Author Topic: EEZ Bench Box 3 (BB3)  (Read 151300 times)

0 Members and 1 Guest are viewing this topic.

Offline jbb

  • Super Contributor
  • ***
  • Posts: 1128
  • Country: nz
Re: EEZ Bench Box 3 (BB3)
« Reply #125 on: July 19, 2019, 02:34:37 am »
Wine bottles are quite long... maybe a can of beer ;)?
 

Offline mrpackethead

  • Super Contributor
  • ***
  • Posts: 2845
  • Country: nz
  • D Size Cell
Re: EEZ Bench Box 3 (BB3)
« Reply #126 on: July 19, 2019, 07:22:52 am »
I'm guessing it would be possible to create custom modules if you so desired given its modular form factor.

Something i'd like to do, would be a POE PSE port, as i do a lot of Powered devices.  It would be handy to be able to measure/control the PSE end.

Ok, but I don't known anything about PSE port. Do you have any useful links to some intro/beginners content?

I am not sure why PoE needs to be integrated into power supply because there are different kinds of power PoE sourcing (standards) and to cover them all, provide all features needed for testing of PoE devices, module needs to be quite complex thus expensive. Those who are ready pay significant money most likely will buy something from SiFos (or similar) anyway.

If you are going to implement PoE module, more or less it shall contain main components of this board:
http://www.ti.com/tool/tps2372-4evm-006

[edit] If it is about popular < 30W Type2 PoE, then this much simpler board a module: http://www.ti.com/tool/tida-00465


you may not see the need for it, and i may not see the need for other types of modules to do various differnet jobs.   Thats not really the point.  The point is that if the interface to be able to plug your module into is stadnard, and its easy enough to add code to it, then it makes it increadibly flexible for people to create things to their own needs.   of course there will be things that will be common for a lot of people, but tehre will be lots of odd ball things that people would make for themself.
On a quest to find increasingly complicated ways to blink things
 
The following users thanked this post: prasimix

Offline Rerouter

  • Super Contributor
  • ***
  • Posts: 4694
  • Country: au
  • Question Everything... Except This Statement
Re: EEZ Bench Box 3 (BB3)
« Reply #127 on: July 19, 2019, 07:54:34 am »
Also there is a lot in 1 given power supply module that can be mostly reused  for other purposes. Even for my SMU thoughts, most of it will be just giving it a bidirectional output stage and current shunt. apart from the level of ideal output stability its just set current with voltage compliance limit or set voltage with current source limit. the smarts come from how you use that, the feedback chain is still analog, the sampling is technically independent,

As such I'm beginning by trying to make a more flexible ADC, then once that is covered, figure out how the output stage could get away with the current tracking approach to keep the output stage regulating with very low dissipation.

The keysight level SMU's seem to have a step response bandwidth close to 50KHz, so nothing too extreme,

Technically with a multi-slope ADC like I'm trying to help work out, you could have as many analog inputs as you can mux in, so having say 6 external ADC inputs would not be expensive to add to things, just slow down how many data points you can pull off a second for a given resolution.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: EEZ Bench Box 3 (BB3)
« Reply #128 on: July 19, 2019, 05:11:30 pm »
Thats not really the point.

That was exactly my point - that this particular function (especially 1-channel), can live it's own life w/o being irreversibly married with power supply.

Quote
The point is that if the interface to be able to plug your module into is stadnard, and its easy enough to add code to it

Yes, good point. Nobody can forbid you to make module *you* want to make and use. Perhaps I missed something - are there any obstacles that are not allowing you to make PoE module or what?
 

Offline mrpackethead

  • Super Contributor
  • ***
  • Posts: 2845
  • Country: nz
  • D Size Cell
Re: EEZ Bench Box 3 (BB3)
« Reply #129 on: July 19, 2019, 09:36:07 pm »
That was exactly my point - that this particular function (especially 1-channel), can live it's own life w/o being irreversibly married with power supply.
Quote

90% of what you need to provide the functionality of a PSE, ( that has all the measurement / control that you want ) is already there.  'Marrying' it with the unit on your desk, is not a problem at all.



Quote
The point is that if the interface to be able to plug your module into is stadnard, and its easy enough to add code to it

Quote
Yes, good point. Nobody can forbid you to make module *you* want to make and use. Perhaps I missed something - are there any obstacles that are not allowing you to make PoE module or what?

I was'nt asking any one to make it for me. It was an example. I was enquiring as the open-ness of the interface that it will connect to. 
There will be no end of special use-cases that people want to build for themselves.




On a quest to find increasingly complicated ways to blink things
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: EEZ Bench Box 3 (BB3)
« Reply #130 on: July 20, 2019, 10:36:33 am »
90% of what you need to provide the functionality of a PSE, ( that has all the measurement / control that you want ) is already there.  'Marrying' it with the unit on your desk, is not a problem at all.

If all what's needed is functionality of PSE, then yes I agree. On the other hand many tests are better done using standalone adapter. Such as lightning surge immunity (if apply), Ethernet signal integrity, looking for transformer core saturation current - all will benefit from standalone PSE board/adapter. If adapter for testing is needed anyway, for me there is no point to have PoE-enabled supply. Disclaimer: I do not require for everybody to agree my point of view obviously [lol].
 

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
DCM220 latest prototype
« Reply #131 on: August 17, 2019, 06:07:19 pm »
I've just finished the latest DCM220 dual power module. A brief info for newcomers: it is built around LT3763 sync buck that offers current programming and monitoring. Unfortunately it does not provide CC mode of operation indication. Therefore I derived it from FB signal. A 4-channel 12-bit DAC (MAX5715) is used for set output voltage and current (times two for two channels). Instead of using dedicated ADC for measuring output voltage and current a STM32F373 MCU is used which comes with attractive multichannel 16-bit ADC! Output terminals are for 4 mm banana plugs and bi-color LED indicator is used to indicate OE (Output Enable) and CC mode of operation. Output voltage is limited to 20 V (but can be much more since Vin is 48 Vdc) and output current to approx. 4 A. Power outputs are "floated" in regard to other module's GND, but two channels share the same GND on board.



Output ripple is pretty good, if we take into account that is SMPS, and e.g. for Iout = 2 A it's below 5 mVpp or 1 mVrms (measured on the output capacitor):



So, with three DCM220 one could count on up to 6 programmable power sources packed into EEZ BB3.

Two yellow wires are correction after so far unsuccessful attempt to use on-board bias supply (+7.5 V) to power LT3763 instead of full +48 Vdc used for powering this module. Idea was to reduce dissipation of its internal regulator, but it seems that in that case output voltage cannot goes beyond voltage that is applied on its Vin (i.e. 7.5 V), at least not without damage (I've lost two of them together with few MOSFETs).

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: EEZ Bench Box 3 (BB3)
« Reply #132 on: August 17, 2019, 08:43:14 pm »
Very nice looking module. Congrats! You did not populate RC snubbers (R31+C52, R54+C75). Why? Do you have switching waveform with/without showing difference or lack of it?
 
The following users thanked this post: prasimix

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 (BB3)
« Reply #133 on: August 17, 2019, 09:24:21 pm »
Actually didn't finished all testing. There is also place for diode in parallel for lo-side MOSFET assistance. I was pretty satisfied with existing results, and something that really bother me is that inability to power the LT3763 with lower voltage (e.g. something that I find possible on many TI/LM controllers).

Offline jbb

  • Super Contributor
  • ***
  • Posts: 1128
  • Country: nz
Re: EEZ Bench Box 3 (BB3)
« Reply #134 on: August 17, 2019, 10:17:01 pm »
Maybe a zener down from your 7.5V rail to the auxiliary Vin pin? (I think those are often around 5V for LT chips.)
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: EEZ Bench Box 3 (BB3)
« Reply #135 on: August 17, 2019, 10:20:49 pm »
Actually didn't finished all testing.

Sure. Testing alone can take lot of effort and time. BTW would be nice to see load regulation step response, like 0.1A -> 2A -> 0.1A

Quote
There is also place for diode in parallel for lo-side MOSFET assistance. I was pretty satisfied with existing results

Hard to see at this time/div, but it seems that there is some ringing. If you have time and option to test with/without snubber - why not.

Quote
something that really bother me is that inability to power the LT3763 with lower voltage (e.g. something that I find possible on many TI/LM controllers).

LT3763 is LED driver controller - it may have limitations & design compromises compared to "proper" power supply IC's. I am afraid that block diagram is not telling whole picture and you can't do anything about this "problem".
 

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 (BB3)
« Reply #136 on: August 18, 2019, 06:05:24 am »
Here is transient response with compensation as it is (check schematics on GitHub, Pg. 3).
A Jim William AN104 transient load tester is used, and step response is 1 A.



Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 (BB3)
« Reply #137 on: August 18, 2019, 06:13:38 am »
Maybe a zener down from your 7.5V rail to the auxiliary Vin pin? (I think those are often around 5V for LT chips.)

Thanks, but I'm not really sure what you mean, and don't know if Vin is "auxiliary" to LT3763 or not. Please find below correction that I had to made that LT3763 behave properly again:


Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: EEZ Bench Box 3 (BB3)
« Reply #138 on: August 18, 2019, 06:41:54 am »
I see that you tied IVINP and IVINN (inputs of amplifier) directly to power rail. Could be reason why you lost IC while you powered it with low voltage, thus creating big difference between IVINP/N inputs and VIN. I know this is speculation - because block diagram does not show how particular amplifier is powered, nor say - it is "over the top input" type or not. [edit] Better idea would be to tie them to VIN pin, through resistor like 1K or so. [edit1] There could be similar problem with output current sense amplifier as well.
« Last Edit: August 18, 2019, 06:45:30 am by ogden »
 
The following users thanked this post: prasimix

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 (BB3)
« Reply #139 on: August 18, 2019, 06:45:28 am »
Hm, that's a good point! Perhaps they should be also tied now to bias power just as Vin. I have to order a few more LT3763 for further testing.

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: EEZ Bench Box 3 (BB3)
« Reply #140 on: August 18, 2019, 06:49:43 am »
For testing purposes I would replace output current sense series R51 & R52 with bigger resistors, like mentioned 1k. It may cause sense error, but could protect amplifier and let you do some testing and tinkering.
 

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 (BB3)
« Reply #141 on: August 18, 2019, 07:14:16 am »
I'll definitely spend some time to explore this possibility. It would be a shame not to use lower Vin, if possible, when we have on disposal that +7.5 V.
 
The following users thanked this post: ogden

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
LT3763 Vin modeling
« Reply #142 on: August 19, 2019, 08:00:47 pm »
Hm, I don't know why I didn't try it before, but here it is. The LTspice model (in attachment) is clearly shows what I've found in practice (with great difference that nothing is damaged). The voltage applied on the Vin pin is set to 8 and 48 V, and with former output voltage cannot reach set value:



 
The following users thanked this post: ogden, AlanS

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 (BB3)
« Reply #143 on: August 21, 2019, 06:02:53 am »
I'll try the different approach here and that is Vout "tracking". LT3763 Vin,min is 6 V therefore we need offset of at least 6 V, i.e. for Vout, Vin has to be 6 V or higher. With tracking a considerable part of total dissipation is moved to external BJT that can be easier to cool. Number of extra parts are minimal and simulation results are promising. Now I have to check that in practice.



Simulation above give us around 250 mW of dissipation for Vout of 5.5 V. That is almost 4 times lower then in case that +48 V is applied to Vin pin.
 
The following users thanked this post: ogden

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
DCM220 Vin hack
« Reply #144 on: August 24, 2019, 08:15:49 am »
I've tested previously mentioned proposal to decrease Vin voltage (and consequently LT3763 power dissipation) and it looks promising. If +48 Vdc is applied to Vin input with Vout=10.5 V and Iout=1.2 A the LT3763 easily goes over 100 oC in a matter of minute (I didn't try to push it to the thermal shutdown):



After that I've added simple regulator as in picture below:



The LT3763's power dissipation drops significantly measured over many minutes (and without forced cooling) and even with higher output current (2 A). On the picture is also visible that temperature of the MOSFET is higher, but in SOT-223 it will be much easier to cool it then LT3763.
In summary, I think that simple hack is a more cheaper then switching from 2- to 4-layer PCB to eventually provide better cooling of LT3763.



EDIT: check important notice in #152
« Last Edit: September 07, 2019, 03:16:46 pm by prasimix »
 
The following users thanked this post: ogden, AlanS

Offline jbb

  • Super Contributor
  • ***
  • Posts: 1128
  • Country: nz
Re: EEZ Bench Box 3 (BB3)
« Reply #145 on: August 24, 2019, 10:30:18 pm »
I'm following along here, and I thoroughly recommend getting some of that linear dissipation out of the driver chip.  But I have a couple of questions:
  • Is having IVINP and ININN greater than VIN OK?
  • What happens when the output voltage is at a maximum?  It looks like you'll still have considerable dissipation across the internal regulator.

It's a shame that you can't feed 5V straight into INTVCC, but the data sheet says "Do not force any voltage on this pin."

If the situation is bad with high output voltage, can you measure the DC supply current into Vin?  I wonder how much of it is gate drive current, which you might be able to get around using an external gate drive supply and BJT buffer pairs to isolate the gate drive power from the on-chip regulator. Or just use MOSFETs with smaller Qg...

 
 

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 (BB3)
« Reply #146 on: August 25, 2019, 06:00:49 am »
Is having IVINP and ININN greater than VIN OK?

Yes, that question still remains, it seems that leaving them tied to +48 Vdc does affect regular operations, but there is still two more options: to connect them together with Vin pin or to leave them connected together but floating. I've asked that question on Analog forum and maybe someone reply.

What happens when the output voltage is at a maximum?  It looks like you'll still have considerable dissipation across the internal regulator.

Obviously that produce max. dissipation, but I'll test that and see if it is still acceptable or not (testing mentioned above was for the middle of the range).

If the situation is bad with high output voltage, can you measure the DC supply current into Vin?  I wonder how much of it is gate drive current, which you might be able to get around using an external gate drive supply and BJT buffer pairs to isolate the gate drive power from the on-chip regulator.

I believe I could and that will be something in line with simulation (i.e. about 20 mA).

Or just use MOSFETs with smaller Qg...

Yup, it still better to move dissipation to MOSFETs (with smaller Qg and consequently higher Rds,on) then to have it on the controller. I'll check what I have on the stock, or I can order something new. If you have any good candidate to suggest that is in DPAK package, let me know.

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 (BB3)
« Reply #147 on: August 25, 2019, 07:20:43 am »
Hm, IRFR7546 really doesn't seems light with its 58 nC (typ) of total gate charge. I'm definitely give chance to some others like NTD5867NLT4G, IPD400N06NG or FDD5612.

I've tried with max. Vout (i.e. 20 V) and for Iout of 2 A and little assistance of cooling fan (which is a part of the deal, together with onboard NTC for measuring temperature) the LT3763 temperature is around 60 oC, that is not bad at all:


Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: EEZ Bench Box 3 (BB3)
« Reply #148 on: August 28, 2019, 02:14:00 pm »
This starts to looks better and better. I got today NTD5867NLT4G and replace initially selected IRFR7546TRPBF MOSFETs. Former is twice as much cheaper has Total Qg of only 15 nC (typ.), but also very low QRR (Reverse Recovery charge) of only 12 nC. The results is obvious, now without fan assistance, for Vout=20 V, and Iout=2 A I have the same results as previous with fan cooling:



... with fan cooling its go below 50 oC. Oh, and currently the ambient temperature is 30 oC!

« Last Edit: August 28, 2019, 02:15:34 pm by prasimix »
 

Offline jbb

  • Super Contributor
  • ***
  • Posts: 1128
  • Country: nz
Re: EEZ Bench Box 3 (BB3)
« Reply #149 on: August 29, 2019, 12:39:43 am »
Now we’re cooking! (Or, rather, not cooking.) Looks like you got a temperature rise of of 32 K down to 18K, which is a big improvement.

Perhaps a quick test at minimum output voltage would be in order to check the temperature rise of your linear dropper MOSFET?

It looks like the MOSFETs run cooler too. It’s easy to get caught up in the on resistance / conduction losses and lose sight of the switching losses.

I would like to clarify one thing: given a maximum output of 20V (which will be totally fine for many applications), why are you using a 48V input rail? Is it for common parts? With a 24V input I think you can get lower Qg MOSFETs again and drive the control chip directly from 24V.
 
The following users thanked this post: AlanS


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf