Measurement vs. Simulation

They can be, but usually aren't. 50/60 Hz isn't fast enough to optimally snap a diode.

10 megs of probe resistance can indeed shift the slow waveform at the transformer secondary. AC coupling the scope can eliminate that.

The snap happens immediately after the diodes have been conducting hard, conducting amps. No scope probe current is going to "build up" during the power diode forward conduction interval.

Sorry, it's not the scope that's confused.

You deny the existance of pn-junction diode reverse recovery. Lots of other people don't.

Have you ever seen a scope with a front-end "protective device that fires"? I've studied schematics of maybe 100 oscilloscopes, and I've never seen one, nor have I ever seen a scope generate a display spike, whether it was overloaded or not.

And a scope with an auto-zero routine??!! Please.

For Pete's sake, do a little math. Compute the energy in the spike, and see if scope probe capacitance can account for it.

John

It might be worthwhile Googling on "soft recovery rectivier diodes application note"

I dug up this - desperately boring - document. There are others.

--
Bill Sloman, Nijmegen

--- LOL, its an HP54602B, a nice little 150MHz DSO, hardly crap.

But, just to try to make you happy, I've posted a few shots from a Tektronix 2465A ,a nice big 350MHz analog scope to abse under the same subject as this thread.

The pictures are a little messy because I wanted to make sure you could see the spikes. They're very hard to see on 001, but I wanted to include that one because the downturned "hook" on the upper trace is actually the result of the secondary floating and, I think, an imbalance between the diode capacitances in the bridge. If I touch the bridge at a certain place or swamp the diodes with parallel caps the hook goes away. Matter of fact, when I touch the bridge the hook drifts slowly upward until it becomes the trailing edge of the waveform, then when I move my finger away the hook slowly drifts into place again.

The probes are both Tektronix P6131's: 10:1 attenuation, 10Megohm Rin paralleled with 10pF.

Anything else you want to bitch about? ;)

---

--- You'll need to take that up with him.

---

Sorry Johns, had to write that!

--- Apology accepted. On my end anyway. ;)

-- John Fields Professional Circuit Designer

Dang, you *are* snapping the diodes hard enough to forward bias the opposite diodes... big time!

It occurs to me, on reflection, that a diode need not be doped to have a "snap" (step-recovery) profile in order to make this sort of power supply spike, it merely has to store a lot of charge.

John

Yeah, but I seem to recall some instrument guy from California talking about putting 40V forward bias on a rectifier.... 1N400x diodes take

*forever* to turn on.

Cheers,

Phil Hobbs

So, in theory, the reverse recovery glitch can more than forward-bias the opposite diodes. JF's waveforms look like that's not happening... the flyback is clamped pretty flat.

In our drift step-recovery thing, we apply +48 througn an inductor to a fast-recovery diode and wait something like 80 ns for the current to build up to 20 amps. Then we reverse the supply to -400 and wait for the diode to snap. It snaps at about -40 amps, and, as I recall, most of the injected charge is recovered, ie there wasn't a lot of spontaneous recombination. Lights neons by proximity.

John

The slope of the current through the conducting diode decreases with

3.1mA/us with the 9 Ohms load. The reverse current into the leakage inductance can not be higher than 1.5mA if the time constant is 1us until all minority charge carriers have been cleared. With the 250pf capacity this corresponds to a 6V "snap".

Now look at the 5817 plot unloaded. Almost all the charge current comes from the probes, only that side where probe1 is sitting provides the load current. This results in a DC current through the transformer. every time the AC-current is in one direction only and doubles the DC-component.

We know you John, good you can see it.

I don't deny anything, I just want to understand the pics.

The internal capacitor is charged up to the common mode limit, so no negative voltage display is possible. The 100nA flowing into the probe will be 1uA or even 10uA(with 100k bias resistor to gnd from the differential amp input) flowing continuously into gnd through the input protection diodes. Since the spikes are synchronized with the mains and only in one polarity, you might even damage the power supply of other appliances plugged into a close outlet. The 15V spikes occur almost in the mains maximum voltage and will put through the transformer a nice 120Vspike of 3ms duration.

Well it seems this particular scope has separate paths for DC and AC signals. The AC coupling cap gets charged up by the unipolar pulses and the reference input will need to follow, either with a DC-servo or it has a cap as well. The resulting unipolar current into the gnd-clips of the scope is a multiple of it.

Well, you didn't get it yet

--
ciao Ban
Apricale, Italy

Well either the operator or the scope, better to say the scope, isn't it. See my other replay where I explain the limiting of the displayed waveform.

You should be happy when someone makes you aware of shortcomings. The bitching we leave to Larkins brain, OK.

--
ciao Ban
Apricale, Italy

--
It\'s better to say what you think instead of muddying the waters
with feigned "niceness".

The fact is, all you\'re doing is trying to shoot the messenger
because the data contradicts what you think it should be.

Since the impedance of the probes will always unbalance the circuit, DC or AC-coupled, we will need to provide a compensation impedance to the positive rail. Here we have a proof that actual measurements are misleading when done without proper care. Of course Fields should have seen this by himself before posting. And I wonder why everybody else gulps those fairy tales of Larkin and Win.

10M || 15p ___ +-----|___|------+ | ___ | | +---|___|------+ | | | Rload | | +-->|--+----+-+-----+ | | | | | | hot -. ,---+-)--+--||--+ | | '-' ' | | | | ' +--|

Now that's funny.

Good grief, the spikes are across the transformer's **leakage inductance!** To couple well into the line, they would have to be

*inside* the leakage inductance. But wait... then they wouldn't exist at all...

The scope trace is entirely on-screen, the signal is benign, and I've personally never seen a scope do anything like this. Besides, JF's waveforms are similar to Helmut's simulations, and Helmut's simulations didn't include a 10:1 probe or a DC servo or input protection diodes. I've seen these spikes in all sorts of rectifier configurations, with all sorts of scopes.

I admit that the rectifier configuration, a 4-diode bridge fed from an untapped secondary, has a very complex transformer secondary waveform, which is why I started the thread, and the complexity confuses the diode recovery effect. It's a lot easier to understand a simple grounded halfwave rectifier (ground one end of the secondary; other end drives a single diode, then a parallel R-C load back to ground.) This simpler circuit will also demonstrate reverse-recovery spikes with properly slow diodes, and nothing limits their amplitude but capacitance and losses in the leakage inductance.

Try it.

John

Please Fields, look again at the pics you have posted, especially that one with the big spikes, where the spike from the top doesn't even get channel2 out of saturation. And then *think*.

--
ciao Ban
Apricale, Italy

positive

done

"Ban" schrieb im Newsbeitrag news:Y5Xhg.24353$ snipped-for-privacy@news.edisontel.com...

rectifier

Hello Ban,

I have done last week some measurements on a rectifier circuit having the discussed bridge rectifier configuration.

I tried with different rectifiers. Two of them were bridges in one package and one used 4 single diodes. The bridge rectifier KBU4B produced no spike. The other one(MDA202) produced a medium height spike and the

4-single diodes of type 1N4007 produced a spike nearly as high as the peak AC voltage in everey cycle. I will send you the picture. It's a photo taken from a fast analog scope screen (BW 250MHz). The medium height spike has had a width of about 1us and the large spike has had 2us. I used a load current of 1.5A. The transformer has a secondary output with 20V/1.5A. Primary is 230V. When I heavily touch the secondary winding(point of measurement) with my fingers, the spike was only decreased by about 5%. It's really the reverse recovery time and the switch off (soft or hard) characteristic of the diode causing this effect. The spike happens exactly at the time when a diode in the bridge switches off. Long recovery time and fast switching off characteristic makes the worst case. You can easily reproduce this spike. Use 1N400x diodes in the bridge to get big spikes. Don't forget to connect a load (e.g. 1A) at the output of the rectifier. No load current, no spike. A 10MHz oscilloscope is good enough to capture the spike.

Best regards, Helmut