Diode PIV

[jaysg]

The FRED saga got me thinking about PIV and how one would replace 1n4007's with FREDS -- one per? There are two applications I know of:
power supply rectification, either full-wave or bridge style, and spike protection on the output tubes, which may not be a good fit because FREDs come in a TO-220 style package.

The traditional Fender SS string in the power supply took 6 diodes for 40 and 80W amps. The Trainwreck notably uses 4 for a 30W amp. Later Fenders like my '83 Concert used a 4 diode bridge. I think the 80 to 100W Fenders went that way too and I recall seeing some doubled up bridges that looked like 8 1N4007s. Is there a consensus?

[Enzo]

The number of diodes in a string has nothing to do with the power output. PIV is PIV. Just because a diode might be faster doesn't change the reverse voltage it faces.

In old Fenders, there were several diodes in series to insure adequate PIV rating for the rectifier circuit. You will see Marshall amps with similar voltages and only a single diode where there would have been three on the Fender.

After certain allowances, you COULD parallel diodes to increase their current rating in total. But what is the point of paralleling a couple 1 amp diodes when a single 3 amp diode would do the job? Diodes are cheap. Besides, a 1 amp diode is enough for most amps we build.

Package type doesn't matter as long as it is properly mounted.

[jaysg]

In old Fenders, there were several diodes in series to insure adequate PIV rating for the rectifier circuit. You will see Marshall amps with similar voltages and only a single diode where there would have been three on the Fender.

One question then, is why did Fender engineers think they needed 3,000V PIV?

[JWK]

They didn't think they needed 3,000 PIV. They put them there for insurance/protection in the case of a diode failure.

[dai h.]

(my guess) the earliest silicon-state rectifiers weren't as reliable as would be later on, so maybe it was insurance(?) then the habit carried on?

[Enzo]

Not only that, I don't know that the original three in a row diodes were 1000PIV.

A quick look at some Twin Reverb schemos shows me in 1976 there were the three in a row, but by 1981 it was a simple bridge of four single diodes.

[Ray Ivers]

jaysg,

They actually needed about 950V - the PT HV secondary's RMS voltage x 1.41 - in most blackface amps using FWCT rectifiers, and could have used 4 x 1kV diodes with a quite adequate safety margin - perhaps as JWK pointed out, they wanted to "keep the lights on" even with a shorted diode. The FWB bridge amps used lower PT secondary voltages, so the PIV requirements were correspondingly less.

I've seen 2 x 1kV, 2 x 2kV, and 4 x 1kV diodes used in Marshall amps with FWCT rectifiers. The 2 x 1kV setup would actually be running just below max in most of these power supplies, and would probably have pretty much a normal lifespan (although good design it ain't, IMO).

Ray

[Rick Erickson]

Parallel Diodes?

After certain allowances, you COULD parallel diodes to increase their current rating in total.

Enzo - I had to read that a couple times before I realized you were aware of the current hogging issue, just not explaining it. For those in the dark:
You can't parallel diodes directly for the same reason you can't parallel transistors - one will always try to take most of the current and will eventually burn out. This is why you see those small value emitter resistors on the outputs of nearly every transistor power amp. The resistors keep the transistors from being greedy. You COULD do the same with doides, but as you pointed out - the hassle & parts count isn't worth it.

RE

[Enzo]

Yes, those small value resistors are termed "ballasts." And just like in a row of power transistors, the ballasts allow them to share.

I didn't go into the explanation because in the next sentence I poo pooed the idea. Didn't want to explain that which I didn't intend someone to do.

But perhaps I should have to avoid confusion.

[jaysg]

I started thinking I was guilty of making this parallel thing up -- found one. Iirc, Majors are like this too.

http://www.freeinfosociety.com/elect...iew.php?id=479

I did get some of those 11A, 1200V FREDS from DigiKey. Now if finding time to tweak was as easy as buying parts...

[Enzo]

Well, they did it, but it is still poor practice. You will note they didn't continue the practice into later models.

[andrew4566]

I had heard and assumed plausible that the rows of diodes were for a redundancy protection. The major problem with that in my mind was that every diode failure I ever had resulted in an open circuit and not a short, plus what would be the chances of two diodes failing? ... I can't see Fender adding extra diodes just-in-case two went bad.

But... if, as mentioned diodes of the day weren't reliable, then maybe.

Or... if the PIVs were lower ... altho I once tried using 3 series n4004s in an amp running 307vdc recified and they popped like crazy.

So...My question as a result of this discussion is: Is there any justification for running series diodes with today's comoponents? I am thinking not?

-Andy

[Enzo]

Not many amps double up diodes these days.

And if you have only seen opens and no shorts in rectifier diodes, consider yourself extremely fortunate. I replace shorted diodes all the time, both singly and in bridge packages. In fact in JBL Eon speakers, I replace shorted main bridges more than all other repairs combined.

Your 4004 trouble could have been surge currents.

[Wild Bill]

It's always a money thing...

Here's where having sold diodes to manufacturers for years counts perhaps less than simply being OLD! :-)

Back in the early Pleistocene Era silicon diodes cost more than today, with much bigger jumps in price as the PIV rating went up. A 1 kv rated diode could cost more than 3-4 times its 400 PIV "cousin".

What's more, like most parts diodes were sold with bigger discounts as you bought in larger volumes. The price for 1 piece might be a quarter while if you bought a thousand you might pay around 10-12 cents.

When a diode maker ran a batch of diodes he would test a sample of the result. If they reached a high enough PIV rating he would label and sell them as 1 kv PIV diodes. If they made only 400 PIV they would be sold at that rating under a different number. That's why you see diode number "families" like 1N4001. 1N4004 or 1N4007. The last digit meant the same style diode met a different PIV rating.

Nowadays quality control has advanced so much they don't have to do much of that anymore and the price difference between PIV "levels" is much less.

So in Leo Fender's day it would often be cheaper to standardize your inventory on a 1N4004 or whatever. That was good enough for lower voltage applications and for higher needs he would put 2 or 3 in series. He could buy in a higher quantity price break by using only the one diode - a diode that was more reasonably priced in the 1st place.

In the VERY early days there were few or no diodes rated above 400 PIV anyway! Putting them in series was the only way to achieve a high enough rating. You also had the problem that diodes varied much more in their forward voltage drop as they passed the current. You can think of this as the diode's forward resistance value. When turned on it always slumps some energy across it, just like a small value dropping resistor.

With diodes in series you could get unequal drops. You might use three 400 PIV diodes in series to get 1200 PIV to replace a tube rectifier. If the drops were not equal you might see 300 across on and 500 across another. The one that saw 500 PIV could pop! So they did the same thing you do with stacking filter caps to get a higher voltage rating. You put equalizing resistors across each diode. You also can get slightly different turn-on points during the AC wave for each diode in the string so you also might put small value ceramic disk caps across each diode as well.

By the time we left the 70's 1 kv diodes had gotten so cheap that no one bothered with the equalizing caps and resistors anymore. Stacking a few more 1N4007's was cheaper than adding the caps and resistors! An extra diode more than made up for any differences between each diode.

"When I was a lad we didn't have diodes like we have today. We had to manually toggle switches REALLY fast to rectify the AC!"

---Wild Bill

[andrew4566]

WB,

Thanks very much for shedding light on this!

I love the idea of building amps inspired by those of the past and the history of these things is very cool.

I am thinking 2 1n4007 in series without equalising caps or resistors for me.

-Andy