Minimizing rectification spikes in transformer secondary?

[leadbelly]

I am working on adding a DC supply to an existing AC circuit.

At first blush, a simple bridge of 4 Schottky diodes and a large filter cap will do the trick, maximizing the rectified DC voltage.

Thinking about it some more, I am worried about the effect of rectification spikes on the AC loop, as there is a component there that might be sensitive to spikes.

I would say as a design criteria minimizing rectification spikes on the AC side would be highest priority, with maximizing DC voltage next priority.

Would someone be so kind as to walk me through what basic additional elements would achieve this? For example, diode type, snubber capacitors, inductors, etc. Not wanting a walkthrough of a scope trace, more like a 1st principles, belt and suspenders answer.

 

[rayma]

Treat the filter capacitor connections as a Kelvin arrangement. The input traces/wires to the pins
go in the opposite direction from the output traces/wires from the pins, and share no traces/wires.
Excessive filter capacitance will increase further the magnitude of the charging current spikes.

 

[jhofland]

The typical transformer secondary and rectification form a resonant circuit that gets excited when the rectifiers turn off. Mark Johnson et al have several postings on the subject. The usual method of controlling it is a capacitor in parallel with a series RC across the AC side to damp out the resonance. Values I have successfully used is a capacitance of 10nF in parallel with a series RC of about 100nF and 150 to 200 ohms. Mark designed a fixture for using a scope to dial in the resistance value to an optimum number to damp out the response. But something even moderately close stops the ringing in a cycle or two.

 

[Elvee]

Would someone be so kind as to walk me through what basic additional elements would achieve this? For example, diode type, snubber capacitors, inductors, etc. Not wanting a walkthrough of a scope trace, more like a 1st principles, belt and suspenders answer.

Inductors or resistors will work, even small ones. Resistors look the simplest, but you probably don't want them, as they turn the excesses into heat.

Inductors spread the charging spike on a longer period, and don't waste heat.

They will reduce the peak voltage slightly, but not like a resistor because the lower peak current is compensated by a longer conduction time.

Different arrangements are possible: common, in the DC output, common in the AC input, individual for each diode or conduction path.

Note that by combining the right topology with the right values of filter cap and inductors, you can even boost the output voltage very slightly, thanks to a pseudo-resonance effect.

 

[leadbelly]

Many thanks for your detailed replies.

 

[Mark Johnson]

An inrush current limiter in series with the primary (such as the CL-60 that Nelson Pass includes in his First Watt designs) will reduce spikes AND limit inrush current. If your transformer's primary current is not anywhere close to 5 amperes then you will want a different ICL than the CL-60; the datasheet offers many alternatives.

An inductor connected in series between the diodes and the filter capacitors, will reduce spikes even further. The free software PSUD2 can help you experiment with different inductor values to see what they do to the output voltage waveform and to the diode current waveform.

And a C+RC snubber, connected across the transformer secondary, will eliminate oscillatory ringing.

 

[tvi]

Simple-math-transformer-snubber-using-quasimodo-test-jig
Heaps of info

 

[trobbins]

Leadbelly, have you tried any DC heaters before, and been able to measure any spikes getting in to the audio chain?

 

[leadbelly]

Leadbelly, have you tried any DC heaters before, and been able to measure any spikes getting in to the audio chain?

This is about as far from DC tube heating as you can get, it's about putting LED lights on my very old dirt bike.

To answer your question, no I haven't, but SY has IIRC, albeit for multiple secondaries, not my case which is single secondary (actually alternator winding).

 

[trobbins]

Ok so it isn't a transformer secondary winding (as would typically be the topic in diyaudio), it's a generator winding. Is it a bicycle generator or motorbike generator?

 

[Elvee]

A generator has a huge leakage inductance: you don't need to add anything.

In fact, if you want to recover a bit more of power you should compensate it with a series capacitor adjusted to resonate at low-mid speeds

 

[sq225917]

And Q-speed diodes for the quietest spike recovery.

 

[Mark Johnson]

The Q speed diode "LQA16T300" had the 4th best test results among high current diodes, in the Linear Audio article about soft recovery and ringing

Soft Recovery Diodes Lower Transformer Ringing by 10-20X | Linear Audio

 

[TonyTecson]

i avoid the full wave center tapped psu designs especially for tube amps, transformer utilization is not so good...a full wave bridge is even better...lower voltages, easier to wind traffos as insulations are not so strict otherwise...

rectification for DC distorted the current waves in the traffo secondary no matter what you do, it is unavoidable consequence of rectification...

a snubber, an NTC to limit inrush when using substantial amounts of filter caps, things you can do to help mitigate some issues..

 

[TonyTecson]

high speed diodes on the 60hz source? really? truly?

 

[trobbins]

The OP is using generator as a source, but hasn't indicated if it is a little 3W bicycle style, or a motorbike generator.

 

[Mark Johnson]

Soft recovery diodes are quite beneficial; they really do lower transformer ringing by 10-20X. Really. Truly. Measured and photographed and published.

The fact that diode manufacturers only sell soft recovery diodes which also happen to be "fast" (low Trr), is in my opinion, an accidental coincidence. A pleasant surprise (or unpleasant depending on your viewpoint).

The downside of soft recovery diodes is their very large forward voltage drop. Most of the very soft recovery diodes, are terrible in this department. Including the LQA16T300. Download its datasheet and compare to MBR10100. You're gonna need a bigger heatsink. apologies to Roy Scheider / Jaws movie 1975.

 

[TonyTecson]

i never heard any of my transformers ringing, maybe because i never really tried,
when you overload your traffo to near saturation you will hear the loud buzzing sound...

if you mean spikes on the rectified dc, a choke can take care of that...or a snubber even....

the secondary current waveforms are distorted as a result of rectification,
and when using either a 50 or 60hz power line, full wave rectification gives you 100 or 120 hz of ripple frequency....

i used ordinary silicon rectifiers and never regretted it...

but here is the deal, i used diodes rated much higher than required, diodes are dirt cheap nowadays, do not skimp...

 

[Mark Johnson]

If you don't want to spend €2,99 on the Linear Audio article, very similar measurements and scope photos and thorough explanations of transformer ringing appear in this newly released (Jan 30, 2020) textbook:

The Art of Electronics: The x Chapters: Paul Horowitz, Winfield Hill: 9781108499941: Amazon.com: Books

Flip to section 9x6

 

[leadbelly]

Thanks for all the replies.

To answer the questions raised, it is a 45W motorcycle alternator. As I already stated, the concern is about protecting the AC devices from spikes; the DC devices are LED lights and are hardy. As I stated, looking for a belt and suspenders design, not a path to ramp up if problems encountered. But forward voltage drop is a big concern.