Using Ultrafast Rectifier in place of 60Hz Bridge

Hi Guys,

I have a number of basic power supply units that use a mains tx with a centre tapped secondry with a 35A, 600V bridge rectifier … and these supply equip with a listed load of a max of 8A 24 V. (being centre tapped they only use half of the bridge)

Clearly you would expect these bridge rectifiers to last forever because they are vastly over rated for their load … but they don’t. Lots of them fail .. go short circuit.

I suspect that they are killed by electrical switching noise around the facility, so am planning on fitting some varistors across the input to the rectifier … but I am also wondering if I can use some high speed dual, common cathode diodes in their place.

Eg. FFAF60UA60DN Ultra Fast Rectifiers.

These are rated at 60A 600V, and are designed for use as clamping and freewheeling diodes in switching power supplies etc. … but my question is … is it ok to use them as basic 60Hz rectifiers in place of the original bridges.

Thanks for any help.

Regards, Wes

I agree, the original rectifier SHOULD last forever.

Could it be that the load causes a very large surge current spike? You know, like a large capacitor bank?
It is only a guess.

To your original question, yes, you can use them as replacements.

Hi,


They only theory that helps to explain this phenomenon is that the original bridge rectifier diodes were under rated to begin with. For example, using a 200piv rectifier in a bridge where the input is 120vac. This sounds ok but it's not.
Now we would expect 120vac to cause 170v peak and so a 200piv diode should work. But during circuit startup slight differences in capacitance of each diode can cause a higher voltage across one or more diodes for a short time period because it acts like a partial voltage doubler. This could cause damage to one or more rectifiers in the bridge which either causes it to fail immediately or in the near future.
So what this means is that a 200piv bridge rectifier is not really suited for long term use with a 120vac input line. Doubling this value leads to longer life, so a 400piv bridge rectifier would be a better idea.
The speed of the rectifier could help here too, as long as the capacitance variation between diodes is lower. But i still think the better solution is to double the peak value and find a diode that is suited for that higher voltage instead of worrying about the capacitance of each diode. Since the overshoot probably doesnt really double though, perhaps a 1.5 x rating would be ok, so for a 240v system a 600v diode would probably be ok.

In your case however you already have overrated diodes, so there must be something else going on here. It would take more than 500v of a spike to kill those diodes, so it could be that they are overheating perhaps. You might want to check the heat sink being used.
You could look for spikes when you turn the device on and off, watching closely on a decent scope.

I have a number of basic power supply units that use a mains tx with a centre tapped secondry with a 35A, 600V bridge rectifier … and these supply equip with a listed load of a max of 8A 24 V. (being centre tapped they only use half of the bridge)

Click to expand...

The highlighted sections in the above quote make it sound to me like you have a transformer that steps down from mains to ~24 Volts, and the rectifier is on the secondary as would be expected. Is this true? If so, chances are it is not overvolting the diodes, but rather they are overheating. You should replace with higher current diodes. The type of diodes you mention are fine. I have used that type in center taped transformer topology before without problems. You should note though that the specific part you have chosen can have up to a 2 Volt forward drop, which will have nearly 20 Watts heat loss at 24 Volts 8 Amps. I would look into a diode with a better fit, i.e. lower forward voltage drop.

If I'm wrong, and you are in fact trying to rectify mains voltage, then yes, you will need to move up in voltage rating from what the original diodes are specified for as MrAl suggests. And it won't hurt to move up in Amps and switching speed either if you can.


Edit: I expect this is about the circuit in question?

You could use a fast (recovery) diode is place of a normal diode, but it has no advantage at line frequencies. Actually the fast recovery (turn-off) could increase the value of any inductive spikes from the diode turning off and increase EMI from the circuit.

As mentioned earlier going short circuit is typically caused by over voltage or over heat or mixed combination of the two. How big or small of heat sink are these connected to if any?

You guys are awesome … thank you

I agree, the original rectifier SHOULD last forever.

Could it be that the load causes a very large surge current spike? You know, like a large capacitor bank?
It is only a guess.

To your original question, yes, you can use them as replacements.

Click to expand...

Thank you Fernando. I’m not really sure about the load … I will ask some more questions.

I have only the heatsink with bridge rect, cct breaker and terminal block in my w/shop.

In your case however you already have overrated diodes, so there must be something else going on here. It would take more than 500v of a spike to kill those diodes, so it could be that they are overheating perhaps. You might want to check the heat sink being used.
You could look for spikes when you turn the device on and off, watching closely on a decent scope.

Click to expand...

Thanks MrAl, the circuit is as suggested by ( )blivion and so I cant see normal supply voltage causing the problem … and as far as overheating, they are mounted on a fairly substantial aluminium plate.

A number of the devices look really new … and don’t look heat stressed at all … you know, with discolouration and hardened wiring etc from a hard life. That makes me think they had been replaced fairly recently and didn't last long.

The highlighted sections in the above quote make it sound to me like you have a transformer that steps down from mains to ~24 Volts, and the rectifier is on the secondary as would be expected. Is this true?

Click to expand...

Hi ( )blivion … Yes … the secondry is about 52V centre tapped. Your diagram is excellent, thank you.

Thanks to crutschow and tcmtech as well ... I really appreciate all of your input.

Cheers, Wes

These diodes are probably better for what your doing.
DSA120C150QB
They should be plenty overrated voltage wise, and they have plenty of current handling capability. But they have a voltage drop of only 0.93 volts. Which will give you half the power wasted in the rectifier compared to the ones you had already chosen. Half the power wasted means half the heat in the part and more directed to the thing you are powering. They are soft switching too, which alleviates the concerns crutschow brought up about EMI.

Edit: Looks like the part above is actually only purchasable in lots of 120 units. My mistake, I'm thinking this is a better part instead.
STPS60SM200C
Or this one
APT60S20B2CTG

Western said:

...
I have a number of basic power supply units that use a mains tx with a centre tapped secondry with a 35A, 600V bridge rectifier … and these supply equip with a listed load of a max of 8A 24 V. (being centre tapped they only use half of the bridge)

Clearly you would expect these bridge rectifiers to last forever because they are vastly over rated for their load … but they don’t. Lots of them fail .. go short circuit.
...

Click to expand...

Toroidal transformer? What size caps after the bridge rectifier?

A properly designed 8A continuous PSU would have deliberate impedance added in series with the bridge diodes to reduce peak currents, reduce startup currents and increase diode conduction angle.

Can you show the circuit? It's likely all you need is some resistance before the caps, and/or increased heatsinking of the diodes. A photo of the diode heatsinking would help too.

Startup current will be an issue if you power the supply down regularly, if so then you might consider a soft start circuit.

A really good thing you can do to increase reliability is to put ZNR's across the filter capacitors.

I fixed an amp that kept coming into a shop before I got it with the same problem every time. One of the amp IC's were blown. Turned out that the supplies would spike over the absolute max limit on power up.

The first time I saw this was on a Auger Spectrophotometer. All of the power supply caps has ZNR's attached to them.

I had to do the same with a power amp that I built because of 40,000 uF of capacitance effectively on a 50 V DC power rail. Without the ZNR's, the optocouplers I had across the supply as part of the slow turn-on circuit would die.

To make things interesting we had a Multi-Channel Analyzer at work (sorts voltage pulses by height) which we had service manuals for and it kept blowing one of the boards. I notice no power line protection stuff at all in the unit. The manufacturer stated that their specifications state 120 V 60 Hz (transients, whatever) and if the power deviates from that it's not their problem.

We had also modified an X-ray goniometer to be computer controlled and the manufacturer was aware that the RS232 drivers would fail every once in a while. Those parts are only like a $1.00 each. The permanent fix was to optoisolate the RS232. With 100 kV power supplies nearby that was the prudent thing to do. No issues.

We also had a thermocouple scanner used nearby 1000 W IR lamps. The scanner would quit scanning, Used elsewhere, the scanner was fine. Turns out the board had h=been designed for a regulated power supply, but the parts were left out. Two transorbs fixed that issue.

There are reasons. In a 30 kW electron beam evaporator, the regulator driver transistor would die regularly. That was basically traced to loose connections on the HV voltage divider. This also caused value changes in the divider. 15 kV at 1.5 Amps shunt tube regulated is nothing to sneeze at. Tightening the connections and using a little Locktite 222 fixed that issue.

Toroidal transformer? What size caps after the bridge rectifier?

Click to expand...

I know this will sound strange, but I don’t really know.

Even stranger is that the equipment has been sent from o/seas for me to fix.

I have done a reasonable number of repairs for this particular company (bit more sophisticated than these things) … and these were in the latest batch.

I have a dickens of a job getting any information out of anyone … so I just do what makes the most sense, send them back and they pay me.

A properly designed 8A continuous PSU would have deliberate impedance added in series with the bridge diodes to reduce peak currents, reduce startup currents and increase diode conduction angle.

Can you show the circuit? It's likely all you need is some resistance before the caps, and/or increased heatsinking of the diodes. A photo of the diode heatsinking would help too.

Click to expand...

The circuit (for the bit I have) is shown above in ( )blivion’s post … except I don’t have the transformer, and there is no capacitor or resistor across the output.

Really it's just a bridge rectifier mounted on an aluminium frame with a couple of terminal blocks and an 8 A circuit breaker.

The same company that made the above mentioned modules, also makes a device that I have been fixing for years.

That is fed with a 52V CT Transformer also … into two 6A diodes and while a small amount of current is bled off (and filtered) to run the timing electronics, the unfiltered output from the diodes is switched by some normal T03 transistors to pulse some 24v solenoids on and off.

So … the 50hz unfiltered supply is switched at around 1Hz and 50% duty cycle.

These things run maybe up to 20 hrs a day for 20 years plus … so despite the lack of sophistication, they run ok.

The first one I ever saw had shorted diodes, and as I didn’t have any suitable ones in stock and the customer was in a desperate hurry … I mounted a 35A 600V bridge rectifier on the side plate and used only half of it.

As I started to see more of them, I just did the same thing and never really thought anything of it. Have fixed maybe 100.

Fast forward 4 or 5 years and I got a couple back in with the bridge diodes shorted !!!

Again, I didn’t really think too much and bolted in a 50A 600V bridge.

Blow me down if 12 months later, a couple of them came back in with shorted bridges again.

At that point I stopped and thought about the issue.

How come the majority of them could run 20 years with absolutely no problem … with 6A diodes in them … and these few were blowing up 50A diodes.

I figured it had to be voltage spikes of some kind getting in the system. Other, electrically noisy equipment or poorly located wiring perhaps.

Now, these original units have been superseded by some later ones, which despite being micro controlled, adjustable via a comms line etc … still use the same system of feeding through raw unfiltered dc to drive the solenoids.

What is different though, is that they use some TO220 D8020L, 20A 800V diodes in the same location as the original type unit with 6A diodes. And … they have a couple of varistors across the input supply to (I assume) take care of any noise coming in.

Having seen all that, I figured I’d do the same thing and promptly started fitting the D8020L diodes to the old units, along with a couple of the same varistors.

I have had virtually none come back in ever since … so I love those diodes and varistors.

It’s also why I wanted to find better diodes to replace the bridges in these latest modules I started this thread about … as well as fit a pair of varistors across the supply input.

If I stuff up, it has the potential to mess up my relationship with my o/seas customer, and I'd prefer not to do that.

Anyway, I finished them off this afternoon and need to pack them all and send them back ... so time will tell.

Thanks everyone for your input, you are a wealth of knowledge.

Hummm... More to the story after all. No load instantaneous overvoltage could still be the problem. If it was me, I would unpack them and do a bit more work.

You can't hurt anything to do what KISS or Mr RB are talking about. Or a softstart as dr pepper recommends. The simplest softstart one can make is just a high wattage, low resistance NTC thermistor. These have a high resistance at first, so limit inrush current. But as the current starts to flow, they heat up. Having negative temperature coefficient (NTC) means they lower their resistance when they heat up. The subsequent drop in resistance allows for full system current with minimal circuit impedance. It's a single cheep part, I see them all the time in CRT displays. The only thing is they can get quite hot. Sometimes they would even get hot enough to unsolder, and thus need to be welded/socketed to be reliable.

*HERE* is what I'm talking about. If you have a business, you can even request free samples from them. But you can also order them from DigiKey, Newark, Farnell, etc. with the links on the left hand side. They make units that you can use to limit current for either the primary or the secondary. Either circuit location should work to limit inrush current through the diodes.

Also a largeish inductor on the o/p, would significantly reduce noise and spikes in either direction.

"loose connections" can do a lot of damage. Most looks like it's taken care of by the Varistors/ZNR's as it should be.

Wrong parts used happen too. This lamp power supply I used everyday had a 10 A bridge in it to operate a 22 A DC lamp. Oops! The mfr fixed it for free, but they had to retrofit a new transformer and half bridge diode assembly.

Western said:

Hi ( )blivion … Yes … the secondry is about 52V centre tapped. Your diagram is excellent, thank you.
Cheers, Wes

Click to expand...

So Why are you using a Full Wave Bridge?
Since only two of these diodes are working, it could give Uneven Heating inside the Bridge Rectifier, causing failure.
I would recommend 2 Single diodes.

As To Those High Speed Diode "MADE FOR HIGH FREQUENCY", I would Not recommend them.
My point is: If High speed diodes were as effective at Low Frequencies, Than why do they make or use low speed diode at all.

And are you using a Silicon Grease between the Diodes and the Heatsink?

So why are you using a Full Wave Bridge?
Since only two of these diodes are working, it could give Uneven Heating inside the Bridge Rectifier, causing failure.
I would recommend 2 Single diodes.

Click to expand...

I agree he should NOT use a full bridge. However, the dual packages are a perfectly acceptable replacement. Single diodes would likely be less well matched. The dual packages are made with the intent that they are very well matched, and they will follow each other as they heat up. That's why they are in a single package to begin with.

Though it's probably true that the OP can get lower forward voltage drop if he goes with single diodes. Which would be better for heat and efficiency.

If High speed diodes were as effective at Low Frequencies, Than why do they make or use low speed diode at all.

Click to expand...

I would say cost. Specifically bulk cost divided between each unit. I mean, extending your logic they don't need to make 100v and 10v capacitors because they make 1000 volt units. Same can be said for high voltage diodes, why make low voltage ones if high voltage diodes exist? Because it costs more to make higher performance parts and keep the rest of the specs the same. The industry revolves around economy of scale. If they can shave 0.5 cents per diode, that adds up over tens of thousands of diodes to be significant savings.

Anyway, as far as I know, diode switching speed, just like max voltage/current/heat, is a limiting factor for certain applications, but it is an open ended requirement. You can use faster diodes in a slower circuit if you want. Just like you can use higher max voltage/current/temp parts. Though as crutschow pointed out, there may be other unforeseen consequences that one may or may not have to deal with.

And are you using a Silicon Grease between the Diodes and the Heatsink?

Click to expand...

This is an excellent point, I would also like to know the answer to this.

And are you using a Silicon Grease between the Diodes and the Heatsink?

Click to expand...

This is an excellent point, I would also like to know the answer to this.

Click to expand...

Ok, I know I've clouded the issue by talking about two distinctly different devices ... albeit from the same manufacturer.

The item I started the thread about ... was designed with a bridge rectifier in it despite the fact it only used 2 of the 4 diodes.

Of the 5 x devices I have here, only one had any heatsink compound ... and that was only a smidgeon! So heat could have been an issue, but as I said earlier (and from my reasonably trained eye), there was no evidence of overheating.

I certainly used an appropriate amount of grease when I fitted the dual diodes.

..............................................................


Now ... speaking about the second device ... in which I initially fitted a bridge rectifier in place of the 2 individual ones ... again, I have always used heatsink compound to be on the safe side, so I dont see overheating being the cause of the bridge failures.

"loose connections" can do a lot of damage. Most looks like it's taken care of by the Varistors/ZNR's as it should be.

Click to expand...

Interestingly enough, as I was checking everything I could think of on the original items ... I did find one that had it's connection from the centre tap to ground that was completely open circuit ... despite the fact it was bolted securely with star washer and two nuts!

Turned out the nice shiny anodising was insulating the solid eye lug being used.

I then had removed the earth lug on each unit, ground back the anodising and rebolted them.

I'm not convinced that would have caused the issue, but at least all five of them now have a definite earth connection to the frame.