Hi again DaBit, This has to be as small a footprint as possible, way smaller than any lamiated core soft iron tranformer and I suspect any serious power UPS would cost 2 ARMS & 3 LEGS to purchase ... not only that I would not gain experience in building SMPS's !!! which is a major part of this exercise...DaBit said:
The IR2113 looks really good for the drive job, do you know of an IC that would control and have internal amps etc. for regulation feedback etc... there seems to be a lot around but once you are not using there internal fets, it starts to get a little more involved with sensing and overcurrent control etc. .TJ
Eva said:
Hi Eva, yes I have already looked at using the same isolated gate drivers as in my circuit diag for the sinwave converter ... on the previous parts of this thread,
with respect to the conversion to 230, I would have to change a number of mechanical pulleys and drives to get the speed correct almost all of the power is driving large synchronous motors with a number of small ones that need to do an exact speed to match the functions of the larger motors and coincide with where each system is up to in its operation!! ... I could most likely get away with it not being a sinewave but it would need to be heading more in that direction for EMC reasons I was thinking do you agree?
SQUARE wave!! That is food for thought, ... maybe I could reduce the amount of pulses in the sinewave generator down to a lower resolution more chunkey waveform, it would certainly reduce MOSFET switching frequency and maybe run lower temp I would expect!!
I will try that in the mean time to see how much I can get away with.
Hmm. I think it is not that easy to make this much smaller than an 800VA toroid. Unless you start putting the control logic on vertical PCB's and the power logic on a 'baseboard'.
If doing the project for fun and learning is the goal, then go ahead
But be prepared that it will take some time to get things working. I think I am not overestimating when I say that 3-6 months would be needed, depending on how much time you are able to spend.No, but driving these HV gate driver IC's from any PWM modulator chip is not that hard. Just make sure that you have a digital signal (pullups on OC outputs for example) and feed that to the driver IC's.
I never used the IR2110/IR2113, but I heard that they are pretty picky on layout. I did use the HIP2500, and I never had problems with it. I have no idea what value to assign to this, but I thought you could use the information.
If I were doing this project, I probably used a full bridge or half bridge driven by the UCC2808 of UCC28065. Power for these chips could be taken from a bias winding on the core (and a startup resistor of course). But there are many, many other control IC's.
Sounds like 50Hz would work; the machine would run only a bit slower.
Generating a sinewave is not that hard, but you might want to create a low-impedance square wave. If you always turn on the MOSFET's in your bridge diagonally, that requirement is fullfilled. Then, filter it with an LC filter (and make sure the motors' inductance does not interfere), and you have a nice sinewave.
Just another weird idea (I'm a master in weird ideas, too bad only 10% of them actually work). Wouldn't it be possible to modulate the DC/DC to generate a half sinewave (using a rectified sinewave as reference voltage for the error amp would do, I suppose)? Then, the switches behind that only need to reverse the polarity. You might be able to use cheap SCR's for that, and a non-critical pulse transformer to turn them on.
The control loop should not have any problems following a 120Hz envelope.
Eva said:
Hi Eva, the motors are switched with relays, when they start, they draw 12 amps for 250mS then back to 4 amps for the duration of there run time, when the motors are not running an 800 watt resistive heating element is switched on for 10 minutes.
If you can suggest some form of closed loop for the sinewave side of things I will start researching that area. Inductive load forms about half the operation time the rest is resistive. Would I be able to feedback to my existing micro? I have lots of spare inputs!! for sinewave corrections.
I am in the middle of reproducing the sinewave part in a full bridge format to see how effective that is running from my existing micro, I assume it wont matter if there is a slight propagation delay thu the fet drivers ! It should only be very small I would think?
The finished item needs to be standalone so it can just be put in line externally without any mods or internal changes to the units that are being fed by it. But must have the 2 to 1 current advantage or the local mains will have to be heavy duty power outlets.
I think that you don't need any voltage conversion: Just use a corrected square wave to drive both the motors and the heater. This is achieved by just switching the 320V bus over the load with a full bridge and a duty cycle of 36% (it may be adjusted according to the bus voltage). Low-pass filtering is nor strictly required.
I suppose that the appliance that you want to power has also some kind of internal power supply to drive the relays and the logic stuff. Is there such an internal PSU? How it works?
I suppose that the appliance that you want to power has also some kind of internal power supply to drive the relays and the logic stuff. Is there such an internal PSU? How it works?
Remember that the amount of EMI would be 1000 times smaller than what a similar circuit working at 120Khz would generate (and this assuming that the same rise and fall times are used).
I think that I've read somewhere that equipment with switching frequencies below 5Khz was not requied to pass any EMC tests. You should get well informed about that matter, though.
I think that I've read somewhere that equipment with switching frequencies below 5Khz was not requied to pass any EMC tests. You should get well informed about that matter, though.
If the PWM is at 60hz the number of high frequency transients is reduced even if all else is kept the same, as per Eva. However, with the lower frequency, the MOSFET drive circuit can also be lightened up a lot with little efficiency loss. It then becomes simpler and cheaper while also giving slower switching and practically complete EMI elimination. But switching losses can still be kept under 2%.
There is always some ringing somewhere without filters. Cable inductance+capacitance for example. And indeed, it does generate 1000x less EMI, but that is 'only' 30dB.
Of course, you can make the edge slower, but a simple LC filter does the job even better. As far as I recall, conducted EMI measurements start at 150kHz.
Of course, you can make the edge slower, but a simple LC filter does the job even better. As far as I recall, conducted EMI measurements start at 150kHz.
subwo1 said:
I can very easily soften the drive to the MOSFETS, but I am a little concerned about insulation breakdown although it really should not be a problem, you would think that an item made for 115VAC would have insulation breakdown of nothing less than at least a 1000 Volts or more.
what I am getting at is, we are considering using the 340VDC rectified from the mains 240VAC (RMS) and shortening the duty cycle to 36% or thereabouts, which of course still has a peak value of 340VDC.
TO SUMMARISE my direction so far due to a few interesting discussions:
MAINS rectified full wave.
Try a full bridge of MOSFETS with a reduced duty cycle direct.
( I hate square waves, but)
If that is not satisfactory then a full bridge converter 2:1 down to 160VDC output and use a full bridge sinwave converter for the correct 60Hz 115VAC RMS output. Here I will have to have some feedback to correct linearity due to Inductive load.
I simply have to take the cheapest option that works .... What are your thoughts on this assumption Guys?
Come on guys, dont leave me high and dry !!!
I need help with TWO things.
1. I built the full bridge set the duty cycle to 36% as per "Eva's comment" and get double the voltage RMS, I cannot recall the correct method of calculating RMS for dut cycle of a Squarewave I expect it needed to be 18% the bus is
240Vrms x 1.414=340VDC ... anyone got the clue?
2. I am using IRFP460's and cannot get the instantaneous on resistance lower than 500mÙ with 14 V switching the gate, I would have thought I would get closer to the specified RDson of 270mÙ, maybe that is only a static figure? any got a better explanation!!!
I need help with TWO things.
1. I built the full bridge set the duty cycle to 36% as per "Eva's comment" and get double the voltage RMS, I cannot recall the correct method of calculating RMS for dut cycle of a Squarewave I expect it needed to be 18% the bus is
240Vrms x 1.414=340VDC ... anyone got the clue?
2. I am using IRFP460's and cannot get the instantaneous on resistance lower than 500mÙ with 14 V switching the gate, I would have thought I would get closer to the specified RDson of 270mÙ, maybe that is only a static figure? any got a better explanation!!!
Relax...
1) Vrms = Vpeak * square root (duty cycle). this applies for square waves only.
2) Rds on is specified at 25 deg C. Rds on increases with temperature. If you are measuring 500 mOhm then your junctions are probably a 130 deg C or higher. Improve your heatsinking and you get an exponential reward.
1) Vrms = Vpeak * square root (duty cycle). this applies for square waves only.
2) Rds on is specified at 25 deg C. Rds on increases with temperature. If you are measuring 500 mOhm then your junctions are probably a 130 deg C or higher. Improve your heatsinking and you get an exponential reward.
I'm sorry. As poobah has mentioned, the effective RMS voltage applied to the load is proportional to the square root of the duty cycle. When I got the 36% figure, I was thinking in power terms and I employed the duty cycle without squaring, so that number is wrong.
The actual duty cycle figure is around 13% ( sqr(320^2*.13)=115 ).
The actual duty cycle figure is around 13% ( sqr(320^2*.13)=115 ).
Eva said:
Not a problem in the world Eva, I should have known that myself, but, could not bring it to mind, I most sincerely appreciate both of you giving me your time.
Poobah..
I have some rather serious heatsinking which only rises to 35deg C. at 5amps, I calculated the Rds on a little crudely by measuring the voltage drop across the drain source a about 30uS after it was turned on.
I have slowed the gate turn on and off to 10uS BUT it still seems to change a little with load changes, e.g. it over shoots slightly with a heavier load!!!
Q1. Do you think It could have something to do with the snubbers?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.