On Tuesday, 14 November 2023 at 00:41:36 UTC+1, john larkin wrote:
> On Mon, 13 Nov 2023 15:29:43 -0800 (PST), Klaus Kragelund
> <klaus.k...@gmail.com> wrote:
>
> >On Monday, 13 November 2023 at 05:49:19 UTC+1, John Larkin wrote:
> >> On Sun, 12 Nov 2023 17:52:25 -0800 (PST), Klaus Kragelund
> >> <klaus.k...@gmail.com> wrote:
> >>
> >> >On Monday, 13 November 2023 at 02:37:13 UTC+1, Klaus Kragelund wrote:
> >> >> On Monday, 13 November 2023 at 01:51:39 UTC+1, John Larkin wrote:
> >> >> > On Sun, 12 Nov 2023 15:54:43 -0800 (PST), Lasse Langwadt Christensen
> >> >> > <lang...@fonz.dk> wrote:
> >> >> >
> >> >> > >mandag den 13. november 2023 kl. 00.32.30 UTC+1 skrev John Larkin:
> >> >> > >> On Sun, 12 Nov 2023 13:53:18 -0800 (PST), Lasse Langwadt Christensen
> >> >> > >> <lang...@fonz.dk> wrote:
> >> >> > >>
> >> >> > >> >s?ndag den 12. november 2023 kl. 21.59.15 UTC+1 skrev John Larkin:
> >> >> > >> >> On Sun, 12 Nov 2023 11:55:40 -0800, John Larkin <j...@997PotHill.com>
> >> >> > >> >> wrote:
> >> >> > >> >> >On Sun, 12 Nov 2023 11:00:03 -0800 (PST), Lasse Langwadt Christensen
> >> >> > >> >> ><lang...@fonz.dk> wrote:
> >> >> > >> >> >
> >> >> > >> >> >>s?ndag den 12. november 2023 kl. 18.15.35 UTC+1 skrev John Larkin:
> >> >> > >> >> >>> On Sun, 12 Nov 2023 07:44:41 -0800, John Larkin <j...@997PotHill.com>
> >> >> > >> >> >>> wrote:
> >> >> > >> >> >>> >On Sun, 12 Nov 2023 03:29:02 -0800 (PST), Klaus Kragelund
> >> >> > >> >> >>> ><klaus.k...@gmail.com> wrote:
> >> >> > >> >> >>> >
> >> >> > >> >> >>> >>On Sunday, 12 November 2023 at 01:02:21 UTC+1, John Larkin wrote:
> >> >> > >> >> >>> >>> On Sat, 11 Nov 2023 14:57:11 -0800 (PST), Klaus Kragelund
> >> >> > >> >> >>> >>> <klaus.k...@gmail.com> wrote:
> >> >> > >> >> >>> >>>
> >> >> > >> >> >>> >>> >On Friday, 10 November 2023 at 16:28:14 UTC+1, John Larkin wrote:
> >> >> > >> >> >>> >>> >> On Thu, 9 Nov 2023 15:14:53 -0800 (PST), Klaus Kragelund
> >> >> > >> >> >>> >>> >> <klaus.k...@gmail.com> wrote:
> >> >> > >> >> >>> >>> >> >Hi
> >> >> > >> >> >>> >>> >> >
> >> >> > >> >> >>> >>> >> >I am on the lookout for a cheap 1:1 pulse transformer. Will be using it to drive a MOSFET based Solid State Switch.
> >> >> > >> >> >>> >>> >> >
> >> >> > >> >> >>> >>> >> >I know Larkin has used the DRQ127
> >> >> > >> >> >>> >>> >> >
> >> >> > >> >> >>> >>> >> >https://www.digikey.dk/en/products/detail/eaton-electronics-division/DRQ127-1R5-R/667262
> >> >> > >> >> >>> >>> >> >
> >> >> > >> >> >>> >>> >> >But, I am looking to see if I can find bottom dollar.
> >> >> > >> >> >>> >>> >> >
> >> >> > >> >> >>> >>> >> >I probably wont go to LCSC, so would prefer a known brand.
> >> >> > >> >> >>> >>> >> >
> >> >> > >> >> >>> >>> >> >Anyone got a recommendation?
> >> >> > >> >> >>> >>> >> >
> >> >> > >> >> >>> >>> >> >Cheers
> >> >> > >> >> >>> >>> >> >
> >> >> > >> >> >>> >>> >> >Klaus
> >> >> > >> >> >>> >>> >> Are you driving the gate of a fet? What fet?
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >STB43N60DM2
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >I did a change to the design, since the 1:10 transformer was bugging me, so used a old fashion CD4093 from 15V supply, to make an isolated supply with a 1:1 transformer.
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >Then an opto to contol the MOSFETs. Emitter followers to lower the impedance driving the gate, since the turn on/turn off losses needs to be kept low due to transient thermal resistance.
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >Schematics, first draft:
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >www.electronicsdesign.dk/tmp/ssr.pdf
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >Some comments:
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >UVLO is handled by the microcontroller in that the PSU starts when power is applied, but the microcontroller is not allowed to turn the FET on until we are sure the PSU is active (milliseconds). Likewise, during shutdown the microcontroller first disables the opto.
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >We need to have UVLO detection on the 15V power supply for the self-running oscillator, in case of a brown-out – lower voltage of the 15V node, the opto needs to be turned off before voltage sags deeper. Normally, if power is lost, then the switch off loss is also zero, but we could have a situation with low mains voltage and we need to take actions accordingly.
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >MOSFET selected is STB43N60DM2, 100mohm/D2PAK, about 40K/W with current layout.
> >> >> > >> >> >>> >>> >At 100mohm/3.7A, loss is 1.3W. 600V device to have 60% derating from peak of 360V.
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >275V varistor SIOV1 is placed across drain/drain of MOSFETs. We might be able to use lower voltage rated MOSFETs due to thje varistor.
> >> >> > >> >> >>> >>> >MOSFETs are avalanche rated for 800mJ.
> >> >> > >> >> >>> >>> >
> >> >> > >> >> >>> >>> >K = 0.015 for less than 10us switching time. Estimated for 1us..
> >> >> > >> >> >>> >>> >Rth = 40K/W. Transient Rth = 0.6K/W
> >> >> > >> >> >>> >>> Since you have that 230 VAC source, you could make a gate power supply
> >> >> > >> >> >>> >>> from that. The little capacitor-limited thing like people use to power
> >> >> > >> >> >>> >>> LEDs from the AC line.
> >> >> > >> >> >>> >>>
> >> >> > >> >> >>> >>Actually the capacitive couplers are not very good when it comes to surges/transients on the mains line, so need to add impedance to reduce peak currents in the circuit.
> >> >> > >> >> >>> >
> >> >> > >> >> >>> >Sure, the LED night-light circuits include a series resistor.
> >> >> > >> >> >>> >
> >> >> > >> >> >>> >The totem-pole opto gate driver needs about zero supply current, so
> >> >> > >> >> >>> >the series resistor can be big. Actually, the series resistor would
> >> >> > >> >> >>> >work without a series cap. I can sketch it if it's not all clear.
> >> >> > >> >> >>> >After this coffee diffuses into my brain.
> >> >> > >> >> >>> >
> >> >> > >> >> >>> >
> >> >> > >> >> >>> You could do your entire isolated gate driver for about 17 cents.
> >> >> > >> >> >>
> >> >> > >> >> >>something like this?
> >> >> > >> >> >
> >> >> > >> >> >Close. I was thinking
> >> >> > >> >> >
> >> >> > >> >> >https://www.dropbox.com/scl/fi/ee9cpnt16nyxotka4zgkw/Isol_Gate_Driver_1.jpg?rlkey=v2o44hxa52eraurnuqpohmcp0&raw=1
> >> >> > >> >> >
> >> >> > >> >> >One could add a zener somewhere without blowing the 17 cent budget.
> >> >> > >> >> >
> >> >> > >> >> >
> >> >> > >> >> That needs to be Spiced. It gets a little weird when the fets are off.
> >> >> > >> >> It may need another resistor.
> >> >> > >> >
> >> >> > >> >it should work, either way is a bit weird to look at in simulation if you don't make the fet sources the ground
> >> >> > >> The source-source node basically floats when both fets are off...
> >> >> > >> battling substrate diodes. So the power supply to the totem-pole opto
> >> >> > >> could fail.
> >> >> > >
> >> >> > >how? C1 gets charged via D1 and Q2s body diode
> >> >> > Yeah, OK, I guess it does work.
> >> >> Very nice solution, and a lot more clean than the transformer one I have been working on.
> >> >>
> >> >> Only worry is noise coupling into the gates, from transients. But that can be remedied by adding a capacitance to the gates of Q1 and Q2. Will slow switching times a little, but more robust.
> >> >I just simulated Lasses solution. C1 needs to be 1uF due to charge transfer, otherwise the switching losses are too high
> >> What sort of switching rates do you need? Entire cycles at 50 Hz? Fast
> >> PWM? The triac solution wouldn't switch fast, and a heater shouldn't
> >> need a lot of switching.
> >
> >Entire cycle, to reduce EMI. It will use something like 1s on, and 1s off, at least many periods between turn on and off.
> >
> >The edge rate is not that critical, it controls the envelope fall rate, which is at a very high frequency. At 50Hz (1Hz in case of 50cycles on/off), the emission is the same.
> >What matters for the MOSFET case is transient dissipation, since that can easily result in very high transient temperature.
> Check the SOAR curves. A mosfet can dissipate a lot of power for a
> millisecond.
For less than 1us, the selected MOSFET can only handle about 100W, and that's without the steady state temperature increase.
>
> Zero-cross switching solves the dynamic dissipation problem.
Yes, can be done. But one rainy day, when the Zero crossing malfunctions and turns off at the peak, we have a problem.