DIY-SMU Project

[RikV]

Part 2 of 2

[RikV]

The Nextion GUI interface in *.TFT format.

[Roehrenonkel]

Hi,
 
does anyone need a transformer for this project?
I'm planning on ordering a custom-made transformer.
Torroid with overall-height under 35mm (still want to put it in a 1 Rackunit (44mm),
2*115V Input, Outputs: 2*120V, 4*15V, 2*8V.
It could be used for the J-SMU also.

[Roehrenonkel]

Hi RikV,
 

Looks like I finally got it working in a stable way, everything running on "real" SPI with the complete keypad decoding.
....

Cool. Thanks for the work.
Can one run the SMU without the TFT attached, as SCPI-only?
Still i want to squeeze the thing in a  1 19"-Rackunit and maybe connect the TFT externaly with a short cable.

[RikV]

Yes, it could work without display, controlled solely by "computer". Apart from using it as data collector you will need to find some software suitable as User Interface to control it in a comfortable way.
Since you are redesigning, have you considered using ADR1399 for voltage reference?

[Kleinstein]

The ciruit uses an ADC with a 2.5 V refererence and thus the logical choice of a ready made 2.5 reference.
A 7 V reference would need an extra divider to get down to some 2.5 V  (may work with 3.5 V if you are lucky) and the divider would add some possible source of drift. So the advantage of a zener ref. with better TC and lower long term drift may not reach the ADC.

The ciruit at the input of the ADC also includes several resistors and other parts not made for ultimate accuracy.
So I don't think it is worth the trouble with using a 7 V ref.  The good grade bandgap references like LTC6655, ADR441, max6325 should be good enough for the circuit.

A 7 V reference may be better used seprate to check the measuring part from time to time for drift. This check would include more parts than just the reference chip.

[Roehrenonkel]

Hi RikV,
 

Yes, it could work without display, controlled solely by "computer". Apart from using it as data collector you will need to find some software suitable as User Interface to control it in a comfortable way.
Since you are redesigning, have you considered using ADR1399 for voltage reference?

For now i'm using HP VEE 6.01 (see attached plot of an Aa triode) but want to move to linux-sw for the raspberry.
 
V-ref.: No, i setteled for the MAX6325. Good inital accuracy (+-1mV), low tc (1ppm), available and not overpriced.
A hi-end V-reference could be a tempting new project for me after using heated-723, 1n823, LH0070/71, LM399 and LT1019s.... ;-)

[jaromir]

I'm planning on ordering a custom-made transformer.
Torroid with overall-height under 35mm (still want to put it in a 1 Rackunit (44mm),
2*115V Input, Outputs: 2*120V, 4*15V, 2*8V.
It could be used for the J-SMU also.

In my design I used double shielded transformer - analog parts were in separate shielded domain - so that common mode leakage is as small as possible; see here https://github.com/jaromir-sukuba/J-SMU/blob/master/pcb/schematics_pdf/smu_interconnect.pdf

I achieved less than 1uA p-p of leakage. Classic toroidal transformers do have one or two more orders of magnitude higher leakage.

[RikV]

ADR1399 is the "updated" version of th LM399. Better, cheaper. But as Jaromir states: a standard toroid transformer is a no-go in this application. You must at least go for "medical grade" and a separate transformer for the digital part.
BTW: Dave showed some sign of activity : http://www.djerickson.com/hp3466a-dmm
Welcome back!

[Bansci]

Marco Reps did a great video on common-mode leakage and mitigation/low-leakage designs just recently (see below)

Jaromir, Marco has a excel file of leakage measurements here: https://docs.google.com/spreadsheets/d/139WT1bAHAmnD6NX4IW7SPVKkw-OCP7QnEriZ-MRHLrw/edit

Your less than 1uA p-p leakage gets you to keithley 2400 levels, so well done! Would be nice to see the exact measurement to see where you land on that table with your/dave's design.

 

[djerickson]

Hi All,
Thanks for all the interest in DIY-SMU. Sorry I haven't checked this blog in a while. I 've been updating my project web site though so hopefully you have seen my progress here. http://www.djerickson.com/diy_smu

Parts shortages: Ugh! In addition to dealing with these, here's a quick summary of my recent changes:
 
The  Voltage and Current cal software (Python) is working very well. I will release it.

I built up 2 new Rev2 main boards with the new Rev3 Amp boards. The main reason for the upgrade is to make the voltage clamps safe in the following conditions:
   Force current mode with voltage clamp
   External voltage is applied that exceeds the clamp voltage setting
   Low current ranges: < 100mA
Before this fix, theses conditions would cause damage to the current range section. The fix is to switch the Amp current limit to 15mA for the 10mA and lower ranges.

I decided to rework my first 2 Rev1 units to apply this fix. The rework is farily simple: I generate the 100mA range inverted signal in firmware to a spare GPIO bit, and add one wire. And upgrade the amplifier boards to Rev3. I decided to upgrade the 2 Rev 1's because between the parts shortages and the labor of hand building these boards,  the built and running Rev1 boards are valuable.

My remaining firmware tasks:
   Better triggering, particularly from SCPI. Requires making the ADC code non-blocking. I have some ideas...
   SCPI command to move cal data to EEPROM. Currently requires a re-compile after calibration
   Finish the keypad value entry code. Keypad (Nextion) is working, but need to parse the output string to set values, test against limits...
   Some basic plotting.
   
I have not yet built the 3-sided sheet-metal cover. Accurate sheet metal bending is not my forte.

I have 4 fully functioning DIY-SMU units and am looking for someone(s) to take 1 or 2 units to do some testing or help with the firmware development. I'm also looking for more python control: curve tracer, plotting,.... Send me a message here to let me know what you can contribute and I'll try to pick the best 2 candidates.

Thanks,
Dave

[RikV]

Hi, Dave,
Great to see you are back!
It is indeed not good looking to upgrade a rev1 main board to rev2. But it works all well. Since it is impossible however to include the extra decoupling around the ADC "noise" becomes a problem in the most sensistive ranges. I would "like " to rebuild my main board to Rev2 to have a cleaner unit. Is the software solution also of use in the Rev2 board or is it switched out by a #def?
You may not have noticed it but I succeeded in implementing native SPI, solving the problem of blocking ADC.

If you find some time (I am retired too, I know very well we have NO time) could you pls publish main board Rev2, including conversion from 595 to 594?

[djerickson]

The ADC rework isn't too terrible. I also changed the 2.5V reference from ADR431 to ADR421. The '421 is a better part: lower noise and more tolerant of the larger bypass caps needed for the ADC. If you want to rework a Rev1 to a Rev2, Change the ADC to ADR421 and add 10uF tantalum caps to the reference and 5V near the ADC. BTW L2 is a ferrite bead, 1K ohms at 100MHz. I did not rework this on my Rev 1 boards.

The software fix to invert the 100MA/ signal and send it to U8-7 has no other effect on a Rev 1 or Rev 2 board since it is unused. The rework is to add a wire, U8-7 to J2-6. The tricky part is to cut away the GND plane to J2 pin 6 first.

I will update the doc package this weekend: new HW docs, Arduino code,  Nextion code, and the Python calibration code, put it on Github...

Thanks for the HW SPI code! I couldn't find your .INO's at first look. Will check again.

In addition to non-blocking SPI, I plan to improve the triggering for SCPI to make timing faster and predictable.  Basically implement 'auto-trigger' so when SCPI MEASure commands occur, they control the ADC. After a short timeout of no SCPI, the main loop will kick in. And yes, it needs an ADC speed (NPLC...) command. Wish me luck:-)

Thanks,
Dave

[djerickson]

RikV:
I cannot see any .INO files in the .ZIP files you uploaded. I see .HMI and some libraries.  Am I missing something?
Thanks,
Dave

[RikV]

Dave,
The files are spread over the blog. I include a .zip with the state as-is today. in the header of the main .ino I included some comment on most of the changes I made. Hope something can be of help.

One thought that came up to me: since it is impossible to really adjust the offsets of the 9 ranges to exactly "0", why not use DAC ch4 in the way Jaromir does: Add a fraction of it to ch.A? He uses a ratio of 1M6/10k (160:1). Could even be much more. This would give a fine step correction without using a potmeter as you did on the 18bit V-ref. Of course, 9 extra cal. values... But in this instrument I believe that it is important that 0 is REALLY 0. How do you feel about this? This could also bea way to correct the DAC output using the ADC, or was this your plan?

[Chen Li]

Hello Mr Erickson,
I anylized your smu design, the most import thing is just dont put mlcc capcitors directly between the output pin of an opamp and gnd. It will oscillate forever.
And not just the opamp on the surface, but also those components have opamp kind thing inside themselves. So the adr431 adr421, if you see their function block diagrams, the output section are basically opamps.
Another thing is dont put dozens capcitors in parellel, when the component is powered up, it need enough electricity, because those capcitors need a lot time to be charged, so maybe sometimes the component just dont get enough energy to mainten the function, so it shut off. You can put one inductor instead.

[djerickson]

Hello Mr Erickson,
I anylized your smu design, the most import thing is just dont put mlcc capcitors directly between the output pin of an opamp and gnd. It will oscillate forever.
And not just the opamp on the surface, but also those components have opamp kind thing inside themselves. So the adr431 adr421, if you see their function block diagrams, the output section are basically opamps.
Another thing is dont put dozens capcitors in parellel, when the component is powered up, it need enough electricity, because those capcitors need a lot time to be charged, so maybe sometimes the component just dont get enough energy to mainten the function, so it shut off. You can put one inductor instead.

Hi, and thanks for the feedback.

I agree, no MLCC caps on op-amp outputs. I'm careful about this. In my reference circuit design I use only tantalums or normal electrolytics: no Low ESR caps. Their higher ESR (1-3 ohms) allows most op-amps and references to drive them without stability issues. Spice it and you will see. If I was concerned about super-low ESR, I'd add about a 2 ohm resistors in series with the caps.

The ADR431 that I originally used has stability issues driving larger ceramic caps. That's one reason I switched to the ADR421.

[RikV]

Hello, Dave
Regarding temperature drift (stability) I have done as follows:
R51-52 and R53-55 are MPMT2002ATS (SOT23,10k+10k,0.1%,0.05%,25ppm/K,2ppm/K) and these fit nicely in the actual layout. I would absolutely have replaced R19-20 by MPM but the layout does not allow it. At this place there is an immediate gain of 20ppm/K to 2ppm/K for a very small cost. Those who are willing to spend more money might use LT5400 for the R4-5-6 and R7-8-9 combinations. not sure if its worth the necessary modif in the layout but for R51-52-53-55 it would be an important -cheap?- improvement.
I installed ADR4525CRZ (1ppm/K) and C19 is 22µF MLLC. I will increase C17 also.
"10K, .1%, 25ppm 0805  are about $0.15. 10ppm ones are about $0.40". Where did you find those resistors that cheap?

Do you have any use in mind for DAC channel D (DAC_SP)? If not, I would use it to improve the offset.

[Kleinstein]

At mouser some of the SMD resistors are in that price range (e.g. Susumu RR / RG or panasonic ERA ) when buying 10 or more. The 25 ppm/K resistors in Qty100 even get to the $0.03 range.

In between single SMD resistors and rather expensive LT5400 there are  ORN / MORN networks.  The MORN type should also fit on a LT5400 footprint, just not thermal pad.

With resistors from the same batch the TC matching is usually better than just random resistors. So the solution with just 0805 resistors is not that bad. An there are enough of the 10 K to warrant buying 100 pieces 25 ppm/K.

edit:
One could use the extra DAC channel D in parallel to channel A to reduce the noise and INL error a little. I don't know if there are other significant noise sources, but the DAC is not that super low noise in x2 gain mode.

[Chen Li]

No sir, I think you misunderstand me. We must see the whole picture.  As I marked on the function diagram, the output pin of adr431 is actually outpin of the opamp function block. So cant put a capcitor directly between this pin and gnd either. And so dose adr421. And I think the adr431 is better than adr421. You didnt find the root cause.
I didnit see pcb file carefully. The stray capacitance must be considered also. This output pin is very sensitive. Put this pin away from gnd. If you pour large area copper gnd plane nearby, the distance at least 60 mil. If you go multilayer pcb, the copper area below this output pin, I mean in the 2nd layer generally gnd plane, just cut off either.

[Kleinstein]

Some of the references, including the ADR421 can work with quite some capacitance at the output. The output stage is no a normal OP but some capacitive loadung is factored in and some even need capacitance. Consider the reference output more like the output of a voltage regulator.  The capacitor at the reference output is not a problem, though it may not be needed.

The capacitor C9 is a bit odd. I don't think it would help and may do more harm than good, though with enough ESR the OP may not oscillate.

[djerickson]

Thanks for the good feedback on resistor networks. I updated the latest BOM to put 10K 10ppm 0805s for the most critical parts.

My drift spreadsheet is for gain drift, but I need to also consider offset drift. Drift is mostly caused by -2.5VREF and the ADC and DAC scaling amps. And some on the V Measure circuit.

Also considered a combined either dual-0805 or SOT23 footprint.  My current layout has most of the 2 10Ks side-by-side, but the common pins are not adjacent Doh! Easy to fix so a SOT23 could be used there. I need to figure out how to use a single footprint with 2 parts: R42A and R42B maybe.

In some places (-2.5VREF and crossover) A vishay ORN 4x 10K at 5ppm could replace 2 expensive SOT23s for  less $$.

On the DAC and ADC scaling circuits (R12/14/15 and R4/5/6) the 2 10K resistors affect offset,  one 10K and the 21.5K affect gain.  I ideally need a 10K/10K/22k network. I could use a Vishay 10K ORN (5ppm) to get 10K/10K/20K and add a 2K 10ppm. The absolute tempco pf ORN is 25ppm so the 2K will degrade the 5ppm just a bit.

I ordered some 10K 5ppm 0805 resistors and 10K SOT23 networks. Will use these parts to build my most drifty unit into a drift 'lab queen' and test its V Force offset and gain tempcos.

[Roehrenonkel]

Hi Dave,
 
thank you for including the NPLC-command (in future versions) and pointing out the MORN- and MPM-arrays.

I have build up the CPU-board to get a feeling for the user-interface.
Inputs over the touch-screen work (a bit slow), Incremental-encoder and switches (except "left" and "right") don't work.

After Reset the "fanhi"-pin should be high for a few seconds to give the fan an initial "kick".
Even my quality Papst-fan is stuck with only 4.3V and makes just noise.

EEPROM-placement: Doesn't have to be on the floating side even if it
belongs on the analog-board since the isolators are on the board as well.

None of the R-Arrays needed are in stock at Mouser, Bürklin nor Digikey.
Why these strange gain-factors anyway? Didn't 1:2 or 1:2.5 work?
I understand that it would lower the amplitude at the ADC to +2.5V +-2V
with a 1:2-ratio instead of +-2.26V in the original design.
Requested a quote for a custom-Caddock T914 with 2*10k, 21k5 and 22k1.
And got an answer right away at 09:00 (AM). At least 250 piecces - no price given.
Big thanks to their distributor AMS-technologies - they care even for one resistor.
No thanks to Vishay-europe they don't even care to reply.
And a big "F... you!" to Conec-conectors. Snobs!

Since i'am layouting my own board i'll probably use the VTF1006/158/212-arrays.
Shunt-resistors will be Caddock USF240 for 5Meg (have an USF340 5ppm left now)
and Vishay S102C/K down to 50 Ohms. Just need a good 556k now.

[Kleinstein]

It should be OK to use a fixed 1:2 gain for the DAC and ADC part. This would reduce the full scale range at the regulator / diode OR part to something like  4.5 V instead of 5 V.  To still adjust the measurement and adjustment range, one would still need one point to adjust to gain, e.g. at R64,R65 So these would have to be a little (some 10%) higher. The current ranges (or shunts) would also change a little.

Skipping the extra gain / attenuation for the DACs / ADC, and running the regulator / cross over part with a 2.5 V level would compromise the regulation a little, but could simplify the circuit a little and a void one gain/attenuation stage.

Even than the measurement part is not made for highest precision, as there are quite some resistive dividers / gain stages in the signal path. So don't expect the SMU to get an accuracy level of a 7 digit DMM in the voltage ranges. The SMU circuit just needs more compromises to get the high grade current measurement and the large voltage range at the same time.

Also the performance at the very low currents is not the same an electrometer - there are quite some switches as parallel path to the 5 M resistor. So there is a limit to what resistor grade makes sense there.  For the lower value shunts good resistors make absolute sense. Here much is about having good power handling capability, as they may run a bit hot. Compared to a more normal DMM the votlage drop is quite large. So the S102 would be too low in power for the 50 Ohms. The 1 W power rating for the 100 Ohms resistors R40, R41 is still on the low side. I would not be ashamed to use 10 W rated parts (e.g. wire wound or with heat sink mounting) there.

For better performance towards higher currents (e.g. > 10 mA) one could consider gain, e.g. a INA  (maybe 2 gain settings) instead of what is now U21.1 + U21.2.

[djerickson]

Hi Dave,
 
thank you for including the NPLC-command (in future versions) and pointing out the MORN- and MPM-arrays.

I have build up the CPU-board to get a feeling for the user-interface.
Inputs over the touch-screen work (a bit slow), Incremental-encoder and switches (except "left" and "right") don't work.

After Reset the "fanhi"-pin should be high for a few seconds to give the fan an initial "kick". Even my quality Papst-fan is stuck with only 4.3V and makes just noise.

EEPROM-placement: Doesn't have to be on the floating side even if it belongs on the analog-board since the isolators are on the board as well.

None of the R-Arrays needed are in stock at Mouser, Bürklin nor Digikey. Why these strange gain-factors anyway? Didn't 1:2 or 1:2.5 work?
I understand that it would lower the amplitude at the ADC to +2.5V +-2V with a 1:2-ratio instead of +-2.26V in the original design.
Requested a quote for a custom-Caddock T914 with 2*10k, 21k5 and 22k1. And got an answer right away at 09:00 (AM). At least 250 piecces - no price given.
Big thanks to their distributor AMS-technologies - they care even for one resistor. No thanks to Vishay-europe they don't even care to reply.
And a big "F... you!" to Conec-conectors. Snobs!

Since i'am layouting my own board i'll probably use the VTF1006/158/212-arrays.
Shunt-resistors will be Caddock USF240 for 5Meg (have an USF340 5ppm left now)
and Vishay S102C/K down to 50 Ohms. Just need a good 556k now.

Hi. The slow control response is probably because with no ADC, the ADC read software is timing out, making the main loop s-l-o-w. Comment out the ADC and things should get nice and fast. The latest firmware has all the switches working except PREV and NEXT.  Maybe you are using old firmware?

I have not finished the Main Board EEPROM HW or SW. It was intended that each main board would carry it's own calibration. Having EEPROM on the CPU is good enough though.

NPLC is not implemented yet, its coming....

I find that most 12V fans will run slow at 5V. Try another. The one I specified works well.

Mouser has Vishay MPM 10K/10K parts in stock. I just ordered some, and also some 0805 5ppm resistors.

Best of luck! If you need a DIY-SMU front panel to hold your CPU, Nextion, and controls, let me know.

Dave