Author Topic: EEZ Bench Box 3 - Sequel to EEZ H24005  (Read 564 times)

0 Members and 1 Guest are viewing this topic.

Offline chriswebb

  • Supporter
  • ****
  • Posts: 132
  • Country: us
EEZ Bench Box 3 - Sequel to EEZ H24005
« on: July 12, 2019, 05:46:27 am »
Thought this forum would be interested in the latest product about to launch from the people who did the EEZ H24005 bench power supply:

Quote
The EEZ Bench Box 3 (BB3) represents a new category of modular Test and Measurement (T&M) equipment. It provides a complete hardware and software framework that bridges the gap between – and combines the best features of – DIY hobbyist tools and professional benchtop equipment. BB3 was inspired by the EEZ H24005 power supply, which attracted many enthusiasts with its broad feature set, rich user interface, DIY friendliness, and fully open source design. BB3 will initially ship with modules that provide the same functionality as the H24005, but those modules will be slotted into a device that also delivers improved modularity, enhanced flexibility, greater capacity, and more processing power.

Our campaign will allow backers with different backgrounds, interests, and budgets to configure their BB3 however they like. Get a bare PCB set or add a custom-made enclosure. Add the three core modules, if you like, and you’ll have an Almost-Ready-To-Run (ARTR) bench box into which you can slot one of our power modules. We’re going for maximum flexibility, here.

Link to the crowd supply.

EEVBlog forum launch thread around the H24005 here.
Always learning. The greatest part of life is that there will always be more to learn.
 
The following users thanked this post: prasimix, Aigor

Offline Kean

  • Supporter
  • ****
  • Posts: 830
  • Country: au
  • Embedded systems & IT consultant
    • Kean Electronics
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #1 on: July 12, 2019, 11:22:04 am »
I have a couple of the H24005, and this new design looks quite good.
Ethernet connection should be on the rear panel IMHO.
Maybe they'll make an SMU plug-in - that would be cool.
 

Offline Rerouter

  • Super Contributor
  • ***
  • Posts: 4103
  • Country: au
  • Question Everything... Except This Statement
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #2 on: July 12, 2019, 11:29:48 am »
Kean. If you have thoughts on how your would implement an SMU plugin. I'm willing to lay it out. Been working piecemeal on making a 4 quadrant output stage but I am limited in knowledge as far as making it accurate enough for SMU level use cases.
 

Offline Kean

  • Supporter
  • ****
  • Posts: 830
  • Country: au
  • Embedded systems & IT consultant
    • Kean Electronics
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #3 on: July 12, 2019, 02:27:01 pm »
Kean. If you have thoughts on how your would implement an SMU plugin. I'm willing to lay it out. Been working piecemeal on making a 4 quadrant output stage but I am limited in knowledge as far as making it accurate enough for SMU level use cases.

Not really.  I had dabbled with a design, but the precision I was aiming at was quite low.
I'm not expecting anything like the performance of a Keithley 2400 of course, but something 4 quadrant and uA or better resolution.

There is the EasySMU reference design/evalutation board from AD, but it is somewhat limited with only ±12V/±40mA
Still many occasions I think I'd find that useful.  Of course, there is no stock anywhere of the required eval board.
https://www.analog.com/en/design-center/evaluation-hardware-and-software/evaluation-boards-kits/dc2591a.html#eb-overview

Marco Reps was working on an open source SMU design called OSMU, but development on that seems to have stalled.  Not sure what his planned specs were.
 

Offline Rerouter

  • Super Contributor
  • ***
  • Posts: 4103
  • Country: au
  • Question Everything... Except This Statement
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #4 on: July 12, 2019, 10:23:10 pm »
I would say dream for the keithley, but lets see how we can approach it economically, the main pain for me is I do not know what the input structure looks like on a SMU, I assume it is similar to an electometer but with higher bandwidth for most of the ranges, equally what voltage and current ranges are you expecting?

So we have this, the keithely 2400 output stage, nothing hard or expensive to make, and can be made a lot more flexible by using switching preregulation to reduce the total power dissipation of the output stage, instead of using all those taps, just keep it working like a bootstrap supply to keep the op amp under low power while leaving headroom for the maximum bandwidth of the transient response
783138-0

Watching Dave's teardown, things of note:
5.5 digit 1uA range at the low end, for measureing, while having a 10nA output resolution so, what can we do to get close to that, e.g. maybe we have a 12-16 bit 1uA full scale range, its not impossible, just needs good layout and decent op amps. (ENOB means we probably end up closer to 3-4 digits)

The keithley does use a switching supply to make its +-225V rails, we could just do it a little more modern. tracking instead of fixed for higher output wattage

E.g. they use an AD847JN for the output op amp, in many ways that is a fairly rubbish op amp compared to what is available today. High bias currents, high offset voltage, only really has speed and gain going for it.

The ADC converter is a multi-slope converter controlled by an FPGA, now from another thread on this forum, the base of operation and op amps required for solid operation does not seem expensive, getting it to play nice over multiple decades of input may be harder, and this is where I would currently struggle, As I would probably be more willing to throw it at a self contained ADC chip, and focus on an analog control loop and managing gain with op amps.

I suppose the multislope does gain you an integrator stage, the easier way to do device power measurement would be to measure the integrator voltage set to a suitable range to get the mA from the integrated mAh allowing the measurement that so many EE's seem to struggle with when devices have uA sleep currents mixed in with mA active currents. just at X samples per second sum to bring the integrator to 0V and measure the charge to cancel, this will give you a wide range device power consuption.

Again throw some thoughts out, the man behind this project is very good at switching converters, so we may just need to step in to guide the linear and measurement side.
 

Offline prasimix

  • Supporter
  • ****
  • Posts: 1236
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #5 on: July 13, 2019, 07:52:05 am »
Oh, it seems that party started here without me :). Thanks @chriswebb for opening this topic and others for followed up.

Ethernet connection should be on the rear panel IMHO.

Yeah, it's not simple at all to keep wire harness (check here) to lowest possible level, minimize PCB dimensions (especially the MCU board that has 4-layers) and have all things on the right places. I was thinking a lot, what to do, and for start I'm very satisfied with minimizing AC wiring (and soldering) to the minimum: currently only AC cables are what connect AUX-PS board (where soft-start/standby control is placed) and AC/DC power modules.
Also I'm strongly believe that Ethernet (e.g. wired connection) should be an integral part of such kind of equipment, and that is better to have it on front panel then to remove it completely or replace with wireless. One can argue that we already have (wired) USB, but we have more ambitious plans for it that I believe will be soon in position to present.

Maybe they'll make an SMU plug-in - that would be cool.

Right, and as the first step in that direction I'd like to build a bidirectional 2-quadrant power module. Then support for all four should follows naturally, and of course we have to expand existing software/firmware that decent SMU functionality would be possible.
 

Offline Teichhermelin

  • Contributor
  • Posts: 8
  • Country: de
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #6 on: July 14, 2019, 08:28:36 pm »
Nice Idea
 

Offline zucca

  • Supporter
  • ****
  • Posts: 1847
  • Country: it
  • EE meid in Itali
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #7 on: July 15, 2019, 10:15:19 pm »
Subbed. I keep my eyes on you. The modularity is a must  :-+ .

Some SMU homework to read for inspiration, K2400:

https://www.eevblog.com/forum/testgear/keithley-2400-sourcemeter-review-and-teardown/
https://dev.xdevs.com/projects/kei2400/repository/raw/service/Keithley%202400%20Service%20Manual.pdf

Last week I was just talking with a Keysight Engineer regarding the N6700 SMU Module

1) The burden voltage for the current measurement is almost 0V
2) The current range switch very fast without losing any data, AKA "Glitch free" (it can analyze the power spike if a device goes out from sleep mode).

It seems there is a lot of demanding right now regarding a DC power supply which can assit the design of low power eg. IoT devices.

I would more than happy to see a module to have not more than 20W of power but can do something like (best case of course!):

Quote
Voltage, 20 V range: 0.025% + 1.8 mV
Voltage, 6 V range: 0.025% + 600 µV
Current, 3 A and 1 A ranges: 0.04% + 300 µA
Current, 10 mA range: 0.25% + 5 µA

Here also the feature to simulate a Li-Ion battery could be handy (it could be just software jazz.....)

I can only suggest to have in mind the low power design process.

PS: Also a 100W DC electric load module would be fantastic, with some feature to analyse battery discharge process.

Now back to work, I have something to finish up.
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 
The following users thanked this post: prasimix

Offline prasimix

  • Supporter
  • ****
  • Posts: 1236
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #8 on: July 15, 2019, 10:24:28 pm »
2) The current range switch very fast without losing any data, AKA "Glitch free" (it can analyze the power spike if a device goes out from sleep mode).

I'd like to learn how "very fast" is really fast: do we talk about something in a range of microseconds, tens, hundreds (or more) microseconds?
 

Offline zucca

  • Supporter
  • ****
  • Posts: 1847
  • Country: it
  • EE meid in Itali
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #9 on: July 15, 2019, 11:31:32 pm »
I'd like to learn how "very fast" is really fast: do we talk about something in a range of microseconds, tens, hundreds (or more) microseconds?

I asked the very same question. He told me something like:"The two current ranges are always on, you don't switch from one to the other, you just switch what you display"  :-//
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline exe

  • Supporter
  • ****
  • Posts: 1194
  • Country: nl
  • self-educated hobbyist
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #10 on: July 16, 2019, 12:15:34 am »
I asked the very same question. He told me something like:"The two current ranges are always on, you don't switch from one to the other, you just switch what you display"  :-//

To me it sounds like there are two power supplies in parallel, the lower range has slightly higher voltage. When a current spike occurs, the first power supply saturates and drops voltage (CC mode), then second starts working. But that's just a guess :)

Something like this: https://www.keysight.com/main/editorial.jspx?cc=US&lc=eng&ckey=520808&nid=-11143.0.00&id=520808 ("Learn to connect power supplies in parallel for higher current output").

UP: relevant video:
« Last Edit: July 16, 2019, 12:17:35 am by exe »
 
The following users thanked this post: thm_w

Offline prasimix

  • Supporter
  • ****
  • Posts: 1236
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #11 on: July 16, 2019, 12:55:00 am »
Hm, that makes more sense the using multiple resistor's based current sense :). If current sense resistors are used you have to switch in one moment of time from one to another, or in case of just two ranges you can bypass low-range (high resistance) shunt with MOSFET switch that is connected in series with hi-range shunt. Switching speed is finite, it could be in a range of microseconds or perhaps less, but still more then zero.

So, the existing H24005 and new BB3 (with two DCP405 power modules inserted into slot #1 and #2) can be under firmware control coupled in parallel (using internal power relay!) and it should be possible to add in firmware such "auto-ranging" mode of operation that user don't need to think a lot about settings, and proper measurement will be presented on the screen (or send to PC).
 
The following users thanked this post: exe

Offline 3roomlab

  • Frequent Contributor
  • **
  • Posts: 985
  • Country: 00
  • can you tell me how to cal sesame street meter?
Re: EEZ Bench Box 3 - Sequel to EEZ H24005
« Reply #12 on: July 16, 2019, 03:05:27 am »
I merge my 2 posts into this 1

I have been quite fascinated at tek Keysight 6705, the ability to scale 7 to 8 decades of current measurement is amazing
on some occasion of googling, I found this article (about range compression on op amp)
https://www.digikey.com/en/articles/techzone/2019/mar/the-fundamentals-of-logarithmic-amplifiers



Last week I was just talking with a Keysight Engineer regarding the N6700 SMU Module

1) The burden voltage for the current measurement is almost 0V

(edit some corrections)
the way keysight "hacked" this problem is to operate like a powersupply (see Andreas Spiess #245  video). the usual voltage feedback CHEATING  :-DD
but to span many decades, the log amp problem remains. I didnt understand the log amp concept until recently.

in the usual shunt, the way is to shift shunts around. shorting parrallel and series switching trick. every switching = 1 glitch. I think if the MCU is clever, it could blank out the glitch and present a current graph with some missing samples

but the always on method is to have many paths of current in different magnitudes

the attach very crude LTspice diagram shows the typical linear PSU but in 2 parts/magnitude A+B each trying to supply a target voltage

opamp feedback drives 2 JFET (or BJT), D is the sink part (to make a half H bridge supply). C is a variable load drawing current thru R15 @ 5.5v. point R0 is 5.5v ref for o/p drive at 5.5v.
A has a sense resistor 500k, it supplies up to around 6uA before it saturates [plot v(s01) dips too low] and B tries to follow up.
at 500k 6uA, the sense returns 0.5V/uA or 0.5mV/nA
B has a sense of 777 (I just put this rubbish value), it goes up to around 1mA before giving up and requires another section with higher current.
in B, the sense return is 0.777V/mA (0.777mV/uA).

at the red netpoint R11 is where both AB tries to combine and compete to supply 5.5v
we could read 6 decades from 1 nA to 1mA (in theory of cos), by playing with 2 ADC with 10-12bit detail
(the simulation is just a smash mashed messy thing, the +v rail is +9 and the -ve rail is -1  :-DD pls ignore the mess)

not sure how this pans out in real life, but "log op amp" appears to be not a new concept
(I trust that anyone could copy the pic and redo this all with LT models, too much non LT models creates many errors)

so in theory again, I assume with the right design, one could use a 10meg sense in one of those decade channels to sense lower than 1nA? but the ADC biasing will need crazy board resistivity layouts
the noise isolation of the top level supply will need to be so quiet, maybe the whole thing can only run on the LTC30xx series. maybe you need to use batteries as a "buffer capacitor" = infinite farads = nV noise floor

it would be an interesting project to watch if anybody "log amps current sense" to the level of keysight 6705, a $100 arduino logamp PSU 1nA to 10A capability?
(edit looks like someone did something https://www.eevblog.com/forum/testgear/iot-power-profiler-and-battery-life-estimation/msg2553039/#msg2553039 ??? )

*bonus pic of another log scale op amp LTC60xx, maybe there is space to insert a resistor to feed some voltage to ADC, but I think the magnitude width is too wide
keeping width to 2 decades seem to allow use of el cheapo 10bit ADC and arduino has them. it could marry worlds of scaling, cheap hardware and low fuss code fiddling
« Last Edit: July 18, 2019, 06:37:32 am by 3roomlab »
true-volt true-rms true-ohm true-lumen true-ampere true-decibel ... what else was false last year?
 
The following users thanked this post: zucca, prasimix, exe

Offline zucca

  • Supporter
  • ****
  • Posts: 1847
  • Country: it
  • EE meid in Itali
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf