Hantek DS5000 series - 300MHz input stage mod

[tatus1969]

This is a spin off from another thread: https://www.eevblog.com/forum/testgear/hantek-tekway-dso-hack-get-200mhz-bw-for-free/

I was still not satisfied with the impulse response that Tinhead could substantially improve, and went out to see if I can improve it further. I originally posted my results in the above thread, but I realized that it can easly get lost in that 116+ pages.

For anyone interested, I have collected my results in the attached PDF. Please tell me what you think about it, and if someone applies it to his scope I would be happy to see the results. Even better: if someone has a suitable signal generator to make a frequency response graph.

This post shows my adaptation of ebay'ed TEK P6205 active probes to the Hantek: https://www.eevblog.com/forum/testgear/hantek-tekway-dso-hack-get-200mhz-bw-for-free/msg988835/#msg988835

After having put all this together, the Hantek achieves 300MHz analog bandwidth (250MHz+ with 50 Ohm feed) and does not show any overshoot anymore. A picture showing the result (50 Ohm feed, using a DIY 500ns pulse generator):

[tinhead]

This is a spin off from another thread ... I realized that it can easly get lost in that 116+ pages.

argh, click destroyer

After having put all this together, the Hantek achieves 300MHz analog bandwidth (250MHz+ with 50 Ohm feed)

i have two questions:
- did you compared your results in all attenuator positions?
- what is with self calibration, is it working again?

[tatus1969]

argh, click destroyer

sorry about that on the other hand, we should be glad that the forum does not have a "show all posts" button, that would crash every browser for your thread

i have two questions:
- did you compared your results in all attenuator positions?
- what is with self calibration, is it working again?

I am a bit limited there, because my generator does only 1V roughly. I did test correct behavior with enabled input attenuator and its compensation network, and I tested high gains up to the point that would saturate the input stage. Self calibration is also fine. I took care to do only little manipulation to overall gain. JFET offset is compensated by the input stage anyway.

[twinpeaks]

Hi,

I  own a Voltcraft DSO-1062D, which I modded a long time ago to 200 MHz according to tinhead's instructions.
Now after seeing tatus' improvement I was curious and gave it a try.

At first I built myself the same pulse generator as tatus as I also had an eval board with the same microcontroller laying around.
This shows the rising edge of this pulse generator measured with an Agilent infiniium Scope 54825A (500 MHz, 2 GSa/s) with sin(x)/x inerpolation turnded off:


The other images show this signal as measured by my scope before (i.e. with tinhead's mod) and after applying tatus' mod.
Unfortunately my scope can not be tricked to disable the sin(x)/x interpolation by choosing EquTime and dot display mode as with tatus' scope (my scope is HW 1007 and FPGA 83eb), so i assume interpolation was always enabled.

The overshoot was significantly reduced. In fact it is not overshooting at all, it just reaches the final level and then swings back a bit.

Next thing I will do is to adjust the attenuator, because when switching to 500 mv/div, the overshoot is a bit higher than on the 100 and 200 mV setting (see last attachment).

[bbjk7]

Excellent work here tatus, this mod more than doubles the value of the scope! Was about to start ordering components to do mine as well, but you left out the size of them (0603? 0805?)

[tatus1969]

Excellent work here tatus, this mod more than doubles the value of the scope! Was about to start ordering components to do mine as well, but you left out the size of them (0603? 0805?)

thanks :-) All resistors are 0603. I'm not so sure about the resistor networks, but from the pictures I think they are 0603 x4 concave.

[bbjk7]

thanks, another question.. My soldering is pretty good (did IPC Aero courses) and I'd like to replace Q01_3 and Q01_4 with the RF JFET as well (with a spin (a literal 110deg CCW)  on your recommended "nasty, free-air, single-leg soldering magic").. would these require any additional resistor changes also?

What about Q01_2? Could that do with the JFET upgrade as well? And what resistors would need to change?

[tatus1969]

thanks, another question.. My soldering is pretty good (did IPC Aero courses) and I'd like to replace Q01_3 and Q01_4 with the RF JFET as well (with a spin (a literal 110deg CCW)  on your recommended "nasty, free-air, single-leg soldering magic").. would these require any additional resistor changes also?

I've tested MMBFJ309 JFETs on Q01_3/4 but that didn't give any additional improvement. And yes, you'd need to change a large part of the surrounding components as well including the bias voltages. I happen to have dumped my records of this change after having cancelled it...

What about Q01_2? Could that do with the JFET upgrade as well? And what resistors would need to change?

No, that circuit is designed as BJT emitter follower and placing a JFET there would not work out of the box. That transistor is not limiting the bandwidth, because it is used with a gain of 1.

[bbjk7]

so i've followed all the instructions... and managed to somehow stuff it up 

the resting signals (with nothing connected) on both channels are way off the top of the scale, regardless of what the range knob is on or how far the offset knobs are wound down to. the only way to get signals showing is to couple them to ground (both ac and dc coupling don't work)

any ideas as to where I could've gone wrong?

[tatus1969]

so i've followed all the instructions... and managed to somehow stuff it up 

the resting signals (with nothing connected) on both channels are way off the top of the scale, regardless of what the range knob is on or how far the offset knobs are wound down to. the only way to get signals showing is to couple them to ground (both ac and dc coupling don't work)

any ideas as to where I could've gone wrong?

I'd trace out the signal as it travels through. With a 0V input signal, you should see the following (CH1):
TP101 = 0V
TP103 = 0V
RX1/1 and RX1/2 = same voltage as C01_15 (OFFSET2_CH1)
RA01_1 and RA01_2 = same voltage as C01_15 (OFFSET2_CH1)
RL04_01 = same voltage as C01_15 (OFFSET2_CH1)

This is referring to the schematic https://elinux.org/File:Das_oszi_schematic.pdf

You can also check if there is a difference between the vertical ranges, you can hear a relay klick as the attenuator kicks in at 500mV/DIV (%1).

[bbjk7]

thanks tatus.

voltage @ TP101 seems fine: 0.00V
but voltage @ TP103 is at 6.13V with nothing connected
beginning to suspect the transistors are somehow permanently biased on/off
unfortunately don't have a lot of time to work on this, but i'm now without a working scope

[tatus1969]

It looks like that you have replaced Q01_2, whereas Q01_1 looks untouched. Is it possible that you have replaced the wrong part? At least that would explain your result, because that would be replacing a BJT with a JFET.

[bbjk7]

that's... so stupid that it's probably exactly what i've done thanks again tatus.. will fix it and let you know how it goes by some time tomorrow (who needs sleep anyway, right?)

[bbjk7]

Yep, that fixed it..
Only one minor issue left: Channel 2 overshoots by a lot on its calibration signal - looks like the probe isn't compensated correctly
I have fiddled with the trimmer cap on the PCB, which does seem to have an optimal position for minimizing the spike.. but it's still quite large. Also, it only does this for the higher ranges.. which would imply that it has something to do with the attenuator circuit

[tatus1969]

how does it look on channel 1, and how does it look at a vertical gain of < 500mV/DIV (make sure the probe divisor is set to %1)?