First impressions with the GW Instek MDO-2204EX

[SWR]

I'm very satisfied with the first impressions using this scope. I received it last weekend and had my first experiences using it this week. This is a short description of a board test session. It may come across as a bit of a commercial, but I guess it's because I haven't run into the first problem or annoyance yet. Our company develops and manufactures electrohydraulic steering units for off-highway machines, and I'm currently involved with developing a new hardware platform for our future products.

After testing that the PSU's on the evaluation board were running as expected I moved to the test the main interface to the hydraulics. The main circuit elements are four processor controlled solenoid drivers controlling four hydraulic valves in a typical H-bridge configuration. They control the position of a hydraulic spool with position feedback by LVDT. Instead of waiting for the software guys to write me some HW test SW or going to the lab with a selection of instruments, the scope integrated all I needed to do the testing at my desk which was nice.

Connecting the first generator output with a square wave to the gate of a solenoid driver and a scope channel across the current measuring transistor i could test the circuit function with a small external electronic load powered by the USB output on the front panel. The hydraulic valves operate at 50Hz but after pull-in the hold current is regulated with a 20KHz PWM modulation of the 50Hz pulse. With the built in waveform editor it was quick to generate a bunch of waveforms with different pull-in time (100% on) and hold current (duty cycle). Both will be regulated based on the supply voltage level.

One issue we face is that the solenoid coils are placed physically close to the LVDT coils so we expected electromagnetic coupling from the coil switching frequencies to the LVDT measurement circuit. I turned the first generator off and connected the second generator with a 5KHz sine output to the LVDT tx coil, and the two rx coils to channel 1 and 2. In the time domain I could confirm that I got a clean 5KHz signal with varying amplitude depending on the position of the LVDT. In the frequency domain I could see that the signal was quite pure with a few harmonics several decades down. Zooming in on the frequencies of interest is very easy in the Spectrum Analyzer mode using start/stop or centre/span. On the Keysights we normally use I often have to fiddle with the sample rate and horizontal position to get a nice FFT picture. The Keysight will do the job but on the Instek is easier and quicker to use in the frequency domain.

Switching the first generator back on it was clear that the solenoid coils was coupling to the LVDT coils. It looked messy in the time domain, but in the frequency domain it was easy to spot the 50Hz base switching frequency with harmonics falling off up to about 1KHz and the 20KHz holding current switching frequency with harmonics up past 200KHz. We expected this and have planned to implement a digital bandpass filter followed by a rectifier and low pass filter in SW. The position is then calculated by (A-B)/(A+B). The nice thing with this scope is that I could test it without waiting for the SW guys to finish writing the embedded code.

I configured a 5KHz bandpass filter on channel 1 & 2 and watched the signals clean up in both the frequency and time domain. The frequency domain is a nice confirmation because it can be hard to see how much the different noise frequencies are surpressed when they're all on top of each other. I could see that the amplitude of the 5KHz dropped quite a bit enabling the BP filter, so I moved the HP to 5500Hz and the LP to 4500Hz. The signal was still clean but the amplitude came back up. It's really nice to be able to do that real-time without having to write SW. We can't do that on our Keysight scopes because they don't support BP filters (only LP and HP).

Enabling the math function on the bandpass filtered inputs i punched in the formula (RMS(Ch1)-RMS(Ch2))/(RMS(Ch1)+RMS(Ch2)) and got a DC level reflecting the LVDT position. Then I could play around with the BP cut-off frequencies and watch the impact on the LVDT in real-time on the screen. It was quite easy to find the optimal filter cut-off frequencies and pass the spec to the SW guys. I guess the SW guys could do the same optimization when they get their SW running, but it's just so much quicker to tweak a knob on a scope than calculating a new filter, change the coefficients, compile and run the new SW. It's a bit like having a lab-in-a-box. It won't replace the lab but I think it will reduce the number of trips to the lab significantly. It will also be easy to bring it to the hydraulic lab where they do metal work. It's not a nice environment to bring a bunch of individual instruments.

All-in-all it was a nice first experience with this scope. The board for this project also has CAN bus for vehicle and internal main/safe processor communication, I²C for external EEPROM and SPI for inertial sensors, so I'll get to test a bunch of serial decoding pretty soon. I may run into issues later when I dig deeper into those functionalities? That remains to be seen, but for now I'm quite pleased. I may be an untypical user? but I actually like that it doesn't have a touch screen. Call me old fashioned, but I hate it when menus keep popping up left and right on the Keysights when I try to point out an observation on the screen to a colleague. I'm probalbly just an old grumpy guy that wasn't brought up by an iPad, but I rather like buttons and knobs. I guess it's just a matter of getting used to a different way, but old habits die hard.

On top of the improved ease of use and performance (double analog bandwidth, 10x memory, dual AWG, DMM and dual PSU to name a few) i got it for €1411. The Keysight DSOX3014T is all right but cost us €3720 and the full-option bundle costs €2444 (which is the cheapest way to get the serial decoding we need). Finally we need the LAN option at €378 to connect to our LabVIEW test system. The total cost of the Keysight is 4,6 times what we paid for the GW Instek, so until I run into trouble with the Instek it does come across as quite a bargain.