Two pairs of 1N4007 in series really necessary in a 12w amp's rectifier?

[Iplayloud]

I'm installing an SS - tube rec switch in my princeton clone. Seen a few good examples around the net. Now I never fully understood why put 2 diodes in series. Tried to read about it and it doesn't add up the amp rating.. The voltage is far from 1kv..?!?

[Alan0354]

It just give you insurance. Diodes burn short, so if one short out, you still have the second one functioning. Yes, it is safe to use only one. The big Marshall Plexi used only one instead of two in series. They do burn, I had my amp burned twice at the time. Those diodes are cheap anyway. In my design, I use 1N5408 1000V 3A diode rather 1N4007 1A diode. I use two in series also.

[Iplayloud]

Cool thx! My oddball PT only cranks 380v so I'm hoping to defibrillate these 6v6s up to a nicer 410v..

[Iplayloud]

In fact none of the examples showed exactly what I wanted to do so here's a picture (except with 2 diodes in series):

Edit: Also the other side of the DPDT will be used for a slight bias adjustment..

[pdf64]

The series connection is to assist with the reverse voltage capability.
For a bi-phase rectifier, the diodes are subject to the peak voltage of the whole B+ winding, eg 650V x 1.414 = 919V
Consider the condition at initial power up before the tubes are up to temperature, eg the unloaded VB+ may rise 10%, eg 1011V, for at least 30 seconds every power up cycle.
What if the line is at the high end of its allowed tolerance, eg 5%, at the time, eg 1061V?
See The Valve Wizard

[J M Fahey]

The voltage is far from 1kv..?!?

Are you sure you are FAR?

My oddball PT only cranks 380v so I'm hoping to defibrillate these 6v6s up to a nicer 410v..

"Only"?

Although you do not mention transformer AC voltage, your hinting at +V=410V (lodaded) so probably 450V peak unloaded meaning the AC voltage has a peak of 450V.

Now remember it's alternating current: on the positive swing peak voltage will charge the supply cap to 450V ; the diode cathode is soldered to the capacitor + terminal meaning it's always connected to it and the +450V

The anode is soldered, meaning always connected to the transformer winding, which in 1/50 or 1/60 of a second will swing negative, up to -450V peak .

So if cathode is at +450V and anode is at -450V , the diode has to stand 900V

Do you think that being within 10% of 1000V , nominal destruction voltage, is a big margin?

I guess not.

I think you will understand now why often 2 1N4007 in series are used , and AB763 era (late 60's) Fenders used three in series.

Nowadays we get by with just one 1N4007 simply because modern parts are VERY good, (sorry NOS junkies ) and probably most run of the mill 1N4007 will self destroy at a much higher value, probably 30 to 50% higher .

Why don't they sell them as "1500V diodes"?

Because then they would have to measure one by one and discard those which don't meet the spec.

Why waste so much time ($$$$$$$$$$$) to charge a couple cents extra?

Better make them better than spec'd , churn them out by the zillions and sell them untested (except usual QC: a couple tested per batch of thousands).

Producing stuff is very cheap, adjusting/calibrating/testing is expensive so avoid it as much as possible.

Flying higher than expected is a good way to avoid crashing on some odd mountain peak.

EDIT: and why did Leo use three?

I guess way back then 1000V meant exactly that, production just met that goal and probably you could find a bad apple now and then, so Leo "spent more" to actually spend less (in Servicing, customer bad Karma, etc.)

He was no fool; besides excellent sound, his amps were probably the most reliable in the Industry (Ampeg too, but at quite higher cost).

Peavey is similarly trusty, and they were the first to use way more output transistors than apparently needed.

[Chuck H]

The dual diodes are to account for voltage, not overall power. So the wattage of the amp isn't part of the consideration. Others have covered the voltage potential for a typical power supply.

You may do fine with single diodes using a small-ish main filter. The charge time for a larger filter should probably cause the voltage to peak higher. Sort of the way falling off a stool isn't as bad as falling off an 8' ladder. You get closer to terminal velocity falling from the greater height. Think of the size of the cap as the height and the speed of the fall as the voltage. Since you're tube rectified you're probably not using a large reservoir cap. Still, for a few cents you can just use two diodes per side. Other than ease of layout, which is pretty small taters, I can't see a good reason not to just use two diodes per side.

[Tom Phillips]

...and why did Leo use three?

I guess way back then 1000V meant exactly that, production just met that goal and probably you could find a bad apple now and then, so Leo "spent more" to actually spend less (in Servicing, customer bad Karma, etc.)...

Some trivia information.
The rectifier diodes used in that era by Fender were usually a metal case package device marked "RCA 35604." I have found them used in Fender amps produced up through at least 1972.

Out of curiosity I have tried to find data on that part but have had no success. I suspect that they were probably rated between 600 and 800 PIV each. That's my guess based on the type of parts that were available in the 1960's. If they were rated 600 to 800V then a series string of three would be good for 1800 to 2400PIV. However, that would be true only if the parts were perfectly matched to insure that the voltage was distributed evenly across each of the three. Good design practice would have been to install balancing resistors across each diode. If that was done then maybe a two diode string would have been sufficient but that would have cost more in parts and labor. As we know, Leo decided to use the three diode string and it has proven to be a good choice by the test of time.

Anyway. That's some interesting trivia and fun speculation. We are lucky that no one has said that the old diodes were responsible for the vintage sound.
Tom

[Jazz P Bass]

Also not having any luck finding info on the RCA 35604.
I believe it may be a house number.

I did find a site that crosses it to an NTE diode.
400 volt/ 3 amp.

35604 NTE Equivalent NTE5804 400V 3A diode... - Wholesale Electronics

[Tom Phillips]

...I did find a site that crosses it to an NTE diode.
400 volt/ 3 amp.

35604 NTE Equivalent NTE5804 400V 3A diode... - Wholesale Electronics

Cool. That's more info than I ever found.
Thanks,
Tom

[J M Fahey]

I don't trust NTE "equivalences" and FWIW, the Fender ones were of unknown voltage but definitely not 3A so there goes the "NTE equivalence" through the window

We are lucky that no one has said that the old diodes were responsible for the vintage sound.

Shhhhhhh!!!!! or they will hear you !!!!!!
It's already bad enough people replacing 1N4007 with ultrafast UF4007 "to reduce amp buzz"

[Alan0354]

I have been using 1N5408 on all my stuffsHTTP 301 This page has been moved. Cheap, very common. 1000V, 3A and 200A surge current. Two in series, I doubt you'll have problem.

[R.G.]

And now for the hyper-cautious belt-and-suspenders view.

Diodes in reverse bias share voltage by the inverse of their LEAKAGE CURRENTS. If one is leakier than the other, the non-leaky one will have the vast majority of the voltage across it. Two in series suddenly becomes 1.1 in series, as leakages are not well specified, just "less than X".

You can stop this somewhat by putting high value resistors across the diodes. The resistors are vastly more "leakage" than the diodes themselves have, and they force voltage sharing in reverse voltage situations because the diode leakages are too small to matter. Some 10M to 22M resistors in parallel with the diodes will promptly force equal sharing so you can get nearly two times the rating of one diode.

If you do this, be aware that resistors have voltage ratings, too, and you may have to use several resistors rated at only 300-400V each in series for the paralleling resistors.

Insurance is always expensive, not least in that you have to think about it to make good choices.

[Alan0354]

From my understanding, I don't think it is particular important to have resistors across the diodes. I don't think diodes are like capacitors. It's the power that burn the diode. Diode is nothing more than a zener diode that above a certain voltage, it starts to conduct. Zener diodes are designed to breakdown in more precise voltage as specified. I remember when I was designing analog bipolar IC, the 6.3V zener diode is just the reverse of the base/emitter junction......literally take the NPN and flip the emitter and base and it become a zener diode. There is no aging that I know of at the time that it gets weaker through time. Just like zener, it's the current times voltage that burn the zener diode just like when over voltage a normal diode. So as long as the diode is not breaking down and conducting current, it's all good. from my experience, spec of the SS device is quite reliable. from my designing of HV circuits, we don't have the luxury of having a lot of spare headroom. We use the transistor, diode, MOSFET to their spec limit and we never have problem. In fact we use TVS and actually count on them to turn on for clamping in normal operation. It survived like a champ.

In the example one diode leak a little and put most of the voltage across the second diode. As long as the second diode does not conduct, there should not be stress. If the second diode start leaking, then the voltage drop across the diode will decrease and both diodes share the voltage.

That's my understanding in semiconductor.

[potatofarmer]

And now for the hyper-cautious belt-and-suspenders view.

So if I suggested ditching FWCT rectifiers in favor of bridge rectifiers that would be, what, the overalls view?