Author Topic: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)  (Read 345007 times)

0 Members and 1 Guest are viewing this topic.

Offline VEGETA

  • Super Contributor
  • ***
  • Posts: 1437
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #475 on: November 16, 2016, 06:10:43 pm »
Well, 40mV is like 0v to me with no problems.

What are the issues with dynamic loads? you mean when load changes a lot?

let's say I put it to 2v... will that require a heatsink? 2v*2A (max for me) = 4W which is not small but is heatsink a must here? The housing I am going to use doesn't have an opening to mount heatsink and make it outside the box to send heat to the outside air. And a heatsink that is inside is no heatsink at all xD right? Not to mention that I will go all surface mount.

Offline prasimix

  • Supporter
  • ****
  • Posts: 1853
  • Country: hr
    • EEZ
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #476 on: November 16, 2016, 10:44:37 pm »
What are the issues with dynamic loads? you mean when load changes a lot?

Yes.

let's say I put it to 2v... will that require a heatsink? 2v*2A (max for me) = 4W which is not small but is heatsink a must here? The housing I am going to use doesn't have an opening to mount heatsink and make it outside the box to send heat to the outside air. And a heatsink that is inside is no heatsink at all xD right? Not to mention that I will go all surface mount.

4 W is big enough to deserve some kind of heatsink. But that is not all. You have to take into account dissipation on the switching IC, power mosfet, diode and inductor. Don't mount switching parts on the same heatsink with linear (post-regulator) parts otherwise you can expect increased ripple on the output that is coupled over heatsink. But if that is inevitable use silicon thermal pad beneath PCB sections of SMPS pre-regulator and main cooling surface (some kind heatsink or metal enclosure plate).

Offline VEGETA

  • Super Contributor
  • ***
  • Posts: 1437
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #477 on: November 17, 2016, 11:59:03 am »
Ok, so how about using this: http://www.digikey.com/product-detail/en/aavid-thermalloy/530614B00000G/HS278-ND/373766  ??

Put it for the Linear regulator (MOSFET) and use another one for the switching regulator too, will this be good enough without a fan or any air holes? putting the 2 of them on the same heatsink will short them for sure.

Maybe I will pick a package that has an exposed pad from the bottom which can be soldered on the board, then I will expand the copper for it so it can dissipate in the PCB itself. Connecting it to a small heatsink is going to be good enough I guess.

Finally, where did you get your enclosures custom drilled and made? it will be so expensive and that is why I chose a nice ready enclosure (plastic = can not be used as heatsink).

I guess a drop voltage of 1v is good enough, you think?

Offline prasimix

  • Supporter
  • ****
  • Posts: 1853
  • Country: hr
    • EEZ
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #478 on: November 17, 2016, 12:23:19 pm »
For the beginning I don't know what circuit you are going to build. Is it the Power board that is part of this project or something else?
With heatsink that you suggested you'll get over 85°C for dissipated 4 W on it's surface without forced air cooling since it's thermal resistance is high (16.7°C/W). Doesn't sounds good, but if it's connected with some other metal surface or additionally cooled with forced air that could be something completely different.

Metal enclosure used here is custom made by Varisom in Portugal.

Offline VEGETA

  • Super Contributor
  • ***
  • Posts: 1437
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #479 on: November 18, 2016, 01:57:45 am »
For the beginning I don't know what circuit you are going to build. Is it the Power board that is part of this project or something else?
With heatsink that you suggested you'll get over 85°C for dissipated 4 W on it's surface without forced air cooling since it's thermal resistance is high (16.7°C/W). Doesn't sounds good, but if it's connected with some other metal surface or additionally cooled with forced air that could be something completely different.

Metal enclosure used here is custom made by Varisom in Portugal.

Well, I want to make a linear lab supply with 2 li-ion batteries inside it... so it is like what you did here, only smaller and more specific to my vision. I have a plastic enclosure and all my stuff will be surface mount, including the linear and switching regulator themselves. What do you suggest for a heatsink or some other similar solution?

I didn't suggest that particular heatsink but I asked you if this is the only affordable and suitable solution for heatsinking smd parts. putting a fan will make it so noisy and maybe heavier than I wanted.... or not?

Offline prasimix

  • Supporter
  • ****
  • Posts: 1853
  • Country: hr
    • EEZ
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #480 on: November 18, 2016, 07:58:14 am »
If you are going to power it with batteries then fan doesn't look like a good option since it will shorter battery life to some extent. You are also not to going use a metal enclosure, that makes cooling even more challenging otherwise you could do something that Dave presented in video #744 (another video on this topic is #105). Maybe you will find a solution with piece of Aluminum plate and silicon thermal tab in-between PCB and plate.

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 8493
  • Country: de
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #481 on: November 18, 2016, 01:32:06 pm »
With battery operation one could go a little lower in the drop for the linear stage, especially if the pre-regulator is relatively fast. For a mains operated version efficiency is not that critical.
For cooling there would still be the option to have a fan, but only operate it if the temperature of the heat sink is high. Even a small fan can make a lot of difference.
 

Offline VEGETA

  • Super Contributor
  • ***
  • Posts: 1437
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #482 on: November 18, 2016, 09:14:27 pm »
I want something that is a standard, not a one-time hack. As I plan for the future to sell these units if I am lucky enough to make it worth buying xD. Anyway, I think I can make it 1v linear difference which will make a maximum power dissipation of around 2W.

Offline VEGETA

  • Super Contributor
  • ***
  • Posts: 1437
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #483 on: November 19, 2016, 12:55:54 am »
I read about using a MOSFET (N type) and there seems to be a disadvantage of using it over the darlington npn bjt, which is the dropout voltage. The gate must be higher that the input for this to work which won't be good for a 1v dropout right? I read that your mosfet has a Vsd (diode forward) of 1.3v, is that related to what I said?

I know that if drop voltage is made to be 1v then a max current is 2A then max power dissipation will be 2W which is just great. But what about Rdson parameter here?

So maybe the core of my question is what determines the minimum achievable dropout for a certain mosfet? is it the amount of Rdson or some other parameter?

Offline prasimix

  • Supporter
  • ****
  • Posts: 1853
  • Country: hr
    • EEZ
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #484 on: November 19, 2016, 08:37:40 am »
If you are going to build portable power supply with battery source then I don't know if post-regulator based on NMOS is good choice. It's gate voltage and input voltage (on the drain) cannot be the same. Actually if it's the same then max. output voltage will be for 4-5 V lower then input because that is how NMOS works. So if you make e.g. boost pre-regulator that provide 1-16 V for 0-15 V on the output (your 1 V difference) then you need to generate in parallel gate bias that is 21 V or higher or "floating" +5 V or higher that will added on top of pre-regulator output voltage. Otherwise for 10 V on the post-regulator input (drain) you'll get 6 V or lower (depends of load) on the output, and yes that difference has to be dissipated on the NMOS.
In my case I don't have that problem, I have on the input +48 V that is provided to Q7 (DC_OUT in schematic) that give me about +6 V difference in the worst case (when max voltage is programmed).

Question is why you are selected this design for your project? If that is for analog part or in the first place for the digital control? If digital control is attractive for you, you can still use it to work with completely different analog/linear part of the solution.

Offline technix

  • Super Contributor
  • ***
  • Posts: 3336
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #485 on: November 19, 2016, 05:49:23 pm »
If you are going to build portable power supply with battery source then I don't know if post-regulator based on NMOS is good choice. It's gate voltage and input voltage (on the drain) cannot be the same. Actually if it's the same then max. output voltage will be for 4-5 V lower then input because that is how NMOS works. So if you make e.g. boost pre-regulator that provide 1-16 V for 0-15 V on the output (your 1 V difference) then you need to generate in parallel gate bias that is 21 V or higher or "floating" +5 V or higher that will added on top of pre-regulator output voltage. Otherwise for 10 V on the post-regulator input (drain) you'll get 6 V or lower (depends of load) on the output, and yes that difference has to be dissipated on the NMOS.
In my case I don't have that problem, I have on the input +48 V that is provided to Q7 (DC_OUT in schematic) that give me about +6 V difference in the worst case (when max voltage is programmed).

Question is why you are selected this design for your project? If that is for analog part or in the first place for the digital control? If digital control is attractive for you, you can still use it to work with completely different analog/linear part of the solution.

You can use a separate gate drive rail, either at a fixed higher voltage, or using a tracking regulator to give you a few volts above output.
 

Offline VEGETA

  • Super Contributor
  • ***
  • Posts: 1437
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #486 on: November 19, 2016, 09:46:43 pm »
If you are going to build portable power supply with battery source then I don't know if post-regulator based on NMOS is good choice. It's gate voltage and input voltage (on the drain) cannot be the same. Actually if it's the same then max. output voltage will be for 4-5 V lower then input because that is how NMOS works. So if you make e.g. boost pre-regulator that provide 1-16 V for 0-15 V on the output (your 1 V difference) then you need to generate in parallel gate bias that is 21 V or higher or "floating" +5 V or higher that will added on top of pre-regulator output voltage. Otherwise for 10 V on the post-regulator input (drain) you'll get 6 V or lower (depends of load) on the output, and yes that difference has to be dissipated on the NMOS.
In my case I don't have that problem, I have on the input +48 V that is provided to Q7 (DC_OUT in schematic) that give me about +6 V difference in the worst case (when max voltage is programmed).

Question is why you are selected this design for your project? If that is for analog part or in the first place for the digital control? If digital control is attractive for you, you can still use it to work with completely different analog/linear part of the solution.

You can use a separate gate drive rail, either at a fixed higher voltage, or using a tracking regulator to give you a few volts above output.

YES! this was my idea that I hoped someone will say it (So I won't look like an idiot xD). I thought of using a NMOSFET as a switch but the pre-regulator is a SEPIC one to be able to go above or below the 2 18650 batteries voltage (8.4v = nearly 8v) which will be the input of the mosfet linear regulator. This will be say 1v more than the required output and will be given using the microcontroller.

Now for the mosfet gate, I won't take it from the same input rail (the SEPIC one), but rather an external one. This one is a little boost converter making a 30v or so voltage to power op-amps and such stuff in the circuit, which is not that high current one so any cheapy will be enough here.

You said it has to be more than the mosfet input level, which is the case here as 30v will be more than the maximum of this power supply which is 20v. However, should it also be tracked? I mean will it generate significant losses and power dissipation?

Offline prasimix

  • Supporter
  • ****
  • Posts: 1853
  • Country: hr
    • EEZ
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #487 on: November 19, 2016, 10:23:18 pm »
You said it has to be more than the mosfet input level, which is the case here as 30v will be more than the maximum of this power supply which is 20v. However, should it also be tracked? I mean will it generate significant losses and power dissipation?

No, the gate bias don't need to be tracked.

Offline VEGETA

  • Super Contributor
  • ***
  • Posts: 1437
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #488 on: November 20, 2016, 06:20:21 am »
Ok so a 25v constant voltage is suitable (assuming the mosfet Vgate can handle it) for this, and by doing that I can achieve 1v dropout right? Now the rest is to get a circuit to activate the gate.

How about connecting the 25v to a transistor BJT type and connecting its other end to the gate of the mosfet, now the gate of the bjt is connected to voltage set op-amp (0-20v for 0-20 output)... I have this plan in mind, will it work?

Offline VEGETA

  • Super Contributor
  • ***
  • Posts: 1437
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #489 on: November 21, 2016, 12:00:28 am »
Ok so here is my trial:



it gives awesome results in LTSpice BUT WITHOUT LOAD!!

When I put loads from 1R to 10K it won't regulate at all! it outputs 4mv or something. Can you describe what did i miss? thanks!

Offline Strada916

  • Frequent Contributor
  • **
  • Posts: 252
  • Country: au
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #490 on: November 21, 2016, 12:06:45 am »
Vegeta: I'm all for seeking advice and all. I just think you should have your own thread about this...... Just saying.

Sent from my D6503 using Tapatalk

The Bone, the Off-White, the Ivory or the Beige?
 

Offline void_error

  • Frequent Contributor
  • **
  • Posts: 673
  • Country: ro
  • I can transistor...
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #491 on: November 21, 2016, 03:14:26 am »
I just think you should have your own thread about this...... Just saying.
Agreed. More people would be able to help then. Maybe start a thread in the beginners section of the forum?
Trust me, I'm NOT an engineer.
 

Offline VEGETA

  • Super Contributor
  • ***
  • Posts: 1437
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #492 on: November 21, 2016, 07:03:41 pm »
Ok, but I asked about the core things in this subject and did my best not to stray out. Because I wasn't active the time these circuits were discussed here. If I get the answers for the last questions above, I will start my own for future questions. xD

Offline prasimix

  • Supporter
  • ****
  • Posts: 1853
  • Country: hr
    • EEZ
3rd channel for digital circuits and USB charger
« Reply #493 on: November 22, 2016, 01:07:27 pm »
While firmware is progressing to support all hardware functionality I was thinking about some new features that can be added to the existing circuits. On of them is to add 3rd channel that could be used for supplying digital circuits with up to 5 V and 25 W. That low voltage channel could be used as "high power" USB charger that means up to 2 A of charging current with detection of plugged device to offer max. possible output current.

One of the challenge was to try to house it in already pretty packed enclosure. I found that is indeed possible if existing AUX PS module is upgraded with more capable AC/DC module and that Arduino Shield is expanded to carry 3rd channel based on sync buck for lower power dissipation (and easier cooling).
Mean Well has 30 W module IRM-30-12 that is compact and affordable (shown below):



Of course it is much bigger then existing 5 W module therefore a new PCB layout is required. Since AUX PS module is mounted directly on the rear panel with 5 screws we need to maintain the same positions of the mounting holes and new PCB should looks like this:







Now +12 V will be shared between +5 V buck for powering Arduino Shield and rest of 25 W will be delivered to the 3rd channel. For that I changed existing 10-pin IDC connector with bigger one (16-pin) that is pin compatible with previous design. Also I decided to select smaller and cheaper power inductor for the +5V buck and use separate AC output connectors for Ch1 and Ch2 AC/DC modules. Since IRM-30-12 is bigger then previous AC/DC module I also put pads for 5 W and 7.5 W modules.

I'll present the 3rd channel circuit in the next post.



Offline void_error

  • Frequent Contributor
  • **
  • Posts: 673
  • Country: ro
  • I can transistor...
Re: 3rd channel for digital circuits and USB charger
« Reply #494 on: November 22, 2016, 03:48:31 pm »
While firmware is progressing to support all hardware functionality I was thinking about some new features that can be added to the existing circuits. On of them is to add 3rd channel that could be used for supplying digital circuits with up to 5 V and 25 W. That low voltage channel could be used as "high power" USB charger that means up to 2 A of charging current with detection of plugged device to offer max. possible output current.

Neat! I was also thinking of adding an extra channel with an USB charging port to my own PSU which is still far from completed but mine wouldn't be isolated from mains earth. I'm curious how you'll do the plugged-in device detection. From my experience some devices increase their drawn current until the voltage drops to a predetermined level and they assume that's the maximum current they can draw.
Trust me, I'm NOT an engineer.
 

Offline prasimix

  • Supporter
  • ****
  • Posts: 1853
  • Country: hr
    • EEZ
Re: 3rd channel for digital circuits and USB charger
« Reply #495 on: November 22, 2016, 03:57:02 pm »
I'm curious how you'll do the plugged-in device detection. From my experience some devices increase their drawn current until the voltage drops to a predetermined level and they assume that's the maximum current they can draw.

Initially I planned to use few analog switches, DAC and ADC inputs of the MCU to accomplish that but then gave up and use standard solution for it such as TPS2511 that comes in MSOP8 package and detect devices such as Apple, Samsung and Sony for charging current of up to 2 A. MCU based solution is more flexible and "future proof" if someone introduce new level on D+, D- lines for charging purpose but I believe that chosen port controller will do the job for a while. Anyway I'm going to test that and let you know.

Offline prasimix

  • Supporter
  • ****
  • Posts: 1853
  • Country: hr
    • EEZ
Re: 3rd channel for digital circuits and USB charger
« Reply #496 on: November 23, 2016, 12:48:42 pm »


Here is proposed circuit for 3rd channel that can be also used as a USB charger. A SMPS controller LT3741 (IC35) is used that offer current limitation and is synchronous buck therefore overall thermal dissipation should be smaller. Selected mosfet (Q6, Q31) comes in interesting LFPAK package that is, pretty small, simple for soldering and have good thermal resistance.
Projected output voltage when calibrated should be 6 V and current up to 5 A on the CH3 binding posts or 2.6 A on the USB A socket while used as a USB charger. Max. power will be limited by software to 25 W.
For voltage and current control I first planned to use Arduino DUE DAC outputs but it's too limited (cannot without extra circuit set voltage down to 0 V) and I ended up with SPI 12-bit dual DAC MCP4822 that has internal voltage reference (2.048 V). It's possible to use the 10-bit version that is pin compatible (MCP4812).

Voltage programming is accomplished by injecting DAC output 2.048 to 0 V for 0 to 6.3 V via R185. For current programming another DAC output is used. Its 0 to 1.5 V limits output current in range from 0 to 5 A. Max. output current is defined with voltage drop on the Rsense that is 50 mV. Therefore R33 of R010 is selected.
For voltage monitoring MCU ADC input is used (AUX_U_MON). Voltage divider R187, R188 is set to provide almost full scale for max output voltage. For current monitoring (AUX_I_MON) INA282 (gain 50V/V) is used that gives 2.5 V for max output current (50 mV x 50). INA282 has zero-drift but only 10 kHz bandwidth that is not critical in our case since its used for passive monitoring and not as a part of the current control loop.

Possibly I will add dual RRIO op-amp (e.g. OPA2344 or similar) into the picture one for buffering voltage monitoring divider output (R187, R188) and another one to extend current shunt monitor output from 2.5 V to MCU's ADC full scale (3.3 V).

Switching frequency is defined by MCU and it will we about 640 kHz or two times higher then what is used for CH1 and CH2 power pre-regulators. Therefore a simple divider by 2 that also offer 180 deg phase shifting will be used:



For USB charging a separate USB A-type socket will be used. Only one output can be selected at the time: binding posts (J7, J8) or X22 connector (vertical, PCB mounted). K_USB relay MCU controlled is used for that purpose. Maybe it can be replaced with two power mosfets.
TPS2511 is used to detect connected USB device and provide up to 2.6 A of output current. Finally as an option is added IC38 for protection.

Adding CH3 require bigger Arduino shield and that is mainly dictated with dimensions and spacing of binding posts and USB-A socket. Additional 50 mm of horizontal space will be enough for that. Because of new AUX PS module and Arduino Shield positions of 4 PCBs on the PCB panel have to be rearranged for better usage:



If someone wants to play with LT3741 here is the LTspice model (see also attached file):



Your comments and ideas are welcome.

Offline Kean

  • Supporter
  • ****
  • Posts: 1186
  • Country: au
  • Embedded systems & IT consultant
    • Kean Electronics
Re: 3rd channel for digital circuits and USB charger
« Reply #497 on: November 23, 2016, 02:04:19 pm »
TPS2511 is used to detect connected USB device and provide up to 2.6 A of output current. Finally as an option is added IC38 for protection.

Nice work!  Very thorough.
A minor note - shouldn't IO2 of IC38 connect to USB DP signal?
 
The following users thanked this post: prasimix

Offline prasimix

  • Supporter
  • ****
  • Posts: 1853
  • Country: hr
    • EEZ
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #498 on: November 23, 2016, 02:05:59 pm »
Oops! That was mistake. I'll fix that. Thanks.

Offline void_error

  • Frequent Contributor
  • **
  • Posts: 673
  • Country: ro
  • I can transistor...
Re: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)
« Reply #499 on: November 24, 2016, 12:10:21 am »
Isn't the INA282 a bit overkill? I mean with 20uV offset it's really accurate but it's also quite pricey. I assume you wanted to share R33 from the LT3741.
Trust me, I'm NOT an engineer.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf