Bridge Rectifier

[trem]

Was looking at Merlin's sight.....
I am not sure what he is talking about regarding Peek Inverse Voltage.
For example...
If you have a PT that has a secondary of 250-250. Does the bridge need to handle 2x250 or 2x500 volts.?
Thank You

[R.G.]

Depends on how you hook it up.

The PT secondary produces 250-0-250. If that's volts RMS, the way transformers are usually specified, then there is 500Vac across the whole thing. Diodes see the instantaneous voltage. 500Vac has peaks of 500* 1.414 = 707V peak. That's from end to end at the peak of the sine wave. If you measure from the middle to each end, you see peaks of half that, or 353.5V.

So what the diodes have to be rated for depends on how you hook it up. If you ground the center tap and use two diodes, then each diode charges up the first filter cap to 353V, and when the voltage reverses it has to withstand the 353V it's still connected to at the filter cap plus the 353V it's being back-biased by its half of the PT secondary, so each diode in on that transformer for a full-wave-center-tap circuit needs to withstand at least 707V or it will pop.

If you use a bridge rectifier, you can hook it up a lot of ways. If you hook it up across one of the half-secondaries, it rectifies that half-secondary to 353Vdc. There are two diodes in series in each direction of blocking in a full wave bridge rectifier, so each diode sees nominally half of the reverse voltage. You can't trust that and use diodes rated for 177V in this application because the diodes are not perfectly matched, but 200V rated diodes connected in a full wave bridge would probably be OK to full wave rectify either half. Risky, but possible.

You could put a full wave bridge on EACH half, each one making 353Vdc, and the total across both outputs would be 707Vdc, but that would be from CT to DC ends, so you could use two capacitors, each one rated for more than 353V, and they'd be fine. And the diodes in each bridge would only need to be the 176 to 200V as noted above.

You could use one full wave bridge across the full outside legs of the secondary. This would make 707V again, but with a bridge, there are two diodes in series each time one polarity is blocked, so they only need to be rated for over half of the total rectified voltage. 400V diodes would work here. Same risks of mismatched diodes, but it would mostly work.

So you need to decide: are you making 353Vdc or 707Vdc? If you're trying to get to 353Vdc, you need two diodes, each one rated for more than twice the DC output, so you need 707Vdc minimum, probably 800 to 1000 to be safe. If you're trying to make 707V out of that transformer (or +/-353, same thing) you need either one full wave bridge rated for at least 800V total. Using two full wave bridges, one on each half-secondary, is a bad idea that I only put in there for illustration.

The two-diode full-wave-center-tapped circuit we know and love from tube rectifiers needs diodes rated for more than twice the DC output voltage; The full-wave-bridge with four diodes can almost get by with diodes rated for half the DC voltage out, but it gets riskier for matching-diodes reasons the closer you get to just half. So you go bigger, but not like with the full wave center tapped circuit.

[trem]

R.G. -
I really appreciate the time it took you (or certainly would have taken me) to write all of this.
The PT I am using does not have a CT on the HT Secondary. So I thought I would Try/Buy one of those bridge rectifiers that are available from Mouser, etc.
So I was wondering what type of Inverse Voltage rating it would need to have for my 250-250 secondary.
If I am reading you properly, I would be OK with a Bridge that was rated at 500 volts inverse.?
Thank You

[R.G.]

Ah. OK, I understand more.

You have to go back to that question - are you trying to make 350+ Vdc or 700+Vdc? Those are pretty much your choices with a capacitor input filter on this transformer. Then you need to figure out whether the current out will be trivially small or nearly as much as the transformer can make.

This transformer was almost certainly set up to make the 350V option with either a tube rectifier or two silicon diodes. It's simpler and cheaper to just wind a single winding if you're going to use a full wave bridge. If you have to use a tube or come from a tube-heritage, then you want to use only two diodes, and you wind twice the voltage-turns and put in a center tap on the transformer to let you do it.

The difference is that the full-wave-center-tap connection uses twice the turns of wire in the transformer, and the wire is thinner because each half of the secondary only conducts every-other half-cycle of AC. If you want 350Vdc and use a bridge on half the winding, you can only pull something less than half the power out of the transformer that it was designed to do, because only one half of the secondary is working.

If you want 700V, kewl, use a full wave bridge and tape/insulate off the center tap, which will be sitting at 350V. If you do this, you can pull the full designed power, but since it's coming out at twice the voltage, you can only pull half the current you'd get with the full wave-center-tapped (FWCT) connection.

But there's something else that trumps all this. Voltage rating in full wave bridges is cheap the last time I looked. Another US$0.50 gets you another 200V or so. I'd go for a minimum of 800V, and probably 1000V if I were buying a full wave bridge (FWB) for this. The amount of money you save by getting just barely enough is not worth the worry.

Here's another sneaky thing to realize. If you get an 800-1000V bridge, it works for either 350V or 700V output. You mount the bridge and connect the two 250V ends of the winding to the "AC" terminals of the bridge. Doesn't matter which way round.

Now if you ground the CT and leave the (-) terminal of the bridge open, you get 350V out of a FWCT, because only the two positive side diodes in the bridge ever do anything. They still have to withstand twice the output voltage, so get the 1kV bridge. The unconnected negative terminal of the bridge sits there not connected to anything, but with -350V on it.

If you want 700V, you disconnect the CT from ground, carefully insulating it, and connect ground and the filter cap negative to the negative terminal of the bridge. This now obligingly makes 700Vdc.

You can even get +350V, ground, -350V by leaving the CT on ground and putting a second capacitor across ground to the minus terminal of the bridge.

The moral of the story is: get the high voltage diodes/bridges. The difference is cheap. You can also use 1N4007 diodes, which are about US$0.10 each, and you only need four to give you a 1kV+ bridge at up to 1A of current, plenty for tube amps. Get the UF4007 diodes which are still 1kV/1A, but fast turn off and quieter. They're about US$0.20 each.

Or if you want to make it easier to wire the high voltage stuff, get an all-in-one 1KV bridge rectifier assembly.

[Steve Conner]

Sorry RG, the diodes in a bridge rectifier must be rated for the full DC output, not half. They are not in series. Whenever two of the diodes are conducting and filling up the filter capacitor, the other two have the full filter cap voltage across them in reverse.

[pdf64]

My feeling is that 1N4007 are borderline for many of the beefier tube amps, in terms of their 1kV PIV rating, eg 350-0-350v will result in 990V across the diodes, so the unloaded voltage may rise 5% to 1039V, and then there's line voltage tolerance to consider.
Pete

[loudthud]

I'll tell you a little secret. Often times underrated parts like the 1N4001 meet or exceed the 1N4007 specs. I used to work for a place that made an 8051 knockoff processor and sold 8, 10 and 12MHz versions. Found out from the test engineer that they were all tested at 12MHz and they all had to pass. When a part is first manufactured, there may be some under performing parts, but once the process is fine tuned, they all meet the highest specs. I'll bet all 1N400X diodes are all made on the same wafer, same mask set, same process.

I recently bought a 50V 25A bridge rectifier at Radio Shack. It was a Fairchild part rated at 50V. Put it on the curve tracer and it didn't breakdown at 1500V. Always a good idea to test the parts however.

[trem]

I am having trouble finding a Bridge Rectifier (800-1k volts) that will screw down to the chassis. I see them at Mouser. But, they are 5-20 bux and rated for way more amps than a guitar amp draws.
There seem to be lots of Rectifiers with the rating I need, that are very inexpensive, but are designed to be inserted into a PCB, or some type of socket. Any harm in buying one of these types, gluing it to the chassis (so the connections are up in the air) and just making my way from there.?
Thank You

[Enzo]

This is for your B+ right? So it needs to handle pretty small currents, say 1 amp tops? WHy must it screw to the chassis?

The little round kind, the flat kind with four in a row leads, or even the square kind can all be mounted on a terminal strip by their leads. And for such small current needs, four 1N4007s wired in a bridge would be simple enough to mount anywhere. 1N4007 sell for 3 or 4 cents each at Mouser, in fact there are some "end of life" Fairchilds there for 2 cents each or $1.60 a hundred. They have over 92,000 of those in stock, so may not run out soon.

The square ones with a hole in the center I don't think happen below maybe 4 amps. I found one at 3A. You can get a square one with a hole and screw it down through its hole, legs up if you like.

They also make a 1000v 1A bridge in a DIP. Same size and shap as an 8-leg op amp, but with pins only on the corners.

1A 1kV flat bridge 65 cents.

[R.G.]

There seem to be lots of Rectifiers with the rating I need, that are very inexpensive, but are designed to be inserted into a PCB, or some type of socket. Any harm in buying one of these types, gluing it to the chassis (so the connections are up in the air) and just making my way from there.?

How about buying a five-tab terminal strip, hooking up four diodes to the four non-chassis tabs, and then wiring to this terminal strip? Gluing axial rectifiers to the chassis seems like a good way to have high voltage shorts.

[R.G.]

Sorry RG, the diodes in a bridge rectifier must be rated for the full DC output, not half. They are not in series. Whenever two of the diodes are conducting and filling up the filter capacitor, the other two have the full filter cap voltage across them in reverse.

Oops. Sorry. Yes, the diodes in a FWB must be rated for the the output DC voltage, not half the voltage. The FWCT diodes must be rated for twice the output voltage, as the diodes withstand the output cap plus the reverse AC. The FWB diodes just hold off the output cap voltage.

[R.G.]

1500V 1.5A diode for $0.31 each

Terminal strip with 5 lugs free

but I like this one:
Terminal strip with 11 tie points

I'd use the 11-place strip, cutting off the sections outside the mounting legs and mounting my kV class rectifiers with two mounting legs.

[trem]

Hey Enzo -
No, you are right. No reason it Has To screw to the chassis. That just makes it real easy.
The guys at Vishay said there would be no problem if I just Glue/Epoxy it upside down to the chassis.
Will order something like that for pretty cheap.
And yes, you are correct again. Current is almost a non-issue 500mA or 1A will be plenty
Thank You

[trem]

1500V 1.5A diode for $0.31 each

Terminal strip with 5 lugs free

but I like this one:
Terminal strip with 11 tie points

I'd use the 11-place strip, cutting off the sections outside the mounting legs and mounting my kV class rectifiers with two mounting legs.

Hey R.G. -
I have built those before. With just what you picture.
This time, I thought I would just glue one of the below (up side down) to the chassis, and run with that.
Thank You

DF10M Diodes Inc. | Mouser

[serkan80]

@ trem. It'd be interesting to see how electronics works when it comes in contact with epoxy. I'd strongly advise against it though. Unless you have properly researched that the glue or epoxy that you're going to be using is an insulator. Even in that case, when you glue something, the electrical connections are disconnected unless you have soldered it properly. Also, gluing it to the chassis is not really a very reliable option. Epoxy and glues are mostly used for wood, steel and glass etc. I've never seen anyone use them for electronics and I strongly suspect that it's because most of them are good electrical and thermal conductors.