For measuring SMPS-primary side voltages I could really use a isolated probe with at least 1mhz bandwith. Thats not DIY-stuff if we look at components like AMC1200 ( 100Khz BW).
BUT, for checking SMPS primary often we dont need DC, we dont need high accuracy and we often dont need high impedance input. - SO could a transformer-coupling work?
I'm guessing that most of my needs could be met with 1(10?)Khz - 1(10?)Mhz +/- 3db and 10K-input impedance IF the setup had 4Kv isolation and low capacitive coupling.
Have anyone been down this path?
Kind regards TroelsM
BUT, for checking SMPS primary often we dont need DC, we dont need high accuracy and we often dont need high impedance input. - SO could a transformer-coupling work?
I'm guessing that most of my needs could be met with 1(10?)Khz - 1(10?)Mhz +/- 3db and 10K-input impedance IF the setup had 4Kv isolation and low capacitive coupling.
Have anyone been down this path?
Kind regards TroelsM
Being aware of the difficulties I never tried this. If you call for Z=10kOhms @10kHz you need 160mH of inductance.
My best bet would be using some 230Vac commone mode chokes with sectional windings as transformers for best isolation / interwinding capacitance.
This results in a stray inductance in the ballpark of 3% thus limiting bandwidth.
Bifilar wound show less stray inductance and higher bandwidth at the expense of higher interwinding capacity and bad insulation.
There is no easy way.
When I dream I dream of a tiny LiIon-driven active probe connected to the scope via fibre-glass conductor.
My best bet would be using some 230Vac commone mode chokes with sectional windings as transformers for best isolation / interwinding capacitance.
This results in a stray inductance in the ballpark of 3% thus limiting bandwidth.
Bifilar wound show less stray inductance and higher bandwidth at the expense of higher interwinding capacity and bad insulation.
There is no easy way.
When I dream I dream of a tiny LiIon-driven active probe connected to the scope via fibre-glass conductor.
Last edited:
I am up against more or less the same problem. An inverter welder misbehaving, with suspicion at present levelled against the main inverter components (snubber components particularly).
The inverter power supply comes via bridge rectifier and big capacitors, giving a good 300v. Two banks of MOSFETs or IGBTs are connected respectively to the +ve and -ve rails, and activated simultaneously at about 100kHz from a 3524 SMPS IC via a coupling transformer. Each bank connects to an end of the (power) transformer primary. When the power semiconductors are on, the primary is fed by about 300v, and when off, the primary current freewheels through diodes.
Various snubber components suffered when the 3524 delivered something like 700kHz because of a failed timing capacitor (one resistor glowed red, with no DC path through it). I have replaced two snubber diodes one of which was acting as a 5Ω resistor in both directions, but it is not clear to me how to test the capacitors in the snubbers under operating conditions. I want to look at the voltage trace ('scope triggered by the 3524 output) at each end of each suspect component and see whether they make sense.
My oscilloscope is an old Telequipment Type 43. (Very much an analogue machine, 2 channel, quite basic but usually adequate for my requirements). I test the welder powered via an isolating transformer followed by a variac. The welder's control circuitry is straightforward to look at with the 'scope, having obvious earth rails. The signal applied to the inverter MOSFETs drive transformer is a reasonably clean square wave, but I have found no way of getting clean traces of anything beyond that transformer primary. The problem is finding something to connect the probe earth to. There is nothing in the inverter which is not oscillating over 100s of volts. If it worked at a low a frequency, and with safety precautions, I could earth to either of the DC rails, and (with the isolating transformer in place) let things float and probably get useable results. But at oscillator frequency (let alone the switching harmonics which the snubbers should suppress) I think that capacitances and inductances of the oscilloscope and the isolating transformer etc. prevent free floating.
My present thought is to arrange a voltage divider across the bridge rectifier (could be a centre tapped transformer, or just resistors), and treat the centre as a reasonably stable and static earth for the 'scope probe.
But I suspect that there is a better way of approaching the problem. Any ideas?
Andrew
The inverter power supply comes via bridge rectifier and big capacitors, giving a good 300v. Two banks of MOSFETs or IGBTs are connected respectively to the +ve and -ve rails, and activated simultaneously at about 100kHz from a 3524 SMPS IC via a coupling transformer. Each bank connects to an end of the (power) transformer primary. When the power semiconductors are on, the primary is fed by about 300v, and when off, the primary current freewheels through diodes.
Various snubber components suffered when the 3524 delivered something like 700kHz because of a failed timing capacitor (one resistor glowed red, with no DC path through it). I have replaced two snubber diodes one of which was acting as a 5Ω resistor in both directions, but it is not clear to me how to test the capacitors in the snubbers under operating conditions. I want to look at the voltage trace ('scope triggered by the 3524 output) at each end of each suspect component and see whether they make sense.
My oscilloscope is an old Telequipment Type 43. (Very much an analogue machine, 2 channel, quite basic but usually adequate for my requirements). I test the welder powered via an isolating transformer followed by a variac. The welder's control circuitry is straightforward to look at with the 'scope, having obvious earth rails. The signal applied to the inverter MOSFETs drive transformer is a reasonably clean square wave, but I have found no way of getting clean traces of anything beyond that transformer primary. The problem is finding something to connect the probe earth to. There is nothing in the inverter which is not oscillating over 100s of volts. If it worked at a low a frequency, and with safety precautions, I could earth to either of the DC rails, and (with the isolating transformer in place) let things float and probably get useable results. But at oscillator frequency (let alone the switching harmonics which the snubbers should suppress) I think that capacitances and inductances of the oscilloscope and the isolating transformer etc. prevent free floating.
My present thought is to arrange a voltage divider across the bridge rectifier (could be a centre tapped transformer, or just resistors), and treat the centre as a reasonably stable and static earth for the 'scope probe.
But I suspect that there is a better way of approaching the problem. Any ideas?
Andrew
I assume your power bank is kind of a half-bridge driver. The easiest way to get some practical measure results is grounding the scope at rectifier bridge minus and probing a low-side gate vs a low-side drain. Btw I do not see how this correlates to DIY audio.
Bucks Bunny: Thanks for the suggestion. That method ought, I agree, to work for the components near (electrically) to the -ve rail, and I should be able to do much the same anchored to the +ve rail for the other set. But there is so much HF everywhere that even anchoring on the rails gives a blurry messy trace whatever the probe touches. (My probes are nothing but wire ends. Might someone say this needs improving?)
Just a supplementary note: I am sure someone will recommend just renewing all passive components — there are only about a dozen — but the construction of the welder is of rather high quality, and the power semiconductor banks were evidently installed as assemblies (with heat sinks) with a 200W or similar soldering iron after the small components were in place. The latter are consequently hard to get at (or even read) and remove, and the two I replaced I ended up soldering them to the back (track side) of the board for convenience. (The sort of temporary measure which becomes permanent.) I don't want to do any unnecessary dismantling.
Just a supplementary note: I am sure someone will recommend just renewing all passive components — there are only about a dozen — but the construction of the welder is of rather high quality, and the power semiconductor banks were evidently installed as assemblies (with heat sinks) with a 200W or similar soldering iron after the small components were in place. The latter are consequently hard to get at (or even read) and remove, and the two I replaced I ended up soldering them to the back (track side) of the board for convenience. (The sort of temporary measure which becomes permanent.) I don't want to do any unnecessary dismantling.
You should use oscilloscope probes with very short connections to gnd, close to minus rail directly at the power devices. No way to do this with open wires.
O scope probes must be in proper operating condition...... period.
Not just some bare wires.
Proper probes on the ends of the wires, calibrated to the scope.
Otherwise, scope readings are worthless wastes of time.
I appreciate the advice about probes. I am a mechanical engineer, and although I do a fair amount of electrical and electronic work, I am less familiar with the niceties than most of you will be. Every electronic repair or build I regard as a learning experience, and at age 72 I value that as much as ever.
May I ask for clarification about probes? I have a 'proper' probe (x1) on one channel of my 'scope and a pair of ordinary thin flexes on the other. I almost always use the latter unless I need both, my thinking being that I am avoiding adding capacitance across the signal. I almost always have the signal earth connections on the 'scope floating. I have assumed that if the pair of wires take approximately the same route from DUT to 'scope, they will experience much the same induced voltages from stray fields (as is done better by a twisted pair), and the 'scope amplifier will act as a differential amplifier between the two wires.
If I use a 'proper' probe, with the signal conductor shielded by the sheath, what happens when the cable passes through a magnetic field? The shield will have voltage induced from end to end, the high impedance of the 'scope means that practically no current flows, and so the sheath allows the field to affect the core equally, and the same voltage is induced in the core. So, as regards stray fields, the situation is similar to but better than my pair of loose wires. (Am I right?) But what about the usual few inches of wire connecting the sheath to the DUT? It is exposed to stray fields. Is its effect any different from a probe with a few inches of unsheathed conductor? I am aware that in critical situations this few inches of earth wire is reduced to next-to-nothing if possible.
Is my (welding inverter) situation critical? Currents and field strengths are high, but frequencies are low (by radio standards). The fundamental is about 100kHz, and even if the snubbers are seriously out of order, would there be anything worth bothering about much above a few MHz?
I look forward to guidance on probes.
When I have time I will try the 'proper' probe with the sheath and core joined that their tip, and see whether touching the joined end here and there produces a signal at the 'scope. (It should not?)
Andrew
May I ask for clarification about probes? I have a 'proper' probe (x1) on one channel of my 'scope and a pair of ordinary thin flexes on the other. I almost always use the latter unless I need both, my thinking being that I am avoiding adding capacitance across the signal. I almost always have the signal earth connections on the 'scope floating. I have assumed that if the pair of wires take approximately the same route from DUT to 'scope, they will experience much the same induced voltages from stray fields (as is done better by a twisted pair), and the 'scope amplifier will act as a differential amplifier between the two wires.
If I use a 'proper' probe, with the signal conductor shielded by the sheath, what happens when the cable passes through a magnetic field? The shield will have voltage induced from end to end, the high impedance of the 'scope means that practically no current flows, and so the sheath allows the field to affect the core equally, and the same voltage is induced in the core. So, as regards stray fields, the situation is similar to but better than my pair of loose wires. (Am I right?) But what about the usual few inches of wire connecting the sheath to the DUT? It is exposed to stray fields. Is its effect any different from a probe with a few inches of unsheathed conductor? I am aware that in critical situations this few inches of earth wire is reduced to next-to-nothing if possible.
Is my (welding inverter) situation critical? Currents and field strengths are high, but frequencies are low (by radio standards). The fundamental is about 100kHz, and even if the snubbers are seriously out of order, would there be anything worth bothering about much above a few MHz?
I look forward to guidance on probes.
When I have time I will try the 'proper' probe with the sheath and core joined that their tip, and see whether touching the joined end here and there produces a signal at the 'scope. (It should not?)
Andrew
There is a "calibration procedure" involved with the probe - a small screw adjustment on the probe - done with a testing procedure for the scope.
This usually involves an internal square wave generated in the scope, and the adjustment screw is "tuned" for a proper square wave on the scope's screen.
Perhaps some scopes use other methods for calibration, not knowing the specific scope, I cannot advise further.
This usually involves an internal square wave generated in the scope, and the adjustment screw is "tuned" for a proper square wave on the scope's screen.
Perhaps some scopes use other methods for calibration, not knowing the specific scope, I cannot advise further.
Yes, I know that there should be an adjustable cap to balance the 'scope's input impedance arrangements. I am afraid that my 'proper' probe is not quite proper enough to have this. But the square wave does look good enough for present purposes. I have to clear away masses of paperwork before I can get back to probing the welder.
Transformers have problems at the top and the bottom of their ranges and that range rarely covers the whole audio band, ie 20-20KHz. At the bottom, the (lack of) transformer inductance begins to short out the source. You can extend the frequency range by using a transformer wound for a higher impedance but with higher wire losses. At the high end, particularly the secondary becomes a parallel resonant tank circuit that rings like crazy, unless you load the secondary to damper that resonance. Again, the solution involves losses/attenuation, and you can only extent the limit a bit. Keysight etc sell differential boxes that are an fast op-amp with input networks for about 1 meg impedance and about 1MHz bandwidth. Of course, if you have a 2+ channel scope, you just use two probes with the scope set to differential mode. Any time you are probing a power device, you should remove the ground wires because the scope ground is probably well grounded and connecting the ground wires will probably burn out your probe, ie use the safety ground as the only ground connection. Also, scopes are not built for high voltage. 100:1 probes are required for line voltages.
https://circuitcellar.com/research-design-hub/high-voltage-differential-probe/
https://www.instructables.com/High-Voltage-Differential-Probe-for-Oscilloscopes-/
https://circuitcellar.com/research-design-hub/high-voltage-differential-probe/
https://www.instructables.com/High-Voltage-Differential-Probe-for-Oscilloscopes-/
Last edited:
Transformers have lots of issues for instrumentation purposes. I have had to build special types, covering 10kHz to 500MHz but they were ones off, impossibly difficult to make, and having severe limitations: low frequency voltage handling was extremely low, and the insulation voltage was limited to the silk/enamel of the wire.
Active solutions are much preferable: I have described a very simple, yet useful one here: https://www.diyaudio.com/community/...direct-mains-measurements.285001/post-4575143
I have built a number of higher performance ones, including a 30MHz one, but they are more complicated, require a careful construction and adjustment which is why I didn't describe them in the forum.
If you are ready to pay a normal price, usual vendors have them in their listings
Active solutions are much preferable: I have described a very simple, yet useful one here: https://www.diyaudio.com/community/...direct-mains-measurements.285001/post-4575143
I have built a number of higher performance ones, including a 30MHz one, but they are more complicated, require a careful construction and adjustment which is why I didn't describe them in the forum.
If you are ready to pay a normal price, usual vendors have them in their listings