Hantek - Tekway - DSO hack - get 200MHz bw for free

[tatus1969]

@Fixup: BNC t connector + term did also not eliminate that residual ringing. My measurements above show that the signal is clean at the BNC socket, but is ringing at the JFET gate, so I am sure that the distortion is caused by the signal path going long ways and through a relay. I will keep you updated.

To the LMH6552 line driver: my idea is to eliminate the series termination and let the opamp directly drive the line. This way I also completely eliminate that RC lowpass that you mention, which will give me more bandwidth. The crux with series termination in general is that you either have it low-ohmic enough for good bandwidth but not matched with the wave impedance (and I doubt that 33/22 ohms is well matched), OR you have good match but too little bandwidth. This is the reason why I want to terminate at the end of the transmission lines. I hope the LMH6552 can drive the required load without overheating, will keep you updated.

400ps sounds really promising :-)

[tatus1969]

yesterday I made some tests regarding that issue with "toggling" bandwidth (https://www.eevblog.com/forum/testgear/hantek-tekway-dso-hack-get-200mhz-bw-for-free/msg965398/#msg965398):
- removed varactor: not better
- updated to latest firmware (got this directly from Hantek): not better
- tried /tst file with "[filter] 35": not better
- tried /tst file with "[filter] 1000": not better (is that appropriate anyway?)

This must be happening in digital domain, stuck here... As if the scope cannot decide if it wants to be a 60MHz or a 200MHz device. I will try and address this to Hantek, but there is chance that they say that I have bought the 60MHz variant, so where's the problem...

I made some progress to my input stage work (flat pulse response, more bandwidth):
- 50 ohm termination (connected to left side of C01_5 according to "das_oszi_schematic.pdf"): works like a charm and completely eliminates all overshoot at JFET input, but I realized that I lose 1:10 attenuation and self calibration of course, so this is not a practical and permanent solution.
- I found that the LMH6552 wildly oscillates when I short RA01_1/2. A look in the datasheet tells me that it is not capable of driving the high capacitive load of our transmission line. On page 20, the datasheet suggests 21 ohms external series resistance at its output for CL=39pF. Hantek has put RA01_1/2=4x22 ohms at the output in my case. Coincidence? So I put that resistor(s) back in.
- the LM6552 datasheet also suggests to choose not too low resistor values for the feedback network (page 10, fig. 9) when I want gain flatness. So I will probably keep that 440ohms that Hantek put in, maybe lower to 390.
- the LMH6552 still seems to have a hard time driving that transmission line, so I put 1.5pF on top of RX1_3 and RX1_4 (that resistors have 440 ohms in my case). This reduces amplifier bandwidth to ~ 250MHz, and completely eliminates overshoot. Reducing this to 1.0pF raises BW to 360MHz but adds a little overshoot, could be a good compromise.
- after that 1.5pF fix, the reflection due to {missing termination at receiver / unmatched termination at transmitter} is still there, and it shows up as a dip of ~2ns, just after the signal slope has ended. So I added three termination resistors: one across R03_29/R03_32 (not on top of them but such that it terminates the end of that t-line), and one each across the inputs of RL04_1 and RL04_2 (to the bottom of the PCB, soldered directly to the corresponding vias... quite nasty). The traces going from the relays to the ADCs cannot be terminated because of that 500MSPS/1GSPS switching/paralleling. I am not right there yet, but a resistor value of 70 ohms each already shows good improvement. I will need to find the optimum by experimenting more. Later, I will need to raise LMH6552 gain a bit such that it compensates the attenuation that I have introduced here (resistive divider RA01_1/2  <--> termination).

I am still waiting for parts for my 500ps pulse generator, and for my 750MHz active Tek probe, so this task will sleep until then. Keep you updated 

[tatus1969]

Reached my goal: Gaussian response, as much bandwidth as possible.

The setup is
- 500ps square wave generator (self made, using a high speed clock output pin from a spare LPC4330 Xpresso board, verified to be free from overshoot with LeCroy 1GHz scope + active probe)
- a used TEK P6205 JFET probe (750MHz) that I power directly from Hantek scope (needs +-5V)
- extensive modifications to the input stage, I will summarize that in a separate post once I have cleaned up the mess that I have created

The waveform was taken in equivalent time acquisition mode, combined with dot display mode. This combination is the only one that prevents the firmware from adding some nasty digital post filtering.

Yes, this is a (hacked) Hantek DSO5062B with 250MHz analog bandwidth 

Thanks to Fixup for the fruitful discussion 

[Badwater]

Hi,

I bought the 4072c model, it is very similar to the 5000 series, but the detail to connect the RS232 console differs.
Anybody already figured out where to connect?

DSO-Tools are basically working, but neither backup (Backup_B_models) nor and the BW changer.

Root directory holds the known files:

Code: [Select]

drwxr-xr-x    2 root     root           880 Jul  2 20:32 OurLanguages
drwxr-xr-x    2 root     root         24016 Jun  8  2015 bin
-rwxrwxrwx    1 root     root            43 Jan  1  2009 chk_base_volt
-rw-r--r--    1 root     root             4 Jan  1  2000 cur_acq.type
drwxr-xr-x    4 root     root             0 Jul  4 00:11 dev
-rw-r--r--    1 root     root           125 Jul  2 19:50 disk_sta.info
drwxr-xr-x    5 root     root           488 Mar  7 09:00 dso
-rwxr-xr-x    1 root     root       4529432 Jul  2 20:32 dso_bin
-rwxr-xr-x    1 root     root             0 Mar  7 09:00 dst1072b
-rwxr-xr-x    1 root     root             0 Jun  8  2015 dst1102b
drwxr-xr-x    5 root     root           808 Jul  2 20:30 etc
-rwxr-xr-x    1 root     root        924672 Jun  8  2015 help.db
-rw-r--r--    1 root     root          8192 Jun 12  2015 i2c.log
drwxr-xr-x    2 root     root         11096 Jun  8  2015 icon
-rwxr-xr-x    1 root     root          1131 Jun  8  2015 keyprotocol.inf
-rw-r--r--    1 root     root             0 Jun 12  2015 language.img
drwxr-xr-x    6 root     root         15120 Jul  2 20:32 lib
lrwxrwxrwx    1 root     root            11 Jun  8  2015 linuxrc -> bin/busybox
-rw-r--r--    1 root     root             8 Jul  4 00:11 logotype
-rwxr-xr-x    1 root     root            15 Mar  7 09:00 logotype.dis
-rw-r--r--    1 root     root        524288 Jul  4 00:11 misc.bin
drwxr-xr-x    5 root     root           352 Jun  8  2015 mnt
-rw-r--r--    1 root     root             0 Jul  4 00:27 msg
-rw-r--r--    1 root     root        317477 Jul  2 20:55 mult_adc.log
drwxr-xr-x    3 root     root           224 Jan  1  2000 param
dr-xr-xr-x   52 root     root             0 Jan  1  1970 proc
-rwxr-xr-x    1 root     root          4448 Jun  8  2015 protocol.inf
drwxr-xr-x    2 root     root           160 Jun  8  2015 root
drwxr-xr-x    2 root     root          5264 Jul  2 20:32 sbin
drwxr-xr-x   11 root     root             0 Jul  4 00:11 sys
-rw-r--r--    1 root     root           105 Jul  4 00:11 sys.inf
-rw-r--r--    1 root     root          2400 Jan  1  2009 tdc.log
-rw-r--r--    1 root     root          1113 Jan  1  2009 tdc_edge125M
-rw-r--r--    1 root     root          1126 Jan  1  2009 tdc_overtime125M
-rw-r--r--    1 root     root          1114 Jan  1  2009 tdc_pulse125M
drwxr-xr-x    2 root     root             0 Jul  4 00:11 tmp
drwxr-xr-x    8 root     root           552 Jun  8  2015 usr
drwxr-xr-x    2 root     root             0 Jul  4 00:11 var

Edit: OK,  found the RS232 header thing; it is near the USB plug as depicted elsewhere in this thread.

[Fixup]

  tatus1969

That signal looks really good and without any overshoot.
It's been a pleasure to share some thoughts about the input stage in order to optimize it and in the same time learn something new.

Looking forward to hear more about which setup you ended up to use.

P.S
Have you tested that the AC signal and DC offset is stable over different temperatures (cold oscilloscope / warmed up)?

[tatus1969]

@Fixup: thanks for that, same for me 

Modding description is almost complete, maybe this evening After my last post I realized that in fact I didn't completely reach my goal, after seeing the input stage literally oscillate when connecting the stock passive probe In fact I probably had created a delay line oscillator or sort of But that should be fixed now.

I will check for gain and offset drift. I also plan to measure the frequency response, I'm curious how that will look like.

[tatus1969]

I think I finally got it right. Bandwidth is even just above 250MHz now. I have collected everything in a document, please let me know what you think about it. Also critical comments are welcome (well, except asking why I did it :-)

[Macbeth]

Sorry tinhead, but... 

THE KING IS DEAD, LONG LIVE THE KING!...

Welcome, tatus1969! 

[aandrew]

Wow, thank you tatus1969!

I've got an MSO5074FG which is the 4 channel version of the DSO5000 series with function generator and logic analyzer. I really appreciate your analysis and critique of the component selection on the analog front end.

One thing I've noticed with my particular unit is that the mainboard was definitely not washed; the board is a little sticky to the touch and I was wondering whether it's worth the risk to take some flux remover and give it a good scrub, followed with some isoprop to get the surface of the board squeaky clean. The digital stuff won't care so much but the input circuitry seems like it might appreciate this kind of treatment. Do you have any comment on that?

[tatus1969]

@Macbeth: thanks  But as some famous theoretical physicist would have said, everything is relative. I guess tinhead and I have to share that cake  My attempt to compensate that excessive input inductance works well for 50 ohm feeds, but it also adds another 22pF of input capacitance. This makes compensation harder for passive probes, and also reduces their bandwidth (lucky me here, I only got 60MHz probes with my Hantek anyway). And I didn't succeed yet getting the scope's attenuator compensation right, there is always at least one of the four options (direct signal vs attenuator, 1:1 vs 1:10 probe setting) that isn't working out. Not sure if there is a solution at all with standard probes...

If someone has a better idea how to handle that inductance (I think I can believe my minVNA reading of 48nH here, because simulation and reality went very close), I will give away one of two slices of my cake

[tatus1969]

One thing I've noticed with my particular unit is that the mainboard was definitely not washed; the board is a little sticky to the touch and I was wondering whether it's worth the risk to take some flux remover and give it a good scrub, followed with some isoprop to get the surface of the board squeaky clean. The digital stuff won't care so much but the input circuitry seems like it might appreciate this kind of treatment. Do you have any comment on that?

I use to clean boards like this
- flux remover, take some time and use a lot of it at the end  in order to dilute the dissolved flux enough for the next step
- bath the entire board in distilled water
- rinse after taking out of bath
- dry with compressed air

This way there will not be any flux residue left. (Compare: just clean with flux remover and let it dry - looks awful.)

This is good for boards that contain only reflow solderable parts. They have already survived more aggressive situations like vapor phase soldering... The only part that I would be worried about for my Hantek mainboard would be the RTC backup battery. Perhaps better remove it before cleaning.

[Fixup]

  tatus1969

Great document about your modifications. I've read it several times

[tatus1969]

this time I really (?) have it. Working with 50 ohms feed as well as passive probes, with input attenuator bypassed and used. I am hoping for some more "resonance" (as opposed to my scope), as the length of this thread suggests that there should be quite a few owners of that Hantek/Tekway 

[tatus1969]

yesterday the last puzzle piece arrived: Tek P6106 passive probes. Got them for EUR35 from ebay. These are 250MHz 10:1 probes with both LF and HF compensation capabilities. In total there are five (!) trimmers. One of them allows you adjusting capacitive/resistive crossover (the LF compensation that you normally do using the scope's built-in test signal) as usual. The other four allow fine-trimming of overshoot and ringing of fast transients. The probes are for 20pF/1Meg scope inputs, which is perfect for my modded Hantek input stage.

Here are some pictures 

[llg]

hey tatus,
the post-filtering thing, is it specific to "B" units ?
The reason I'm asking this is I have a 200Mhz "P" unit and it's not showing the behaviour you mention in your document.
Perhaps it's harder to notice/reproduce ?

[tatus1969]

can you post a screen shot of a signal with very fast rise time (less than 1.5ns)? That problem only shows up when going to the limits.

[llg]

I see now.. so it's probably the same behaviour.

no, unfortunately I do not have anything close to 1.5ns and not entirely sure I have parts to build a test generator either.

[tatus1969]

it you have something like 74AVC logic or a microcontroller with maybe >150MHz speed you can probably see it.

The post filtering is not a bad thing in general, because it allows complete reconstruction of a sine wave up to the nyquist frequency. It is just totally annoying when you want to examine fast transitions, therefore normally this is a feature that you can switch off in a scope. There are nice articles here https://www.eevblog.com/forum/testgear/sin(x)x-interpolation-and-digital-filters-in-oscilloscopes/ and here http://cdn.teledynelecroy.com/files/whitepapers/wp_interpolation_102203.pdf

[tatus1969]

finished my active probe mod with two TEK P6205 (you can get them for $40 from ebay). The two receive their power directly from the scope. Here are some photos

[Macbeth]

tatus1969, It gives me a laugh to see how pimped out your Hantek is but surely you could have powered the probes with a custom home made TekProbe™ BNC Interface. A software mod for the automatic x10 ID would be the icing on the cake.   

Those P6205 are 750MHz... what are you really able to achieve?

[tatus1969]

and it has been by far cheaper than that one below O0 Thought about making a Tekprobe adapter board, but then I would have had no reason to open the probe's casing, right? 

That combination roughly does 300MHz, check out my PDF summary a few posts above, there are some charts on the last page. Dont have the right equipment to properly do the frequency sweep, maybe that will be the second next project? 

p.s. thats 300MHz for around $400 (not counting some fresh gray hairs) 

p.p.s. did you notice these nice inline terminators? Copied from this great idea https://www.eevblog.com/forum/projects/home-made-feed-through-terminator-for-bnc/

[frohro]

• Can anyone confirm what you can definitely hack on the Hantek DSO5102P for 200 MHz?
  Has anyone added LAN to a DSO5000P series instrument?
  Has Hantek done any nasty things to stop these modifications?

I'm looking at buying a Hantek 5102P but would like confirmation before I do.

Thanks,

Rob

[pascal_sweden]

What is the noise floor of the Hantek oscilloscope? It is in the same range as Rigol or Siglent?

[CustomEngineerer]

Is it in the same range as the Siglent or Rigol  what? They both make  more than 1 model. Also, instead firing off questions for every little thing, is there some reason you can't look up most of this information yourself?

[Badwater]

I can confirm that the LAN hack is working on the DSO4072; got all stuff from China for less than 5 bucks.
I set the IP for my network in the rcS: ifconfig eth0 192.168.xxx.xxx
The driver is complaining while bootup w.r.t. the MAC but things are working.
Fixed MAC may also be set in rcS: ifconfig eth0 hw ether xx:xx:...
I can telnet the DSO but did not find out what else?!
TTscope is not connecting via LAN.

As mentioned elsewhere in this thread, the DSO-USB-TOOL has some issues with the IP connection.