Author Topic: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)  (Read 285665 times)

0 Members and 3 Guests are viewing this topic.

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #50 on: March 07, 2015, 11:42:00 pm »
Had the similar problem with my power supply.
Solved it by using  the output set voltage to control the pre-regulator when the power supply output is switched off.

Are you using op-amp or transistor based tracking control? Can you share it with us?
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #51 on: March 07, 2015, 11:44:57 pm »
My idea is to leave TO220 device soldered on the PCB, drill i.e 4mm "thru-hole" on back panel and mounting them with M3 screw to heatsink

I'm using thick (6 mm) aluminium plate to mount transistors. The plate mounting to external heatsink through rear panel window. It's easy to disassemble - no need to unscrew every TO-220 case. And I can briefly turn on device even without external heatsink in debug state - plate is sufficient heatsink.

Looks excellent and quite practical! I also like this "debug" feature  :-+
 

Offline mij59

  • Frequent Contributor
  • **
  • Posts: 693
  • Country: nl
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #52 on: March 08, 2015, 12:39:48 am »
Had the similar problem with my power supply.
Solved it by using  the output set voltage to control the pre-regulator when the power supply output is switched off.

Are you using op-amp or transistor based tracking control? Can you share it with us?

My circuit differs a lot from yours.
The pre-regulator is derived from a SCR control, instead of SCR's I use a MOSFET.
I use a separate dac to control the pre-regulator. ( and output of the power supply )
The output voltage of the pre-regulator is the sum of the dac output and the output voltage of the power supply.

Currently building version 4, I have attached version 3 of the power supply.
 

Offline Liv

  • Regular Contributor
  • *
  • Posts: 133
  • Country: by
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #53 on: March 08, 2015, 09:41:23 am »
The pre-regulator is derived from a SCR control, instead of SCR's I use a MOSFET.

Similar pre-regulator used in Agilent U8002A circuit.
PSU PSL-3604 Pulse gen. PG-872 Freq. cnt. FC-510
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
New post-regulator à la Liv
« Reply #54 on: March 22, 2015, 07:02:47 am »
Many things have happened since my last post. I continued my quest for workable tracking control and tried one that is e.g. used in LTC’s DC2132A. It’s interesting that they have two schematic revision where tracking control section is updated. Both of them contains RC filter without values. Don’t know why possibly because a whole thing is not so exact and require some experimenting. I tried that circuit and can say that it works fine up to ~25V and after that it starts to produce instability in SMPS pre-regulator especially after 30V which makes this solution unacceptable for my project. I tried to isolate PNP from FB pin using regular optocoupler (4N35), it works a little bit better but still not acceptable and finally when I used LDR coupled with LED it works nice up to 50V. So lets say that this part of the story is closed.
After that I decide to test Liv’s design which he is using in all of PSU’s what he built and presented here and here. Finally I started to use spice software, first Simetrix (thanks void_error) but shortly came to free version limitation :(. That’s pity since this software works really nice and I started to use it almost instantly. Anyway simple CV control loop model for Simetrix is enclosed in attachment. Next station was Ltspice. Thanks to massive and excellent assistance from Liv I ended up with model which is also enclosed. Actually three of them. First is with LTC2057 as a current shunt monitor which I selected since I have few of them on disposal for breadboarding.



Second model is derived from previous one as an example of CV control loop only in case that you have the same problem that I noticed – simulation lasts possibly forever (I left it overnight but without success). It’s possibly a problem with my environment since I’m running LTspice on Ubuntu with Wine. Due to that I also send in attachment another model with AD8610 used as current shunt monitor. I currently have no plan to test AD8610 since it doesn’t look attractive due to its high price. Final candidate for this function is still not selected, but there is no need to stick with only one since there is many nice op-amps that can be used. Here is the table with some of them that I found interesting:



Any your inputs is highly welcomed. In the final step when a new PCB has to be made I need to decide which package to use SO8 or MSOP8. Again, your inputs is welcomed.

« Last Edit: March 22, 2015, 10:20:55 pm by prasimix »
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Downprogrammer...
« Reply #55 on: March 22, 2015, 07:14:25 am »
New post-regulator include so-called down-programmer. I found this feature (check i.e. this and this Agilent documents) useful and decide to definitely test it and eventually add to the next PCB revision. Difference cannot be noticed with huge load (small impedance) but it’s clearly visible with smaller load or no load. Please find below two screenshots from simulations with down-programmer (DP) switched off and on:




Here is how it looks like when Vout is controlled by sequence 5-7.5-10-7.5V sent using DAC. Notice difference in fall time:



« Last Edit: March 22, 2015, 07:15:56 am by prasimix »
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Rise time...
« Reply #56 on: March 22, 2015, 07:23:07 am »
LTspice model presented recently will produce for some reason not so attractive rise time for U_SET step function (tested with PWL(0 0 110u 0 120u 1 2.78m 1 2.8m 50m 3m 50m):



In real life it looks much better (115us!):



« Last Edit: March 22, 2015, 07:30:21 am by prasimix »
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Power board redesign...
« Reply #57 on: March 28, 2015, 11:46:39 pm »
From some previous posts became obvious that Power board which is hosting pre-regulator, post-regulator, bias powers and ADC/DAC sections will require massive redesign. New post-regulator is tested on breadboard and looks very promising. Also I made some testing with new power pre-regulator based on LM5088 device which makes a whole thing cheaper. Bias pre-regulator based on LM5574 works nice and will stay but for bias post-regulators I'd like to use more affordable items such as LM317, LM337 and LP2951.
I also starts to think that maybe splitting this board in two section will simplify a whole assembly and testing and make it a more flexible for the future modification when i.e. a modification on the pre-regulator side will not require to thrown away the post-regulator section.
This approach rise a new question: how to interconnect that two section that was previously reside on the same PCB? My first choice is to use board-to-board connectors such as Samtec HTSW-105-08-F-D-RA (Farnell: 1926980) and SSW-105-02-T-D-RA (Farnell: 2308470). A multiple pins will be used for power out and ground. According to Samtec power rating is at least 5.7A per pin. Actually I don't understand what does it mean: 1 pin powered per row in datasheet. Anyway I hope that four of them for power out will be more then enough for continuous 3A (or even 5A).
I'd like to ask you if you can suggest some better but still "down to Earth" solution.
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
New model with improved OE circuit...
« Reply #58 on: April 06, 2015, 06:53:55 pm »
The model presented in post #55 include Output enable circuit which is not optimal. It has two issues: when it is active it drops Vout below zero (up to ~-0.45V) and when power is going down with OE activated some ugly artifacts appears on the output. Here is the measurement of the extreme case when no load is connected:



New model incorporate current mirror for controlling power mosfet bias and it resolve successfully before mentioned issues. Here is the schematics and .asc file is in the attachment. Note that if you have a problem with long calculation time for the whole model, simply remove CC/CV indicators section (down left).

« Last Edit: April 06, 2015, 08:30:14 pm by prasimix »
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
New power PCB...
« Reply #59 on: April 08, 2015, 11:56:16 pm »
Here is the latest revision of power pre-regulator and post-regulator with new PCBs where that two circuits are separated. Thanks to that I decide to play a little bit with pre-regulator and made two PCB: SMPS version based on LM5088 and another with p-ch mosfet driven by thyristor (will be presented in the post that follow). Pre-regulator PCBs now also include floating charge pump to provide proper power mosfet bias on post-regulator board.

New power board is still comply with requirement that it can works without MCU (ADC, DAC and I/O expander are located on post-regulator PCB but are not mandatory for "manual" operation).



 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Mosfet pre-regulator PCB...
« Reply #60 on: April 09, 2015, 12:00:34 am »
Mosfet pre-regulator is based on blackdog design presented here. It looks promising at least in spice simulation that is also attached.
Since I'm going to make a new power PCB I think that one is worth some effort and could be used at least for comparison with SMPS version.

EDIT 2015-05-27: Blackdog inform me that original circuit comes from an English designer.


« Last Edit: May 27, 2015, 06:46:10 pm by prasimix »
 

Offline ali80

  • Contributor
  • Posts: 48
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #61 on: April 24, 2015, 10:56:55 pm »
great work, thanks for sharing
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Floating charge pump for nmos gate bias
« Reply #62 on: April 30, 2015, 06:21:24 pm »
Inspired by question asked here by void_error and not to hijack that thread I'm adding here model of floating charge pump that I'm going to use in the latest revision of the PSU. It's intended to add on top of Vpreg voltage supplied from pre-regulator min. 5V to insure proper operation of power mosfet in the post-regulator. I made a basic testing on the breadboard and it seems that it works at least when DP (down-programmer) circuit is disabled: a whole gate driver that include OE (output enable) circuit in that case consume ~7mA. With DP enabled according to LTspice simulation another 5mA will be required. That is something that I currently cannot produce on breadboard - consumption goes crazy up to 55mA (!) which basic charge pump cannot deliver. If this problem persist possibly circuit presented in second model (555 charge pump v2 (booster).asc) could resolve this problem. This require further investigation.

Current charge pump float on top of the input voltage for the post-regulator (Vpreg). That is required since TLC555 circuit max. supply voltage is 18V. That is accomplished with D3, R3 and filtered with C4 and in that way set to max. 15V. R3 is a problem here if you'd like to cover a whole Vpreg range (2-52V). Even if its value is zero timer cannot work properly from 2V. Due to that auxiliary supply is added (+15V) which is already present in post-regulator circuit. The most critical range is when Vpreg is a little over 15V. If R3 is too high it will be too limiting, if it's too small it will dissipate to much heat. Chosen value of 1K2 is a compromise and in worst case (Vpreg=52V) it will dissipate ~1W.

Maybe R3 could be replaced with some active circuit to provide better limitation/regulation over the range of 15-52V. Any suggestion is welcome.






There is also another approach: to fix bias voltage to max. Vpreg (52V) + 5-12V and don't bother with mentioned auxiliary supply and issue when Vpreg is close to 15V.

The last possibility is to completely remove bias supply generated by charge pump. How that could be possible? If I use 48VAC on the input that rectified and filtered gives ~67VDC and put pre-regulator in-between which deliver 2-52V for the post-regulator it will act as some sort of "isolator" between Vin (67VDC) post-regulator where in the worst case (52V) difference for bias will be even in the case of max. load at least 12V.

To test mentioned scenarios I added some jumpers (JP2 and JP3 on Sheet 1). on the latest revision of the pre-regulator PCB.

« Last Edit: April 30, 2015, 06:25:16 pm by prasimix »
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Charge pump revised
« Reply #63 on: May 02, 2015, 07:17:46 pm »
Here is another attempt to have a charge pump that can follow pre-regulator output voltage (Vpreg) and provide in the worst case 5V higher voltage as a power mosfet bias in the post-regulator. I didn't test it yet on the breadboard since I don't have DN2535 or similar depletion mosfet but circuit looks correct in LTspice. Simulated case shows how Vpreg changes from 5 to 32 and 52 and back to 5V. When Vpreg drops below 15V Vaux is used as a supply. According to this simulation Vaux could go down to 9V and that the 5V difference is preserved. DN2535 power dissipation is in the worst case ~1W that shouldn't be a problem to manage with tiny TO-220 clip-on heatsink.



 

Online harrimansat

  • Regular Contributor
  • *
  • Posts: 78
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #64 on: May 03, 2015, 06:55:02 am »
Well done!, a lot of job!
Which efficiency do you achieved?
Why don´t directly use mains 220V as power?
My recommendation to make prototypes, especially for power electronics is to use solder breadboard. I don't like jumper boards, especially when aged, connections often faults
 

Online harrimansat

  • Regular Contributor
  • *
  • Posts: 78
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #65 on: May 03, 2015, 07:08:23 am »
Modern cheap switching power supply, 0-30V, 0-5A, with multiturn pots :





« Last Edit: May 03, 2015, 07:15:09 am by harrimansat »
 

Offline Rerouter

  • Super Contributor
  • ***
  • Posts: 4013
  • Country: au
  • Question Everything... Except This Statement
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #66 on: May 03, 2015, 08:24:36 am »
I can see that this thread has been a bit quiet apart from the op, all i can say is its looking like its shaping up to be a really nice piece of kit, :)  :clap:
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #67 on: May 03, 2015, 06:01:39 pm »
Well done!, a lot of job!
Which efficiency do you achieved?
Why don´t directly use mains 220V as power?
My recommendation to make prototypes, especially for power electronics is to use solder breadboard. I don't like jumper boards, especially when aged, connections often faults
Thanks. I'm still not ready for 220VAC adventure, but at least in the latest revision I made a clear cut between pre-regulator and post-regulator part of the power PCB. That will open a more room in the future for various scenarios including AC/DC converter without or even with PFC circuit or buck/boost SMPS for battery (car or solar) solution.
PSU should be easily downsized to i.e. 0-20 or 0-30V range and for higher current (up to 5A per channel).

I still cannot talk about overall efficiency. I made some measurement of pre-regulator based on LM5118 in post #36.
I'm planning to explore three another type of pre-regulators: based on LM5088, LM5116 and one proposed by blackdog (see post #61). The latest one probably has the worst efficiency but also the cleanest output.

Modern cheap switching power supply, 0-30V, 0-5A, with multiturn pots :
Looks nice and compact. Still don't like to see a fan ;). I even cannot think about price since to this moment I spent enough time and money to afford the most expensive PSU from brands such as Keysight (Agilent) :). I cannot imagine that my final BOM can be in anyway competitive with mentioned MCH-K305D. From the other side I hope that achieved flexibility and feature set (especially with MCU board added) will make some sense for anyone who eventually decide to follow this project just as I decided to follow Liv's PSU from where I borrow the "final stage" which in LTspice (see #59) and even horrible breadboard looks very attractive.
 

Offline Agape

  • Newbie
  • Posts: 1
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #68 on: May 11, 2015, 08:19:00 am »
I found your design thread when googling DIY lab power supply ideas and read through it in one go. Really like what you are doing and I am definitely going to watch it develop further. I will have to go back and study more about power supply designs as of now I don't have enough knowledge to really understand this but hopefully I will come from this educated and not just confused.

So I have couple notes from reading the thread (if some of them are laughable or silly, please just bear with me):

  • Funny how you consider THT more DIY friendly than SMT when I think the exact opposite. There is no need to flip flop the board which I find tedious when doing THT and I personally find soldering 0603 easier than THT. Also, I don't think that there was mention about tolerance of used components? ('Normal' THT resistors are 5%, 'normal' SMD are 1%.)
  • Kudos for modular design. That is really interesting. It got me thinking thought with connection to chain-ability / is there anything stopping to put together more than two power stages? Not that I would need more that 100 volts but sometimes (for motor control applications or electrolysis) I need more that 6A (10A would be nice). So far I was using power supply with few discrete output voltages but something more continuous would be better (for example for DIY thought hole copper plating where you need to control current based on area of the board).
  • What kind of toroidal transformer are you using? With your requirement of no fan (no noise) I would assume some lower core gain type (usually sold as AUDIO) to minimalize risk of rustling which is sometimes a big problem (and quite annoying).
  • It's quite interesting that you would make analog only capability one of requirements of the design. I can understand the theoretical mental exercise of it but really who would in today's age lab power supply without MCU. Moreover one with quite complicated design and some serious high quality requirements. The design then has to accommodate two ways for controlling voltage and current. Rails on PCB are longer and there is bigger room for interference and error. (At least the lab PSU I have seen up until now usually had some form of shielding of control signal to DAC/ADC. But on other hand you are only shooting for 1 cV/cA and these had 1 mV/mA accuracy so maybe that is OK.)
  • Interesting at your choice to add ethernet connection. That is quite industrial. Small note on your panel design. I don't remember when I lastly disconnected ethernet connection on one of my devices. It usually stay connected all the time (non portable devices), but I usually disconnect USB after use. So ergonomically I see the point to have USB connector on front panel, I don't see the point in ethernet's case, quite the opposite.
  • Ha at Arduino. That will make it easier for many people to do some open source extending of PSU's firmware.
  • I have never had in my hands that display or at least I'm not aware of it. But at that photo it looks positively horrible. I hope that it is only bad photo-effect. (But then it's arduino everybody's dog can change it.)
  • You have the ADS1120 in your schematics as 15-bit, but by the TI it's 16-bit.
  • Do you have plans for some form of distribution? I have seen the OSHW logos on boards but then they are 110x60 mm. Given that Chinese fab houses work with discrete sizes it will get pretty inefficient pretty quick. So if there was way to buy set of PCBs it would be great.
  • On the same note: Are the boards from Printed.CZ not tinned? Looks like it.
  • You could always use more than one pin to connect line from one board to the next. Look at what is doing guy from Granite Devices with his new IONI motor driver: he is using standard PCIe connector for high power input/output.
  • Do you plan on some protection for cases when for some reason there is problem in connection of the boards?

Well I hope I didn't make complete ass of myself.
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #69 on: May 11, 2015, 09:21:37 pm »
Funny how you consider THT more DIY friendly than SMT when I think the exact opposite. There is no need to flip flop the board which I find tedious when doing THT and I personally find soldering 0603 easier than THT. Also, I don't think that there was mention about tolerance of used components? ('Normal' THT resistors are 5%, 'normal' SMD are 1%.)

Ha, when I started this project I had zero experience with SMT. With simple lamp magnifier (x3, x8) it started to be a really joyful experience. Personally I can say that I'm now ready for "SMT only" solution even without stencils and oven :) As you can see there is a just a few responses to this project and I'd like to receive more feedback what to use for the "final" PCB revision.

Kudos for modular design. That is really interesting. It got me thinking thought with connection to chain-ability / is there anything stopping to put together more than two power stages? Not that I would need more that 100 volts but sometimes (for motor control applications or electrolysis) I need more that 6A (10A would be nice). So far I was using power supply with few discrete output voltages but something more continuous would be better (for example for DIY thought hole copper plating where you need to control current based on area of the board).

5A per module shouldn't be a problem and I already made some testing with CC loop disconnected.
I set limit to 2 modules due to MCU board and BP post board, but of course with bigger MCU then currently selected Atmel32u4 (for more I/O pins) hopefully that is achievable.

What kind of toroidal transformer are you using? With your requirement of no fan (no noise) I would assume some lower core gain type (usually sold as AUDIO) to minimalize risk of rustling which is sometimes a big problem (and quite annoying).

I'm currently using a custom made by local manufacturer but, I hope that something like this Breve Tufvassons TTS300/Z230/48-48V or this one Talema 55188-P1S2 will be fine. Another possible source (that also included audio grade shuff) with quite attractive prices is Toroidy.pl. Please note that I currently have no experience with any of mentioned products.

It's quite interesting that you would make analog only capability one of requirements of the design. I can understand the theoretical mental exercise of it but really who would in today's age lab power supply without MCU. Moreover one with quite complicated design and some serious high quality requirements. The design then has to accommodate two ways for controlling voltage and current. Rails on PCB are longer and there is bigger room for interference and error. (At least the lab PSU I have seen up until now usually had some form of shielding of control signal to DAC/ADC. But on other hand you are only shooting for 1 cV/cA and these had 1 mV/mA accuracy so maybe that is OK.)

I think that possibility to have functional unit without coping in the first step with digital part is beneficial for general DIY audience. That together with modular design simplify testing and more gradual approach to the final product. It's not even real (or serious) analog only approach since at least tracking for dual channels is not implemented.
BTW, ADC/DAC control traces on the power PCB is really short and hopefully it will not picking up substantial noise.

Interesting at your choice to add ethernet connection. That is quite industrial. Small note on your panel design. I don't remember when I lastly disconnected ethernet connection on one of my devices. It usually stay connected all the time (non portable devices), but I usually disconnect USB after use. So ergonomically I see the point to have USB connector on front panel, I don't see the point in ethernet's case, quite the opposite.

Yes, I put as much as possible features on the MCU as an exercise :) We'll see to which extent it will works. I'm agree with you about ethernet and USB. Need some rethinking and possibly USB connector could be something like this USB 2.0 B female panel mount to 5Pin F 0.1" header and small "breakout" PCB for ethernet has to go on the rear panel.

I have never had in my hands that display or at least I'm not aware of it. But at that photo it looks positively horrible. I hope that it is only bad photo-effect. (But then it's arduino everybody's dog can change it.)
Huh, it doesn't looks so horrible and it's really cheap :). Anyway as you can see a DOGL128 is also supported by MCU board.

You have the ADS1120 in your schematics as 15-bit, but by the TI it's 16-bit.
Yes, but bipolar, for unipolar you can use only 15 bit.

Do you have plans for some form of distribution? I have seen the OSHW logos on boards but then they are 110x60 mm. Given that Chinese fab houses work with discrete sizes it will get pretty inefficient pretty quick. So if there was way to buy set of PCBs it would be great.
My idea is to publish Eagle and Gerber files of first workable version. Thanks to modular approach that can be done in more steps (i.e. various pre-regulators, post-regulator, MCU board etc.). I can also organize group buy of PCB only (I closed one successfully last year on diyaudio forum) if such thing is not against policy of this forum.

On the same note: Are the boards from Printed.CZ not tinned? Looks like it.
This one is not since I asked for "galvanic Au" surface instead. You can play with different versions on their online calculator.

Do you plan on some protection for cases when for some reason there is problem in connection of the boards?
What kind of connection problem you have on your mind?

Well I hope I didn't make complete ass of myself.
Not at all :). I believe that people is here to share some ideas and assist each other not to compete who is the smartest (anyway even the most brilliant mind for electronic could still be a complete 'life idiot' so what at the end could be a "RMS value" of such person?  :-//).
 

Offline m100

  • Regular Contributor
  • *
  • Posts: 143
  • Country: gb
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #70 on: May 11, 2015, 11:01:26 pm »
Funny how you consider THT more DIY friendly than SMT when I think the exact opposite. There is no need to flip flop the board which I find tedious when doing THT and I personally find soldering 0603 easier than THT.

Many aged over 40 really struggle with seeing and positioning even large SMD passives.  IMHO It's the biggest barrier to repair and construction of modern electronics and magnification with a stereo scope and really good lighting isn't always enough.  SMD is, I guess, fine for those in their 20's who seem to cope even after a heavy weekend and zero sleep.  I know I saw far more and could work at a much smaller scale with a greater degree of dexterity a few decades ago.  Through hole is IMHO far more accessible.  I'd choose it every time because SMD really is a complete PITA for those with less than perfect eyesight.
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #71 on: May 11, 2015, 11:15:30 pm »
Funny how you consider THT more DIY friendly than SMT when I think the exact opposite. There is no need to flip flop the board which I find tedious when doing THT and I personally find soldering 0603 easier than THT.

Many aged over 40 really struggle with seeing and positioning even large SMD passives.  IMHO It's the biggest barrier to repair and construction of modern electronics and magnification with a stereo scope and really good lighting isn't always enough.  SMD is, I guess, fine for those in their 20's who seem to cope even after a heavy weekend and zero sleep.  I know I saw far more and could work at a much smaller scale with a greater degree of dexterity a few decades ago.  Through hole is IMHO far more accessible.  I'd choose it every time because SMD really is a complete PITA for those with less than perfect eyesight.

Thanks m100 for elaboration. That's something that leads me in decision to have as much as possible THT components. If this project ever ends up with some group buy I'll be glad to prepare two sets of PCBs if we meet required minimum quantity targets for both tracks.
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
PSU with Mosfet pre-regulator first results...
« Reply #72 on: May 16, 2015, 12:16:04 am »
New PCBs arrived and I decided to start with testing blackdog's mosfet pre-regulator. It looks pretty good. It can deliver even with a single mosfet over 6A. With two of them it shouldn't be a problem to deliver declared 10A. I tested it with Vin=48VAC (67VDC) and single 10.000uF capacitor. I didn't measure capacitor current but according to spice model it goes far beyond what manufacturer declared as limit. A rise of its temperature can be detected but nothing alarming (so far).

As a post-regulator I used a new design based on Liv circuit and everything is tested strictly in CV mode. I found two issues: if Vin is mentioned 48VAC output enable circuit cannot bring Vout down to zero but to ~5V. If Vin is e.g. 40VAC it works perfectly. Another issue (or "it's a feature not bug" type of event) is possibly connected with pre-regulator: depends of Vout with the same load (16R4/100W resistor is used) transformer start buzzing. Mostly in the middle range (20-30V). See on the picture 2015-05-15 Mosfet pre-regulator, Vout=25V, Load=16R4.png how output from pre-regulator looks like in that case. Don't know is it normal that transformer is audiable with such type of output (it is completely silent with e.g. 2015-05-15 Mosfet pre-regulator, Vout=10V, Load=16R4.png or 2015-05-15 Mosfet pre-regulator, Vout=50V, Iout=3A.png shape of the output voltage).

I tried to measure overall efficiency of such combination of post-regulator board (r3B36) and blackdog's pre-regulator (r1B2). Cheap DMM has been used for measuring rms values so some significant error can be expected. But regardless of used instrument I noticed a considerable fluctuation in the middle range which I think is connected with transformer buzzing (and accompanied pre-regulator signal shape). Please find below results of three measurements (within 15 minutes time frame).

 
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Mosfet pre-regulator board latest revision...
« Reply #73 on: May 16, 2015, 12:23:52 am »
Few small changes are made on the mosfet pre-regulator board. Floating charge pump is removed since post-regulator works in full range, full load (0-50V, 3A and up to 5A!) with Vin=48VAC without it. The most important thing is adding of D3 to isolate C3 from mosfet input since witout that it will be in a real trouble trying to deliver something what is not intended :). Also sync isolator is added when more SMPS is presented (like in dual channel version or when MCU board is deployed). The latest schematic is attached.
« Last Edit: May 16, 2015, 12:26:15 am by prasimix »
 

Online prasimix

  • Supporter
  • ****
  • Posts: 1218
  • Country: hr
    • EEZ
Re: PSU with Mosfet pre-regulator first results...
« Reply #74 on: May 17, 2015, 08:56:39 pm »
if Vin is mentioned 48VAC output enable circuit cannot bring Vout down to zero but to ~5V. If Vin is e.g. 40VAC it works perfectly.
I fixed this one by changing BC557B in OE (output enable) current mirror with BC640. Actually with Vin=48VAC emitter-collector voltage goes beyond BC557B's allowed maximum (if max. Vceo=50V). With BC640 that limit is rised to 80V.

EDIT: BC556B works even better and comes with the same pinout :).
« Last Edit: May 19, 2015, 07:48:52 pm by prasimix »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf