EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: rwgast_lowlevellogicdesin on June 30, 2024, 06:35:09 pm
-
I had a 25mhz Siglent DSO I got second hand back in 2012 or so, it was my first real scope. I blew channel 1 testing the ripple of a PSU, i don't remember the exact circumstances, I don't even think I knew the channel was broken till a day or two later. The signal had some weird attenuation issue. At the time I just ordered a new Siglent SDS1104x-e in 2016ish. Later I took the 25mhz scope apart and there was a blown resistor in the front end. I think I had shorted the power supply I was testing with the probe in 1x mode. Shouldn't the 10Mohm impendence stop the scope from drawing any of the current if the psu gets shorted?
Last night I blew out channel 1 of my SDS1104x-e, I was testing some custom filtering on the output of a cheap switched lab PSU. The PSU was loaded down to about 720ma @ 5V. The scope was in 1x mode, and I was using a BNC to alligator clip lead across the load resistor. I got most of the noise out of the thing down from 70mv PK to PK to 12mv PK to PK. I was messing around though and changed the last inductor in the filter(the inductor is a series inductor on the + side of the DC output right before the load) from a 10microHenery to a 1500microHenery I wound on a #26 core years ago. It killed all of the noise on the wave form except a slower periodic ripple that looked to be bjt/fet overshoot, I looked away for a second and then looked back and my trace was gone.... I reset the scope and re caled it, and there is still no trace on channel 1. I can see the trace if I couple the channel to ground, but it disappears when using DC or AC coupling. I can also see it briefly if I scroll the voltage range knob quickly but then it disappears in a split second. Im pretty sure I must have blown a resistor or clamping diode in the front end use to bias the input. Now I may have accidently shorted the PSU output for a second and didnt realize it, but is there anyway that big inductor could have blown the channel out?
Im pretty sure I can open the thing up and find the problem in the front end section and replace the 0202 passives or whatever, but ive had some other issues that have made me want to upgrade to a new 1104x hd or 2204x hd. The thing is I don't want to spend that kind of money on a scope and blow the front end up again.. im not sure why i keep blowing front ends and why aren't they equipped with fuses, signal integrity?
-
Never, Never, Never use 1X probe. throw it away. Use 10X.
There are very few uses for 1X. You have just discoverer the most common use.
-
Well i was only using a 1x becuase #1 clips were more convient, but #2 i watched a video about ripple and common mode noise and i remmber dave saying you want to make the measurements at 1x.
It looks like i ended up blowing a jfet and a few uhf transistors, not sure if thats it since i dont have a scope to trace the signal path.
-
The scope is rated for 400V pk-pk so 1X seems reasonable, but maybe not once you factor in a voltage spike from the inductor.
-
+1
Momentarily mechanically interrupt a DC circuit that has a large inductor, and you can expect a transient of several kV. The scope input breakdown voltage is lower than that so all the energy got dumped into the scope's input circuit with predictable results.
I'd suggest getting a couple of these https://www.amazon.com/Coaxial-Lightning-Arrestor-Protector-Protection/dp/B07WLNL76F (https://www.amazon.com/Coaxial-Lightning-Arrestor-Protector-Protection/dp/B07WLNL76F)
and using one on each scope input whenever you are messing with PSUs and inductors. Their nom. breakdown voltage is 90V, (min. 75V) so if the DC rail is above approx. 35V (to maintain a factor of two margin), use a x10 probe, not a x1 one, or use an appropriately rated external DC blocking capacitor.
-
Yeah a 1x probe should generally be fine for measuring ripple on a 5v power supply. Thats a pretty common use case. But a huge inductor can make big spike that delivers a lot of energy.
It's weird it broke just sitting there. Normally this would happen if the load is disconnected and the indicator current has to go somewhere.
Keep in mind this tells you something about your supply: if it blew out your scope input it can probably blow out the circuit you are going to power with it.
-
Wont those 50ohm arrestors attenuate my voltage readings? Thats pretty rough when your on the 10mv range. What is the safest way to break the load from the inductor? I was just pulling the load resistor out by hand, guess thats a bad idea.
The scope had some 50v zeners clamping the input of a jfet, im not sure if the zeners are bad but i think so, still have to put psus back togather to make a 50v source so i can properly test them.
-
Surely at this point, disconnect the scope first!
Next thought would be just avoid using the inductor altogether.
But if you must keep it, maybe direct short then disconnect.
-
BNC to alligator clip lead
Unterminated BNC cable? Could that have caused issues?
-
Wont those 50ohm arrestors attenuate my voltage readings? Thats pretty rough when your on the 10mv range.
They are GDT surge arrestors. Well below their breakdown voltage their DC resistance (center pin to shell) is very high, many times higher than the scope's input resistance, so you wont see significant attenuation of DC and low frequency signals. However they will contribute a little extra capacitance, which may be a problem for higher frequencies (and fast edges) with x10 or x100 probes if your probes have insufficient compensation adjustment range on the low side.
N.B. Once they breakdown they can shunt tens of thousands of amps, so don't use one when x1 probing a high current DC rail too close to their min. breakdown voltage. That could easily result in a melted probe and surge protector! Also many scope inputs don't provide a path to ground when AC coupled, so don't set the scope input to AC when using one with a x10 or x100 probe, and probing a rail above their min. breakdown voltage, or you will get to see *interesting* but unhelpful behaviour of the GDT. I don't think the input capacitance will be sufficient to sustain relaxation oscillator behaviour, but shunting the input with a current limited glow discharge wont do you any favours for signal integrity!
-
Never, Never, Never use 1X probe. throw it away. Use 10X.
Mantra like this isn't helpful without giving the "why". rwgast has watched some Dave videos suggesting to use 1x but you say to use 10x. Confusion results.
10x probes burden the circuit less, and they have lower capacitance which allows higher frequency operation.
1x probes do not have the full penalty of the oscilloscope's input amplifier noise to contend with, so they are better for low signal levels.
Don't forget that probes have a voltage vs frequency derating as well.
I looked away for a second and then looked back and my trace was gone....
I was just pulling the load resistor out by hand, guess thats a bad idea.
Ahh.. while looking away did you happen to pull the load resistor? Now we are getting closer to the real sequence of events :)
10uH is a fairly small inductance, but 1500uH is a fair bit of energy storage. Inductors can generate quite impressive voltages when you open-circuit them. In this case, you had 700mA flowing through the inductor and then disconnected the load. The inductor will generate a substantial voltage across itself in order to keep 700mA flowing, and once it reaches a high voltage the only place left to dump that energy is into your probe. So your scope probably copped >1kV and popped the front end.
You could avoid the voltage spike by putting a flyback diode across the inductor (very common practice for suppressing relay coil transients). Generally for power supply filtering it's safer to use capacitors to do the bulk of the filtering work and use smaller inductor values to avoid voltage spikes like this. Have a capacitor as the last filter element, as well, and it will improve voltage regulation of the output.
-
What is a safe value for inductors? About how big of a cap should i parallel after the inductor, to make sure and surpress transients? I usually just model a filter in elsie and go with it, can programs like lt spice and microcap model the transients when the inductors magnetic field collapses?
Lastly would this have happened in 10x mode?
-
I think they are correct about the inductor, If this is the case and you used a 10X probe I believe that you would blow the probe and it is possible the scope would not be damaged. Passive Probes are quite inexpensive these days and usually fixable.
A Gas Discharge tube or a MOV inside the PS might prevent this but maybe they would not be fast enough, lightening arrestors do 'fire" but the equipment is not always saved. Any addition to the probe changes capacitance but I just measures a 150 V gas discharge and it measured ONE pF at one MHz. MOV have higher capacitance. HP used gas discharge on some of their different inputs, like the 3456. But I do not know of any of these devices used on scopes. I do not think fuses are fast enough.
Perhaps someone else knows if these devices are fast enough under these circumstances.
-
Surely a good piece of advice is do not work on a device under test while the scope is connected to it? Detach the scope, make the desired alterations, and then reattach the scope probes. For that matter, do not try to modify a device while it is powered up. There is always the risk of a slip of the fingers and an inadvertent short circuit that can damage things.
-
What is a safe value for inductors? About how big of a cap should i parallel after the inductor, to make sure and surpress transients? I usually just model a filter in elsie and go with it, can programs like lt spice and microcap model the transients when the inductors magnetic field collapses?
Lastly would this have happened in 10x mode?
Vinduced = Ldi/dt. So work out the current change in the time interval.
Spice will correctly calculate the transients in model circuit you specify. That should give you a general feel for what's happening. The issue is that you have to correctly specify all components in the model; that is non-trivial. Knowing what is necessary and sufficient requires skill, judgement and experience.
You do realise that L+C=>resonant circuit - don't you?
-
Lc is only resonent in a parallel config right? The capacitor was in parallel across the dc and the inductor was in series on + side of the dc supply in an lc filter config.
Yes disconnecting the scope when changing the circut is a safe way to make measurements like this and i wish i had done that.. but then lets say i used that circut on the output of my PSU becuase the scope made it look acceptable, i would have blown up whatever i had the psu connected too if i unplugged the board without first powering down the psu. The only way to really test that is to use those lightening arrestor or simulation? I guess fast flyback diodes?
-
Oh you can definitely have a series-LC resonance too.
Parallel LC has a high impedance at resonance, series LC has a low impedance at resonance.
Vinduced = Ldi/dt. So work out the current change in the time interval.
In theory, for any circuit regardless of inductance, the voltage induced is infinite for a sudden disconnect because current goes from some non-zero value to zero instantly.
In practice devices don't turn off immediately, a circuit that's physically opened will draw an arc (the resistance of which will act to decrease the current) etc. Or a device will momentarily go into breakdown to close the circuit.
Whether a circuit is damaged or not really depends on how much stored energy is trying to dissipate. If you have a 10uH inductance at <1A, the stored energy is low, it'll likely safely dissipate into various imperfections in the real devices surrounding it. If you have 1.5mH, you start blowing things up. Of course if you had 10uH at 1000A then you're back to some hefty inductive kicks again.
General comment regarding simulation, it's a great tool, but it's only as good as the information you put into it. Part of 'experience' is knowing what information is relevant, and what can be skipped for expedience.
-
Lc is only resonent in a parallel config right? The capacitor was in parallel across the dc and the inductor was in series on + side of the dc supply in an lc filter config.
Yes disconnecting the scope when changing the circut is a safe way to make measurements like this and i wish i had done that.. but then lets say i used that circut on the output of my PSU becuase the scope made it look acceptable, i would have blown up whatever i had the psu connected too if i unplugged the board without first powering down the psu. The only way to really test that is to use those lightening arrestor or simulation? I guess fast flyback diodes?
A cursory bit of research will indicate that parallel and series resonance exists - and is the basis for many types of filter.
You should also realise that a straight piece of wire has inductance, the rule of thumb being 1nH/mm. Hence if you have a, say, 1ft wire then you should model it as its DC resistance in series with 300nH. If it is coiled, even slightly, the inductance will be higher.
Given your moniker, I would expect you to have seen the consequences of that when using a standard "high" impedance scope probe on a logic circuit. For the basic theory and indication of consequences, see https://entertaininghacks.wordpress.com/2015/04/23/scope-probe-accessory-improves-signal-fidelity/
You will also note that some subsystems make a point that they can be "hot plugged" or "hot swapped", i.e. they can be inserted/removed while power is applied. That's a big hint that most subsystems don't guarantee that.
Finally, realise that a regulated PSU (i.e. any modern PSU) employs internal feedback to keep the output voltage constant as the load current varies. Different PSUs do a better or worse job, especially where the load is reactive - and all loads are reactive. Look at the PSU's specifications; if none, be suspicious.
-
I had a 25mhz Siglent DSO I got second hand back in 2012 or so, it was my first real scope. I blew channel 1 testing the ripple of a PSU, i don't remember the exact circumstances, I don't even think I knew the channel was broken till a day or two later. The signal had some weird attenuation issue. At the time I just ordered a new Siglent SDS1104x-e in 2016ish. Later I took the 25mhz scope apart and there was a blown resistor in the front end. I think I had shorted the power supply I was testing with the probe in 1x mode. Shouldn't the 10Mohm impendence stop the scope from drawing any of the current if the psu gets shorted?
Last night I blew out channel 1 of my SDS1104x-e, I was testing some custom filtering on the output of a cheap switched lab PSU. The PSU was loaded down to about 720ma @ 5V. The scope was in 1x mode, and I was using a BNC to alligator clip lead across the load resistor. I got most of the noise out of the thing down from 70mv PK to PK to 12mv PK to PK. I was messing around though and changed the last inductor in the filter(the inductor is a series inductor on the + side of the DC output right before the load) from a 10microHenery to a 1500microHenery I wound on a #26 core years ago. It killed all of the noise on the wave form except a slower periodic ripple that looked to be bjt/fet overshoot, I looked away for a second and then looked back and my trace was gone.... I reset the scope and re caled it, and there is still no trace on channel 1. I can see the trace if I couple the channel to ground, but it disappears when using DC or AC coupling. I can also see it briefly if I scroll the voltage range knob quickly but then it disappears in a split second. Im pretty sure I must have blown a resistor or clamping diode in the front end use to bias the input. Now I may have accidently shorted the PSU output for a second and didnt realize it, but is there anyway that big inductor could have blown the channel out?
Im pretty sure I can open the thing up and find the problem in the front end section and replace the 0202 passives or whatever, but ive had some other issues that have made me want to upgrade to a new 1104x hd or 2204x hd. The thing is I don't want to spend that kind of money on a scope and blow the front end up again.. im not sure why i keep blowing front ends and why aren't they equipped with fuses, signal integrity?
Welcome to the world of Back EMF !
How high can the voltage of a collapsing field go, as high as it needs to to find a discharge path ! ! !
Disconnecting current sources while probing something is never a good idea and when inductors are involved, bad things can happen.
Some protection is available by using 10x scope probes instead of 1x BNC to grabber connections but safe measurement technique dominates.