Author Topic: DIY-SMU Project (Read 79237 times)

RikV · « **Reply #75 on:** July 18, 2021, 09:31:32 pm »

Could someone hint me to where I can find the Nextion .HMI file for this project? According to what Dave says it should be included in his latest update but I don't see it. I would love to proceed on this project.

My fault: just found the Github page: https://github.com/djerickson/diy-smu
Thanks again for the interesting work!

RikV · « **Reply #76 on:** July 19, 2021, 03:06:19 pm »

Where can I find those exotic USB-to-panel cables?

prasimix · « **Reply #77 on:** July 19, 2021, 03:13:03 pm »

All kind of "impossible" and "exotic" cables could be found on aliexpress

RikV · « **Reply #78 on:** July 19, 2021, 09:29:25 pm »

No doubt about that! All kinds of stuff you can imagine and a lot more! But... You must kno how to ask kindlyk. "Cabinet mounted USB-B to USB-Bmicro cable" doesn't return anything useful. how do you describe such a cable in a way ebay or other search engines understand what you mean? Is there a dictionary somewhere? I have the same problem when seaching componenet in an on-line catalogue. "What is the F** name othis f** thing??". I must be getting too old for this I guess.

prasimix · « **Reply #79 on:** July 20, 2021, 05:27:25 am »

What's about "screw lock panel mounted usb cable"? https://www.aliexpress.com/wholesale?catId=0&initiative_id=SB_20210719212607&SearchText=screw+lock+panel+mounted+usb+cable

poorchava · « **Reply #80 on:** July 20, 2021, 08:31:41 am »

As for GUI I'd say: TouchGFX is the best by a large margin. At least when it comes to performance & capabilities vs cost (which is 0). It used to be horrendously expensive, but ST bought them and provides it for free. Obviously works only with STM32 now. It can provide a modern Android-like experience.

LVGL is nice, but waaaay more rudimentary than TGFX and waaaay lower performance, despite TGFX being written in C++ (and not the reduced kind, full blown with virtual methods and abstract classes). I also found, that LVGL requires much more coding to obtain same basic functionality as compared to TGFX.

But I'm one of those people who can't stand the Arduino crowd (maybe because of number of customers begging their inquiry with 'i've done it with Arduino so 3 weeks is plenty to go into series production for market, right? That's what we told the investors....'). Ohh, and the questions along the lines of 'how to run my AVR at 100Mhz' and 'I CAN HAZ HDMI ON AT MEGA?' </flame>

RikV · « **Reply #81 on:** July 20, 2021, 04:44:29 pm »

@Prasimix: not only creative with chips but also with words.
Thanks for the extra hint. I had already found what I was looking for, only took waaay tooo long. Must be the age.
As lon as I can handsolder a 0603, you won't hear me complain.

ducreux92 · « **Reply #82 on:** July 26, 2021, 02:24:52 pm »

Hi,
Is this project reproducible ?

How many are working properly in the world ?

Thanks.

ducreux92 · « **Reply #83 on:** July 28, 2021, 02:20:31 pm »

Hi,

I tested a prototype on breadboard with TL074 with sucess.

I used +-15V powers instead +-170 V and only two powers mosfet ( doesn't works with 4 mosfets at 15V)
At output i have about 3 mVpp noise.

Why not replace the circuits OPA2140 with cheaper circuits like ADA4522-X for example ?
idem for AD5686 ?

Kleinstein · « **Reply #84 on:** July 28, 2021, 03:46:02 pm »

I had asked about the OPA2140 before - it just was available and cost was not such an issue for just a prototype.
There are different places the chip is used and different places may really want that chip or could use different alternatives (e.g. the ADA4522, MCP6V51, maybe OPA2202 or maybe even TL032 at some places).

ducreux92 · « **Reply #85 on:** August 09, 2021, 08:47:23 am »

Hi,
I find opa2197 (opa197) which is much cheaper that opa2140 with similar caracteristics, but
architecture is CMOS not JFET amplifier - Is this amp suitable for the project ?

Kleinstein · « **Reply #86 on:** August 09, 2021, 09:19:29 am »

The OPA2197 has more low frequency noise, but is still a good OP. I like the low power consupltion for the speed.
Another cheaper replacement relatively close to the OPA2140 is the OPA1642 (mainly higher offset). It really depends on the positions - the OPA2140 may not even be a good choice in some places.

With hobby projects it is not so uncommon to use the parts you have at hand and know and this may than look odd.
Currently some parts may also not be avialable (or not from the favorite source) - this can also cause strange looking choices.

djerickson · « **Reply #87 on:** August 26, 2021, 11:24:25 am »

Thanks for the op amp substitute ideas for the OPA140/2140. I think JFET is the right choice, particularly for low bias current stages. Bipolar is generally higher input current, and CMOS is challenging for +/-15V, generally higher drift and noise. I don't trust the linearity (CMRR) of CMOS: they need circuit tricks to get wide input voltage range that sometimes compromise linearity.

I settled on OPA145/2145 as my go-to precision amp. Similar DC and noise specs to the OPA140/2140, slower (5.5MHz vs 11MHZ), better availability, and ~1/2 the cost.

I have built 2 complete DIY-SMUs so far. I'm getting ready to build the 'final' Rev2 Main boards and Rev3 Amp boards.

Jaromir has designed a very nice take on this project and has built a couple of units. His design uses an off-the shelf enclosure, and a custom AC transformer. And a simpler LED / buttons front panel similar to the original K236 instead of a touch screen.

If there is interest, I can get the new boards machine-assembled and can sell a few. At least the main board with its 175 parts. The Amp board (75 parts) is not bad to hand-build.

There are firmware additions and fixes that I plan:
EEPROM based calibration. Currently requires recompile. A Python app does the hard part.
Control of remote sense. Currently SCPI only
Better Clamp control. Basically works, not accurate.
Basic plotting page

Thanks,
Dave Erickson

DIY-SMU: www.djerickson.com/diy-smu
Jaromir's DIY-SMU: https://www.eevblog.com/forum/projects/how-i-made-my-smu/new/#new

ducreux92 · « **Reply #88 on:** August 28, 2021, 10:37:50 am »

Hi,
Today impossible to find AD5686XX in tssop format.

Do you know an equivalent ?

djerickson · « **Reply #89 on:** August 28, 2021, 01:38:07 pm »

Thanks for the heads-up on the AD5686. Last I checked 2 months ago some were available in TSSOP. Just not the most accurate grade.
I guess some options are to change to the harder-to-solder QFN, or add a footprint for the similar 8 channel AD5676.

Dave

stephanm · « **Reply #90 on:** September 12, 2021, 10:11:05 am »

Hi,

first of all: awesome project! Using LTspice and looking a bit into Dave's curcuit, I saw that the output voltage can overshoot when the instrument comes out of compliance, depending on the set voltage and compliance. Here's an example: Force voltage is set to 1V, and the current clamp is set to 10mA. The load is just a 1Meg resistor. At 5ms within the simulation I am shorting the 1Meg RLoad resistor with a 1 Milliohms resistor, and keep it shorted until 7.5ms. During the short, the circuit goes into compliance as expected, limiting the output current to the programmed 10mA.

However, there are two transients on the output when applying and, respectively, removing the short:

When applying the short, there is a short and large current spike (above 100mA), which is ok because I am basically shorting the 135pF output capacitor of the SMU. There is also some oscillation visible which even drives the output voltage below 0V(!), but this does not concern me so much at the moment.
What concerns me more is that when removing the short, the output volage overshoots quite a lot, it goes up to 7V, after which it takes 1.5ms to recover to the programmed 1V.

From my understanding of the DIY-SMU circuit, a voltage overshoot after removing a short on the output is hard or even impossible to avoid due to the fact that the circuit has to deal with the voltage drop on the current sense resistor. As I had set the current sense resistor to 500 Ohms, the voltage drop on that resistor is 5V when the instrument is in compliance, and these 5V plus some additional voltage will end up on the output when the load current suddenly drops from 10mA to a few microamperes due to removing the short.

Looking at the Keithley 236 schematics, I was looking for some circuitry to deal with this situation and reduce the overshoot, but so far I have not found anything that looks like doing so. So, what I am curious of is whether such overshoot is also encountered on other SMUs, such as the Keithley 236?

Thanks and kind regards,

Stephan.

djerickson · « **Reply #91 on:** October 04, 2021, 03:26:25 pm »

Thanks for the feedback. I tried the same test on 2 units. For the same conditions: 10mA, 15V range, 1V force, switching from short to 1Mohm load, I get a similar amplitude, but the response is faster, 10uS wide, not the ~1mS you see in simulation.

Send me your .asc file and I'll try to figure out what's up.
Dave

Johnny10 · « **Reply #92 on:** October 04, 2021, 04:46:29 pm »

stephanm · « **Reply #93 on:** October 05, 2021, 07:08:19 pm »

Hi Dave,

apologies first, as I saw when looking through my last simulation again a part of the problem was self-made, I was running the circuit in 150V mode (R4=300K!), which is not what I wanted for a 1V output voltage. The attached .asc file has this corrected, I have set R4=20K (that roughly matches R64||R65 on the schematic you published on your web site) and V3=0.333V (which gives Vout=1V as we now have a voltage gain of 3). I also stopped playing with C5, it does not have an effect on what I see.

Please correct me if I am wrong, but from my understanding these settings for R4 and V3 should be what I need for 1V output in +/- 15V mode.

With the fixed LTspice schematic file I see the same behaviour as previously described, but now the transient when removing the short is considerably shorter, in the same ballpark as what you measured (thanks for that, very interesting!). However, the spike amplitude has grown to 16V in the simulation...

Thanks and kind regards

Stephan.

8334455 · « **Reply #94 on:** November 15, 2021, 06:40:12 am »

great work, but what is next update time in you blog? expecting the new update!

djerickson · « **Reply #95 on:** November 17, 2021, 10:19:09 pm »

There's new stuff on my DIY-SMU web page now. I got the new rev. PC boards in and they are working fine.
Addressed the current limit issue.
Fun with parts shortages (NOT!)
(Mostly) automatic calibration working
New chassis wiring schematic

I will release the latest HW docs soon.

Thanks,
Dave
www.djerickson.com/diy_smu

RikV · « **Reply #96 on:** November 23, 2021, 08:02:24 pm »

Dave, thanks for this project! Also thanks for the continuation to keep this thread up-to-date. But I unfortunately could not find any updates? Am I too inpatient or did I look in the wrong place (happens more and more to me)?

guymo · « **Reply #97 on:** November 25, 2021, 01:27:21 pm »

I believe the updates are at the bottom of this page:
http://www.djerickson.com/diy_smu/smu-bringup2.html

RikV · « **Reply #98 on:** November 29, 2021, 04:24:04 pm »

I made some updates to the Nextion main screen:
Suppress the update of the MV while Iset or Vse menu is on screen.
In these menues I colored the background of the active selection
Make Iselect menu disappear when Vselect is activated and vice versa.
some alignment on the screen layout.

RikV · « **Reply #99 on:** December 01, 2021, 07:20:40 pm »

Hi, Dave,
Thanks and respect for your continuing effort on this project!
I am in the phase of testing my build of your Rev.1 main board. So far not much result. I noticed that DAC pin RSTSEL should be connected to VCC but is left floating on the PCB. Has this been corrected yet? Is corrected in Rev2 as I can see on the Gerbers.
I also cannot see the pictures (Nextion Keyboard) associated with you recent update. Am I missing some authorisation?

Seems to be OK now. Thanks
Any other errors in R1 PCB I missed?


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: DIY-SMU Project (Read 79237 times)

Share me