On 2023/02/20 at 6:07 PM, BritishRacingGreen said:

 a Word of Warning when testing the Bus Soft Start Circuit Manually

In previous posts I have suggested that one can debug a faulty bus soft start circuit by removing the control board and exciting the soft start circuit manually .  However one must keep in mind that when the circuit suddenly start to work , the bus voltage will reach some uncomfortable high voltages. The reason is the control board knows where to stop the soft start , but when we are enabling the soft start manually , there no stopping . I have fortunately had a meter on the bus , and when it reached close to 500V ! , I switched off.  So one way of ensuring safety is to remove the diode in the high voltage transformer secondary , then debug your faulty circuit in comfort.

 

have  seen youtube  video  where the author  explains that  some  output of the  soft start is fed to its  input  and  with new  higher  input   the  output goes   even higher   ......according to this autor its   an infinite process but i doubt that  

there have to  be   a point  where the uc3845  and stop all   output 

but that limit will be  so high that   a  damage  will happen   if  not stopped  by  cpu  much earlier 

 

6 hours ago, wael_fathe said:

the author  explains that  some  output of the  soft start is fed to its  input  and  with new  higher  input   the  output goes   even higher 

What model is this schematic trace from? It seems unusual.

6 hours ago, wael_fathe said:

there have to  be   a point  where the uc3845  and stop all   output 

It depends on the design. In the versions I've seen, there is no voltage feedback.

12 hours ago, Coulomb said:

It depends on the design. In the versions I've seen, there is no voltage feedback.

but  then if  cpu   failed  to  observe what  going on the  smps could  go  to   insanely high voltages?

so it is still possible to make an open loop  ucxx design  ,,,,that seems tempting ....

the current sense resistor  only  limit the  output  should the current at secondery become exessive  while the voltage  regulation is only  done  by the turn  ratio    ,,,i think this is  done  only for the simple reason that there is  one load  all the time   ,,,for other  ucxx design the voltage  regulation  is must as the load  keep changing   for exampe  in devices with fans  fans on   requires more  current then  fan  off requires less    current  the voltage  will  fluctuate  at moment fan on  so  voltage  regulation is needed 

 

those inverters   are great lesson in electronics   i have  seens all ideas   done in them  ,,,,soft start  voltage dooblers,,,,snch rectifiers  ...digital electonics   ,,,,,,you know them  you conquer  electronics   

12 hours ago, Coulomb said:

19 hours ago, wael_fathe said:

the author  explains that  some  output of the  soft start is fed to its  input  and  with new  higher  input   the  output goes   even higher 

 

please check the video he only  hold  a paper  and  describing i wonder  if whathe said  is truth 

if  what he is saying is not  truth  then how  the smps  is  supposed  to raise the voltage  across the caps  if it does that  in small time the smps  will be  stressed  it  can easily  break 

what  the method of  bring  bus voltage   "gradually"   in the ucxx based soft starter smps?

1 hour ago, wael_fathe said:

what  the method of  bring  bus voltage   "gradually"   in the ucxx based soft starter smps?

My understanding is that the relatively weak power of the bus soft start circuit coupled with the flyback topology means that the output is approximately constant current. Constant current into a capacitor results in a constant dV/dt, i.e. a constant rate of change of voltage with respect to time.

1 hour ago, wael_fathe said:

please check the video he only  hold  a paper  and  describing i wonder  if whathe said  is truth

What he has on his piece of paper is quite different to what I and others have traced. That's why I was curious as to what model this is supposed to be from.

Compare his sketch:

with what I'm used to seeing:

Note the complete absence of anything other than the diode (and its snubber capacitor) on the output of the "transformer". My first guess is that the video author was unaware of previous schematic traces, and mistraced the circuit. But the other guess is that this is a different circuit from a model that I'm not familiar with.  The board at the bottom, which I assume is a solar charger, is not familiar to me. Edit: Though it's obviously a high PV voltage solar charger, which I know little about.

Edit 2: It's also possible that I and others have missed this feedback part of the circuit, but I doubt that.

Edited October 11, 20232 yr by Coulomb

3 hours ago, Coulomb said:

My understanding is that the relatively weak power of the bus soft start circuit coupled with the flyback topology means that the output is approximately constant current. Constant current into a capacitor results in a constant dV/dt, i.e. a constant rate of change of voltage with respect to time.

yes i totally  aware with this phenomena  when  smps  subected to load that is too high such as the 

capacitor of the bus  it turn  into  a  curret source reducing its  voltage   and  pushs  in constant current into the load which is a cap ...cap will charge over time ,    yes  true no need for  gradual incrase of voltage  

i think the smps  can handle  that   

the author is from third  world  so mistake or unaccuracy possible  ,  also  possible  that he  refer 

to  clone  models  with this unsuual  design   ,  

 

 

On 2023/09/03 at 10:54 PM, JustinSchoeman said:

The lower currents on the HV side mean that you have to move much less charge to turn the devices on and off, and you can get away with driving directly from the transformer.

interesting    so does the drive  requirement of  fet or  igbt  differ  according to the current  being  driven  in drain or   collector  of   a fet or an igbt

what i know  that gates  are voltage  driven  give that spefic voltage  to gate  and it will turn on 

fet  or  igbt  ?

On 2023/09/07 at 8:29 AM, JustinSchoeman said:

D32 bypasses the gate resistor for discharge (turn off) - so turn off is faster and less damped. You need to manage dI/dt during turn on, which is why you have R91. Don't have the same problem during turn-off, so bypass R91 with D32 for turn off.

amazing  ....as soon as low  appear on pin 8 of drive transformer d32  becomes  forwarde baised 

due to fact that it has low infront of it and  voltage back of it (gate capacitance stored voltage) 

it   speeds  up turn off process   and  skip  the gate resistor  ...

 

thanks get it 

23 minutes ago, wael_fathe said:

what i know  that gates  are voltage  driven  give that spefic voltage  to gate  and it will turn on 

fet  or  igbt  ?

The gates of both FETs and IGBTs are capacitors (sort of - actually non-linear compound capacitors, but it is an accurate enough simplification for this discussion).

So, with a gate capacitance of C and a turn on voltage of V, we require a charge of Q = C * V to turn the device on.

In truth, while we say the devices are voltage controlled, they are actually charge controlled. Because of the above equation, they are largely interchangeable.

But it is the charge on the gate capacitor which provides the electrons in the channel/base of the device. And it is also the number of electrons (and therefore charge) that limit the charge carrying capability of the device.

So the more current you need to carry, the bigger the charge required.

But we also have Q = I * t. So for a specific turn on/off time 't' we require I = Q / t.  So the required gate current is proportional to the gate charge, which is proportional to the current carrying capability.

This is a little over-simplified, but it should give a good, basic understanding of the device parameters.

On 2023/10/10 at 7:25 PM, wael_fathe said:

sir  your method is great but i think this method of colomb is easier 

1  rmeove all fets  and dc-dc igbt(though for igbts  its ok  to leave them if they not  shorted)
2  short u9 colpler to enable sg3525
3 remove cpu board
4  connect the inverter to dc power supply 48vdc 24vdc etc
5 hit the ac on switch on
6 scope for signals at mosfet and dc dc igbts

Sorry for my late reply , but rather late than never!  I do appreciate you evaluation of subjects we are touching here  and your reporting thereof. 

Yes its easier to power the 48V from the battery terminals , but I would like the mosfet/igbt drivers to be debugged with no battery or HV dc-bus voltages. Hence the need to power up the system supplies only (+12V,-12V and +5V).  So this can can be done with the transistor fitted or not . The cpu card must be removed , as the gate driver signal are manually activated by jumpers or switches fitted to the controller card connectors.

On 2023/10/10 at 7:25 PM, wael_fathe said:

but i think your method is  a must  if

1  cpu is built in 
2  you have not removed the mosfets yet 

No , the cpu card must be removed.  Sadly , my procedures will not work where the CPU  is an inseparable part of the main board , eg. in the case of 3kw main board. This gives you the freedom to activate drive signals as and when required during debugging. The transistors may be fitted or not . It must be understood though that in order to test the final capability of the gate drive , a transistor must be fitted , even temporarily , even a low voltage low current device can be fitted for testing , e.g. IRF640 etc. But this can only be done when there are no battery terminal voltage , or the battery voltage must be in the scope of the transistor max operating region. I have not yet published my Part 2 of the Bringing Up Inverter in Softest Possible manner Chapter , but I will try to publish it over the coming weekend. This will provide details of how you can bring up the dc-dc converter manually by adding small controlled voltages to the battery bus , and check for corresponding results on the HV dc bus. Its actually very easy , and has helped me a lot lately . 

 

On 2023/10/10 at 7:25 PM, wael_fathe said:

am i right or there is  more hidden  benifits of your   method?

I hope so , but you be the judge of that when you read Part 2 . The idea is not to blow away expensive transistors as I have done in the past.

On 2023/10/10 at 8:13 PM, wael_fathe said:

have  seen youtube  video  where the author  explains that  some  output of the  soft start is fed to its  input  and  with new  higher  input   the  output goes   even higher   ......according to this autor its   an infinite process but i doubt that  

there have to  be   a point  where the uc3845  and stop all   output 

but that limit will be  so high that   a  damage  will happen   if  not stopped  by  cpu  much earlier 

 

I have studied the video , the language barrier is making it difficult , but I think I do get the drift.

It interest me that the soft start circuit discussed has analogue feedback , The soft start circuits I have tested and seen working in the 5kW and MAX Axperts has no feedback. The circuit has an open loop forward path , where the output of the driver transformer is actually weak and therefore has a high output impedance . When this output is driving the dc bus upon startup evaluation , the driver is confronted with an very small load impedance via the two big capacitors in parallel (2x470uF 500V) . Therefore the RC time constant is in the order of seconds to  charge these capacitors up to test level. The DSP will continuously check this voltage via its feedback measurement of the dc bus voltage and will disable the soft start when the voltage reaches 350v or so. This also pre-charges the capacitors , ready to accept the feed from the dc-dc converter as soon as that is switched on at a later stage.

Care does needs to be exercise when you manually enable the soft start , in order to prevent the dc bus voltage from approaching 500V . I have seen the voltage go beyond 450V by leaving the jumper on for too long ! So I aborted.

 

On 2023/10/11 at 3:53 PM, wael_fathe said:

but  then if  cpu   failed  to  observe what  going on the  smps could  go  to   insanely high voltages?

so it is still possible to make an open loop  ucxx design  ,,,,that seems tempting ....

Technically the circuit is functionally closed-loop via the DSP and its measurement of the increasing bus voltage . And when the terminal voltage is reached , the DSP switches the soft start off .  Of course this is slow and crude control , but fit for this particular purpose.

 

On 2023/10/11 at 3:53 PM, wael_fathe said:

those inverters   are great lesson in electronics   i have  seens all ideas   done in them  ,,,,soft start  voltage dooblers,,,,snch rectifiers  ...digital electonics   ,,,,,,you know them  you conquer  electronics   

Yes 👌

On 2023/10/11 at 3:53 PM, wael_fathe said:

if  what he is saying is not  truth  then how  the smps  is  supposed  to raise the voltage  across the caps  if it does that  in small time the smps  will be  stressed  it  can easily  break 

what  the method of  bring  bus voltage   "gradually"   in the ucxx based soft starter smps?

Again , the output is stable because of the very large bus capacitors value (1000uF) and the high source impedance of the soft start output. Instability and overshoots will not be present as it takes a very long time (seconds) for the output to raise to high levels.  The voltage increase gradually only by virtue of this RC time constant. And when the terminal voltage is reach the DSP switches the circuit off.

20 hours ago, JustinSchoeman said:

The gates of both FETs and IGBTs are capacitors (sort of - actually non-linear compound capacitors, but it is an accurate enough simplification for this discussion).

So, with a gate capacitance of C and a turn on voltage of V, we require a charge of Q = C * V to turn the device on.

In truth, while we say the devices are voltage controlled, they are actually charge controlled. Because of the above equation, they are largely interchangeable.

But it is the charge on the gate capacitor which provides the electrons in the channel/base of the device. And it is also the number of electrons (and therefore charge) that limit the charge carrying capability of the device.

So the more current you need to carry, the bigger the charge required.

But we also have Q = I * t. So for a specific turn on/off time 't' we require I = Q / t.  So the required gate current is proportional to the gate charge, which is proportional to the current carrying capability.

This is a little over-simplified, but it should give a good, basic understanding of the device parameters.

@wael_fathe to add what @JustinSchoeman is pointing out , yes, the mosfet and igbt has very high input impedance and therefore  is a voltage controlled device in terms of DC applied to the gate. But it possesses an unwanted parasitic capacitance which  will delay the switch-on threshold voltage of the transistor according to the RC constant where R is the drive source resistance of the gate driver. Now in order to decrease this RC time to acceptable levels , we cannot decrease the gate capacitance , but we can decrease the drive resistance . It is for this reason that the drive resistor is typically in the order of 22R and 47R . 

 

Below is a trace I have taken over the 22R drive resistor . It actually represents the current thru this resistor , and therefore the current thru the gate of the MOSFET. Unfortunately I have not shown the gate voltage as a second channel , but be assured that the gate voltage is turned on where the positive going pulse is , and turned off where the negative going pulse is. 

You can clearly see the effect of the MOSFET gate capacitance . There is a sharp rise in current as the cap is charged up to max turn-on voltage . Same during turn-off . The sharp rise is courtesy of the low 22R resistance . If this value was 2 kilo ohm , for example , you can imagine it would take longer to charge the gate cap up to the desired threshold voltage for switch-on.  Also note that in between the the switching current pulses , there is zero drive current. This is because of the very high input impedance of the MOSFET gate.

It is for the reason of the high impedance of MOS transistors that MOS circuits draws very little current. But it is also for the switching current reason that the circuit's power consumption will rise with higher frequency  .

Edited October 12, 20232 yr by BritishRacingGreen

On 2023/10/11 at 10:15 PM, JustinSchoeman said:

required gate current is proportional to the gate charge, which is proportional to the current carrying capability.

and i always  though of a total  independence  between   voltage applied to the gate- source   to turn 

the mosfet on   

and  the amount  of the current passing  from drain to source

and therefor   for  a mosfet that intended to switch on and off  10a  

the amount of  charge  needed  to   do that  is much lesser 

than   same mosfet  in same  circuit that siwtch on and off 100a

viewing the phenomena form  gate side  we  face  a "sudden raise of capacitance  if the current at drain   to be switched on and off changed from 10a to 100a"

 

am i right 

Edited October 13, 20232 yr by wael_fathe

8 minutes ago, wael_fathe said:

and i always  though of a total  independence  between   voltage applied to the gate- source   to turn 

the mosfet on   

and  the amount  of the current passing  from drain to source

and therefor   for  a mosfet that intended to switch on and off  10a  

the amount of  charge  needed  to   do that  is much lesser 

than   same mosfet  in same  circuit that siwtch on and off 100a

viewing the phenomena form  gate side  we  face  a "sudden raise of capacitance  if the current at drain   to be switched on and off changed from 10a to 100a"

 

am i right 

Nope.  It depends on the size (current capacity) of the device, not the actual current.

If you want to switch on a transistor with a continuous drain current rating of 100A, it would require about 10x more current than to turn on a transistor with a continuous drain current rating of 10A (at the same switching frequency).

When that 100A transistor is on, it can conduct 10A or 100A, but it would have taken the same current to turn it on. Once it is on, it requires no current to keep it on.

On 2023/10/12 at 6:25 PM, BritishRacingGreen said:

power consumption will rise with higher frequency  .

when we switch fet  once a second  it  make its transition from off to on    and pass through that resistive  region  once

with higher  frequency   it pass throug  that region many many times

each passing it   emit heat out  of its body  

too much frequency= too much heat

 

that is  mainly  why scientist  invented  the   resonance  converter 

in the resonance converter  the mosfet  switched only few  times    the rest of the  cycles  are done

by the resonance components   , this way they can achive  high  frequency  with out actually  turn on and off   fets   at the same  frequncy  of that converter?

so in  future   we may be repair   those  and find the fet   only work  on 20khz    while the frequency of 

the  converter is  1 mega hz  

 

some  call them   "  soft  switchs"   as most of the switching happen due to  resonanct  componnets 

rather than the   switching fets 

6 minutes ago, JustinSchoeman said:

When that 100A transistor is on, it can conduct 10A or 100A, but it would have taken the same current to turn it on. Once it is on, it requires no current to keep it on.

what  triggers  my question in the first place  

is your  discussion with  britchgreenracing   about why drive in primary fets

is  different  than the drive  of  dc-dc  igbts  in scecondery   in terms of compoenent used for the drive

so  now   with  your answer  i conclude that   you mean   diffrent devices  used  different   drives  due to

different  current rating which  effects  the  gate  charge   and thus  component used for  drive   

 

that is logical now   ,,,and i think  i observe  your theory  by downloading  diffrent mosfets  of different currents and observe  the diffrence  in gate  charge 

On 2023/10/12 at 5:37 PM, BritishRacingGreen said:

bus , and check for corresponding results on the HV dc bus. Its actually very easy , and has helped me a lot lately . 

 

that will make  a  rovolution in servicing this  inverters....  i appreciate your  work ...

a

On 2023/10/12 at 6:25 PM, BritishRacingGreen said:

The sharp rise is courtesy of the low 22R resistance

you observed the current in resitor by observing the voltage  accross it

so   at last you  show  us  gate capacitance "in action"  studing something is  not

as nice as  seeing it    bravo sir 

 

 

On 2023/10/12 at 5:37 PM, BritishRacingGreen said:

No , the cpu card must be removed.  Sadly , my procedures will not work where the CPU  is an inseparable part of the main board , eg.

you and colomb  made too many inventive  ways to test  all in one  solar inverters thank you guys

but with new designs   as you said  voltronic  now  make cpu  responisble  for  driving the primary fets

and no more sg3525  buiness 

do you have ideas in mind  for the these new designs 

Hello sir ,

i am recently facing the Error 03 , battery high voltage error. Measured with multimeter reads 54V but inverter shows around 90V which gradually come down and inverter starts outputting 230V ac. Recently changed the -12v regulator and capacitor c132 (general capacitor). All +12V (around 12.04v) ,-12 (11.95v) and 5v (5.01v) seems ok. What could be the problem. Do i have to use low ESR capacitor. 

I also checked R29-R32 and R36-R33 ,seems ok. But in R36-R33 section total resistance value shows around 3.7M which gradually increases to 3.87M while in R29-R32 section it shows 3.95M without fluctuating. I individually checked after de-soldering but seems fine. Lastly interchanged the resistor form R29-R32  and R36-R33. But same result, showing fluctuation in negative side value. (Checking with control card removed). Checked the OPAMP and its surrounding components i.e. resistor and capacitor  in control card all seemed ok.  

Any ideas where to look. 

Recently changed the gate driver transformer ( self wounded in toroid core). Also changed KA3525 IC with SG3525 . all waveform looked ok. Also changed c141 and c20 (thought it may be dropping the voltage of -12v). Earlier I used to get -12.05v but now its showing -11.95v , checked with thermo-vision camera only showed max heating in 5v regulator and c27 . Changed the C27 too. Earlier facing some noise in IC side , thought I may be due to ringing in gate drive transformer ( During dismantling of it for calculating winding turns bobbin broke :)) . But latter found the one Q15 was slightly heating which was causing the noise. Replaced all driver transistor . Now all is Ok. no heating.  

Edited October 14, 20232 yr by razmasoom

50 minutes ago, razmasoom said:

Recently changed the -12v regulator and capacitor c132 (general capacitor

Ok good work, well done. 

Your error code 03 is mainly due to unstable -12v . The kingpin capacitor that typically fails is C116, the filter cap. Have you replaced C116? . C132 is not so much exposed to ripple currents as C116 and C116 therefore deteriorates quicker. 

 

1 hour ago, razmasoom said:

self wounded in toroid core)

Brilliant! 

I had to rewind one too for a MAX machine about a year ago. Had to initially  soften it up by cooking  it on a gas braai! But it wss fun! 

Edited October 14, 20232 yr by BritishRacingGreen

9 minutes ago, BritishRacingGreen said:

Ok good work, well done. 

Your error code 03 is mainly due to unstable -12v . The kingpin capacitor that typically fails is C116, the filter cap. Have you replaced C116? . C132 is not so much exposed to ripple currents as C116 and C116 therefore deteriorates quicker. 

 

No sir , Earlier I thought same , checked with ESR meter , was showing around 100uf and low loss. So I re-soldered it as currently i don't have Low ESR capacitor in stock.

6 minutes ago, razmasoom said:

No sir , Earlier I thought same , checked with ESR meter , was showing around 100uf and low loss. So I re-soldered it as currently i don't have Low ESR capacitor in stock.

Ok, even a  normal esr and  85 deg celcius rating  will be ok to test. And dont be afraid to use up to 220uF if you have. 

Per @Coulomb advice it is also good to check your battery voltage resistor string for value and dry joints. These are the dual strings made up of 3 x 1M and 1 x 0R resistors. Check for good continuity. 

Edited October 14, 20232 yr by BritishRacingGreen

3 minutes ago, BritishRacingGreen said:

Ok, even a  normal esr and  85 deg celcius rating  will be ok to test. And dont be afraid to use up to 220uF if you have. 

Per @Coulomb advice it is also good to check your battery voltage resistor string for value and dry joints. These are the dual strings made up of 3 x 1M and 1 x 0R resistors. Check for good continuity. 

There is another condition that loads the -12v rail . Its when you change 3525 ic then certain passive components must be removed for SG3525A. Its discussed earlier on in this thread, but if you struggle let me know. 

1 hour ago, razmasoom said:

I also checked R29-R32 and R36-R33 ,seems ok. But in R36-R33 section total resistance value shows around 3.7M which gradually increases to 3.87M while in R29-R32 section it shows 3.95M without fluctuating. I individually checked after de-soldering but seems fine. Lastly interchanged the resistor form R29-R32  and R36-R33. But same result, showing fluctuation in negative side value. (Checking with control card removed). Checked the OPAMP and its surrounding components i.e. resistor and capacitor  in control card all seemed ok.  

 

checked and resolder as if any dry joint may cause the issue but still same. Will definitely try  with the higher capacitor value as suggested and will report back . Thank you for quick response sir.

2 minutes ago, BritishRacingGreen said:

There is another condition that loads the -12v rail . Its when you change 3525 ic then certain passive components must be removed for SG3525A. Its discussed earlier on in this thread, but if you struggle let me know. 

tried with KA3525A but no change. Error at starting 03 and which goes away after some time.   i do have U3525a ic too. will try with that too.  Sir could you share the post link so i can study it for better understanding.  Thank you