EEZ Bench Box 3 (BB3)

[krho]

I found that the USB 3 PD functionality can be provided using MCU from STM32F0 family for which ST offers USB-PD stack. I believe that could speed up a whole development of something that looks rather complex.

You probably mean G0, as this are the CPUs that have CC pins.. however they don't provide protection on those pins.. so you have to somehow get the voltage which can go up to 25V down to safe levels e.g somewhere between 3.3 and 5V.

[prasimix]

No, I did not. I'm referring on what ST specified. There is few possibilities with F0, but also more broader support for F0/F4 families. MCU is just a part of solution and seems that it replaces Type-Port Manager (TCPM) with hard-coded protocol. ST is mentioning that their solution is tested with ONsemi's FUSB307B Type-C Port Controller (TCPC). PD looks like a science for itself, but promise flexible bidirectional power distribution.

[krho]

Ah ok. Well they also provide the PD libraries for G0 and L5, but G0 is way cheaper as it's m0+

[jbb]

Hi prasimix

I’m a bit behind on the thread and just caught up to the range switching CC issue. When I looked at the schematics with the three slope current setting I thought “that looks complicated, there might be ‘bumps’ moving between ranges for CC mode.”  Then I saw your post where you discussed the issue.

I think the simplest solution is what you’ve done already: just have one CC range and call it a day.

 But there may be a useful compromise between simplicity and performance. How does the following sound?
- CC split into 2 ranges (3 is too complicated)
- coarse range is 45mA - 5A
- fine range is 0 - 50mA
- the 45mA vs 50mA is for hysteresis and fewer surprises.
- when using fine range, just force the range switching circuit to 50mA range, because you already know what the current range should be.
- for improved accuracy, you could use the signal out of the switched current sense amplifier for feedback (but you may need different current control gains for stability)

[prasimix]

Thanks for suggestion. Splitting CC in two ranges could include some guess work in operation, i.e. if load in sleep mode drawn few mA, and few hundreds of mA when wake up, then setting lower CC range will affect load operation. Anyway, I have to rethink auto-ranger design together with complete new proposed power module (DCP505).

[Rerouter]

It seems some other designs rely on overkill DACC resolution in combination with a log amplifier for the input current to make the current control work over the full range. I imagine mainly it would just make it harder to calibrate at the trade off only needing 1 range.

[jbb]

So I had a go at implementing a similar circuit in LTSpice.  Hopefully some of the concepts will be helpful.

Basic concepts:

• Have a 5A shunt resistor always in circuit.  As you pointed out a while ago, this makes it much easier to manage the range switching.  It's also used for Constant Current (CC) control.
• Use a separate amplifier for 500mA and 50mA range.  No need to measure the 5A range again.  Also allows the use of a different gain which can improve Vos drift sensitivity.
• Use simplest comparator chain to select gain.  Some hysteresis will be required, because switching the range FETs injects a bit of current into the system (due to gate charge).
• Separate CC ranges from ADC range.  We don't want automatic switching of CC range, because switching glitches could disturb the main control loop and jam pulses into delicate loads.  But we do want the current sensing to autorange.
• Provide 2 CC ranges with 'manual' switching.  This will avoid automatic switching bouncing around, because the main control system controls the CC range override and the CC DAC.

Note: I'd expect to use a quad analog mux chip of some description.

In terms of devices with widely varying current limits, there are 3 approaches we could take:

• The user just selects the CC range before turning the supply on.  If the 50mA CC range was selected, then the supply could only go to 50mA.  If the 5A CC range was selected then current limit steps would be coarse.  Note that current sense would still autorange.
• Basic CC range switching during operation.  This would probably cause some kind of bump in CC mode because the CC DAC and the ForceLow signal might not be perfectly synchronised.
• Seamless CC range switching during operation.  Add a second CC DAC&EA so that there is a 5A range CC DAC + EA and a low range CC DAC + EA.  Preset the DAC (and allow it to settle) before flipping between ranges.

[electromotive]

I just stumbled across this today. I like the wide input, wide output. I'm working on familiarizing myself with github and getting myself up to speed. I think you're really onto something great here.

[alex-sh]

I found that the USB 3 PD functionality can be provided using MCU from STM32F0 family for which ST offers USB-PD stack. I believe that could speed up a whole development of something that looks rather complex.

I have one of these PD Buddy boards with USB-C PD (5-20V 3A max).
I think it is based on STM chip. Please check it out:
https://www.tindie.com/products/clarahobbs/pd-buddy-sink/

[alex-sh]

@prasimix

Where are you going to print out your front panel please?
I did use Schaeffer AG in the past, but they are expensive.

Thanks
Alex

[prasimix]

@prasimix

Where are you going to print out your front panel please?
I did use Schaeffer AG in the past, but they are expensive.

Thanks
Alex

Yes, Schaeffer is very nice but expensive, if I remember correctly only front panel would cost almost as twice as much as complete custom made enclosure (painted and with printed front and rear) for EEZ H24005 made by Varisom. I've ordered first prototypes from Varisom again (see #68) that is going to paint and print out front and rear panels, too. I'll post some picture when it arrives.

[prasimix]

I found that the USB 3 PD functionality can be provided using MCU from STM32F0 family for which ST offers USB-PD stack. I believe that could speed up a whole development of something that looks rather complex.

I have one of these PD Buddy boards with USB-C PD (5-20V 3A max).
I think it is based on STM chip. Please check it out:
https://www.tindie.com/products/clarahobbs/pd-buddy-sink/

I'll try to build a capable buck stage with CV and CC control, first. Next step could be addition of STM based PD control. The buck should accept 48 Vdc on input and I'm looking to give a try to LT3763 controller. It seems that it has all what is needed even some extra stuff such as PWM dimming that could be useful for powering/testing power LEDs and laser diodes.

[blakme]

I came across this project while searching for open source Bench power-supply. I am interested in contributing to this work. I have background in FPGA/Digital HW and software design. Thanks! Lakshmi Aiyer

[prasimix]

Many thanks Lakshmi for your interest. Yes, FPGA is something that is definitely coming into picture (after I failed to make something with XMOS xCORE ). Since development team of ULX3S project (based on Lattice ECP5) is in Zagreb, Croatia and I met them in person, we'd like to use it as a core for some new modules. Perhaps you should take a look on it, and we can continue discussion via PM, email or discord.

[prasimix]

A small update regarding backplane. A new prototype now can host up to four power relays for combining output terminals on the first two slots (i.e. without external wiring) in the one of the following ways:



Power relays are now controlled with 4-port I2C I/O expander. The populated PCB (except LEDs for indication which coupling is active) looks like this:



Backplane pin mappings can be found in attached PDF.

[alex-sh]

You are really making it modular.
How many main modules do you intend having and do you foresee any add-on (optional) modules please?

[prasimix]

The number of modules could be up to 7 with current addressing of the shared I2C (#0 is reserved for MCU board) that is intended to be used for module identification and module specific info (e.g. calibration params, working hours counters, etc.).
This backplane (BP3C) obliviously can carry up to three peripheral modules, and its dimensions are defined with Bench Box 3 enclosure presented in #68. For larger module counts you'll need a bigger enclosure, and on the backplane you have to decide which of SPI channels you'd like to share since currently each of three slot has dedicated SPI channel which is not really necessary at least when we talk about power board functionality.
So far, I've made two power modules: DCP505 (0-50V/0-5A max. 250 W) that works in combination with CF-DIC, and DCP405 (0-40V/0-5A max. 155 W limited by Mean Well AC/DC module). I'm now playing another module which should offer two simplified, more "crude" DC outputs i.e. without linear post-regulator stage based on LT3763 controller as announced recently.

[prasimix]

Finally, after many months of waiting the first enclosure prototype arrived few days ago. Unfortunately with more issues that I expected, but all of them are really minor and related to the front panel. The biggest and obvious one is that hole for encoder's know is not positioned correctly







View inside the enclosure reveals that number of wires are minimum that will simplify assembly and servicing:



Area allocated to AC/DC modules are huge enough to host a toroidal transformer of respectable size:



Feet that I found in Turkey (Altinkaya) looks pretty convenient and give us possibility to position the enclosure with an angle that will improve viewing angle on the TFT touchscreen display:



I'd like to hear your opinion about carrying handle mounted on the top cover. Some of my friends ridiculed it on the first sight as something extremely ugly. I was put it there to avoid increasing enclosure width and because with it one can conveniently and safely carry BB3 around using only one hand (another question is how often that could be a case).

Finally, here is a short overview of the first prototype:



The project also got its web pages: https://www.envox.hr/eez/eez-bench-box-3/introduction.html

[Sparky49]

Your attention to detail is astounding as it is inspirational. Than you for your impeccably detailed posts, I've enjoyed reading the backlog over years across all your projects.

[alex-sh]

This looks very good and professional!
The handle - some people like it some people not. Personally I think it’s ok. Alternatively you can do a handle on the side, but this means the handle has to come off first before the cover?

[jbb]

It’s good to see this project coming along nicely! Things like enclosures always take a lot of work to get right.

I have some questions on the enclosure:
- what are the dimensions of the current prototype?
- which way does the air flow? Is the box normally at positive or negative pressure?
- with the carry handle in place, can you stack two units on top of each other?

[nimish]

Is this rack mountable?


Sent from my iPhone using Tapatalk

[prasimix]

It’s good to see this project coming along nicely! Things like enclosures always take a lot of work to get right.

I have some questions on the enclosure:
- what are the dimensions of the current prototype?

290 (W) x 123 (H) x 240 (D) mm without feet mounted. Feet are 15 mm high.
Talking about width, that is more then a half of 19" rack. If we'd like to reach a full width, additional four modules could be added (seven in total).

- which way does the air flow? Is the box normally at positive or negative pressure?

That's the good question, the fan is currently mounted to provide positive pressure, but I presume it could be also negative. As far as I know each option has its pro's and con's. What would you like to suggest?

- with the carry handle in place, can you stack two units on top of each other?

Yes, stacking is possible:

[prasimix]

Is this rack mountable?

It could be, but we should make some sort of mounting brackets (and few additional holes on the sides, too), and probably in that case feet have to be removed.

[nimish]

Hmm for a "pro" version something that could reuse a 4u server case and pcie style expansions to save costs would be cool...

Looks good for a compact half-width thing


Sent from my iPhone using Tapatalk