...Now to functionality. This is where it really gets bad, especially if you initially believe the banner specs. I tested on two different computers (a laptop and a desktop, both fairly recent models, and both running Windows 7 64-bit home edition) but the results were identical on both....
Actually, the software may be more XP friendly than Win7 then. I use XP and the software ran well (no lags, smooth and snappy) on even a very slow but very small 1GHz laptop I often use for space reasons.
The software is portable (don't need to install - just copy the installed folder. You do need to install the driver). I just copy the folder over and without fuss the software is running. (Without having to run install means I am having less junk clogging up the system and adding junk to the registry. I like software that just copy-and-go.)
- no AC coupling (but at least the user can add a capacitor himself to get this.
Yeah, I hate that too. I have a couple of capacitors at hand just because of that.
- uses two USB ports (but at least on my machine it worked just as well with a standard single-port cable. I did not measure the instrument's current draw but I'd be surprised if it exceeds 500mA. Probably they just re-used the same cable as for the other Hantek models.
Not to confuse others, it adds two devices via one usb connection - one for each channel. So, one single USB cable does it.
- Self cal did not execute, no matter what I tried
This one confused me a bit at the start too. Until I read the manual that both channel need to be grounded. It does execute but no feedback. My volt readout is still high (by about 2-5%) but not as much over as before I ran the self calibration (by about 5-7%).
- I did not see any way of displaying pretrigger wafeforms. Did I miss anything? The specs claim it can do pretrigger acquisition.
After using it for a while, I got the trick worked out now:
- pause the display
- kill the cursor (otherwise, mouse click sets cursor)
- now point at wave form, press mouse button, and drag the wave from left to right
Then you can see the part of the wave before the trigger. Problem is, the amount of info before the trigger is random. What I guess is this: The software appear to treat the memory buffer as a ring. The data just get written in and once it gets to the end, it returns to the start but clears the entire buffer. So, when the trigger occurs when the WRITE pointer is near the end, you captured lots of pre-trigger but little post trigger wave form when the buffer is plotted start to finish linearly. However, when the WRITE pointer is near the start of the buffer, you captured darn little pre-trigger info.
- Trigger seems very noisy, i.e. it's hard to get a truly stable waveform on the screen; it will often trigger on the wrong edge (e.g. falling edge when you selected rising).
- Trigger instability seems to get worse when you scroll the waveform far to le left (takes many, many mouse moves as stated above).
No kidding here. The noise is awful and it does cause trigger to hit. It is very annoying but at least is is still useable. I was able to find the 50mV resonance wave to measure an inductor's inductance. I do find a reboot sometimes helps. So it could be picking up the noise from other connected devices.
- At 10 us/div and faster the scope no longer acquires full 1M data sets, but rather short ones (maybe 1000 or 2000 points).
- At 5 us/div the acquisition gets completely messed up. Actually that sometimes happened already at 10 us/div. The acquired waveform is completely unstable. Single shot aquisition shows many short sections seemingly randomly joined together. Result is different every time you re-acquire.
- At 2us/div (sampling at 8 MHz) and faster the display shows random sections of the waveform, most of the time there is empty space on the left or right side. Single shot acquisition shows that the scope seems to acquire a short data record (one screen width worth of data, maybe 1000 points or so), without any triggering, and then randomly misaligns it on the screen, so only a portion of the screen is filled. I can scroll back or forth to display the full record, but again, it does not seem to trigger on anything constant... so these sample speeds are completely useless. It could be the instrument simply fills its FIFO (4K, so that would be 2000 8-bit samples per channel) and without bothering with triggering, and then dumps it onto the screen on a random start location). Hard to believe they wouldn't have found this even with the tiniest bit of product testing.
I think I was caught with this same confusion too, but here is what I found out:
At 2us or below is when it captures at 48msps and at merely 1060 datapoints. At 5us or slower, it goes to over 100K datapoints at 16msps. The slower the more datapoints.
BUT...
When you freeze the display and zoom in/out (say capture at 5us and zoom to view at 100ns), the display does not show your capture frequency (5us=100,000+ samples) and instead it shows what it would be at 100ns (1060 samples at 100ns). After I realized that, I can see that it works. Of course, the noise still make that difficult.
- Vertical sensitivity (part 1): The unit's maximum range is only +/-5V. So in 2V/div you can only use +/-2.5 divisions around zero. And even with the probes in 1:10 mode you can't display anything larger than 50V. Maybe that's Hantek's way of making sure you stay away from any dangerous voltage :=)
- Vertical sensitivity (part 2): 20mV/div and 50mV/div are fake: Thge waveform gets simply blown up (like digital zoom on a digital camera), so the actual vertical resolution is very coarse. Becomes very obvious when you switch to "display dots", you can clearly see the large quantization steps. So the true smallest resolution is a poor 100 mV/div.
- Unit is not galvanically isolated from the PC, but I would say this is to be expected given the low price. Even the cheapest galvanic isolation (data & power) would add a couple $$ to the component cost. Given the 50V limit (see above) the lack of isolation seems the least of my worries with this instrument; it's clearly intended for low-voltage work anyway.
- Analog bandwidth: At least here actual performance does not seem to be orders of magnitudes off from what the vendor claims. Of course it is not trivial to get a useful acquisition where you can measure signal rise time, given the problems at fast sample rates above. But it looks like the true bandwidth is somewhere around 17 MHz for large signals (5V pk-pk, acquired at 1V/div), dropping to a bit below 10 MHz for smaller (100 mV/div).
- Noise: there are spikes on the acquired signal, around +/-20mV. I suspect the switching regulator that is used to supply the analog input section. Visible but less concerning than other shortcomings, especially given the limited actual vertica, resolution (see above). Averaging would be a good way to make these spikes disappear in the display, as they are not correlated to the input signal - but as said, the software does not offer averaging...
I agree with you 100% here... Only 8 bit resolution to +- 5Volts, so it is about 40mV per ADC bits. I am not sure if they have a way (such as lowering their reference voltage) to get the lower readings.
At best sampling rate (48msps), it is about 21ns between samples. With a 10mhz wave, you have 100ns per period, 50ns high and 50ns low. Best case is 2 samples during wave-high, and 2 samples during wave-low with PERHAPS another sample somewhere in the middle. Can't measure rise time with that.
Overall though, I am please with the scope. For merely $70, it does more than I expect.