Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Bridge Rectifier won't Rectify

Status
Not open for further replies.

Swend

Full Member level 4
Joined
May 14, 2019
Messages
237
Helped
10
Reputation
20
Reaction score
9
Trophy points
18
Location
Roskilde, Denmark
Activity points
2,304
Hi friends!

I made a simple diode bridge rectifier, except that each diodes is made up of ten STTH812 in series, and those are ultrafast 1200V 8A diodes with 100nS rrt and 400nS frt. But it doesn't seem to want to rectify properly please see below oscillograms and multiply v/div by 1000 to get the correct voltage.

INPUT__________________OUT 100Mohm load___________OUT 30Kohm load

SDS00001.png

SDS00003.png

SDS00002.png

I can't figure what is going on here and why it won't work.
 

Maybe you should tell us what we’re looking at. Maybe you should tell us what your input is. Maybe you should post a schematic.
 

Maybe you should tell us what we’re looking at. Maybe you should tell us what your input is. Maybe you should post a schematic.

The three oscillograms were side by side when I submitted the post, but now I see they are on top of each other, so the original post has lost it's meaning. So let's start again from the beginning.

schematic:
Screenshot from 2019-08-17 20-58-45.png

Each diode in the above bridge rectifier is made up of ten STTH812 in series, and those are ultrafast 1200V 8A diodes with 100nS rrt and 400nS frt.

Here is the input of the bridge rectifier (or the output of the secondary winding) you have to multiply v/div with 1000 to get the correct voltage.
SDS00001.png

Here is the output of the bridge rectifier when load resistance is 100Mohm. you have to multiply v/div with 1000 to get the correct voltage.
SDS00003.png

Here is the output of the bridge rectifier when load resistance is 30Kohm. you have to multiply v/div with 1000 to get the correct voltage.
SDS00002.png
 

What exactly is driving them, the waveform suggests it is a single pulse and the transformer is 'ringing'.
Are the tabs mounted on a a heatsink? I appreciate they are insulated but the accumulated capacitance of 40 diodes will have some effect at those frequencies.
I would also like to know how you are measuring 16KV with the scope, obviously some probe with a very high input impedance, does it have the bandwidth to handle ~500KHz?

Brian.
 

Hi Brian.

What exactly is driving them, the waveform suggests it is a single pulse and the transformer is 'ringing'.

It is a single pulse, here is the oscillogram from the primary side, you have to multiply v/div with 10.000 (TEN THOUSAND) to get the correct voltage.
SDS00005.png

Are the tabs mounted on a a heatsink? I appreciate they are insulated but the accumulated capacitance of 40 diodes will have some effect at those frequencies.

I'm sure there is capacitance the way I mounted them, but I did not know a better way.
IMG_20190817_221442.png

I would also like to know how you are measuring 16KV with the scope, obviously some probe with a very high input impedance, does it have the bandwidth to handle ~500KHz?

I'm measuring with a resistive divider 100M/100K as seen in the above picture

I'm using these 100 Mhz probes set at x10
 

It's not surprizing to see large ripple at the rectifier output. Negative voltage as shown is however not feasible. Most likely explanation is the waveform distortion by the uncompensated voltage divider.

The voltage divider tranfer function should be checked with a square wave generator and corrected by compensation capacitors.
 
  • Like
Reactions: Swend

    Swend

    Points: 2
    Helpful Answer Positive Rating
capacitance will dominate for hi-Z loads ... and when the source is hi-Z

- - - Updated - - -

you can buy special diodes for EHV and higher currents - not that cheap - but do work ...
 

I suspect that the voltage is not getting shared equally by the diodes. You need to put resistors in parallel with the diodes so that the applied voltage is divided equally.

The value of the resistors to be decided by the reverse leakage current of individual diodes.

Does it work with lower frequency? Say 1kHz or 10kHz?

Is your scope probe frequency compensated?

In post #5, you show the primary side single pulse. The ringing is coming from the probe or the transformer?
 

capacitance will dominate for hi-Z loads ... and when the source is hi-Z

I calculated Z for the secondary winding to about 60ohm, and the resistive load is 30Kohm (2nd measurement) - does that qualify as hi-Z?

you can buy special diodes for EHV and higher currents - not that cheap - but do work ...

Could you suggest a specific one?

- - - Updated - - -

It's not surprizing to see large ripple at the rectifier output. Negative voltage as shown is however not feasible. Most likely explanation is the waveform distortion by the uncompensated voltage divider.

The voltage divider tranfer function should be checked with a square wave generator and corrected by compensation capacitors.

You are right, I will do that. But what makes you say negative voltages are not feasible here? I'm asking out of ignorance and for educational purposes.

- - - Updated - - -

I suspect that the voltage is not getting shared equally by the diodes. You need to put resistors in parallel with the diodes so that the applied voltage is divided equally.
The value of the resistors to be decided by the reverse leakage current of individual diodes.

Is that how diodes in series are usually done?

Does it work with lower frequency? Say 1kHz or 10kHz?

I don't know, I really can't adjust the frequency that much.

Is your scope probe frequency compensated?

You mean by feeding it a square wave and turn the trim-cap? Yes

In post #5, you show the primary side single pulse. The ringing is coming from the probe or the transformer?

Neither, it's the pulse.
 

But what makes you say negative voltages are not feasible here? I'm asking out of ignorance and for educational purposes.
Rectifier output voltage is clamped to n*diode forward voltage, e.g. a few 10 V in your design.
 
  • Like
Reactions: Swend

    Swend

    Points: 2
    Helpful Answer Positive Rating
You mean by feeding it a square wave and turn the trim-cap? Yes
Are you compensating the probe alone or with the HV divider resistors attached? You need to apply the square wave through the divider then do the compensation. I suspect you have inadvertently added an RC low pass filter in your measurement.

Yes, adding voltage sharing resistors is the normal way to protect the diodes. There is no issue when they are in forward conduction, they clamp their own forward voltage but in the reversed condition there is no guarantee they each 'hold back' the same voltage and it is easy to see some may exceed their PIV rating and cause a cascade failure of the others. Obviously use equal value, high voltage rated resistors, their value should be high but isn't critical.

Brian.
 
  • Like
Reactions: Swend

    Swend

    Points: 2
    Helpful Answer Positive Rating
Here is the output of the bridge rectifier when load resistance is 30Kohm. you have to multiply v/div with 1000 to get the correct voltage.

The first peak is around 8 kv; the current should be around 270mA. The diodes are not rectifying.

In the previous graph, with a load of 100Mohm, you get currents of the order of microamps. Small currents for small times can come from junction capacitances.

With a large current of 200mA (more than that) you see new features in the waveforms. Some diodes are punching through.
 

Is that how diodes in series are usually done?

For low frequency (say 50-60Hz), you can protect the diodes with capacitors and resistors. They help suppress spikes.

The capacitor cannot be used for high frequency. I do not know other ways to protect the diode from the reverse breakdown. It becomes really important when you have diodes in series.
 
  • Like
Reactions: Swend

    Swend

    Points: 2
    Helpful Answer Positive Rating
For low frequency (say 50-60Hz), you can protect the diodes with capacitors and resistors. They help suppress spikes.

The capacitor cannot be used for high frequency. I do not know other ways to protect the diode from the reverse breakdown. It becomes really important when you have diodes in series.

OK, I will order some 1Megs and try your suggestion.

- - - Updated - - -

Are you compensating the probe alone or with the HV divider resistors attached? You need to apply the square wave through the divider then do the compensation. I suspect you have inadvertently added an RC low pass filter in your measurement.

Yes FvM did mention that, and that's what I'm going to do today.

- - - Updated - - -

I checked the supposed 1000:1 divider

This is the uncompensated trace, with the function generator setting displayed
uncompensated.png
What really surprises me here is that its more of a 60:1 divider, 6Vp-p in should yield 6mv but instead close to 100mv.

And this is the compensated trace, I added a 33pF across the lower 100K resistor, I would call that a square.
compensated.png
6Vp-p in and 25mV out, that's a 240:1 probe, and why is it offset with +8mV? Or is that just the small voltages that the scope has problems with?

- - - Updated - - -

And here is the supposed 10000:1 divider

This is the uncompensated one, if I just look at p-p values its a 67:1 divider
uncompensated10k.png

And this is the compensated trace, I added a 55pF across the lower 10K resistor, I would suspect that a square, but the scope doesn't go any lower. With 3.4mV it's a 1764:1 divider.
compensated10k.png
 
Last edited:

The scope traces look suspicious to me. Given that the scope, probe and stray capacitance is effectively the load after 100M of resistance, I would have expected the opposite to have happened, the trace would have a slow rising and falling edge. Adding extra capacitance at the bottom would make it worse, not better. Possibly the 100M resistor itself has a relatively high capacitance so it is effectively more conductive than Ohms law would suggest at that frequency.

Brian.
 

The scope traces look suspicious to me. Given that the scope, probe and stray capacitance is effectively the load after 100M of resistance, I would have expected the opposite to have happened, the trace would have a slow rising and falling edge. Adding extra capacitance at the bottom would make it worse, not better. Possibly the 100M resistor itself has a relatively high capacitance so it is effectively more conductive than Ohms law would suggest at that frequency.

Well, I tried adding capacitance (ranging from 33p to 2.2n) across the 100M, and what I saw on the scope was that the full 6Vp-p from the signal generator appeared across the lower 100K resistor, nice clean traces but I thought that defeated the purpose of a voltage divider, so I tried the lower leg instead and the result can be seen in the traces shown.
 

I remember seeing the schematic for a Tektronix HV probe similar to the one in the photo. The compensation network was really complex, as you can tell from the several visible adjustments.

I concur with the rest of the posters: your uncompensated voltage divider is distorting the waveform.

https://i.ebayimg.com/images/i/222168262591-0-1/s-l1000.jpg
 

Compensation ideally works only for a fixed frequency. Why the frequencies seen in post 14 for the square pulses are so low?
 

I remember seeing the schematic for a Tektronix HV probe similar to the one in the photo. The compensation network was really complex, as you can tell from the several visible adjustments.

That's a nice probe, I will wish for one.

I concur with the rest of the posters: your uncompensated voltage divider is distorting the waveform.

Me too, so I tried to make them compensated, see post #14

- - - Updated - - -

Compensation ideally works only for a fixed frequency. Why the frequencies seen in post 14 for the square pulses are so low?

it's 500KHz - "low" relative to what you mean? I checked the high end frequency, it worked without attenuation up to about 2MHz - but wait, you gave me an idea, maybe I should check the low end too...because the circuit is oscillating at about 33KHz.
 

I remember seeing the schematic for a Tektronix HV probe similar to the one in the photo. The compensation network was really complex, as you can tell from the several visible adjustments.

Compensation ideally works only for a fixed frequency. Why the frequencies seen in post 14 for the square pulses are so low?

It really depends on the intended probe frequency range, which wasn't specified along this thread. The shown resistor divider primarily suffers from HV resistor parallel capacitance, which is causing large error already in the kHz frequency range. A first order compensation improves the probe function a lot, it's probably sufficient for the relative low frequency signal this thread is dealing with.

The setup in this thread looks pretty much like "my first DIY HV probe". Don't expect same performance as with calibrated professional products or other expert designs.
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top