As you increase the frequency of the machine, you are effectively creating more on time, which creates more heat, that needs to be effectively dissipated in order to reduce the duty cycle to a reasonable level, so that the welder is more useful.
When the multiple paralleled MOSFETs get too hot, one will eventually fail and start a chain reaction of failures amongst the remaining MOSFETs that cannot handle the extra load/strain. The resulting sound is akin to machine gun fire or pop corn popping.
The number 1 enemy of any electronic circuit is heat. Keep your electronics cool and they will not only last longer, they will allow for a much better duty cycle.
That is not correct.
As you increase the frequency, the off time also goes up in line with the on time! If you change nothing but the frequency the over heat load is (in theory) the same. However there can be switching delays in the devices as they go from fully ON to fully OFF. In these in between states you will start to see, for a very brief period of time a resistance in the device which is higher than the very low ON resistance, these can add up if the frequency is very very high, (hundreds of kilo herz) but in the scope of running at say 500hz instead of 100hz the difference is NOTHING.
As you increase the frequency, the off time also goes up in line with the on time! If you change nothing but the frequency the over heat load is (in theory) the same. However there can be switching delays in the devices as they go from fully ON to fully OFF. In these in between states you will start to see, for a very brief period of time a resistance in the device which is higher than the very low ON resistance, these can add up if the frequency is very very high, (hundreds of kilo herz) but in the scope of running at say 500hz instead of 100hz the difference is NOTHING.
WerkSpace wrote:As you increase the frequency of the machine, you are effectively creating more on time, which creates more heat, that needs to be effectively dissipated in order to reduce the duty cycle to a reasonable level, so that the welder is more useful.
When the multiple paralleled MOSFETs get too hot, one will eventually fail and start a chain reaction of failures amongst the remaining MOSFETs that cannot handle the extra load/strain. The resulting sound is akin to machine gun fire or pop corn popping.
The number 1 enemy of any electronic circuit is heat. Keep your electronics cool and they will not only last longer, they will allow for a much better duty cycle.
- Otto Nobedder
-
Weldmonger
-
Posts:
-
Joined:Thu Jan 06, 2011 11:40 pm
-
Location:Near New Orleans
You're ignoring the "fixed" heat displacement time of the heat sinks involved, and the soak they suffer. The heat produced is an RMS value, not some imaginary instantaneous figure.Pavlo wrote:That is not correct.
As you increase the frequency, the off time also goes up in line with the on time! If you change nothing but the frequency the over heat load is (in theory) the same. However there can be switching delays in the devices as they go from fully ON to fully OFF. In these in between states you will start to see, for a very brief period of time a resistance in the device which is higher than the very low ON resistance, these can add up if the frequency is very very high, (hundreds of kilo herz) but in the scope of running at say 500hz instead of 100hz the difference is NOTHING.
Swithching, also, is not instantaneous, until someone invents a quantum switch device.
Steve S
Power dissipated (as heat) by a resistor or arc in our case, is equal to the current squared multiplied by the resistance.
In very basic terms, if you were to consider each Hz as a pulse (of set length) through the arc, then the more Hz the longer there is current going through the component per second, meaning more Mhz = more heat.
Frequency is a little more complex and somewhat weird. Now instead of simple resistance, we have impedences. These change their 'resistance' based on the frequency. 1MHz can actually seem like a lot of change for some impedences and they will suddenly jump up very high in their 'resistance'. Actual values are hard to pin down because the equations can be pretty ugly, and aren't always very straight. They also can do very weird things to the 'shape' of your wave.
If you really want to see things in a different light.
Check out Larry's atomic theories. https://teslatech.info/ttstore/articles ... tomthe.htm
In very basic terms, if you were to consider each Hz as a pulse (of set length) through the arc, then the more Hz the longer there is current going through the component per second, meaning more Mhz = more heat.
Frequency is a little more complex and somewhat weird. Now instead of simple resistance, we have impedences. These change their 'resistance' based on the frequency. 1MHz can actually seem like a lot of change for some impedences and they will suddenly jump up very high in their 'resistance'. Actual values are hard to pin down because the equations can be pretty ugly, and aren't always very straight. They also can do very weird things to the 'shape' of your wave.
If you really want to see things in a different light.
Check out Larry's atomic theories. https://teslatech.info/ttstore/articles ... tomthe.htm
I'm not ignoring it, in fact I did mention it in lay mans terms "for a very brief period of time..."
The power loss difference when looking at 100hz vs 500hz is insignificant.
A quick google searth for MOSFET and the very first specific device is this one:
https://www.sparkfun.com/products/10213
The rise time and fall time combined is 660ns. That is 0.00000066 seconds, at 500hz the period is 0.002 seconds so the switching time is 0.033% of the period.
If we were switching this at 500khz then it would be different, but that's just the very first device I found in a google search.
I'm not sure I follow you regarding the heat output though, since the RMS value of a true square wave (which this is not I realise, since I just mentioned the switching speed above) is independent of the frequency. If we are running 50% duty it doesn't really matter how fast we do it, even with the switching times while we are under 1000hz it's less then 0.1%.
The heat outputs is therefore massively dominated by the RDS on ohms, the current and the duty cycle. Looking at our lucky google find, it's only good for 32amps, but that just means we need more of them, so lets put 8 devices on (for once half of the AC wave) for about 180amps capacity assuming we can't cool them to 25ºC in use, and we have a combined RDS on resistance of about 0.006 ohms when at working temperature (lets call it 100ºC). Power dissipation is duty*I*r^2 = 0.5*180*0.006*0.006=.003 Watts. The switching losses at 100hz would be negligible,
In other words if we were using our chosen at random device in that way (and I'm not suggesting that's necessarily the case), to throttle a power source, it would consume less power than than the solder joints and tracks of the PCB.
But. I'm pretty sure an inverter welder will be running those devices at many Khz anyway, and the 125, 500 or whatever Hz we see in our AC frequency, or pulsed DC at the torch is just a variation of the duty cycle at a very high frequency to give us the current regulation we are after. In which case this is all a very pointless argument as the switching losses would be much higher, and also be independent of the apparent frequency we see at the torch because it's always at (say) 100khz.
This article gives a much better insight into switching loses:
http://www.eetimes.com/document.asp?doc_id=1278970
The power loss difference when looking at 100hz vs 500hz is insignificant.
A quick google searth for MOSFET and the very first specific device is this one:
https://www.sparkfun.com/products/10213
The rise time and fall time combined is 660ns. That is 0.00000066 seconds, at 500hz the period is 0.002 seconds so the switching time is 0.033% of the period.
If we were switching this at 500khz then it would be different, but that's just the very first device I found in a google search.
I'm not sure I follow you regarding the heat output though, since the RMS value of a true square wave (which this is not I realise, since I just mentioned the switching speed above) is independent of the frequency. If we are running 50% duty it doesn't really matter how fast we do it, even with the switching times while we are under 1000hz it's less then 0.1%.
The heat outputs is therefore massively dominated by the RDS on ohms, the current and the duty cycle. Looking at our lucky google find, it's only good for 32amps, but that just means we need more of them, so lets put 8 devices on (for once half of the AC wave) for about 180amps capacity assuming we can't cool them to 25ºC in use, and we have a combined RDS on resistance of about 0.006 ohms when at working temperature (lets call it 100ºC). Power dissipation is duty*I*r^2 = 0.5*180*0.006*0.006=.003 Watts. The switching losses at 100hz would be negligible,
In other words if we were using our chosen at random device in that way (and I'm not suggesting that's necessarily the case), to throttle a power source, it would consume less power than than the solder joints and tracks of the PCB.
But. I'm pretty sure an inverter welder will be running those devices at many Khz anyway, and the 125, 500 or whatever Hz we see in our AC frequency, or pulsed DC at the torch is just a variation of the duty cycle at a very high frequency to give us the current regulation we are after. In which case this is all a very pointless argument as the switching losses would be much higher, and also be independent of the apparent frequency we see at the torch because it's always at (say) 100khz.
This article gives a much better insight into switching loses:
http://www.eetimes.com/document.asp?doc_id=1278970
Otto Nobedder wrote:You're ignoring the "fixed" heat displacement time of the heat sinks involved, and the soak they suffer. The heat produced is an RMS value, not some imaginary instantaneous figure.Pavlo wrote:That is not correct.
As you increase the frequency, the off time also goes up in line with the on time! If you change nothing but the frequency the over heat load is (in theory) the same. However there can be switching delays in the devices as they go from fully ON to fully OFF. In these in between states you will start to see, for a very brief period of time a resistance in the device which is higher than the very low ON resistance, these can add up if the frequency is very very high, (hundreds of kilo herz) but in the scope of running at say 500hz instead of 100hz the difference is NOTHING.
Swithching, also, is not instantaneous, until someone invents a quantum switch device.
Steve S
No!
More hz doesn't mean more heat!
Because you can't think of a pulse of a set length! If you have a 50% duty cycle at 100hz, that's a pulse 0.005 seconds long.
If we keep the duty the same, you can't have anything but 100hz with 0.005 second pulses at 50% duty cycle.
If we ran it at 500hz, and 50% duty, the on time would be 0.001 seconds!
In both cases the total on time is always 50%, and the power dissipation (ignoring switching loses) the same.
More hz doesn't mean more heat!
Because you can't think of a pulse of a set length! If you have a 50% duty cycle at 100hz, that's a pulse 0.005 seconds long.
If we keep the duty the same, you can't have anything but 100hz with 0.005 second pulses at 50% duty cycle.
If we ran it at 500hz, and 50% duty, the on time would be 0.001 seconds!
In both cases the total on time is always 50%, and the power dissipation (ignoring switching loses) the same.
WerkSpace wrote:Power dissipated (as heat) by a resistor or arc in our case, is equal to the current squared multiplied by the resistance.
In very basic terms, if you were to consider each Hz as a pulse (of set length) through the arc, then the more Hz the longer there is current going through the component per second, meaning more Mhz = more heat.
Frequency is a little more complex and somewhat weird. Now instead of simple resistance, we have impedences. These change their 'resistance' based on the frequency. 1MHz can actually seem like a lot of change for some impedences and they will suddenly jump up very high in their 'resistance'. Actual values are hard to pin down because the equations can be pretty ugly, and aren't always very straight. They also can do very weird things to the 'shape' of your wave.
If you really want to see things in a different light.
Check out Larry's atomic theories. https://teslatech.info/ttstore/articles ... tomthe.htm
noddybrian
- noddybrian
-
Weldmonger
-
Posts:
-
Joined:Thu Jan 24, 2013 12:13 pm
@ Nissan20det
I'm not sure if it's OK to post the importers Email address here - but I have just found it - I've lost his phone number as I've not spoken to him in ages - if anyone can sort out the WSME welders I think he'd be your guy - he had many highly qualified nerds looking into the design for him & he has schematics of them as supplied & a re-draw to improve them - let me know if you'd like to try him ( assuming he's still using the same one )
As to the ongoing question of heat - I think it may help to qualify your desire to modify the machine - as I understand you'd like the pulse to go down lower ( no problem there ) but you'd like AC frequency to go up significantly - this I see as a causing issues - but in general when I've played with similar if basic circuits the higher the frequency the more heat & the more problems - much like the problems found when trying to replicate any of the Stan Meyer circuits or scaling up the current from the basic drawing of the small demo cell as used in the patent information.
I'm not sure if it's OK to post the importers Email address here - but I have just found it - I've lost his phone number as I've not spoken to him in ages - if anyone can sort out the WSME welders I think he'd be your guy - he had many highly qualified nerds looking into the design for him & he has schematics of them as supplied & a re-draw to improve them - let me know if you'd like to try him ( assuming he's still using the same one )
As to the ongoing question of heat - I think it may help to qualify your desire to modify the machine - as I understand you'd like the pulse to go down lower ( no problem there ) but you'd like AC frequency to go up significantly - this I see as a causing issues - but in general when I've played with similar if basic circuits the higher the frequency the more heat & the more problems - much like the problems found when trying to replicate any of the Stan Meyer circuits or scaling up the current from the basic drawing of the small demo cell as used in the patent information.