes. The preamp in the 3-series organs uses 6X4s, and the first stage is 40uF. Current draw is higher, around 50mA, but, again, they tend to last a long time.

[R.G.]

And let me step in here with an heretical opinion.

It's difficult for me to imagine a less musical thing than running AC through rectifiers and filtering it in a capacitor. There are people who will swear that they can hear the difference between a tube rectifier and a closely matched set of silicon diodes and some resistance. I do not discount that this may be possible. But I suspect that there are very few people who can really do this. I think many more "hear" this because they have been told and believe that music is forever damaged by having any silicon junctions within 50 meters of where the music is made, and worse yet, THEIR hearing is suspect if they cannot.

So be it. I've tried and I cannot tell the difference between silicon rectification and other parts to match the conduction to be like a tube rectifier and an actual tube rectifier. And although I've never done a broad study, I have never encountered someone who could notice the difference without being asked.

There is always the option to use silicon diodes and some resistance to replace the 6X5. There is the further option to use more parts than one power resistor to fake the 6X5 conduction more closely.

Maybe you could hear the difference, maybe not.

Maybe.