Example Project: 20 Channel solid state scan card for K2000 DMM

[voltsandjolts]

I have previously posted a 10 channel armature scan card here:
https://www.eevblog.com/forum/circuit-studio/example-project-relay-scan-card-for-k2000-dmm/

I needed more channels for a particular job where a little SSR channel leakage wouldn't be a problem, here is the result.
It's been tested in K2000 and DMM6500.
It's switchable, 10 or 20 channels; some 2000 series gear might only support 10 channels (?), and saves some money if you only need 10 channels.
Initially I thought my K2000 would only support 10 channels but to my surprise the 20 channel card is supported too - at least it is in the latest firmware.

The PIC MCU firmware has a known issue when using with a DMM6500.
Switching between 2W and 4W resistance on the same channel causes the channel relay to open.
It works fine changing between channels in 2W and 4W.
I just need voltage measurements so didn't spend time fixing it, there's a note in the firmware source.

Thanks to forum members alm and cozdas for providing some background information and inspiration.

Edit:
I expect this card will NOT work in a K2001, K2002 or K2010 DMM for reasons described in this post.

FYI I adapted this card design to function in the Keysight 34970A/34972A DAQs, see here.

[voltsandjolts]

CircuitStudio Project attached in 7-Zip format

[jaudo]

Looks amazing!

I'm starting with Circuit Studio. I've seen you are manufacturing with JLPCB. Could you export the .rul file for us (if it's designed for JLPCB manufacturing?) this would save me a lot of time.

[voltsandjolts]

I was quite lazy about rules in that project, probably better to use this:
https://github.com/ayberkozgur/jlcpcb-design-rules-stackups

[Yansi]

The "spare SSR" area on the board is funny. 

Also quite surprised about the claimed <1nA typical leakage of the SSR in off state.  (but still it doesn't beat a proper reed relay)

[voltsandjolts]

The "spare SSR" area on the board is funny. 

Hah, yeh, lots of board area going to waste so why not!
I'm quite experienced in letting the smoke out of SSRs 

[mtnzap73]

The DMM6500 datasheet on  page 12 has the following:

Temperature Characteristics
.....
Thermocouple Reference Junction External (CJC on 2001-TCSCAN or user-provided with 2000-SCAN) or simulated (fixed)
......

The 'user-provided with 2000-SCAN': does this mean that a user can connect a Thermocouple Reference Junction to a 2000-SCAN card? (like the 2001-TCSCAN card which comes with the reference)

Most if not all temperature measurement info leads to the 2001-TCSCAN card.

[voltsandjolts]

You could make a reference junction but its probably simpler to use a PRT and 4W resistance measurement.
The Ohms to degc calculation is shown in this post:
https://www.eevblog.com/forum/circuit-studio/example-project-relay-scan-card-for-k2000-dmm/msg2884320/#msg2884320

[Anders Petersson]

Many thanks, I'd be happy to get away from the clicking of 2000-SCAN, but I wonder about the downsides of using solid-state relays?
The specified Vishay VOR1142M4 (http://www.farnell.com/datasheets/2614513.pdf) lists typical on-resistance as 22 ohm.

And at least DMM6500 lists max 4-wire ohms lead resistance as

5 Ω per lead for 1 Ω range
10% of range per lead for 10 Ω, 100 Ω, 1 kΩ, and 10 kΩ ranges
1 kΩ per lead for 100 kΩ, 1 MΩ, 10 MΩ, and 100 MΩ ranges

Does this mean this scan card is only suitable to measure 1 kΩ and higher?

[voltsandjolts]

Yes, you're correct.

SSR selection is a compromise between leakage current and on-state resistance. Since the more common scenario is scanning for voltage measurements, where on-state resistance doesn't matter much, I chose the VOR1142 for its primarily for its low off-state leakage current. It also has good contact voltage rating and isolation rating.

You could look around for other SSRs in same package, with a lower on-resistance which would suit your use case better.

[Anders Petersson]

I looked around and found Omron G3VM-61VY2 (https://omronfs.omron.com/en_US/ecb/products/pdf/en-g3vm_61g_61vy1.pdf)
On-state resistance typ. 1 ohm, max 2 ohm
Off-state leakage typ. 0.1 nA, max 1 uA
The only downside I see is the rating for 60 V, where VOR1142M4 has 400 V.

And depending on how lucky one feels, the worst-case G3VM-61VY2 leakage of 1 uA is higher than VOR1142M4 0.1 uA (at up to 100V).
Wonder why Omron specifies a worst-case leakage 10000x higher than the typical leakage.

[voltsandjolts]

For 60V max applications that Omron looks like a fair choice.
It's a little slow at 2ms max turn-on but I doubt the keithley firmware will be scanning anywhere near that fast even at low resolutions.
For manufacturing, the wide typ/max range helps with semiconductor yield (less failures) and 1uA makes device test faster/easier = cheaper.
Most will be way closer to typical than max.
Of course, you could do your own selection testing to pick the best of a batch if critical.

[Anders Petersson]

Thanks!
A faster variant, but a bit higher resistance and leakage, is https://www.vishay.com/docs/84724/vo1400aeftr.pdf
leakage typ. 2 nA (max 1 uA)
On-state resistance typ 2.3 ohm, max 5 ohm
Turn-on max 500 us
Rated 60 V

With these two options, I'm confident to try it out.

By the way, the closest commercial equivalency I find is Keithley 7710 (http://www.farnell.com/datasheets/1884807.pdf) that states "Actuation time <0.5ms" and "Scanning speeds of up to 500 channels/second". Also, it's 60 VDC and 5 ohm.
...and about 10x the price of DIY.

[KedasProbe]

What is the cost of the PCB?
Anyone in Europe with some PCB order experience wants to order a few more empty PCB's and sell it?
(this will reduce the fixed cost much)

edit:
I tried https://cart.jlcpcb.com/quote  with the provided zip file.
It's like 20 euro for 10 plus shipping, is it that cheap or am I missing something here?
4 SSR components will cost me more than 1 PCB.
And they have the vo1400aeftr in their library if you want them to assemble, probably also cheaper than buying it separately. (only one side can be done though, maybe they should al be moved to the top side.)

[KedasProbe]

Is there is 3D enclosure model for 3D printing or is it not really required to hold it in place in a DMM6500?

[voltsandjolts]

No enclosure required, it fits as-is.
Some (all?) of the OEM modules have covers to reduce risk of electric shock.

[KedasProbe]

This may also be good alternative for the VO1400AEF
https://jlcpcb.com/parts/componentSearch?searchTxt=CPC1014N
Is more expensive but can handle more current with a bit lower resistance and has a low typical leakage.

I installed the trial of CircuitStudio, is there a way to export this project to a free software package?

[voltsandjolts]

I think this importer is a work-in-progress but might be of some help, never tried it.
https://forum.kicad.info/t/work-in-progress-native-altium-importer/20712

[KedasProbe]

Tried to see what it would cost with their (jlcpcb) free assembly coupon but they don't have any component in stock (even if it's in their database)
So I will just order 5 PCB's without any modifications.

Their Analysis gives:
layers : 2
minimum trace width : >=10 mil
minimum trace spacing : >=10 mil
minimum drill size : 0.406 mm
width : 228.6 mm
height : 71.89 mm

So I guess the provided zip file is ok for them.

edit: for 5 PCBs  (I could order 10 but then they will just not be used)
Merchandise Total: €9.34
Shipping Charge: €15.81
Order Total: €25.15  (including taxes)

[KedasProbe]

Some comparison, since I have at least 2 weeks to choose components
I may order 2 or 3 to measure them.

[KedasProbe]

Since I haven't setup a scanner card before, how is the delay between 'change channel' and the measurement on that channel arranged?
Is that a set delay by the user or a fixed value of the device?

Edit:
Since nobody is reacting I assume the devices are usually slower than the SSR relays to start a measurement hence no delay actions required, 'how slow' is something that will need to be tested I guess.

I'm planning to make an 10 channel with VOR1142M4  and an 10 channel with G3VM-61VY3
(why not 20, well I don't see myself using it yet so I can always upgrade later)

[KedasProbe]

I received the PCBs from JLCPCB today, for those interested, pictures attached.
The only changes I did from default was lead free and blue instead of green.

[KedasProbe]

Since I didn't noticed the link mentioned here, this is another project based on the work already done by others.
This one is with an STM32
https://www.christidis.info/index.php/personal-projects/keithley-dmm-scanner-card-with-ssr

[czecht]

I would like to buy couple of the PCB's from you and also, if you have all the parts for it, just let me know what you need from me and the price. I live in Lincoln, Nebraska, USA 68505 and my personal email is czecht@gmail.com
Do you have youtube channel?
Or some other place to see your videos?
Did you make a video how to assemble the PCB?
Is it working now?
My Keithley is DMM6500 and I just purchased it few days ago.
Thank you very much.
Tony

[KedasProbe]

This is what I ordered but I haven't assembled it yet it's an "important but not urgent" project so I don't know yet if I'm missing something or not.
To order the PCBs it's 25 euro for 5 so there isn't really much profit in shipping spare PCBs around.
For components it's about 70 euro for one 10 channels card. (this BOM is for 2 different cards)


It also includes a pickit programmer since I didn't have one so you may want to remove that, and maybe buy a few less spares.
The STM32 version above is maybe a better option.

[KedasProbe]

Had some free time to work on this
I tested a few G3VM-61VY3 SSR to see if there are outliers that I should not use.
Screenshot below is of 3 SSR. (next to each other in the packaging)

Led current was 5mA 1.26V  (yellow)
Voltage over the output (blue)  serie with 2.1 Ohm and 0.5v supply and 0.3 Ohm wire gives about 0.5 Ohm that is above the typical 0.15 Ohm (max. spec: 2 Ohm), probably some other small resistances in the test setup are making it difficult since it's not a soldered setup.
'Vp-' gives an idea about how equal the resistance is. (assuming the setup stays the same)
'vpp' would be the voltage over the resistor.

Timing ON is about 1.4ms edge to edge but to 'ON' you need to add 0.2ms (see screenshot), that is below the typical 2ms.
Trigger OFF is very fast 0.078ms so that's good, below the typical 0.1, so you don't have to wait before enabling the other relays. (in theory you could do it at the same time without creating shorts, having 1ms margin)

Edit:
With 7.5mA 1.3V Led I get about 1.0+0.2ms on time, off time is the same <0.1ms (there were even a few with 0.90+0.2ms on time.)
7.5mA is also the advised current for this SSR according to the spec sheet.
So for this SSR I will change the 470 Ohm on the PCB to 300 Ohm.

If someone knows what the reason is of this small hiccup near complete open (blue falling edge), feel free to let me know.

Edit2:
1) The hiccup was the power supply who couldn't follow the fast current step. (about 150mA , <1ms)

2) I also did some measurements with 10 Mohm in serie (like Voltage DMM6500, I used the scope probe 10Mohm in serie) with the SSR, Supply 0.5V, then it's faster on:
ON 0.53 ms with 7.5mA led
ON 0.45ms with 10mA led
OFF 0.10 - 0.11ms is typical

Edit3:
17 * G3VM-61VY3,  3 * VOR1142M4T, both 1V 1kOhm load (1mA), 7.5mA led.
yellow is again led voltage, blue is SSR out voltage.
First timing is ON edge time, second timing is OFF edge time.

[Anders Petersson]

Thanks for sharing, Kedas
I use G3VM-61VY2 for 4-wire resistance measurements with good result.

[KedasProbe]

Hi,

I do have a spare PCB but I don't want to go to the trouble of shipping for this price.
Have you tried ordering the PCB? (The price isn't very high, most of the cost is in shipping)
Just upload the provided zip file from the thread starter in https://cart.jlcpcb.com/quote

The pictures I provided are so you would see what you receive.

P.S. 'ghg' Your PM inbox is full.

[ddavidebor]

Hi,

Has anyone been working on the thermocouple card?

[voltsandjolts]

I just use PT100 4W with the standard card.
CJ comp is a bit of a pain because you need an iso-thermal block for the contact block, like the copper one in 2001-TSCAN.

[aronake]

I am thinking about building a scanning card for DMM6500.

What are experiences from those of you who have built this model? Recommended? Tried any other DIY scan card? How was that in comparison to this?

[voltsandjolts]

Well, I designed it, so I'll leave others to provide critique.
I don't think many have been built, maybe Anders Petersson has some feedback (ping'd him).

I think the 10 channel card (which doesn't need an mcu) is more popular, as long as you don't mind the relay clicking noise.

[Anders Petersson]

I actually modified and built another PCB, since that one was drawn in KiCAD so I could cad in an edge connector to plug in a whole set of wires. Here's the one I started from: https://github.com/cozdas/CozScan2020

Mine is working well with DMM6500. It doesn't work with Keithley 2010 however -- possibly solvable with some firmware modification. Maybe voltsandjolts version works with older DMMs.

With lots of channels, even the touchscreen of DMM6500 is inconvenient when setting up channels and navigating the data. I'll make PC scripts instead.

I also have an official 2000-SCAN card with 10 channels and clicking relays. I don't need the electrical advantages of the electromechanic relays and hearing the ticking wearing on the relays gets on my nerves. 2000-SCAN works with Keithley 2010 though.

[aronake]

Well, I designed it, so I'll leave others to provide critique.
I don't think many have been built, maybe Anders Petersson has some feedback (ping'd him).

I think the 10 channel card (which doesn't need an mcu) is more popular, as long as you don't mind the relay clicking noise.

Could you elaborate on:
"Switching between 2W and 4W resistance on the same channel causes the channel relay to open.
It works fine changing between channels in 2W and 4W."

In the first post? So possible to do 4W meassurement but need to set up in a certain way?

otherwise all works well for you?

[voltsandjolts]

"Switching between 2W and 4W resistance on the same channel causes the channel relay to open. It works fine changing between channels in 2W and 4W."

When my PIC firmware receives a command (from the DMM) to open a relay, it opens all relays. I do this because it makes it simple to behave in a safe way and ensure there are no unintended conduction paths between channels.

This method works fine on the K2000 DMM, because it always sends commands to opens all relays before closing any.

But the DMM6500 doesn't do that - it remembers which relays are open and which are closed. So, when changing between 2W and 4W it just sends a command to open or close the sense channel relay. My firmware sees the open command and opens all relays. So its a safe failure mode. Workaround is to select 4W mode, then change to the channel you want to measure.

Firmware can be changed to fix this, but you would be best to do some checks on which relays are closed, to be safe.
It doesn't affect me because I just do 2W voltage scanning, so I did firmware in simplest/safest way which suits me.

[aronake]

I have been reading though lots of SSR data sheets.

Seems G3VM-81GR1 is a good choice for voltage measurement up to 80 V. Leakage current max 1 nA (close to 0 typically). Downside is resistance up to 25 ohn (16 ohm typical), but that does not matter for voltage.

https://www.mouser.kr/datasheet/2/307/en_g3vm_81gr-2583855.pdf

It also have very low "Capacity between terminals".

For a low ohm version, G3VM-61GR2 seems good. Leak current max 100 nA (0.2 nA typically), 0.7 ohm max and 0.25 ohm typically.

https://omronfs.omron.com/en_US/ecb/products/pdf/en-g3vm_61gr1.pdf

My plan is to make a new card of this kind (already have one with G3VM-61VY3). Short out the sense and 2 wire relays (so signal only go though 1 relay) and have 16 channels with G3VM-81GR1 for voltage measurement and 4 channels with G3VM-61GR2 for temperature (with termistor) and 2 wire ohm measurement

[voltsandjolts]

Seems G3VM-81GR1 is a good choice for voltage measurement up to 80 V. Leakage current max 1 nA (close to 0 typically).

Nice find. Indeed the datasheet graph shows typical leakage as 10pA at 80V, 5pA at 40V

[Doctorandus_P]

Except maybe that some of those G3VM chips go over USD20

https://octopart.com/search?q=g3vm

Is that considered "normal" for a package with two fet's and a diode, or is it a result of the chip shortage?

Ever since "marketing" has been invented some 150 years ago the correlation between what it costs to make something, and what you need to pay that same something has become ever more fuzzier. In these modern times the difference can be all over the place.

But it does not really feel "right" to me that a simple SMPS IC for example costs EUR3 if you buy it form a "western" brand, while a similar chip costs 30ct or less if it's from a local Chinse manufacturer and you buy it from LCSC.

[voltsandjolts]

Yes, expensive now thanks to chipageddon.
But that is being discussed elsewhere, please keep this thread on topic.

[jrc]

voltsandjolts - many thanks again for developing and sharing the design. I ordered some PCBs from PCBWay using your design files. They have now arrived and I will start assembly as soon as the rest of the parts arrive.
One substitution I have made is Toshiba TLP 241B instead of TLP3558A for the relays.
Max voltage drops from 200V to 100V, but current goes from 0.7A to 2A and most interestingly, max resistance drops from 2 ohm to 200 mohm max, 110 mohm typical.
For my applications it's a worthwhile tradeoff, and others might want to check these out too.
One question on the PCB - I noticed there are no solder pads for the small shunt resistors.
Are they meant to be added as an option across the relay terminals, using their existing solder pads? I probably won't use them but would like to know in case I change my mind later.
Thanks again!

[voltsandjolts]

One question on the PCB - I noticed there are no solder pads for the small shunt resistors.
Are they meant to be added as an option across the relay terminals, using their existing solder pads?

First post, welcome to the forum, jrc.

Yeh, I was focused on voltage scanning and didn't place footprints for current shunts.
For a low dissipation shunt, you could place a small SMD resistor across the SSR pins, as you suggest.
For higher dissipation, use the screw terminal pin pads on underside of PCB with larger SMD resistors (less heating of the SSR).
I tend to use leaded resistor in the screw terminals - but that could be a pain if you have several channels and shunts (needs insulated legs to prevent shorts).

Have fun with the build - we need pictures, or it didn't happen

[ozkarah]

I have a DMM6500.
Has anyone assembled a few of these cards and has an excess one to sell?

[wmattias]

Hi Anders!

Would you mind sharing the KiCad files for your modified version?

[voltsandjolts]

You might need to PM him as he probably doesn't get notified of new posts here.

I think KiCAD now has an Altium import tool, which should work with the CircuitStudio files in the OP (let us know how it goes if you try it). But maybe you are specifically looking for the KiCAD card by cozdas.

[Doctorandus_P]

I think KiCAD now has an Altium import tool, which should work with the CircuitStudio files in the OP (let us know how it goes if you try it).

KiCad does not import altium "projects", but KiCad can import schematic files and PCB files from altium.
I had a bit of spare time so I gave it a try, and it worked quite well, but not perfect.
I had some trouble organizing the files, but I found out it's mostly my stubborn brain. I kept saving files in different directories and I (or KiCad?) got confused by that. In the 4th try I just created all KiCad files in the unzipped circuitstudio directory and later deleted most of the circuitstudio files. KiCad is also a bit stubborn in that it wants the main project schematic to have the same name as the project.

* Schematic import worked for both schematic files.
* Schematic import of inverted labels does not work. J3 pin 2 is missing the CTS label. U45 pin 30 is missing the 2WIRE label, etc.
* KiCad imports the page layout over it's own page layout, this needs some cleanup.
* KiCad imports the PCB outline on an "Outline" layer, but it has to be on Edge.Cuts for KiCad to recognize it.
* KiCad needs a hierarchical sheet, it cannot have all sheets on the same level.
* KiCad's PCB editor complains about some "zero-sized pad". I have not checked into it.

Overall it worked quite well. KiCad even renders the 3D view of the PCB including the step models. Only the big IDC connector on the left seems to be missing. I've attached the converted project below. I did not do the cleanup.
I did the conversion in KiCad V7.0-rc1 to get some experience with it. This does mean you can not open this project in KiCad V6, but if you want a KiCad V6 project, you can convert it yourself from the circuitstudio project.

[Doctorandus_P]

I had a bit more fun with this project, and did some cleanup in the conversion to KiCad.

* Fixed the damaged labels.
* Replaced local labels with global labels (or else KiCad does not recognize the connection.
* Exported PCB footprints to another library.
* Replaced the Ciruitstudio pdf with a KiCad generated one.
* Added link to this website to the schematic.
* Re-linked schematic symbols with PCB footprints, updated the PCB.
* KiCad names the "unnamed nets" differently.
* Re-run ERC and DRC, fixed some more minor issues.
* Deleted the circuitstudio paper and titleblock.

KiCad now recognizes all the PCB tracks and agrees that it's 100% routed.
I did ignore some DRC issues.
I also do not consider this conversion "complete" and you should check it yourself and compare it with the original project.

Overall, it's quite a nice project, but my interest is limited because I do not have a "K2000 DMM".

Some things I noticed:
KiCad flags several "isolated copper" areas and the GND zones around the microcontroller are quite fragmented. I would reassign the uC pins to improve the PCB routing. This would also get rid of the serpentine tracks to some of the relays. I also checked with the original gerber files of this project and under the TPDMM connector, the copper is not touching. One of my (small) annoyances in KiCad is that the THT pads for such connectors are quite big, and I'm relieved to see that KiCad is not the only program here My solution is usually to replace the footprints with footprints with oval pads that leave more room for the GND zone to sneak in between pins.

I would probably also have placed all relays on the topside of the PCB for easier assembly.

Why is "STROBE" connected to two uC pins?

[Fabian]

Hi everyone! Thanks for the nice project first of all.
I build two of the cards, but they don't work. Maybe someone can kick me in the right direction 
What I did so far: I flashed the PIC with a MPLAB ICD 3 using the MPLAB IPE by setting the device to PIC16F18875, switching to advanced mode, importing the hex file and pressing "Program". The card gets detected by my DMM6500, it is set to 20-Channel mode. The OUTA pins are connected to the rear input jacks (I don't want to use 4W). Nevertheless, the DMM only measures zero volts, while I inject 5V to Channel 1 (and Ch1 is enabled as DCV in the DMM). When I feed the voltage directly into the rear jacks, the DMM shows the correct voltage.
So, I guess the switching is not working properly. Is there anything I am missing? Settings of the DMM? Wiring? Can I easily check that the PIC is doing its thing (I am not a PIC person)?
It is not exactly clear to me if I need the USB-Serial adapter to supply the PIC.

[Fabian]

Update: Ch1-Ch10 are working (it was, obviously, my fault and a stupid mistake  )
But Ch11-Ch20 are not working. Switch is set to 20Ch. Any ideas?

[Fabian]

Nevermind, it was a solder issue 

[Doctorandus_P]

Maybe someone can kick me in the right direction 

Now I'm wondering how hard and in which direction you want to be kicked for this soldering mistake. 

[voltsandjolts]

Hi Fabian - you got it working, that's great.

It is not exactly clear to me if I need the USB-Serial adapter to supply the PIC.

No, not needed when using the card with K2000 or DMM6500. The FTDI USB-Serial connection was intended as an option to control the scan card from a PC if using with some other DMM, but the firmware was never implemented for that.