Electrolytic capacitor reverse voltage question. / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Electrolytic capacitor reverse voltage question.

1 2 next

Move Thread

LAN_403

Ash Small

Thu Aug 16 2018, 03:17PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Thanks to a fellow 4HV'er I have 40 or so 40V 4,700uF stud mount electrolytics.

The circuit I intend to use them in has a large inductive load, and some SiC power JFET's.

I anticipate that when power is shut down, the circuit will want to 'ring', so I need some type of 'snubber' (?) circuit to overcome this issue.

The simplest 'protection' I can come up with is doides connected between the bus bars of the capacitor banks, connected in such a way that if the negative side of the electrolytics goes positive, the current is shorted to the positive side of the electrolytics via the diodes.

Numbers involved are 40V electrolytics operating at a steady thirty volts (141,000uF plus a resevoir bank of 47,000uF), resonant frequency with 100mH choke 1.34Hz, 5A steady current.

My question relates to 'how much reverse voltage is safe for the electrolytics?'

I'd prefer to use stud diodes due to their ruggedness, generally superior current specs. and much better reverse current specs, as well as their bulk, but the best forward voltage figures I can find at a reasonable price is 1.1V.

TO220AB diodes can get down to close to 0.6V. Will a reverse voltage of 1.2V harm the electrolytics?

I assume standard switching speed devices presumably designed for 50/60Hz will be fine at 1.34Hz

Datasheet of the 'affordable' stud diodes here :-

Datasheet for capacitors here :-

2Spoons

Fri Aug 17 2018, 03:03AM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

Why not use a Schottky? Plenty to choose from.
eg

Your diode will only have to absorb ~1.25J. Or 5W for ~0.25 seconds (these are only rough approximations)

Oh, and your stud diode Vf is only 0.9V @ 5A - check the graph. 1.1Vf is at 35 A.

Internet says electros can handle -0.5V safely (for short periods) but dont exceed -1.5V or things will break.

Ash Small

Fri Aug 17 2018, 12:25PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Interesting comments 2Spoons.

The one you link to is only rated for 5A up to ~50C, if I've read the graph correctly, so is possibly not ideal for protection purposes.

My research on the internet suggests that -1V to -1.5V can be destructive, I'm asking mostly for clarification, etc.

I'd sooner use a DO5 stud package than axial or TO220AB, it will be fixed to a copper busbar (flattened 28mm copper water pipe), and I believe the DO5 package will disperse any heat more effectively, but I'm still considering all the options.

I've found another at 1.2V forward voltage drop, with much higher current and voltage specs, I'll have to check the graph.

I've also found a trench Shottky in a DO5 package with a Vfd of 0.75V, again I should check the graph.

These are more expensive than the 'affordable' ones I found, but I guess to replace the caps would cost hundreds, so it's worth spending a bit more on protection.

I'm also unsure of what maths to use in this case, where voltage across the cap bank and current through the inductor both start off 'in phase', so to speak, rather than the 'normal' situation where Vc and Il are 90 degrees out of phase, both should fall together, but I anticipate some ringing, if it's not effectively 'snubbed'.

The current proposed design consists of four cap banks, each with ten 4,700uF caps, three banks connected in parallel, as a storage cap, the other bank forming a resevoir cap, with diodes across each bank, so theoretically the current will be shared, but ideally, the 'protection circuit' needs to be designed so each diode can take the full current, with a decent safety factor.

I guess I need to go back and study more data sheets

2Spoons

Sat Aug 18 2018, 01:35AM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

There's really only one place you need the diode and thats across the feed to your inductive load. If you use several diodes they will not share current equally - much better to use one larger diode. Speaking from experience in repairing a fence energiser design that used parallel clamp diodes across the storage caps.

I read that datasheet I linked as 6.5A CONTINUOUS current, without heatsink. Your application is a momentary pulse. At any rate, that was just one I plucked off the list without digging into it too far.
Why don't you like TO220? they'll do 2.5W in free air, and a lot more than that bolted to a heatsink.

Sulaiman

Sat Aug 18 2018, 01:24PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I prefer rectifiers such as the TO247 MUR3020 (or newer/better)
Bolt to +ve bus bar and no insulation required.
Connecting both diodes in parallel is common practice.

Larger stud diodes need nuts. bolts, washers, ring-crimps and most of all
tightening the nut/bolt it is easy to damage the seal.

Ash Small

Sat Aug 18 2018, 03:06PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

There are two inductors in the circuit, one in the 'Pi' filter between resevoir cap and storage cap, and one as part of the load, so both cap banks (resevoir and storage) could 'see' reverse polarity.

One issue I foresee with either axail diode or TO220/247 is soldering the legs to the busbar, this will involve considerable heat, which could damage the diode. This issue is avoided using a stud diode, with a ring connector on the lead. (soldering will involve putting a lot of heat into the copper busbar, the legs on the diode would be subjected to elevated temperatures for a lot longer than normal)

I've pretty much decided on the MBR6045 now, at 5A it has a Vfd of 0.4V (which drops as temperature rises), is a DO5 package, 45V 60A, with not too much leakage current. (according to the sims, voltage never gets above about 30V)

Both inductors will be conducting 5A when it's switched off, so it's reasonable to assume I need protection for each cap bank, minimum two diodes. The caps are what need protecting, so it makes sense (to me) to put the diodes as close to the cap banks as possible.

The larger cap bank consists of three banks of ten 4,700uF caps in parallel, while one 60A diode should provide plenty of protection, diode failure 'could' be catastrophic, if the 'magic smoke' gets released from all 141,000uF simultaneously.

Two diodes is Â£30 plus delivery, four diodes is Â£60 including delivery, it seems to make sense to put one diode on each of the four cap banks, even though the three doides on the larger cap bank are in parallel.

This is a lot more than I originally envisaged spending, I already have hundreds invested in the project (components for tone circuit were Â£100, SiC power JFET's another Â£100, chokes over Â£100, etc). An extra Â£30 for two 'extra' diodes seems trivial.

Also, if they do end up sharing the current, Vfd is even smaller.

Ash Small

Thu Aug 30 2018, 06:31PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

I've been thinking about this some more, regarding current sharing.

The graph in the post above shows how Vfd increases with current.

Now, I guess one will start conducting first, but as more current flows, Vfd will start to rise.

At this point, any that aren't conducting will have a lower Vfd, so will start conducting.

They should settle at the point where all have the same Vfd, even though they 'may' not be sharing the load equally.

They should end up sharing the current fairly equally.

Any perceived flaws with this reasoning?

2Spoons

Thu Aug 30 2018, 11:00PM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

You're also up against the diodes' negative temp coefficient. So the one that conducts first will heat up fastest, and its Vfd will drop fastest. It then becomes a race between Vfd dropping due to heating and Vfd rising as the current rises. And if you are doing this repeatedly then the diode that got hottest on the previous pulse is going to get hit harder the second time around, if it hasn't had time to cool.
Bolting diodes to the same heatsink helps, but you still can't get around the thermal transient problem.
Similar issues occur with other semiconductors - eg paralleled BJTs. It can even occur within a single die - with large thyristors there is a failure mode where part of the die starts conducting first, that bit gets hot, its Vfd drops, and all the current gets focused through that small section of the die resulting in destruction of the device. This is usually the result of inadequate gate drive or dI/dt being too high.

Parallel diodes, while not a total no-no, needs to be approached with caution. Its generally regarded as not being a good idea - except maybe for redundancy I guess.

Dr. Slack

Fri Aug 31 2018, 05:15AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

You could force pulse sharing between diodes with a centre-tapped transformer.

As you are only looking for mV, you don't need many Vs for the core. However, as you're doing this to minimise the voltage you see across the diode shunts, you need exquisitely low leakage inductance, as that ends up in series with the diodes. Helping you in this is that you need neither voltage isolation nor low capacitance between the windings. This means the windings should be made from a bundle of parallel wires lightly twisted together, with alternate wires in the bunch selected for the A or B winding.

Are you familiar with yak-shaving?

Ash Small

Sun Sept 02 2018, 08:55PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Interesting concept with the transformer.

I just like to understand as much as I can about what is going on.

Regarding the thermal coefficients, maximum current in the circuit is around five amps. Each diode is rated for 60A, and bolted to a copper busbar over 60mm wide and at least 2mm thick. I wouldn't expect a lot of heating, but I realise the heat needs to be conducted away from the die.

This is one reason for using such large diodes, in a DO5 package.

Redundancy is also a factor here. I've experienced capacitors blowing in guitar amps before, after overloading it, a fireball came out of the back. While I may be being over-cautious, I'm aware it's not a good idea to reverse charge electrolytics

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob