Help choosing a Hantek USB oscilloscope

[RocketScienceSmurf]

First off I would like to ask you to refrain from suggesting non Hantek oscilloscopes.

I am looking for a 4ch USB connected oscilloscope. I bought the Hantek 1008 a couple of years ago and I like but it has very limited bandwidth at only 100kHz. When fiddling around with the royer induction heater I realized I couldn't view the mosfet gate signals with much detail so I need higher bandwidth.

The problem is that I don't really know what bandwidth I need and what the difference between bandwidth and bandwidth limit is. Hantek has a couple of promising candidates but the Hantek6204EU for example has 200MHz bandwidth but 20MHz bandwidth limit. What does this mean?

I would like to be able to see the ringing on the mosfets gate to help me pick the best gate resistor so I guess that if my switching frequency is let's say 40kHz the ringing has an even higher frequency.

Is perhaps 100MHz enough for most of what I need it for or should I go even higher than 200MHz?

[Mads Barnkob]

I hope Hydron will see this thread and give his experiences with USB oscilloscopes, as I never owned one. I never dared due to the low performance / cost, when compared to Rigol etc.

Here is a nice little quick guide to buying oscilloscopes that explains 12 important points, really 3 of them is the most important as a lot doesn't make sense for a USB oscilloscope and at lower bandwidth, bandwidth is desribed at #1: https://www.mouser.com/pdfdocs/Tektronix12_things_to_consider1.pdf

[Hydron]

I would look at a USB scope to get something portable, or for special purposes (e.g. lots of mathematical manipulation of captured waveforms); for general use I'd get a bench scope. I use my USB/Ethernet scopes for special purposes, and have a Rohde&Schwarz RTB2004 for general purpose bench use. Also worth noting is that the USB interface is absolute garbage for EMC immunity, expect it to have major problems when probing a running tesla coil or similar (Ethernet is vastly superior for this).

As for specs, having a decent memory depth would be at least as important as moving from 100 to 200MHz bandwidth, if not more so. Don't buy anything with memory depth measured in kilo-points, you want the ability to capture at least a million samples in one shot, especially for a USB scope where you're likely to be doing more detailed analysis of captured waveforms, and less prodding about at stuff on the bench.

The 20MHz thing is likely a switchable filter - this is a standard feature of most scopes, and is used when you're looking at a low-frequency signal and want to cut out some of the noise that you get from capturing a higher-than-necessary bandwidth.

Also important is making sure that the scope uses an ADC with a high enough sample rate to make sure that the analogue bandwidth isn't sneaking too close to the nyquist rate, e.g. 200MHz bandwidth is a bit iffy if you're only sampling at 500MSamples/s - this is only just above the Nyquist frequency and you can suffer from poor sin(x)/x reconstruction and aliasing from strong signals >250MHz sneaking through the input filter. If the 100 and 200MHz options are using the same ADC and it's only running at 500MSamples/sec (as per my example), then you're not gaining as much as you might think by the bandwidth increase, and are actually increasing the risk of seeing signal artifacts.

Finally, for poking about at gate drive signals 100MHz should be sufficient - I do like having 300MHz available for probing my QCW coil (which has very fast switching speeds) but in reality a 100MHz scope would do the job nearly as well.

If this doesn't make a whole lot of sense, then have a look at some of the info produced by the major scope manufacturers - this should give an idea of the basics, though you do need to watch for each manufacturer emphasising the importance of the features they are most competitive with (e.g. Keysight has great update rate on their low-midrange scopes and will go on and on about this, but very marginal memory depth compared to their competitors and won't talk about that much).

[RocketScienceSmurf]

Wow thank you both for excellent information!

Yes the 20MHz limit seem to be a switchable feature and I did find some information about it in on of Hanteks user manuals. One of the reason I think I prefer a USB oscilloscope as opposed to a bench model is the compactness. I can bring it with me practically anywhere and connect it to almost any computer I can install the software on.

Absolute accuracy is not extremely important. I need to get a good visual representation of the wave form so I can decide if a change in component value make things better or worse. The specs of the Hantek models is different depending on where on their site you look which makes it hard to compare the different models.

It is a bit outside my price range but their DSO3204 looks promising. However the series comparison specifies the rise time of the 200MHz model at 1,7ns but on the product page it says 3,8ns.

http://www.hantek.com/en/ProductDetail_2_134.html
http://www.hantek.com/en/productdetail_144.html

I like the possibility to have an external trigger and still have 4 usable channels. The network feature could also possibly useful to me.

Is the input capacitance of 25pF considered high and might that be a limiting factor? I tend to use regular RG58 cables and solder them in at my measuring points and I think that could add a lot of capacitance as well. I know it's not optimal but a good scope AND proper probes cost a ton of money so I am aiming for "good enough" but still want a fairly good scope. My dirt cheap 8 channel Hantek has helped me out a lot and I like it but as I said, 100kHz bandwidth is VERY limiting when dealing with frequencies above mains or audio. The fact that it's DC bias only also presents a challenge a lot of times.

EMC immunity is one thing to consider but the another is ground loops. I use a small "galvanic separator" with a built in DC-DC-converter but it has a connector for 5V-power so I could use battery or an external DC-supply of my choice. The built in works well with my Hantek1008B but perhaps I will not be able to power a higher end USB oscilloscope.

[Twospoons]

Is the input capacitance of 25pF considered high and might that be a limiting factor? I tend to use regular RG58 cables and solder them in at my measuring points and I think that could add a lot of capacitance as well.

Thats a reasonable thing to do if you are measuring low impedance nodes, but it could limit your effective bandwidth.  It can also screw up some circuits quite badly with all the extra C (a colleague at work tried this with the switch node of a buck converter - made the poor thing ring like a bastard, and nobody knew why until I pointed out what they were doing).  Its easy enough to build a simple 10:1 probe head you can solder in - google will show you a appropriate circuit ( its just two resistors and a variable capacitor). That can give you ~ 1Mohm and 10pF.

Personally I chose bandwidth over channels when I bought a scope ( a Rigol DS2072a, hacked to 300MHz) - don't want to miss any fast transients. Not keen on USB scopes myself, I like knobs and buttons for control.

[Hydron]

25pF does seem a bit high, but not crazily so. I'd recommend for most purposes using real 10x probes - cheap but reasonable probes can be obtained these days on ebay, just take a look at some of the other forums (e.g. eevblog) for some recommendations. The input specs of the Hantek you're asking about look a bit limiting too - you'll definitely need 10x or 100x probes to measure anything over a few 10s of volts (and you will need to be very careful about using AC coupling mode, as it normally results in the DC voltage at the probe tip _not_ being attenuated by the probe's attenuation ratio due to no DC path to ground for voltage division being present in AC coupling mode).