Cheap buck converter board modification ?

Discussion in 'General Electronics Chat' started by Dyslexicbloke, Jul 27, 2018.

  1. Dyslexicbloke

    Thread Starter Distinguished Member

    Sep 4, 2010
    559
    34
    Hi Folks,
    I should probably know this but I am unsure...

    I want to electrically modify the output of a buck converter board...
    All the boards I have looked at use a monolithic buck chip with an additional diode, conductor and caps externally.

    There is a single sens pin fed by a decider across the output. The values alter depending on the requirement of the chip but the one I messed with last maintains circa 1.4 V at the sens input.

    I noted that current limiting was implemented by adding a diode, fed from an op amp output, that turns on and feeds current onto the sens pin, biasing it higher and thus doping the output voltage.

    However what I want to do is have two distinct voltage set-points, with one as a default and one activated by either a low or high input, preferably a low as I want to control it with a PLC output and pulling it to ground seems like a better plan than driving it with 24V.

    Now I could play with this but I am short on time and was hoping for a bit of guidance to reduce the tinkering time.

    I considered an open NPN collector switch to pull sens lower via a trimmer but then realised that the threshold voltage of the transistor may cause problems...
    Would it ? Can I do it anyway?
    Is there a better way?

    I am looking for outputs of around 22V & 15V with a 24V supply
    I appreciate that some buck converters will require more headroom than that but that is a separate issue.

    The application is to torque limit a small motor, whilst still retaining the ability to start it easily. Or increase the torque for any other reason for that matter.
    Both voltages will need o be adjustable, although not via the PLC, and I may well install a bypass relay to deliver the full 24V rail.

    Any thoughts and suggestions would be much appreciated.
    Thanks,
    Al
     
  2. Vytas Klyvis

    Member

    Dec 5, 2016
    47
    4
    Is the chip by any chance the XL4016?

    I'm using a buck converter to heat a stainless steel rod and I need to control the time that this rod gets heaten.
    I do this by feeding the sense pin with 5v (onboard regulator) through the aforementioned resistor and diode. When I want to turn on the buck converter I simply pull down the resistor to ground. This is at the connection of resistor to diode.
    I can imagine that you do something similar but then using your voltage setpoints.
     
  3. Dyslexicbloke

    Thread Starter Distinguished Member

    Sep 4, 2010
    559
    34
    Not sure to be honest... I have done it, well tested it with several chips, mostly from that well known auction site and it main rival.

    I agree that turning it off will work, and from what I have seen it is stable, ish, when used as a current limit.

    I tried setting up 6 in a bank, with diode separation of the outputs, which worked ok until I fried several....
    It turned out the 15A boards I had purchased had 6A chips on them.... I will refrain from casting any aspersions toward an entire continent here.

    Anyhow, as the on-board current limit is based on a resistor in the ground line that wasn't going to work as the share couldn't be controlled so I retrofitted some hall effect modules and fed their output, board by board, back to the sens pin via the diode.
    That seemed to be working well untill I ramped up the current and fried 4 of the 10 virtually simultaneously, due to the switch being too small.

    I never did get to test stability... I was a little worried that the lack of clock sync could potentially cause issues but didn't get that far.
    Still in a heap on the bench... Trying anther strategy now using PWM from an Arduino to control FET's

    With respect to controlling voltage I was hoping to get a bias, OPAMP Adder esque. Looks like I am going to have to give it a whirl and see what happens.
    Problem is its going into a customer panel so I need to be sure its robust...

    industrial strength DC-DC converter module recommendation anyone, cant find one with sufficient output swing.
    Meanwell do a good range but they only adjust by +/- 5% or so which isnt going to work for me.
    I don't mind setting a fixed output and then switching it in and out, heck I could just use a relay and diode separation.
    Finding something that will swing between 22V and circa 12-15V with a 24V supply is the issue.

    I have to use a DC input because the valve needs to close if the power fails and the UPS is DC
     
  4. ebp

    Well-Known Member

    Feb 8, 2018
    2,332
    811
    A small MOSFET such as the 2N7000 or 7002 can be quite useful for switching in a resistor as long as the source can be connected to circuit common ("ground"). The FET will behave like a small resistance rather than adding a voltage drop as such. It will introduce a bit of capacitance from the error amp input to ground, but usually this is small enough to be of no consequence because the fixed resistor in series more or less isolates the capacitance.

    Shunting the "bottom" resistor in the voltage divider to raise the voltage is definitely preferable - and easiest. Because the voltage across that resistor is fixed, it has no effect on the frequency-dependent gain of the attenuation network. Lots of voltage dividers are just resistors, but in many cases there will be a capacitor or RC network across the "upper" resistor as part of the overall frequency compensation.

    Most commercial switchers offer only a narrow range of voltage adjustment. Making a reliable supply that is adjustable over a wide range can make a lot of extra demands throughout the circuit.

    Most of the low-cost switcher modules I've seen on places like ebay have specs that are far too high for any kind of long term reliability. Now I admit I'm a little prejudiced because most of the switchers I designed were expected to run continuously for 10 or 12 years with only natural convection cooling.

    Attempting to parallel constant voltage power supplies of any sort without added sharing circuitry "never" works well. Typically one supply will deliver 100% of the output current until it reaches its current limit, then one-by-one the others will begin to contribute as the voltage falls slightly with increasing load. With supplies without good current limiting, this is an invitation to burning things up, and even with good current limiting it results in unequal stress on the individual supplies. There are methods to assure good sharing, but they require extra circuitry and supplies that have be designed to support it. If the power supplies are basically well designed with temperature stable parts in the error amps and are operated so their temperatures are all the same, sharing can be improved by use of "ballast" resistors between the individual outputs and the common points. This of course degrades voltage regulation.

    Constant current power supplies can be paralleled and each will contribute according to its current setpoint.

    If the power supplies are non-isolated DC-DC converters, all bets are off with regard to paralleling if both the inputs and outputs are paralleled, even if they are nominally constant-current supplies.

    I recently saw a video of someone adding "constant current" control to a CV switcher. It was done with a comparator driving the input to the voltage error amp. This produces hysteritic control, which is crude but acceptable for some applications. It is certainly vastly simpler than trying to get the frequency compensation right for a proper current loop error amp. It looks pretty ugly if you scope the duty cycle.
     
  5. Dyslexicbloke

    Thread Starter Distinguished Member

    Sep 4, 2010
    559
    34
    That all makes sense...
    I had realised that running the units CV was a lost cause so I didn't try.
    After adding the hall effect current limits, and calibrating each one separately, with the original current limit disconnected, I was able to get a reasonable share.

    I was essentially doing MPPT, delivering current to a battery at a rate that maintained the voltage on the PV in preference to battery voltage.
    If battery voltage or current exceed my limits the same current limit circuit was used to control the battery whilst letting the PV float.

    At least it would have done had the 'Conservative' - NOT as it turned out, 8 A / unit limit not generated smoke.
    I had 8 supplies in parallel.

    And yes I was aware it was a long-shot
     
  6. ebp

    Well-Known Member

    Feb 8, 2018
    2,332
    811
    I've designed a few MPP tracking switchers. I used three error amplifiers - one for output voltage, one for output current and a third for input voltage to directly control the minimum voltage at the input. High performance MPP tracking is rather tricky with just analog circuitry because the MPP moves so much with temperature and you need to find a peak - which you don't know you did until you are on your way down the slope on one side or the other. I have used a sort of "sample and hold" method - interrupt the output current for long enough to let the PV array get to open circuit voltage, then set the MPP input voltage to a fixed fraction of open-circuit. For a particular cell type this works reasonably well, but only where interrupting the output is acceptable.
     
  7. Dyslexicbloke

    Thread Starter Distinguished Member

    Sep 4, 2010
    559
    34
    Oh I wouldn't be going analogue...
    I can deal with complex code easily whereas a bunch of op-amps may be a great learning opportunity but with my skill level it is just as likely to be a smoke generator...

    Perhaps when I am only doing it to play with... I managed a PWN servo positioner a while ago, which I enjoyed doing, but to b honest I could have done it 10 times as fast and at half the cost had I gone digital.
     
  8. Dyslexicbloke

    Thread Starter Distinguished Member

    Sep 4, 2010
    559
    34
    Getting off topic here... and its My fault!!

    RE "pretty ugly if you scope the duty cycle"
    I did and it is...
    Fortunately in this instance what I want to do is bend the feedback set-point rather than override it.

    I like the FET as a resistor switch Idea, may even try using one as a resistor...
    Wondered about electronic POT IC, any thoughts?
    Also found a few analogue switch IC's in my parts bin so considering trying those too as I assume they are built from FET's

    In an Ideal world I would want to set the output voltage using a signal voltage rather than switching between set-points but as I said earlier, for this particular task the set-points will work and I think, based on comments thus far, that it is my best option.

    I hope to have a play with this later in the week and will check back for any further feedback.
    Help and suggestions thus far is much appreciated, it has defiantly stopped me going stuff that wasn't ever going to work.

    Cheers,
    Al
     
Loading...