Measuring voltage and current on the DRSSTC bus

[profdc9]

Another question I have...

I have a digital storage oscilloscope, and I would like to know how to safely measure the voltage on the bus of the DRSSTC (300 to 600V) to see if there any voltage transients that could damage IGBTs.  Does a regular 10x scope probe work, or is there a high voltage probe that you use?  I have a 1 gigaohm Heathkit probe that's good up to 30 kV, but perhaps there's some better way to do it?  I suppose I could build a resistor divider network and clip across that.

As for measuring the current, do you just measure the voltage off of the current transformer on the burden load, or actually place an ammeter clamp around the busbar?

Dan

[Netzpfuscher]

I use a differential probe http://www.warwickts.com/testec-tt-si-9110-active-differential-probe-high-voltage-100mhz-140v-1400v-p1299

But this is costly. The cheapest way is to use a isolation transformer at the bridge or can isolate the scope but this is risky, you can get shocked.

[Hydron]

I was going to write a big post here about how to probe the bus here, but it's very difficult to suggest techniques for safely doing this sort of thing when not knowing who is doing it. I think a few notes will be better:

a) Be careful if you use a 10x probe that it's rated for measuring mains (or higher). There's a good chance that they'll be near the limit of their ratings, and it's unlikely that they'll be rated better than CAT I, which is a bit iffy for any mains work. If you can't find a rating for the probe, don't use it!

b) 100x probes are probably a better option than 10x, and are available affordably from reputable sources.

c) Be careful when using any 10x probe or a 100M ohm input impedance 100x probe in AC coupling mode. The FULL DC voltage will be present on the scope input un-attenuated, blocked only by the voltage rating of the scope's AC input coupling capacitor. Be sure to check the scope's voltage rating - this is what will limit the DC component of any waveform you are probing, NOT the probe voltage rating.

d) Easiest is to use a differential probe as suggested. The only cheap one I know of is Micsig's DP10013 - this has some design issues limiting bandwidth (it's 100MHz rating is bullshit, and the long leads will ring at about 70MHz) and CMRR, but will work fine for this job.

e) If not using a differential probe, you can still check the power rails relative to earth ground with a 100x probe or similar. If you're using a voltage doubling rectifier correctly then each rail will be fixed at ~1.4*Vrms above or below ground; if using a full bridge rectifier then the rails will move up and down at mains frequency, but you can still get a look at their behaviour despite this. This will also work for scoping the output of each half bridge.

d) NEVER FLOAT YOUR SCOPE with an isolation transformer unless you know _exactly_ what you're doing. It's a great way to kill yourself if you forget and touch something. The only acceptable way to use an isolation transformer in this situation is for isolating the tesla coil power input so that you can tie the mains protective earth reference to one of the bus rails (as well as the heatsinks and enclosures etc that should already be earthed) rather than the neutral of the mains supply. Note that this still connects mains protective earth to the coil!

Regardless of how you go about it, if the scope has a good protective earth connection (via mains) and you've set up all your probes etc before turning anything on (and are not touching or moving them live), then most safety risks will be confined to the probes/scope rather than having the potential of hurting you. If you use a decent 100x probe in DC coupling mode (or a differential probe) any risks to equipment should be minimal.

[profdc9]

I was thinking about this, and I came up with might be a satisfactory solution if the goal is to look at transients < 1 MHz.  Perhaps this is not fast enough.

Do you think the following would work?   The idea is to place a 10 k burden resistor across the bus, and then use a ferrite transformer to isolate it.  It would be a 4:1 ratio of turns using the four ethernet pairs as the scope side in a quadrifilar winding, and one heavily insulated wire on the bus side wound bifilar with the ethernet cable.  It would isolate the DC, but should have a good frequency response from about 10 kHz to 1 MHz.   I included the effect of winding capacitance (very high estimate, 100 pF) and because the load impedance is low, it doesn't seem to have much of an effect until the low MHz range.  If a MnZn ferrite is used, eddy current losses are also going to be an issue as well into the MHz range.

The 10k burden resistor could dissipate 36 W if 600 V constantly is put on it, so it should be used in a DRSSTC for only intermittent operation. 

Dan

[profdc9]

By lowering the burden resistor to 1k, it could be useful for looking at gate drive as well:

Dan

[Mads Barnkob]

Everyone else gave you all the advice you need. I would also like to chime in and say get a differential probe if you want precise and calibrated measurements.

I was lucky to buy a 2nd hand Tektronix P5200 in Denmark at a good price, they can also be found on ebay at similar prices, but only buy the ones listed AS WORKING, all the rest are most likely blown from over-voltage

You could also try your luck with a brand new Chinese differential probe: https://www.ebay.co.uk/itm/Micsig-Oscilloscope-DP10013-1300V-100MHz-High-Voltage-Probe-kit/253141750649 before you buy it, look at this huge thread on eevblog about it: http://www.eevblog.com/forum/testgear/new-low-cost-($170)-100mhz-differential-scope-probe-from-micsig/?all

[profdc9]

Is there any other way to adjust the phase lead than monitoring the bus voltage?  That is mainly what I need to do.   I am just looking for transients on the bus.

Dan

[Hydron]

Do you have a variac? As i said you can still scope the bus or bridge output using earth as the reference (just need to ignore the 50Hz movement, which will be slow compared to the wanted signal anyway), and if you're worried about probe voltage rating you can run the bridge at less than full mains voltage for setup and testing using a variac. Phase lead setting will change a little as you turn the voltage up but it shouldn't be enough to worry about as long as you set up at a reasonable point (e.g. half the designed maximum). Also note that scary looking spikes at low testing voltages (e.g. 30V spikes when testing at 50V bus voltage) may not get much worse as you turn the bus voltage up (e.g. they might become 60V spikes at 300V bus).

[profdc9]

Yes I have a variac and I can give that a try.  I was looking at some of 100X probes you can get on Amazon cheap.  I don't know if these can be trusted:

https://www.amazon.com/WINGONEER-P4100-Universal-100-Oscilloscopes/dp/B06XHJT63M

I was also looking into using a coaxial capacitive divider probes.    They apparently can be used well over 100 kV and with transients as fast as 1 ns, and can span ranges with a single probe from about 1kHz - 100 MHz or more.   I actually built just built a primitive one in a few hours, using a wire in a 1/8" thick polyethylene tube as a 5 pF high breakdown probe, and divided this by a 500 pF ceramic.  Getting all the stuff to do the DRSSTC is going to be expensive, and I haven't even bought the MMC capacitors yet, and I don't necessarily want to buy a differential probe right now, especially if its not the greatest probe.

Perhaps you might be interested in this link:

http://www.imajeenyus.com/electronics/20120216_capacitive_hv_probe/index.shtml

Another question: do you have any estimate of the frequency of the spikes?  What is the bandwidth needed of a probe to see them?

Thanks for your help,

Dan

[Steve Ward]

Just another opinion...

While differential probes are nice, I've never owned one and can never seem to justify the expense.  I keep my scope grounded, pretty much exclusively (can't recall ever floating it).  I use a 60hz isolation transformer when i need to scope power circuits.  The only time this gets ugly is when the probe reference is on a circuit that is switched potential, like the high-side gate measurement.  However, you can usually make this measurement at low voltage (so common mode current into the scope is lower) and use the low-side gate measurement at high voltages as a reference for what the high-side gate would look like.

I don't have much experience with differential probes, but i will say that in the last few years, especially when dealing with MOSFET, SiC or GAN devices, that probe bandwidth and proper probing technique can become essential to properly identifying problems with these devices capable of extreme dv/dt, and i don't think affordable differential probes are up to the task. I built a 20kW GAN motor drive that hard switches 150A in ~5nS... my 100MHZ scope was too slow to accurately measure the switching waveforms, so a 500MHZ scope was procured ;-).  The interesting part is that the 100MHZ scope probe was actually ringing and exaggerating a transient that i thought was responsible for device failures, but the 500MHZ setup revealed a waveform that more closely matched my spice model which was totally fine... and then i later found out the actual reason the GAN fets were failing which was completely unrelated (DC bus resonance with exceptionally high Q all-ceramic bus cap... overvolt... splat!).

With respect to IGBTs, 50MHz is probably plenty of bandwidth.  I've seen some diodes ring in the 10s of MHZ range on CM300 modules.  Diode recovery ringing always looks terrible because it generates the highest dv/dt.  I've used several cheap 100X probes (2kVDC rated) from ebay, usually they seem to be mostly fine aside from the fact that they just fall apart or break after enough of my use.  For the $20 i paid, i don't expect much longevity.

 

[ErdaWilt]

Hi..Do you have a balance sort of thing as a main priority? Accept that the decoration has quit developing and expect that its obstruction gets lower when it gets more smoking. Additionally expect that the temperature is simply so its opposition boosts power utilization. At that point the decoration won't tend to get much more sweltering since that requires more force and force input diminishes. It very well may be colder, however. That would be the situation if an expansion in temperature couldn't be supported by the comparing increment in force.

That being stated, I got from the long consistent tail in my estimation a capacitive reactance of 130k and an obstruction of 110k, which is a deviation well inside my estimation mistakes.