TeeJay,
I don't really know your circuit... so I could easily make many BAD suggestions.
You should improve your turn on/turn off times... 10 uSec's is slow... this gives you the high dissipation. You need strong gate drivers, over 1 amp, but you can easily reduce this to 200 - 400 nSec's. 50 nSec's, like the data sheet says, is harder to do and makes more noise.
The best heatsink means nothing if you don't have a good thermal connection between the FET and the sink. I have gotten away from grease/mica and screws... now I use silicone pads and springs... people always use WAAAAY too much grease. If I do use screws... I always use "belleville washers" to act as a spring. The spring is important to maintain pressure... it seems a regular screwed connection always get loose. Google on "belleville washers".
Now, if reducing your switching speed, makes too much noise, you have to work on snubbers. This depends alot on your circuit... but I can say, in general, RCD snubbers work best. You MUST use low inductance resistors, fast diodes, and low ESL caps for them to work right.
And you are lucky to have SMPS
Eva on your thread.
Good luck!
I don't really know your circuit... so I could easily make many BAD suggestions.
You should improve your turn on/turn off times... 10 uSec's is slow... this gives you the high dissipation. You need strong gate drivers, over 1 amp, but you can easily reduce this to 200 - 400 nSec's. 50 nSec's, like the data sheet says, is harder to do and makes more noise.
The best heatsink means nothing if you don't have a good thermal connection between the FET and the sink. I have gotten away from grease/mica and screws... now I use silicone pads and springs... people always use WAAAAY too much grease. If I do use screws... I always use "belleville washers" to act as a spring. The spring is important to maintain pressure... it seems a regular screwed connection always get loose. Google on "belleville washers".
Now, if reducing your switching speed, makes too much noise, you have to work on snubbers. This depends alot on your circuit... but I can say, in general, RCD snubbers work best. You MUST use low inductance resistors, fast diodes, and low ESL caps for them to work right.
And you are lucky to have SMPS
Eva on your thread.Good luck!
RCD snubbers doesn't work for anything other than single switch converters or push-pull. Also, 10uS may seem excessive, but such a long transition time when switching at 120Hz still produces 10 times less dissipation than 100ns transition when switching at 120Khz, so it's fine and EMI-wise.
Concerning clip-mounting and sil-pads, I'm also using them but I'm quite disappointed because my switches are now running up to 30ºC hotter (relative to the heatsink) than when I was using screws, mica and thermal compound (measuring with an IR gun at the tab of TO-220 devices). Don't get fooled, all this stuff was only thought to reduce manufacturing time and costs, it never was intended to improve performance or reliability.
Concerning clip-mounting and sil-pads, I'm also using them but I'm quite disappointed because my switches are now running up to 30ºC hotter (relative to the heatsink) than when I was using screws, mica and thermal compound (measuring with an IR gun at the tab of TO-220 devices). Don't get fooled, all this stuff was only thought to reduce manufacturing time and costs, it never was intended to improve performance or reliability.
Sorry Teejay,
I had to read your whole thread... I take it your building a square generator to power sine driven equipment? You had better check carefully the ratings of all the equipment you intend to power. Especially the motors, because they are not itended to be driven by the harmonic series you will generate. Look at Dayton Electric's website and look at "Inverter Rated" motors.
As far as snubbers are concerned... that IS topology dependant... and it also device dependant.
Regarding silicone insulators; I have observed higher temperature differentials with silicone... nothing on the order of 30 deg C. Perhaps I am using better material. Also I NEVER run junctions hotter than 105 deg C... so the temperature differences are alwasy smaller. The important difference lies in the fact that a silicone/spring joint improves with use, heat , and age. Mica/grease connections, especially those with plain screws (i.e. no springs) are subject to degradation over time.
I had to read your whole thread... I take it your building a square generator to power sine driven equipment? You had better check carefully the ratings of all the equipment you intend to power. Especially the motors, because they are not itended to be driven by the harmonic series you will generate. Look at Dayton Electric's website and look at "Inverter Rated" motors.
As far as snubbers are concerned... that IS topology dependant... and it also device dependant.
Regarding silicone insulators; I have observed higher temperature differentials with silicone... nothing on the order of 30 deg C. Perhaps I am using better material. Also I NEVER run junctions hotter than 105 deg C... so the temperature differences are alwasy smaller. The important difference lies in the fact that a silicone/spring joint improves with use, heat , and age. Mica/grease connections, especially those with plain screws (i.e. no springs) are subject to degradation over time.
poobah said:
Hi Guys, Just to sum up where I am at now after all that!!!
Re: the heatsinks nothing to worry about there remember 30 odd years TV servicing you get rather good at mounting to heatsinks for very good thermal conductivity, no noticeable MOSFET to heat sink differential to worry about on this device..
SO FAR,
I completed the 60Hz device with the 13% which worked out to be very close, less a bit of poor in fact no regulation yet.... I made this device actually work all the components separately on the bench! then connected it nervously to a new one of the machines and CAPOWWWW ...
This NICE KIND OF EXPLOSION has not happened to me for a few years, it took 3 seconds to work out what had happened ... OF COURSE the yanks used a great spike suppressor 180 Volt MOV which certainly did not like a peak voltage of 340V .. sometimes you do not think of the obvious ...... till the bang!
MY FIRST DESTROYED FETS in this project ...
So now fast and furious I changed the bridge that was doing the primary job to now run into the ferrite transformer I had previously started to wind, at 100Khz 2:1 reduction and am now furiously building a second bridge same as primary side to do the sinwave conversion from the rectified 100Khz secondary 170VDC (a little up my sleeve for a bit of regulation head room) ... will be finished in 24 hours and I will probably be asking for your best approach to the snubbers which I suspect are going to be quite critical ... I am going to sample some rectified voltage from the output and feed it to the micro's A to D input varying the duty cycle as a form of regulation ... wish me luck ...
poobah said:
Poobah, I am using HCPL3120 Isolated FET drivers capable of 2A easily, driving IRFP460 MOSFETS, what do you find you are using as gate resistor values for the 460's, I have tried none, but end up with to much ringing, which looks like you would never snub it away,
so I am at 33 Ohms which of course slows the gate to about 300nS which is probably going to be OK I am finding I have to reduce the recomended 95% duty cycle (400nS OFF time) back to 800nS of OFF time at 100Khz... So the FETS are not both on at the same time ... big compromises everywhere ....any coments on this area re snubbers?
poobah said:
Ok Poobah, this is the circuit so far and it seems to work reasonable, I had to rmove the Snubbers used at the 120Hz speed as they were a bit heavy at the 100Khz thats for sure.
It is not very bad with this duty cycle which is 100Khz and a 800nS off time, no load on the seconday total current draw is 90mA so far I have loaded it to 5A and it does not seem to be to upset about much at all, surprising !! I have put an extra turn or 2 on the secondary so I would have a little head room for regulation !!
I assume you regulate the secondary sinewave generator by varying the duty cycle +- is that the right approach?
See circuit attached ....
poobah said:Where do IC10 & IC13 get their power?
What are L2 & L4? There should be NO inductance between the Vee pin in the HP driver and its FET... this will cause huge ringing.
If you are trying to make a sine wave, why the bridge rect. on the output?
Interesting queations !!
Please ignore the different rails, but I did not have time to create new power rails e.g. 15V-1, 15V-2 etc. in my circuit library of parts ...I just used different rails to keep them separated as they are isolated supplies on the high side, as for IC10 & IC13 I do not see a problem with them, they are powered from the isolated supplies above with the matching rail tags! but on the diag I DID leave pin 5 of IC10 & IC13 unconnected to the MOSFETS they are driving !!! ... of course I fixed that after noticing the link to the fet was missing while I was constructing the board, Sorry I did not update the diag with the links to the FETS.
Regards the Inductance I put in the DC rails to the drivers, I was trying to get away from common mode effects back thru the isolated power supplies, as these are Iron cored conventional transformers. I will eventually replace these with low capacitance ferrite cored driver supplies, I made sure I had good bypasses right at the fet drivers themselves, so they had a chance of driving hard, Correct me If I am wrong but I would have thought the inductances only made a high impedance return for high frequencies back to the supply and would not effect the DC getting to the drivers as long as the drivers had GOOD 4.7uF Tantalums with nice short leads at the driver supply pins ....?
There are no signs of overshoot just miinor ringing without any form of snubbers, which I expected I could clean up later, looking at the waveforms of the drivers, they are clean and very fast so far.
It certainly is radiating very naughty amounts of noise from the bottom to halfway up the HF radio band, but that was expected with square waves at 100Khz and parts laying out a little on the workbench...will fix that when all else is sorted.
I am more than Interested in your comment about the rectifier situtaion, my approach was to end up with 160VDC to use as a feed for my PWM controlled bridge which creates the sinwave.
Of course that part is still to be completed, it will be similar to the first bridge part of the existing circuit.... IS THERE AN EASIER WAY TO MAKE A SINEWAVE WITHOUT STARTING WITH DC ? Remebering I needed the 2:1 Current advantage, hence the high frequency tranformer part.
Kind Regards TJ
poobah said:
Yes that is the overall Pix.
====================
The IDEAL would be :
100-250VAC any freq to 115VAC 60Hz 10 Amps SINEWAVE
====================
The expected will be 200-250VAC to 115VAC 60Hz 10AMPS SINEWAVE.
You say there is one too many conversions can you suggest a simpler way to achieve this AND gain the 2:1 stepdown current gain? e.g. with a 10 AMP output load I should only have approx 5 AMP primary load from the 240VAC, this is important as it begins to exceed the standard wall outlet otherwise.
Also what are your comments on the queries I explained!!! do they make sense?
Thanks for your time Poobah.
Gotta think about this a bit.
The idea of an intermediate stage, with constant DC, is appealing but wastefull. It does make the circuit design some what less challenging.
A temptation would be a PFC corrected front end... a much bigger inductor, but a lot less parts. If you are building ONE of these... who cares what size the inductor is? A $4 core plus $10 shipping is little different than a $2 core plus $10 shipping.
As far as your previous queries... my point was, with floating supplies and fet drive; the most important connection is from Vss on the driver chip to the source on the fet... this connection should be as low resistance and inductance as you can make and it should be a "Kelvin/star" type. Don't worry about interwinding coupling with the floating supplies; the connection between source and Vss take care of this... if done well.
I am still concerned about your motor loads... generally Load times 4 is a workable factor for motor starting. Can you be more specific about the motor types?
The idea of an intermediate stage, with constant DC, is appealing but wastefull. It does make the circuit design some what less challenging.
A temptation would be a PFC corrected front end... a much bigger inductor, but a lot less parts. If you are building ONE of these... who cares what size the inductor is? A $4 core plus $10 shipping is little different than a $2 core plus $10 shipping.
As far as your previous queries... my point was, with floating supplies and fet drive; the most important connection is from Vss on the driver chip to the source on the fet... this connection should be as low resistance and inductance as you can make and it should be a "Kelvin/star" type. Don't worry about interwinding coupling with the floating supplies; the connection between source and Vss take care of this... if done well.
I am still concerned about your motor loads... generally Load times 4 is a workable factor for motor starting. Can you be more specific about the motor types?
poobah said:Gotta think about this a bit.
The idea of an intermediate stage, with constant DC, is appealing but wastefull. It does make the circuit design some what less challenging.
A temptation would be a PFC corrected front end... a much bigger inductor, but a lot less parts.
I am still concerned about your motor loads... generally Load times 4 is a workable factor for motor starting. Can you be more specific about the motor types?
You are on the ball,,, the motor draws 15amps for 250mS to start, then only needs 4 to 5 AMPs while running.
I will be building quite a number of these and need to have as fewer parts as possible .....COST & PARTS are VERY IMPORTANT ...
Well then, your motor loads will dominate evrything. You really need to get a machine and quantify your motor loads... and then multiply by 2 or 3 to ensure end-of-life specifications.
I am sorry, but I have a bit of direct experience here building solar powered vaccine refridgerators. There are some tricks though.
I don't mean to throw you back to square one... better now than later.
I am sorry, but I have a bit of direct experience here building solar powered vaccine refridgerators. There are some tricks though.
I don't mean to throw you back to square one... better now than later.
poobah said:
So can you give me an indication of your suggested block diagram?
The 13% square wave ACTUALLY started and ran the motor although the motor did not quite have the gutts that it would have had running from the correct sinewave and voltage.
So I am assumming we were not far from a generally good approach by now deciding to give the correct waveshape and voltage..... What are you suggesting?
At least we are not going back to square one really, as I have learned a hell of a lot getting to this point, I consider.
I will wait for your thoughts on the correct overall picture.
poobah said:
You are working on a formula for disaster... you do not provide direct answers... only the next level of assumptions and YOUR questions that arise from them.
You must test your loads... nameplate ratings ONLY apply when connected to a stiff soure... like a power line.
I have given you correct information, this was derived from on line (30 Amp continous stepdown 115 V output transformer) actual testing with a halleffect current probe (fluke) measuring and recording the current, 250mS is the time that the motor draws start current for then settles down to precisely 5 Amps running under normal load conditions ... HOW ELSE WOULD YOU SUGGEST I MEASURE THIS .. I am all ears ...
I wonder if you could put an LC filter between the output of a 340vdc to 60hz AC converter and the load to make it work well enough. The capacitor could be a film type with a value of around 100uF. The inductor could have a value of around 20mH. Then you would probably adjust the duty cycle a little.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.